Vol1

XGEOM

Sonnet® User’s Manual EM

Release 6.0 Volume 1 Application Notes

Cover: James Clerk Maxwell (1831-1879). A professor at Cambridge University, England, Maxwell established the interdependence of electricity and magnetism. In his classic treatise of 1873, he published the first unified theory of electricity and magnetism and founded the science of electromagnetism.

Sonnet® User’s Manual Volume 1 Printed: April 1999

Release 6.0 Sonnet Software, Inc. 1020 Seventh North Street, Suite 210 Liverpool, NY 13088 Phone: (315) 453-3096 Fax: (315) 451-1694 Technical Support: [email protected] Sales Information: [email protected]

 Copyright 1989,1991,1993, 1995-1999 Sonnet Software, Inc. All Rights Reserved Registration numbers: TX 2-723-907, TX 2-760-739

Copyright Notice

Reproduction of this document in whole or in part, without the prior express written authorization of Sonnet Software, Inc. is prohibited. Documentation and all authorized copies of documentation must remain solely in the possession of the customer at all times, and must remain at the software designated site. The customer shall not, under any circumstances, provide the documentation to any third party without prior written approval from Sonnet Software, Inc. This publication is subject to change at any time and without notice. Any suggestions for improvements in this publication or in the software it describes are welcome.

Trademarks

The program names xgeom, em Control, emvu, patgen and patvu, dxfgeo, ebridge, emgen, emgraph, gds (lower case bold italics), LEVEL1 and LEVEL1plus are trademarks of Sonnet Software, Inc. Sonnet® and em® are registered trademarks of Sonnet Software, Inc. UNIX is a trademark of Unix Systems Labs. X Window System is a trademark of the Massachusetts Institute of Technology. AutoCAD and Drawing Interchange file (DXF) are trademarks of Auto Desk, Inc. SPARCsystem Open Windows, SUN, SUN-4, SunOS, Solaris, SunView, and SPARCstation are trademarks of Sun Microsystems, Inc. HP, HP-UX, Hewlett-Packard, HP-EEsof, Touchstone, Libra, Academy, Series IV, and Apollo are trademarks of Hewlett-Packard Company. Super-Compact is a trademark of Compact Software, Inc. GDSII is a trademark of Calma Company. FLEXlm is a registered trademark of Globetrotter Software, Inc. OSF/Motif is a trademark of the Open Software Foundation. IBM is a registered trademark of International Business Machines Corporation. MS-DOS and Windows are registered trademarks of Microsoft Corporation. DESQview/X is a trademark of Quarterdeck Office Systems. Empipe, OSA90, and OSA90/hope, Datapipe, Geometry Capture and Space Mapping are trademarks of Hewlett-Packard Company. Micro-Stripes® is a registered trademark of Kimberly Communications Consultants (KCC) Ltd.

Xgeom User’s Manual

Xgeom User’s Manual

Table of Contents XGEOM

Table of Contents Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i 1

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 A Brief Overview of xgeom. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Enhancements In Recent Releases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 New Features for Both Windows and UNIX. . . . . . . . . . . . . . 2 New Features for UNIX Only . . . . . . . . . . . . . . . . . . . . . . . . . 3 Changes for UNIX only. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Xgeom User’s Manual Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Using the xgeom Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Tool Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Tool Box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Status Bar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Describing Menu Bar Accesses . . . . . . . . . . . . . . . . . . . . . . . . 7 Shortcut Keys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Shift Selecting Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Invoking Sonnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 UNIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2

Editing Your Circuit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Pointer Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Reshape Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Selecting Objects for Editing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Clicking on an Object or Point . . . . . . . . . . . . . . . . . . . . . . . 14 Multiple Selects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Select by Lassoing an Area . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Unselecting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

vii

Xgeom User’s Manual Multi-Layer Selects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Shift and Control Keys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Points. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Moving Points. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Adding Points to a Polygon . . . . . . . . . . . . . . . . . . . . . . . . . 19 Deleting Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Polygons, Ports and Vias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Moving Polygons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Deleting Polygons. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cutting and Pasting Polygons . . . . . . . . . . . . . . . . . . . . . . . . Moving Polygons to Another Layer . . . . . . . . . . . . . . . . . . . Flipping, Rotating, and Resizing . . . . . . . . . . . . . . . . . . . . . 3

21 21 21 22 22

Using Ports. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Standard Box-Wall Ports. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Standard Ungrounded-Internal Ports. . . . . . . . . . . . . . . . . . . . . . . . . . 26 Via Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Automatic-Grounded Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Manipulating Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Adding Standard Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adding Via Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adding Auto-grounded Ports . . . . . . . . . . . . . . . . . . . . . . . . Deleting Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Changing Port Numbering . . . . . . . . . . . . . . . . . . . . . . . . . . Changing Port Impedance. . . . . . . . . . . . . . . . . . . . . . . . . . . Defining Reference Planes/Calibration Lengths. . . . . . . . . .

30 30 31 31 32 33 34

Creating Auto-grounded Ports: An Example . . . . . . . . . . . . . . . . . . . 36 Auto-grounded Port Reference Planes . . . . . . . . . . . . . . . . . 39 4

Using Vias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Via Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Creating the Polygons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

viii

Table of Contents

Creating an Airbridge: An Example . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Loading in the Example File . . . . . . . . . . . . . . . . . . . . . . . . . 49 Circuit Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Creating the Polygons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Adding Edge Vias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Summary of Vias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 5

Dielectric Bricks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Applications for Dielectric Bricks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Creating a Dielectric Brick . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Defining Dielectric Brick Materials. . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Changing Brick Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Z-Partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

6

Palette of Standard Geometries. . . . . . . . . . . . . . . . . . . . . . . . . 59 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Rectangle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 Interdigital Capacitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 Donut . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Meander . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Round Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Rectangular Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 Parallel Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Fan Stub . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

ix

XGEOM

Edge Vias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Via Posts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Adding a Via to Ground . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Making Entire Polygons Vias . . . . . . . . . . . . . . . . . . . . . . . . 47 Deleting Vias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Via Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Shorted Vias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Via Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Xgeom User’s Manual Lange Coupler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 7

Function Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 The Startup Menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 The File Menu. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 The Help Menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 The Main Menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 The File Menu. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Typing in Coordinates From the Keyboard. . . . . . . . . . . . . . . . . . . . 147 An Example of Keying in Coordinates. . . . . . . . . . . . . . . . 147 Panning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

x

Chapter 1 Introduction XGEOM

Chapter 1

Introduction

Xgeom is a window based program used to capture circuit geometry for input to the electromagnetic analysis program, em and is compatible with UNIX, Windows95/98 and WindowsNT 4.0.

A Brief Overview of xgeom Xgeom provides you with a straightforward interface which allows specification of all necessary information concerning the circuit to be analyzed. First, you capture the circuit with xgeom and save the resulting file to disk, with a name customarily ending in “.geo”. You may then run em using the geometry file as input. Em automatically subsections the circuit, using variable size subsections, and performs a complete electromagnetic analysis. Em saves the resulting Sparameters in a file format selected by you, ready for input into subsequent microwave design programs.

11

Xgeom User’s Manual

Enhancements In Recent Releases The major change incorporated in Version 6.0 for UNIX systems is the new interface which is WindowsNT, Windows95/98, and UNIX compliant. The most notable additions for Windows systems is the Palette of Standard Geometries and the addition of the Undo and Redo commands. Listed below are the new features and changes introduced in xgeom for Version 6.0. For what’s new in other Sonnet products, please refer to the appropriate User’s Manual.

New Features for Both Windows and UNIX Undo and Redo: The Undo and Redo command have been added to the Edit commands in xgeom. For details, see “Edit - Undo,” page 93 and “Edit - Redo,” page 93. Palette of Standard Geometries: The palette of standard geometries provides a set of commonly used circuit elements to add to your design. Choices include rectangles, meanders, donuts, spirals, interdigital capacitors and Lange couplers. Once added to your circuit, the standard geometry is converted to its component polygons which may subsequently be edited. For details, see Chapter 6, “Palette of Standard Geometries” on page 69. Launch em analysis: The item Analyze has been added to the File menu to allow you to launch the em analysis program from xgeom. An Analyze button has also been added to the tool bar. See “File - Analyze,” page 89. Status Bar: The status bar appears at the bottom of the xgeom display window to provide you with information on xgeom’s operation. It has four fields: messages, zoom level, cursor position, and mode. The cursor position now provides the dimensions of any selected objects and the delta change if they are moved. This is useful for measuring objects and distances between objects without having to view the measuring tool. Undo and Redo: The Undo and Redo command have been added to the Edit commands in xgeom. For UNIX, only the Redo command is new. For details, see “Edit - Undo,” page 93 and “Edit - Redo,” page 93.

12

Chapter 1 Introduction XGEOM

New Features for UNIX Only File Export: You now have the capability to output your geometry file in DXF or GDSII format. For details, see “File - Export,” page 89. Layer Visibility and Locking: You now have the capability to lock (i.e., to protect from editing commands) or unlock a metalization level and choose the visibility of that level. For details, see “View - Metalization Levels,” page 101. New View: You now have the capability to have two views of the same circuit. For details, see “View - New View,” page 100. Object Visibility: You can now selectively choose which types of objects you wish to display at any given time. For details, see “View - Object Visibility,” page 103. Orthogonal Mode: This mode allows xgeom to operate only in the horizontal and vertical planes when adding or moving polygons and points. For details, see “Tools - Ortho,” page 114. Selection Filter: You can disable selection of object types. For details, see “Edit - Select Filter,” page 97. Status Bar: The status bar appears at the bottom of the xgeom display window to provide you with information on xgeom’s operation. It has four fields: messages, zoom level, cursor position, and mode. The message field will indicate what you need to do to perform a specific action. Tool Bar: Just below the menu bar in the xgeom display appears a tool bar consisting of a series of small buttons. These buttons allow quicker access to more frequently used commands. For details, see “Tool Bar,” page 152. Tool Box: A tool box appears on your display when the xgeom program is started. This allows you to quickly access more commonly used functions. For details, see “Tool Box,” page 155. Dielectric Bricks: You can now add objects of dielectric brick material to your circuit. For details, see Chapter 5, “Dielectric Bricks”. Moving Polygons using Keyboard Entry of Coordinates: When a polygon is selected, entering coordinates moves the polygon to a new location. For details, see “An Example of Keying in Coordinates,” page 157.

13

Xgeom User’s Manual

Changes for UNIX only Shift Selecting Modes: Modes now default to pointer mode after one use. However, when selecting a mode, use of the shift key will allow you to remain in the mode until you explicitly exit it. This is useful for adding multiple objects, such as ports or vias. For details, see “Shift Selecting Modes,” page 18. You may also set preferences to xgeom 4.0 mode. For details, see “File - Preferences,” page 85. Comments: You will no longer be prompted for comments when saving your files. To attach comments to a file, select the File ⇒ Comments menu item, and enter your comments in the dialog box. For details, see “File - Comments,” page 91. Keyboard Entry of Objects: When you are in many of the modes, simply enter coordinates. As you type, your input will appear in the messages field of the status bar and the polygon will be added. For details, see “An Example of Keying in Coordinates,” page 157. Menus and Dialog Boxes: The menus have been reorganized, and dialog boxes have been implemented. The presence of a dialog box is indicated by an ellipsis (...) following an item on a pull-down menu. For details about the dialog boxes, see “Dialog Boxes,” page 75 in the Sonnet Tutorial. Multi-Layer Selects: Selecting Edit ⇒ Single Layer toggles between single layer editing mode and multi layer editing mode. When in multi layer mode, you may select an object that does not appear on the level that you are presently viewing. Origin: The origin is now on the bottom left only. Panning: Panning is no longer a menu item, but is now accomplished through the use of scroll bars in the xgeom window. Ports: It is now possible to individually select Ports to delete them or modify their attributes. This can be used to change the port number, port parameters and define the port as autogrounded or standard. You can also modify all port impedance values in one operation. For a detailed discussion of ports, see Chapter 3, “Using Ports”

14

Chapter 1 Introduction

Top and Bottom of Box Parameters: The box top and box bottom parameters are no longer directly entered; instead, metal types are assigned. You then select a metal type for the box top and box bottom from a pull down menu when defining the box parameters, which will include the user defined metal types, and the predefined metal types, Free Space and WG Load. For details, see “Parameters Box,” page 129 and “Parameters - Metal types,” page 138. Box Top and Bottom Metal Types: Two new predefined metal types have been added for the box top or bottom, available in the Box Parameters dialog box. The first is WG Load which models a perfect matched wave guide load. The other is Free Space which models removing the top or bottom box cover. For more details, see “Parameters - Box,” page 129. Vias: Through use of the editing tools, it is now possible to individually select vias while in pointer mode in order to delete them.

Xgeom User’s Manual Layout Chapter 2 through Chapter 5 forms a guide to using xgeom, to be read once you have tried the tutorials in the Sonnet Tutorial. The last portion, Chapter 7, of the manual forms a concise reference for xgeom’s functionality, to be used once you have a working knowledge of the program. You should also note that the illustrations alternate between Windows and UNIX examples as this manual is used for both applications.

Using the xgeom Environment The following section discusses conventions used in this manual and basic instructions in the use of the xgeom environment.

15

XGEOM

Shift Selecting Objects: Allows you to add additional objects, or delete objects, to or from a group of previously selected objects. For details, see “Shift and Control Keys,” page 28.

Xgeom User’s Manual

Tool Bar Just below the menu bar in the xgeom display appears a tool bar consisting of a series of small buttons. These buttons allow quicker access to more frequently used commands. Placing the cursor over these buttons will cause a brief description of their function to appear in the message field of the status bar at the bottom of the xgeom window. For a complete description of the these buttons, see “Tool Bar,” page 152.

Tool Box A tool box, shown below, appears on your display when a geometry file is opened, or when you select View ⇒ Tool Box from the menu. This allows you to quickly access more commonly used functions. Placing the cursor over a button in the tool box will cause a brief description of the function to appear in the status bar at the bottom of the xgeom window. For a complete description of the these buttons, see “Tool Box,” page 155. Xgeom’s Tool Box

Status Bar The status bar, shown below, appears at the bottom of the xgeom display. It displays information pertaining to cursor position, mode, zoom level and prompts.

Status Bar

Messages

16

Zoom level

Cursor Position

Mode

Chapter 1 Introduction XGEOM

The status bar is also useful to measure dimensions in your circuit. When a polygon or set of points is selected, the cursor position on the status bar will display a readout of the bounding box around the points or polygon.

TIP As you move the cursor over buttons in either the Tool Box or the Tool Bar a brief description of its function appears in the Messages section of the Status Bar.

Describing Menu Bar Accesses In this manual, we describe accessing the menu bar of xgeom using a “pointer” description to illustrate selecting the desired menu buttons. For example, Tools ⇒ Add metalization ⇒ Draw Polygon means to move the cursor to Tools on the menu bar, press and hold down the left mouse button, drag the cursor down the menu which appears until Add Metalization is highlighted, causing another pull-down menu to appear. Continuing to hold down the mouse button, drag the cursor until Draw Polygon is highlighted and release the mouse button. The new mode will be indicated on the status bar.

17

Xgeom User’s Manual If a (Shift) appears before the flow description, you should hold the shift key down while performing the menu selection. This enables you to remain in the mode selected, until you explicitly choose to exit it. For a complete explanation, see "Shift Selecting Modes" below.

Shortcut Keys Some menu items have a shortcut associated with them. These menu items can be selected by typing the associated shortcut key. Shortcuts are shown on the menu bar at the end of the name of the menu item. Typing the shortcut key has the same effect as selecting the menu item with the mouse, but with no need to go through the menus. Most of the shortcut keys are “control” keys, i.e. you must first hold down the Control key, and then press the key. In the menus the control key is indicated by “Ctrl +” before the letter. The “control” keys are denoted in the manual with a “^” symbol. Any time you see a “^” symbol before a letter, e.g., “^U”, you should interpret this as meaning a control key. As an example, to type a Control-U (^U), hold down the Control key and press the U key. A list of the shortcut keys is found in section “Keyboard Shortcuts,” page 150 of this manual.

Shift Selecting Modes When selecting a mode, such as Tools ⇒ Add Via, you will remain in the mode until you add one via to the circuit. After adding a single via, xgeom will then automatically return to pointer mode. If you wish to continue in the Add Via mode in order to add multiple vias, then hold the shift key down while selecting Add Via mode from the menu with the mouse. You will continue in this mode until you select another mode from the menu bar or press the Escape key which will place you back in pointer mode, the default startup mode, which allows you to edit your circuit. Note that shift selecting a mode applies to all methods of entering that mode, including selection of a button from the tool bar or tool box, or use of a shortcut key. Also, when using a button to enter a mode, double clicking on the button acts the same as pushing the shift key.

18

Chapter 1 Introduction XGEOM

Invoking Sonnet You use the Sonnet task bar, shown below, to access all the modules in the em Suite. Opening the Sonnet task bar, for both Windows and UNIX systems is detailed below.

UNIX 1

Open a terminal. If you do not know how to do this, please see your system administrator.

2

Enter “sonnet” at the prompt. The Sonnet task bar appears on your display.

19

Xgeom User’s Manual

Windows 1

Select Start ⇒ Programs ⇒ Sonnet ⇒ Sonnet from the Windows desktop Start menu.

The Sonnet task bar appears on your display.

20

Chapter 1 Introduction

Table 1 Sonnet Task Bar Buttons Button

Button Name

Sonnet Program

Edit Circuit

xgeom

Analyze Circuit

em Control

View Response

emgraph

View Current

emvu

View Far Field

patvu

Online Manuals

Adobe Acrobat

21

XGEOM

Once the Sonnet task bar is open, for UNIX or Windows systems, clicking on any given button opens the appropriate module. The table below shows which modules are invoked by each button.

Xgeom User’s Manual The translation programs, dxfgeo and gds, are accessed through the Sonnet task bar main menu, as shown below. Select Convert Dxf for dxfgeo and select Convert Gds for gds.

For details on each program, please refer to the appropriate user’s manual.

22

Chapter 2 Editing Your Circuit XGEOM

Chapter 2

Editing Your Circuit

This chapter provides an overview of xgeom modes of operation, and basic editing tasks. Changing the circuit is accomplished with the Edit, Modify and Tools options described in this section.

Pointer Mode The pointer mode allows you to select objects for editing. This is the default mode of xgeom upon opening a file. To return to this mode from another mode, press the ESC key or choose Tools ⇒ Pointer from the menu. When you are in pointer mode, the cursor appears as shown to the left.

23

Xgeom User’s Manual

Reshape Mode The reshape mode allows you to manipulate points in polygons. To enter this mode click on the reshape button in the tool box or select Tools ⇒ Reshape from the menu. You will remain in this mode until selecting another mode. Pressing the ESC key will return you to pointer mode.When you are in pointer mode, the cursor appears as shown to the left.

Selecting Objects for Editing Some of the xgeom functions require you to “select” a point or object prior to choosing the menu item. In order to select points you must be in the reshape mode. In order to select objects, you must be in the pointer mode. Once in a particular mode it is possible to select objects in a variety of ways which are explained below. When an object or point is selected it is highlighted on your display.

Clicking on an Object or Point Clicking on an object while in pointer mode will select the object. If the object is a polygon, clicking inside the polygon or the edge will select the polygon. The polygon will become highlighted when selected. To select a point, you must first be in the reshape mode. Then you select a point by clicking on that point. The selected point should become highlighted.

24

Chapter 2 Editing Your Circuit

Selected polygon Click inside the polygon to select it.

Dimensions of bounding box

Separate polygons sometimes have several vertices in common. If you want to select a vertex of, say, polygon A, but not the coincident vertex of polygon B, be sure the cursor is close to the vertex and just inside polygon A. The polygon A vertex is then selected.

Multiple Selects You may select multiple objects while in pointer mode and multiple points while in reshape mode. Objects include metal polygons, brick polygons, ports, and vias. Multiple polygon selects are very useful when you want to change the attributes of many polygons at once. The following menu items operate on multiple polygons: •

Edit

•

Edit

•

Edit

•

Edit

⇒ Cut ⇒ Copy ⇒ Duplicate ⇒ Delete 25

XGEOM

When a polygon or set of points is selected, the dimensions of the bounding box around the object(s) will appear in the cursor portion of the status bar. This provides a readily accessible measuring device.

Xgeom User’s Manual •

The Modify menu

The Modify ⇒ Attributes menu item can be used for metal polygons, dielectric brick polygons or ports. This menu item will open the Metalization Attributes dialog box, the Dielectric Brick Attributes box and the Port Attributes box, depending on which type of object is selected when the menu item is invoked. If all three types are selected, each type of dialog box will appear in succession in response to Modify ⇒ Attributes. For more details, see “Modify - Attributes,” page 118. Ports are attached to polygon edges; therefore, the port moves with the polygon. Also, if points are selected on a polygon, and a port appears between them, the port will move with the points. However, if those points are cut, the port will remain on the polygon. This also holds true for edge vias. If you are not familiar with edge vias, refer to “Edge Vias,” page 52 for a definition.

Select by Lassoing an Area One way to select multiple points while in reshape mode or multiple objects while in pointer mode, is to “lasso” an area which contains these objects. To lasso an area while in the pointer or reshape mode, move the cursor to the upper left corner of the region you want to select and press the left mouse button without releasing it. Hold the mouse button down and drag it to the right and down. A rectangular

26

Chapter 2 Editing Your Circuit

Lassoed area in pointer mode.

Selected polygons from lasso.

Lassoed area in reshape mode.

Selected points from lasso.

Unselecting Edit ⇒ Unselect will unselect any objects that are presently selected. This item is unavailable when nothing is selected.

TIP Clicking on the substrate, outside all selected objects, will also unselect all objects.

Multi-Layer Selects If your circuit requires more than one layer of metalization, you may want to modify the objects or points on some or all of these layers simultaneously.

27

XGEOM

box shows the area that will be selected. Now release the mouse button. All points or objects in the region are selected. An example of a lasso select in each mode, pointer and reshape, is shown below.

Xgeom User’s Manual To select points or polygons on more than one layer at a time, toggle single layer select to the off position by selecting Edit ⇒ Single Layer Select. This puts you in the Multi-Layer select mode. When in this mode, the select by lassoing or select by pointing applies for all layers. This mode is usually used for moving and copying multilayer structures, such as vias.

Shift and Control Keys When performing multiple selects, it may be desirable to add items to items already selected or to remove items from previously selected items. In order to add items, without unselecting any other item, press the Shift key while performing the select. To unselect items, which are already selected, without unselecting any other item, press the Control key while performing the unselect. Note that the Shift key allows you to select additional objects, while the Control key toggles the state of the objects. If it is currently selected, pressing the Control key while performing a selection action will unselect the object. If the object is not currently selected, then pressing the Control key while performing a selection action will select the object. The above applies whether you are performing selections by clicking on a single item or by “lassoing” a group of items.

Points The following sections describe how to move, add and delete points.

Moving Points If you need to move a point, you must first be in the reshape mode as described in the previous section. If you wish to move a single point, then position the cursor on the point and press, without releasing, the left mouse button. With the button held down, move the mouse. The point moves as you move the mouse. Drag the point to the desired location and release the button. If you wish to move multiple points, select the points by lassoing them as described above. Then place the

28

Chapter 2 Editing Your Circuit

Notice that the points that you are moving always “snap” to the nearest grid point. If you want to move a point between your grid points, you must change a property called the “snap distance”. See “Tools - Snap Setup,” page 116 for information on how to do this.

Adding Points to a Polygon Sometimes you may want to change the number of points that make up a polygon. For example, you might wish to change a triangle to a square or a pentagon. To change a triangle to a rectangle, select Tools ⇒ Add Points to Polygon from the menu or click on the Add Points to Polygons button in the tool box. Then click on one of the sides of the triangle that you want to modify.

Starting triangle polygon

Added point.

29

XGEOM

cursor on any one of the highlighted points and drag the mouse. Note that, in both cases, the polygon will be reshaped to accommodate the new position(s) of the point(s).

Xgeom User’s Manual A highlighted point appears. Place your cursor over the highlighted point, press the left mouse button, then drag the point to the desired location. Two new line segments connect the new point to the existing endpoints.

New point

Final rectangle after new point has been moved.

You may keep adding points in this fashion for multi-sided polygons. If you make an error, you may press the Delete or Backspace key and the last point entered is deleted. When done, push the ESC key to go to pointer mode or select another menu action to exit the Add Points to Polygon mode.

Deleting Points If you wish to delete points from an existing polygon, invoke the reshape mode by selecting Tools ⇒ Reshape or clicking on the Reshape button in the tool box. Then select the points you wish to delete by lassoing an area that contains them or by clicking on a single point. Then delete the points by pressing the Delete key or by selecting Edit ⇒ Delete on the menu bar.

Polygons, Ports and Vias The following sections describe how to edit and modify polygons. For more information about handling ports, see “Manipulating Ports,” page 40. For more information about handling vias, see Chapter 4, “Using Vias”.

30

Chapter 2 Editing Your Circuit XGEOM

Moving Polygons If you need to move a polygon, you must first be in the pointer mode as described above. Then position the cursor inside the polygon and press, without releasing, the left mouse button. The polygon is highlighted, indicating selection. With the button held down, move the mouse. The polygon moves as you move the mouse. Drag the polygon to the desired location and release the button. Notice that the polygons that you are moving always “snap” to the nearest grid point. If you want to move a point or polygon between your grid points, you must change a property called the “snap distance”. See “Tools - Snap Setup,” page 116 for information on how to do this. If you want to move a group of polygons in unison, start by selecting all of the polygons you want included in the move. Each polygon is highlighted. When all polygons are selected, click on one of the polygons. Holding the button down, drag the polygons to their new location.

Deleting Polygons To delete a polygon or polygons, go to the pointer mode by pressing the ESC key or selecting Tools ⇒ Pointer. Then, select the polygon(s) that you wish to delete, and press the Delete key, the Backspace key or select Edit ⇒ Delete.

Cutting and Pasting Polygons When you delete polygons using Edit ⇒ Cut, they are removed from your circuit and inserted into the xgeom clipboard. You can cut and paste from one layer to another or within the same layer or even to another circuit. You can cut or copy polygons from one xgeom window and paste them into any other xgeom window. The polygons remain in the clipboard until you replace them by cutting or copying again. Polygons, with their ports and edge vias are copied into the buffer; i.e., single points, or individual ports or vias are not copied into the buffer. To insert polygons into the xgeom clipboard and remove them from your circuit, first select the polygons that you wish to cut. Now choose Edit ⇒ Cut or click on the Cut button on the tool bar. The selected polygons are deleted. To retrieve 31

Xgeom User’s Manual the polygons from the xgeom clipboard, choose Edit ⇒ Paste or click on the Paste button on the tool bar. The polygons are copied from the clipboard to your circuit. Also note that the polygons just pasted in are still selected. This makes it easy to move them to a new location. Cutting and pasting using Edit ⇒ Cut and Edit ⇒ Paste can be used to move polygons from one layer to another layer, or from one xgeom window to another. Edit ⇒ Copy is the same as Edit ⇒ Cut but the selected polygons are placed into the clipboard without being deleted from the circuit. Use this menu item when you want to keep the original polygons where they are and make a copy of them somewhere else. Edit ⇒ Duplicate combines Edit ⇒ Copy and Edit ⇒ Paste into one menu item. This menu item provides a fast way to copy polygons to another location on the same layer. To do this, first select the polygons and then choose Edit ⇒ Duplicate. The selected polygons are inserted into the buffer and duplicates of the polygons are then pasted on top of the original polygons. The pasted polygons are highlighted to remind you that they are still selected.

Moving Polygons to Another Layer To move polygons to another layer, first select the polygons that you wish to move. Then choose Edit ⇒ Cut. Now move to the new layer where you wish to place the polygons. This may be done by using the shortcut keys ^D (down) and ^U (up). Now choose Edit ⇒ Paste and the polygons are placed on the new layer.

Flipping, Rotating, and Resizing Xgeom provides you with a way of flipping, rotating, and resizing polygons. To use these features, first select the polygon(s) that you wish to change, and then choose the appropriate menu item. You are then prompted for more information. For example, to create the mirror image of a polygon, first select the polygon, and then choose Edit ⇒ Duplicate so that the original polygon remains unchanged. Now choose Modify ⇒ Flip to open the Flip dialog box and select a Pivot Point. The polygon is flipped about an axis passing through the pivot point you have

32

Chapter 2 Editing Your Circuit

33

XGEOM

chosen. Then choose either Left-Right or Up-Down direction. For a complete description of the Flip, Rotate, and Resize functions, see the appropriate entries in "The Modify Menu" on page 118.

Xgeom User’s Manual

34

Chapter 3 Using Ports XGEOM

Chapter 3

Using Ports

Introduction All ports are two-terminal devices. In most applications, the first terminal is attached to a metal polygon, and the second terminal is attached to ground. Such ports are referred to as grounded ports. Occasionally, however, it is useful to attach the two terminals of a port between two abutted polygons. These ports are referred to as ungrounded ports. Ports are not allowed on diagonal lines and ports are not allowed to overlap each other (just touching is permitted). Em tests for these error conditions. When analyzing multi-port circuits to find S, Y, or Z parameters, all of the ports in the circuit are normally grounded. An ungrounded port can have a different ground reference from other ports in the circuit, so it is important to exercise care when using ungrounded ports to avoid corrupting the analysis results. For more details on the use of grounded and ungrounded ports, see Chapter 5, “Ports” in the Em User’s Manual. 35

Xgeom User’s Manual In addition to being either grounded or ungrounded, ports can be further characterized by their location in a circuit, and by whether or not em can de-embed them. Each port type is described below.

Standard Box-Wall Ports A standard box-wall port is a grounded port, with one terminal attached to a polygon edge coincident with a box wall, and the second terminal attached to ground. An example of a standard box-wall port is shown below. Standard boxwall ports can be de-embedded.

Page 18

Standard Ungrounded-Internal Ports A standard ungrounded-internal port is located in the interior of a circuit, and has its two terminals connected between abutted metal polygons. An ungroundedinternal port is illustrated in the next figure on page 37. Ungrounded-internal ports can be de-embedded by em.

36

Chapter 3 Using Ports

-

+

Example of a circuit with an ungrounded internal port.

In the figure above, the ungrounded-internal port is attached between two polygons which have equal widths. This is not a necessary condition for ungrounded-internal ports. These ports can also be attached between polygons which are still abutted, but have unequal widths, as shown below. The only difference between the two conditions is that de-embedding requires the use of more standards (and therefore more time) when the polygons have unequal widths.

37

XGEOM

Ports on the edge of a single polygon are not allowed, with em generating a “Port is not connected between two polygons” error message (see “Error Messages,” page 308 of the Em User’s Manual). Extreme care should be taken in interpreting the results since the port has no access to ground.

Xgeom User’s Manual

Via Ports A via port has one terminal connected to a polygon on a given circuit level, and the other terminal connected to a second polygon on a circuit level above or below the first polygon. For more details on vias, see Chapter 4, “Using Vias”. Thus, when ports are desired on the interior of a circuit, capture a via between two layers and add a port to the edge via. An example of this port type is shown in the figure below. Note that the triangle symbols in the figure represent a via. Level 0

Level 1

Example of a circuit with a via port (both levels shown.)

Em cannot de-embed via ports. However, in a circuit which contains a combination of via ports and other port types, the other port types can still be deembedded. Em will automatically identify all of the other ports present in the circuit and de-embed them, but leave the via ports unde-embedded. In most cases where you need grounded ports, your first choice would be to use automatic-grounded ports as discussed in the next section. If you need a port with the flexibility to be connected between any two layers of your dielectric, you will want to use the via port. The example “patch.geo” included with your software has an example of using a via port.

38

Chapter 3 Using Ports XGEOM

Automatic-Grounded Ports An automatic-grounded port is a special type of port used in the interior of a circuit. This port type has one terminal attached to the edge of a metal polygon located inside the box and the other terminal attached to the ground plane through all intervening dielectric layers. An auto-grounded port is illustrated below.

page 21

In many circuits, the addition of auto-grounded ports has little influence on the total analysis time of the em job. However for some circuits, auto-grounded ports may require some extra overhead calculations, thus increasing the total analysis time. Therefore, they should be used only when they provide an advantage over standard box-wall ports. Auto-grounded ports provide advantages over standard box-wall ports when: •

the layout of your circuit does not allow a direct path for a feed line to be connected between the port and the box wall, as in the figure above, or

•

your circuit requires a large feed structure to reach the box wall. If all or part of your feed structure can be eliminated, using an autogrounded port could reduce the total number of subsections in your circuit, thus decreasing the analysis time and/or memory requirements.

Auto-grounded ports are similar to via ports with the exception of the following characteristics: •

Via ports require you to manually create vias that extend upward through the dielectric to the edge of a metal polygon. This is not the 39

Xgeom User’s Manual case with auto-grounded ports. You can simply place autogrounded ports anywhere a grounded port is needed. Em automatically detects the presence of auto-grounded ports in the circuit and connects the port terminals appropriately. •

Auto-grounded ports connect directly through all dielectric layers to the ground plane. Via ports allow the flexibility of connecting between any two adjacent dielectric layers.

•

Auto-grounded ports are de-embedded when the de-embedding option for em is used, while via ports are not.

•

Reference Planes may be set with auto-grounded ports but cannot be set for via ports.

Manipulating Ports The following sections explain the basics of manipulating ports; how to add and delete them and change their characteristics.

Adding Standard Ports 1

To add a standard port to a circuit, click on the Add Port button in the tool box, or select Tools ⇒ Add Port from the main menu.

2

Then click on the polygon edge at the desired position for a port. A small box with a number in the center will appear on your circuit, indicating the position of the port. Refer to the figure on page 36.

Ports are numbered automatically, in the order in which they are added to your circuit, starting at the number one. You may later choose to change a port number; this operation is discussed below.

Adding Via Ports 1

40

To add a via port, you must first have two polygons on adjacent levels with a via connecting them. For more details on how to create vias, see Chapter 4, “Using Vias”.

Chapter 3 Using Ports Then click on the Add Port button in the tool box, to put you in Add Port mode.

3

Then click on the edge via at the lower level polygon to add the port. An example is shown in the figure on page 38.

Adding Auto-grounded Ports 1

To add an auto-grounded port, proceed as you would to add a standard port, as described above.

2

Click on the port to select it. This will enable the Modify menu option.

3

Select Modify ⇒ Attributes to open the Port Attributes dialog box which will resemble the figure below.

4

Now the port can be changed from a standard port to an auto-grounded port by choosing Autognd from the drop list in the Type field. The port should now be similar to the figure on page 39.

page 23

Deleting Ports All types of ports are deleted in the same manner. Simply select the port, then perform a Cut or Delete operation, either through the tool bar or menu, to remove the port from the circuit.

41

XGEOM

2

Xgeom User’s Manual

Changing Port Numbering 1

Click on the port or ports to select it/them. This will enable the Modify menu option.

2

Select Modify ⇒ Attributes to open the Port Attributes dialog box, shown in the figure on page 41.

3

Now the port number can be changed by typing the desired number in the Number text entry box of the dialog box. Any nonzero integer, negative or positive, is valid. Note that if multiple ports are selected, “Mixed” will appear in the Number text entry box. Editing this field will apply the same number to all selected ports.

Ports are numbered in the order that they are entered into the circuit. There is no limit on the number of ports and the number of ports has absolutely no impact on analysis time. As many physical ports as desired may be given the same numeric label, and all ports with the same label are electrically connected together as illustrated below and have identical parameters. Such ports are called “push-push” ports and have many uses such as simulating thick metal. See Chapter 18, “Thick Metal with Arbitrary Cross-Section” in the Em User’s Manual for additional details.

Ports may also have negative labels as shown in the figure on page 43. This feature can be used to redefine ground. Strictly speaking, em sums the total current going into all the positive ports with the same port number and sets that equal to the total current going out of all the ports with that same negative port number. For example, for a circuit with a +1 port and a -1 port, em sets current flowing into

42

Chapter 3 Using Ports

An example of push-pull ports.

Ports may be non-sequential, i.e., you may have only 2 ports, one labeled “1” and the other labeled “4”. The port order for the S,Y, or Z parameters will be listed in increasing numeric order. For the example of a two-port with ports labeled “1” and “4”, the output would be as follows: S11, S41, S14, S44.

Changing Port Impedance There are two methods for changing the impedance of a port. If you wish to change the impedance of a given port, and do not need to see the impedance values of other ports, take the following steps: 1

Click on the port or ports to select it/them. This will enable the Modify menu option.

2

Select Modify ⇒ Attributes to open the Port Attributes dialog box which is pictured in the figure on page 41.

3

Now the impedance values can be changed by typing the desired values in the Resistance, Reactance, Inductance and Capacitance text boxes in the dialog box. This changes the parameters on all ports selected and all ports with the same number as the ports selected.

If you wish to change the impedance of a given port, and wish to see the impedance values of other ports while doing so, proceed as follows:

43

XGEOM

port +1 to be equal to the current flowing out of port -1. Thus the name “balanced”, or “push-pull” port. See Chapter 17, “Coplanar Waveguide Discontinuities and Balanced Ports” in the Em User’s Manual for more details.

Xgeom User’s Manual 1

Select Parameters ⇒ Ports from the main menu to open the Port Impedance dialog box.

2

Now the impedance values for any port can be changed by typing the desired values in the Resistance, Reactance, Inductance and Capacitance fields in the row labeled with the desired port number.

TIP Note that the impedance of multiple ports may be changed at the same time through the first method by selecting multiple ports before selecting Modify ⇒ Attributes, and by the second method, by modifying all the desired port values while the Port Impedance dialog box is open.

Defining Reference Planes/Calibration Lengths Reference Planes or Calibration Lengths, which are mutually exclusive, can be set for most types of ports, but the method differs according to the port type. Reference planes cannot be set for via ports since em cannot de-embed them. Detailed below are the methods for standard ports and auto-grounded ports.

Standard Ports For standard ports, which normally reside on the box wall, the Reference Plane or Calibration Length refers to the distance to the box wall. To set the Reference Planes/Calibration lengths, do the following:

44

Chapter 3 Using Ports XGEOM

1

Select Parameters ⇒ Ref. Planes/Cal. Lengths from the main menu. This will open the Reference Planes/Calibration Lengths dialog box.

2

Select the side from which the reference plane or calibration length is to extend (Top, Left, Right, or Bottom) and choose Cal or Ref from the drop list in the Type field.

3

Then enter the desired length in the Length field from the keyboard. Alternatively, you may use the mouse by clicking the Use Mouse button and then clicking in your circuit at the desired location. The Length entry will be updated to reflect your selection.

4

To remove a reference plane or calibration length, enter a value of zero in the Length field. Alternately, click on the Use Mouse button, then click outside the substrate. There is only one plane or length per box side. The same reference plane or calibration length is used on all ports on all levels on that side of the box.

45

Xgeom User’s Manual Auto-ground Ports For autoground ports, the reference plane or calibration length is defined as the distance from a particular port. To set a reference plane or calibration length for an auto-grounded port, do the following; 1

Click on the port to select it. This will enable the Modify menu option.

2

Select Modify ⇒ Attributes to open the Port Attributes dialog box depicted in the figure on page 41.

3

Then enter the desired length in the Ref. Plane or Cal. Length field in the Autognd Port Data box. Alternatively, you may use the mouse for setting the reference plane by clicking the Use Mouse button and then clicking in your circuit at the desired location. The Ref. Plane entry will be updated to reflect your selection.

4

To remove a reference plane or calibration length, enter a value of zero in the respective field.

TIP Changing a port to an autoground type and setting up a reference plane or calibration length for the port can be accomplished at the same time in the Port Attributes dialog box. It is also possible to set calibration lengths for multiple ports by selecting the desired ports, selecting Modify ⇒ Attributes and inputting a value in the calibration length text entry box in the Port Attributes dialog box.

Creating Auto-grounded Ports: An Example This next section details how to add auto-grounded ports to a circuit, and set up Reference planes/Calibration lengths to the port. You add an auto-grounded port to a circuit in the same way that you add a “standard” box-wall or ungrounded-internal port. The only difference is that after adding the port, you must change its type to “auto-grounded.”

46

Chapter 3 Using Ports

Place Standard Box-Wall Ports Here

Place Auto-Grounded Ports Here

This circuit is available in the Sonnet examples directory. To obtain a copy of the geometry file, type the following command: copyex autonopt.geo The copyex command will copy the file “autonopt.geo” from the examples directory to your present directory. Then, if you would like to load the file into xgeom, type: xgeom autonopt Now we will go through the steps you would use to create two “standard” boxwall ports and two “auto-grounded” ports at the locations indicated in the figure above. 1 2

Select (Shift)Tools ⇒ Add Port. Click with the mouse on each of the indicated polygon edges. The resulting circuit will be similar to that shown in the figure on page 48. Note that all four ports are “standard” port types.

47

XGEOM

To illustrate the procedure for adding auto-grounded ports to a circuit, let’s begin with the example circuit shown below.

Xgeom User’s Manual

Example circuit with “standard” ports added to polygon edges.

3

Press ESC to return to pointer mode.

4

Use the mouse to lasso ports #3 and #4. This action will highlight the ports you select.

5

Select Modify ⇒ Attributes. This will open the Port Attributes dialog box shown in the figure on page 41.

6

Now the ports can be changed from standard ports to auto-grounded ports by choosing Autognd from the drop list in the Type field.

7

Click on the OK button to exit the Port Attributes dialog box.

After converting ports #3 and #4 in this manner, the circuit will appear as shown below.

The circuit now has the two box-wall and two auto-grounded ports in the desired locations.

48

Chapter 3 Using Ports

copyex autoport.geo

Auto-grounded Port Reference Planes Just as it is possible for you to define reference planes for the de-embedding of box-wall ports, you can also define reference planes for the de-embedding of autogrounded ports. We will again consider the example circuit shown in the figure on page 48. To add reference planes of length 508 µM for Port #3 and length 1016 µM for Port #4, perform the following steps. 1

Click on Port #3 to select it and enable the Modify menu item.

2

Select Modify ⇒ Attributes to open the Port Attributes dialog box, shown in the figure on page 41.

3

Then enter 508 in the Ref. Plane field of the dialog box.

4

Click on the OK button to close the dialog box. The reference plane will be drawn on the circuit.

5

Click on Port #4 to select it and enable the Modify menu item.

6

Select Modify ⇒ Attributes to open the Port Attributes dialog box, shown in the figure on page 41.

7

Then enter 1016 in the Ref. Plane field of the dialog box.

8

Click on the OK button to close the dialog box. The reference plane will be drawn on the circuit.

49

XGEOM

This circuit is available in the Sonnet examples directory and may be obtained by typing:

Xgeom User’s Manual Upon completion of these steps, the circuit layout now appears as shown below.

50

Chapter 4 Using Vias XGEOM

Chapter 4

Using Vias

The examples in the first few chapters of this manual show circuits that use only planar (X-Y) currents. Vias allow current to flow in the Z-direction. Vias can extend from ground to the substrate surface, a ground via, as well as between levels in a multilayer structure. Note, however, that vias, by definition, always extend upward. They can also extend to the top cover of the box.

Via Concepts To create a via that connects one layer to another layer, you must follow these steps: 1

Draw a polygon on each of the two layers that you want to connect.

2

Define the position of the via by creating an “edge via” (defined later).

3

If needed, modify the polygons so that vias are as desired.

51

Xgeom User’s Manual

Creating the Polygons First, you must create the two polygons that you want to connect with a via. This is true even if you want to connect a via to ground. The polygons must be on adjacent levels. If they are not, you will need to create a series of vias - each one connecting polygons on adjacent levels. Type ^U and ^D to move up and down between the two levels that you want to use, and create a polygon on each level. If the via is going to ground, create a “dummy” polygon on the ground level which represents the bottom of the xgeom box. It is called a “dummy” polygon because the ground level is already completely metallized and the polygon does not change this characteristic.

Edge Vias To connect two polygons with a via, you must first create an “edge via”. An edge via is an edge of a polygon that a via originates from. To create an edge via, go to the lower of the two polygons to which you want to connect a via. Now use Tools ⇒Add Via and click on the edge of the polygon where you want the via to go. If the cell fill is on, the edge via will appear in inverse coloring with two triangles shown representing the via.

52

Chapter 4 Using Vias

a) Cell Fill “Off”

b) Cell Fill “On”

Xgeom uses edge vias to define the position of via posts. When the cell fill is turned “On”, the via posts can be seen as reverse video metalization cells.

Via Posts With the metalization turned on, by setting View ⇒ Cell Fill to “On”, the via subsections, called “via posts”, are also displayed in reverse video. See b. When em subsections the circuit, it subsections each edge via, which you specify, into subsectional vias called “via posts”. Each via post is a rectangular cylinder of current, extending between the present level to the next level above. A via post has a horizontal cross-sectional area equal to one cell and a height equal to the thickness of the dielectric layer. If you change the cell size, then the edge via is resubsectioned into via posts with the new cell size. Xgeom places enough via posts to cover the entire length of the edge via. We will refer to this series of via posts as a “via fence.” See and . To view vias as they are being captured, it is convenient to be able to change the viewed level in xgeom quickly. To do so, just type ^U (Control-U) to go up one level, towards the box top, or ^D to go down one level.You may also click on the Up One Level or Down One Level button on the tool bar.

53

XGEOM

The “up” via symbol indicates that the via post connects this level to the next level above as shown below. Move up one level. The vias are shown on this level with a “down” via symbol which is a “down” triangle. The “down” via symbol was generated automatically when the via-edge was added on the lower level.

Xgeom User’s Manual If you want a level to be displayed as a “ghost” outline whenever you are not on that level, make the level visible in the Levels dialog box which appears in response to selecting View ⇒ Metalization Levels. Then you can see how different levels of metalization line up. You may also use the Levels dialog box to turn off the visibility of any given level. By default, xgeom starts with all levels visible.

Adding a Via to Ground To add a via to ground, you should go to ground level, the bottom of the box and add a polygon. Then add an edge via to the polygon edge. The polygon on ground level is not subsectioned and does not change the characteristics of the ground.

54

Chapter 4 Using Vias XGEOM

The ground level is completely metallized, with or without the polygon. The polygon is created only to provide a polygon edge so a via can connect to the ground level. A via to ground is depicted below. Edge-via is added to this polygon. Polygon

Down Via Post

Up Via Post

Ground Level

Level 0

Via

Level 0 Metallization

Dielectric Layer Ground Level Metallization

The lower part of the figure depicts a via going from the single metalization level to ground. The same via is shown in the top of the figure as it appears in xgeom. The via post is a metal filled column which penetrates through the dielectric layer and connects the metalization level to ground.

55

Xgeom User’s Manual The illustration below depicts a via going to ground. However, this time the via is wider than a cell, so a via fence, made up of adjacent via posts, is used.

Edge-via is added to this polygon.

Level 0

Ground Level

Via Fence

Level 0 Metallization Via Post

Dielectric Layer

Ground Level Metallization

The lower part of the figure depicts a via fence going from the single metalization level to ground. The same via is shown in the top of the figure as it appears in xgeom.

56

Chapter 4 Using Vias XGEOM

Making Entire Polygons Vias To turn entire polygon(s) into via(s), first select the polygon(s), then choose Modify ⇒ Add Vias to All. Edge vias will be attached to all sides of the polygon simultaneously. If your polygon has diagonal lines or points not on the grid, xgeom does a staircase fit to the polygon. Xgeom only attaches via posts to the edges of the polygon, creating a hollow “cylinder” of metal. Since current tends to flow on the surfaces of conductors, creating a solid cylinder would just waste subsections.

Deleting Vias To delete an edge via, select the via while in pointer mode by clicking on a triangle of the edge via that you want deleted. Then select Edit ⇒ Cut from the menu bar or the Delete key to delete it. Make sure that you are on the level where the via post starts, and emanates upward, and not on the level above where the via post ends. Deleting an edge via deletes the via-fence associated with it.

Via Loss The loss for the via post is determined by the metalization of the polygon that the via is associated with. See “Parameters - Metal types,” page 138 for an explanation on how to set the metalization loss.

57

Xgeom User’s Manual

Shorted Vias One commonly occurring error using vias is to create an edge via that is adjacent to a port. An example is shown below. This shorts the port to the box wall. In general, never allow a via post to touch the edge of the box. Right

Wrong

Via Ports Vias may also be ports. This is a special case of the internal port described in “Via Ports,” page 38. The port is inserted between two levels. An example of a via port is shown below. The example “patch.geo” included with your software has an example of using a via port.

A via port with both levels shown.

58

Chapter 4 Using Vias XGEOM

Creating an Airbridge: An Example A good example of the use of vias is an airbridge. This section is a tutorial on creating a transmission line that bridges over another transmission line.

Loading in the Example File The example file for the airbridge is “bridge.geo.” Obtain a copy of “bridge.geo” using copyex. If you are not familiar with copyex, see “Obtaining the Example Files,” page 7 in the Sonnet Tutorial. Now load the example file into xgeom. To do so, perform the following: 1

Select xgeom from the Sonnet menu in the Windows desktop Start menu.

2

Select File ⇒ Open from the xgeom main menu. The Open File dialog box will appear.

3

Double-click on “bridge.geo” in the file list to open the file in xgeom. If “bridge.geo” does not appear, use the browse button to locate and open the file.

Your screen should look like this:

a) Upper level (Level 0)

b) Lower level (Level 1)

Unfinished airbridge example. The text describes how to complete the bridge.

59

Xgeom User’s Manual

Circuit Description This example file contains a 6-port circuit. However, the ports are on another level. Type ^D key or down arrow to go down one level. Your screen should now look like b. This level shows a transmission line going from ports 1 to 2, another transmission line going from port 3 to a via post (notice the arrow-up symbol), a similar line at port 4, and two open stubs at ports 5 and 6. Let’s concentrate on what happens on the line connected to port 3. This line goes to a via post, indicated by the arrow-up symbol. The via post connects the line to another line on the level above this one, which appears as a ghost image. Type ^U to go up one level. You are now looking at the highest level again, level 0. There is a single rectangle of metal that acts as the “bridge” of the airbridge. Notice the two arrow-down symbols. These were automatically created by xgeom and indicate that there is a via post coming up from the level below this level. Remember, via posts always project upward from the level that they are specified on.

Creating the Polygons Now let’s connect the two open stubs connected to ports 5 and 6 to another airbridge identical to the one that is already there. As a general rule, it is best to create the polygons on both layers first. In this example, the polygons on level 1 have already been entered for you and the only polygon left is the actual airbridge. Type ^M to turn the cell fill off and create a new polygon similar to the one that is already there by using Tools ⇒ Add Metalization ⇒ Draw Rectangle or ^R for example. It is good practice to create polygons such that the edges are in line

60

Chapter 4 Using Vias

a) Upper level (Level 0)

b) Lower level (Level 1)

Adding Edge Vias Now type ^D to go back to level 1 shown in on the right in the figure above. Select (Shift) Tools ⇒ Add Vias and click on the end of the each of the open stubs. Three things show up for each stub. First, a dark dashed line is displayed. This is the edge via. Second, an arrow-up symbol is displayed, signifying a via post that goes up to the next layer. Third, a new metallized polygon shows up. This shows you the actual size and position of the via post. See the next illustration. Now go up to

61

XGEOM

with each other. This makes it possible to change your cell size without having to change your edge vias. When you are done, turn the cell fill back on. Your circuit should now look like this:

Xgeom User’s Manual level 0. Your circuit should look appear as below. An arrow-down symbol and an extra metallized square polygon were automatically created by xgeom for each edge via that you entered. Press the ESC key to return to Pointer mode.

a) Upper level (Level 0)

b) Lower level (Level 1)

Completed airbridge example after adding two edge vias.

Summary of Vias Here is a summary of how to use vias:

62

•

Xgeom uses edge vias to determine the position of a via post.

•

When creating edge vias, start on the lowest level and work up.

•

To create a via to ground, start by specifying a polygon on the ground layer.

•

Create your polygons before adding vias.

Chapter 5 Dielectric Bricks XGEOM

Chapter 5

Dielectric Bricks

Although em is a “primarily planar” electromagnetic simulator, it also has the capability to add “dielectric brick” material anywhere in your circuit. A dielectric brick is a solid volume of dielectric material embedded within a circuit layer. Dielectric bricks can be made from any dielectric material, including air, and can be placed in circuit layers made from any other dielectric material, also including air. All realizable values for the dielectric constant, loss tangent, and bulk conductivity can be used to define a dielectric brick. Furthermore, it is possible to set these parameters independently in each dimension to create anisotropic dielectric bricks. Keep in mind, however, that dielectric bricks add a large number of subsections to your circuit, thus substantially increasing memory requirements and analysis times.

63

Xgeom User’s Manual

Applications for Dielectric Bricks The use of dielectric bricks is appropriate for applications where the effects of dielectric discontinuities or anisotropic dielectric materials are important. Examples of such applications include dielectric resonators, dielectric overlays, air bridges, microstrip-to-stripline transitions, dielectric bridges and crossovers, microslab transmission lines, capacitors, and module walls.

Creating a Dielectric Brick To create a dielectric brick in xgeom, first move to the circuit level where the base of the dielectric brick is to be located. The dielectric brick that is created will rest on this circuit level, and will extend upward to the next level. Dielectric bricks can be placed on any level, including the ground plane. If a brick is placed on the highest circuit level (level 0), it will extend up to the top cover of the metal box. When on the circuit layer where the base of the dielectric brick is to be located, the next step is to create a base polygon which defines the cross-section of the brick. This is done in xgeom by selecting either Tools ⇒ Add Dielectric Brick ⇒ Draw Rectangle or Tools ⇒ Add Dielectric Brick ⇒ Draw Polygon. The first option allows the vertices of arbitrarily shaped base polygons to be entered on a point by point basis. This option can always be used to create dielectric bricks with any cross-sectional shape. However, if the cross-section is rectangular in shape, it is often quicker to create dielectric bricks using the second option. Once a dielectric brick has been created in xgeom, it is possible to “see” the brick from both the circuit layer where the base of the brick is located and the circuit layer where the top of the brick is located. On both levels, you will see a polygon which defines the cross-sectional shape of the dielectric brick. The brightness of the polygon, however, will vary depending upon whether you are on the top level, where you will see a “dim” polygon, or the base level, where you will see a “bright” polygon. Note that while it is possible to “see” a brick from two different circuit levels, “selecting” a brick, for cutting, copying, moving, changing attributes, etc., can only be done from the circuit level where the base of the brick is located if you are in Single layer edit mode. The polygon can be selected on either level if you are in multilayer select.

64

Chapter 5 Dielectric Bricks

Defining Dielectric Brick Materials Just as it is possible to define a variety of metal types, each with different properties, it is also possible to define a variety of dielectric brick materials, each with different values for the dielectric constant, loss tangent, and bulk conductivity. To define a new dielectric brick material, or to modify the characteristics of an existing material, select Parameters ⇒ Brick Materials. This will bring up the Brick Materials dialog box, shown on page 66, which shows all the dielectric brick materials previously defined, the xgeom color/fill pattern assigned to each brick material, and whether the material is isotropic or anisotropic. To modify the settings for a particular dielectric brick material, edit

65

XGEOM

Finally, it is possible to turn dielectric bricks “on” or “off” in xgeom by selecting View ⇒ Object Visibility. This will open the Object Visibility dialog box shown below. Click on the Only Objects Checked Below radio button. The object choices will now be available. Click on the Dielectric Bricks checkbox to turn it off. This will make any bricks present in the circuit invisible and unselectable, but does not remove them from the circuit. The dielectric bricks can be turned back “on” by once again selecting View ⇒ Object Visibility and clicking the Dielectric Bricks checkbox or the All Objects radio button. Occasionally, when a circuit contains many layers, with overlapping metal polygons and dielectric bricks, it may be somewhat difficult to distinguish the metal polygons and dielectric bricks from one another. The ability to turn dielectric bricks “off” usually makes it easier to view such circuits.

Xgeom User’s Manual that materials text entry boxes. Note that for anisotropic materials all the parameters do not fit in the dialog box simultaneously, so that it is necessary to use the scroll bars to access all settings. The brick materi als dialog box.

If the brick type is isotropic only one set of parameters, X, will be set. Conversely, if the brick material is set to anisotropic, each parameter is defined separately for the X, Y, and Z dimensions. If you wish to make a brick material anisotropic, click on the Ani checkbox. The “default” material used when new dielectric bricks are created can also be set in the Brick Materials dialog box. Select a brick type from the Default for add bricks drop list. Once the default material has been set, all bricks created thereafter will be made of that material.

Changing Brick Materials The material type for bricks that already exist in a circuit can be changed by following the procedure given below: 1

66

Select the brick(s) by clicking on it or lassoing it.

Chapter 5 Dielectric Bricks XGEOM

2

Select Modify ⇒ Attributes. This will open the Dielectric Brick attributes dialog box, shown below.

3

Select the brick material you desire from the drop list labeled Brick.

4

Click on the OK button to apply your selection and close the dialog box.

Z-Partitioning A dielectric brick simulates a volume of dielectric material. Because a brick simulates a volume, it must be subsectioned in the X, Y, and Z dimensions. The more subsections (better resolution) used in each dimension, the more accurate the analysis. X/Y subsectioning of dielectric bricks is identical to X/Y subsectioning of metal polygons. You can control the X/Y subsectioning of both through your choice of grid size, XMIN, YMIN, XMAX, YMAX, and subsections-per-lambda. Z subsectioning of dielectric bricks is controlled by the “Z-Parts for bricks” parameter under Parameters ⇒ Dielectric Layers. This parameter specifies the number of Z partitions for all dielectric bricks on a particular circuit layer.

67

Xgeom User’s Manual To set this parameter in xgeom, select Parameters ⇒ Dielectric Layers. Xgeom will then display the Dielectric Layers dialog box, shown below.

Listed in the far right column of text entry boxes labeled Z-Parts. is the number of Z partitions for each circuit layer. To change a value in this column, edit the applicable text entry box. Note that the “number of Z partitions” parameter only affects dielectric bricks. Changing this value for a particular layer will have absolutely no affect on the analysis if there are no bricks on the layer. If there are multiple bricks on the layer, the Z subsectioning for all of those bricks will be identical. It is not possible to apply different Z partitions to brick polygons which appear on the same layer. For additional information on dielectric bricks, refer to Chapter 14, “Dielectric Bricks” of the Em User’s Manual.

68

Chapter 6 Palette of Standard Geometries XGEOM

Chapter 6

Palette of Standard Geometries

Introduction Xgeom includes a palette of standard geometries, comprised of commonly used circuit elements, available under the Add Metalization menu. Each item has an associated dialog box which allows you to input the desired parameters for the element. Once the element is added to the circuit, it is converted to its component polygons. These polygons may then be manipulated using all the normal xgeom functions. An example of a parameter dialog box, for the fan stub, is shown below. To change the parameter values, edit the corresponding text entry box.

69

Xgeom User’s Manual .

The table below lists the available geometries, and the associated command for adding it to your circuit. For a description of the associated parameters, see the entry for the geometry in this chapter. Geometry Rectangle

Command Tools ⇒ Add Metalization ⇒ Rectangle

Interdigital Capacitor Tools ⇒ Add Metalization ⇒ InterDigCap

70

Entry Page 71 Page 72

Donut

Tools ⇒ Add Metalization ⇒ Donut

Page 74

Meander

Tools ⇒ Add Metalization ⇒ Meander

Page 76

Round Spiral

Tools ⇒ Add Metalization ⇒ Round Spiral

Page 77

Rectangular Spiral

Tools ⇒ Add Metalization ⇒ Rectangular Spiral

Page 78

Parallel Lines

Tools ⇒ Add Metalization ⇒ Parallel Lines

Page 79

Fan Stub

Tools ⇒ Add Metalization ⇒ Fan Stub

Page 80

Lange Coupler

Tools ⇒ Add Metalization ⇒ Lange

Page 81

Chapter 6 Palette of Standard Geometries XGEOM

Rectangle ILLUSTRATION:

Width

Height

Y

X

PARAMETERS Parameter

Definition

Width

Width of the rectangle in length units.

Height

Height of the rectangle in length units.

71

Xgeom User’s Manual

Interdigital Capacitor ILLUSTRATION: Terminal Width

Number of Finger Pairs = 4 Finger Pair F inger S pacing F inger W id th

Y

X

End Gap Overlap

PARAMETERS Parameter

Definition

Finger Width

The thickness of the finger in the Y plane.

Finger Spacing

The space in between fingers in the Y plane.

Overlap

The length of the finger in the X plane as measured from the ending of the End Gap to the end of the finger. The Finger Length = Overlap + End Gap.

End Gap

The distance between the terminal and the finger. Note that the gap is metal on the side where a finger connects to the terminal and is open space between the finger and the opposite terminal.

72

Chapter 6 Palette of Standard Geometries XGEOM

Parameter

Definition

Number of Finger Number of finger pairs present in the capacitor. The example above has 4 finger Pairs pairs. Terminal Width

The width of the terminal which is measured from the outer edge of the capacitor to the beginning of the End Gap. There is a terminal on either side of the capacitor.

73

Xgeom User’s Manual

Donut ILLUSTRATION: Inner Radius

Ending Angle = 300°

Number of Sides = 72

Y Start Angle = 45° X Outer Radius PARAMETERS Parameter

Definition

Outer Radius

The distance from the center of the donut to the outside edge of the donut.

Inner Radius

The distance from the center of the donut to the inner edge of the donut. Setting this value to zero produces a donut with no center opening.

Start Angle

The angle, measured in the clockwise direction and referenced to the X plane which passes through the center of the donut, at which the polygon is started.

End Angle

The angle, measured in the clockwise direction and referenced to the X plane which passes through the center of the donut, at which the polygon is terminated.

74

Chapter 6 Palette of Standard Geometries

Number of Sides

XGEOM

Parameter

Definition This value controls the curvature of the inner and outer edges of the donut. The higher the value the smoother the edge.The value of 8 yields an octagon for a donut which starts at 0° and ends at 360°. The range for this value is from 3 to 360.

75

Xgeom User’s Manual

Meander ILLUSTRATION: Length First Length

Number of Legs = 5

Conductor Width Conductor Spacing

Conductor Width Y

X

Last Length

PARAMETERS Parameter Number of Legs

Definition Number of legs in the meander. The example above has 5 legs.

Conductor Width The width of the conductor in length units. Conductor Spacing

The space between legs in length units.

Length

Length of the leg in length units.

First Length

Length of the first (top) leg in length units.

Last Length

Length of the last (bottom) leg in length units.

76

Chapter 6 Palette of Standard Geometries XGEOM

Round Spiral ILLUSTRATION:

Conductor Spacing

Number of Turns = 3

Conductor Width

Y

X

Inner Radius

PARAMETERS Parameter

Definition

Number of Turns

The number of turns in the spiral. The example shown above has 3 turns.

Conductor Width

The width of the conductor in length units.

Conductor Spacing The spacing between turns in the conductor in length units. Inner Radius

The radius of the inside of the spiral in length units.

77

Xgeom User’s Manual

Rectangular Spiral ILLUSTRATION: Conductor Spacing

First Length

Number of Turns = 2

Conductor Width

Y

X Second Length

PARAMETERS Parameter

Definition

Number of Turns

The number of turns in the spiral. The example shown above has 3 turns.

Conductor Width

The width of the conductor in length units.

Conductor Spacing The spacing between turns in the conductor in length units. First Length

The length of the first segment in length units.

Second Length

The length of the second segment in length units.

78

Chapter 6 Palette of Standard Geometries XGEOM

Parallel Lines ILLUSTRATION:

Conductor Spacing Conductor Width

Number of Lines = 4 Y

X Length PARAMETERS Parameter

Definition

Number of Lines The number of lines in the parallel set. The example above has 4 lines. Length

The length of each line (conductor) in length units. This value is the same for all the lines.

Conductor Width The width of each line (conductor) in length units. This value is the same for all the lines. Conductor Spacing

The space between parallel lines (conductors) in length units. This spacing is the same between all pairs of conductors.

79

Xgeom User’s Manual

Fan Stub ILLUSTRATION:

Width

Angle

Y Length X

PARAMETERS Parameter

Definition

Width

The width of the base of the fan stub in length units.

Length

The length of the straight edges of the fan stub in length units.

Angle

Angle subtended by circular sector, in angle units.

80

Chapter 6 Palette of Standard Geometries XGEOM

Lange Coupler ILLUSTRATION: Feed Width Finger Length Finger Spacing Finger Width

Y

X

PARAMETERS Parameter

Definition

Number of Fingers

The number of fingers in the Lange coupler. The example above has six fingers. Only even numbers are allowed and you must specify a minimum of 4 fingers.

Finger Width

The width of the fingers in length units.

Finger Spacing

The spacing between fingers in length units.

Finger Length

The length of the finger in the Y plane as measured from the end of the finger to the feed line.

Feed Width

The width of the feed line in length units.

NOTE

The Lange coupler does not include wire bonds or airbridges. These must be added manually.

81

Xgeom User’s Manual

82

Chapter 7 Function Reference XGEOM

Chapter 7

Function Reference

This section contains a brief, yet detailed, description of xgeom’s functionality. It assumes you have read the previous tutorial and understand xgeom’s basic framework and purpose.

83

Xgeom User’s Manual

The Startup Menu The Startup Menu appears on the display when the xgeom program is invoked with no specified file or when all xgeom windows are closed without exiting the xgeom program. The startup windows for both UNIX and Windows, with no file specified are shown below. Actions possible from this menu are described below.

Startup menu for UNIX

Startup menu for Windows

The File Menu The File menu allows you to load a file from disk, open a new file, or exit the xgeom program.

File - New This performs the same function as File ⇒ New on the main menu in an xgeom window. See section "File - New" on page 86.

84

Chapter 7 Function Reference

This performs the same function as File ⇒ Open on the main menu in an xgeom window. See section "File - Open" on page 86.

File - Preferences This performs the same function as File ⇒ Preferences on the main menu in an xgeom window. See section "File - Preferences" on page 91.

File - Exit This performs the same function as File ⇒ Exit on the main menu in an xgeom window. See section "File - Exit" on page 92.

The Help Menu The Help menu provides you with basic operating information such as the xgeom version and licensing.

Help - License Info This performs the same function as Help ⇒ License Info on the main menu in an xgeom window. See section "Help - License Info" on page 148.

Help - System Info This performs the same function as Help ⇒ System Info on the main menu in an xgeom window. See section "Help - System Info" on page 148.

Help - About This performs the same function as Help ⇒ About on the main menu in an xgeom window. See section "Help - About" on page 149.

85

XGEOM

File - Open

Xgeom User’s Manual

The Main Menu The main menu, shown below, appears at the top of the xgeom window and allows you to access all of xgeom’s functions. The menu selections and their functions are explained, in detail, below.

The File Menu The File menu allows you to open a new file, load a file from disk, save a file to disk, close a file, print the file, view the file comments, and exit the xgeom program. It will also allow you to select environmental preferences such as an autosave file.

File - New Select File ⇒ New to open another xgeom window for a new file. The window will have a blank substrate to indicate that no circuit has been input yet. The ^N key is a shortcut for this command.To name the file, select the File ⇒ Save As. See below for a description of this option.

File - Open Select File ⇒ Open to load a file in from disk. Xgeom opens the Open File dialog box, which will resemble the dialog boxes shown below, in which you enter the directory and filename of the file you wish to open. The appearance of the

86

Chapter 7 Function Reference

The Open File dialog box for Windows

The Open File dialog box for UNIX

To open the “.geo” file you desire in one step, double-click the filename in the scroll list.

87

XGEOM

dialog box is system specific; therefore the functionality described will apply although some details may differ. You may also access this command by use of the shortcut key, ^O.

Xgeom User’s Manual If the file you want to open is not in the scroll list, change to another directory by double-clicking a directory name in the scroll list. For UNIX, you may also change to another directory by typing a different name in the text box for directory entry followed by a RETURN. Alternately, you may type in the complete path and file name in the File text box and click the OK button. The Windows version will also allow you to select a drive and directory from a drop list.

File - Close Select File ⇒ Close when you want to exit the particular file, but not necessarily, the xgeom program. If you have not saved your file to disk, a confirmation popup box will appear to allow you to save or discard your work. If the file you are closing is the last xgeom window open, it may also cause you to exit the xgeom program.

!

WARNING Closing the last open file may also cause you to exit the program.

xgeom

File - Save You may use File ⇒ Save to save your circuit to disk under the current name. We recommend that you use a name that ends in “.geo”. If you do not use a file extension, xgeom will automatically use an extension of “.geo”. You may also access this command by using the shortcut key, ^S.

File - Save As You may use File ⇒ Save As to save your circuit to disk under a different name. A dialog box appears, similar to the Open File dialog boxes shown in the figure on page 87, in which you can specify the desired filename. See the description of the Open File dialog boxes starting on page 86 for how to use the Save As dialog box.

88

Chapter 7 Function Reference XGEOM

File - Revert to Saved You may use File ⇒ Revert to Saved to close a circuit without saving it and then open the most recently saved version. This is useful if you make changes to your circuit and then decide you do not want to keep the changes. Xgeom opens the version of the document you last saved when you used the Save or Save As command. Xgeom does not open its own automatically saved version.

File - Analyze You select File ⇒ Analyze to launch the electromagnetic analysis program, em with the circuit being edited in xgeom input as the geometry file to be analyzed. Any changes to the circuit must be saved before invoking em. For more information about using em, see the Em User’s Manual.

File - Export Select File ⇒ Export when you want to create and save a bitmap of your current view, a DXF formatted file of your circuit or a GDS formatted file of your circuit. File - Export Picture Select File ⇒ Export Picture when you want to create and save a bitmap of your current view. A dialog box similar to the Save As File dialog box appears, in which you can specify the directory and filename to which you wish to export the file. File - Export DXF Select File ⇒ Export DXF when you want to create and save a DXF format copy of your circuit. A dialog box similar to the Save As File dialog box appears, in which you can specify the directory and filename to which you wish to export the file. If you have trouble reading the export file, you may wish to set the Simplified Write checkbox to “on” in the Preferences dialog box. This is accessed by selecting File ⇒ Preferences.

89

Xgeom User’s Manual File - Export GDS Select File ⇒ Export GDS when you want to create and save a GDS format copy of your circuit. A dialog box similar to the Save As File dialog box appears, in which you can specify the directory and filename to which you wish to export the file.

File - Print Setup You may use File ⇒ Print Setup to open the Print Setup dialog box, shown below. This allows you to select a printer, printer properties, paper size and source, and orientation. These options will be used when the print command is executed.

File - Print You select File ⇒ Print to print a copy of your circuit as specified in the Print Setup command discussed above. The Print command is not yet available on UNIX systems.

90

Chapter 7 Function Reference XGEOM

File - Comments You select File ⇒ Comments to view existing comments or to enter new comments. You accomplish this via the Comments dialog box detailed below.

Edit the text entry box with the comments you wish to include in the file. Em reads these comments and transfers them to the output file so that the resulting analysis can be easily traced to the “.geo” file from which it came.

File - Preferences You select File ⇒ Preferences to open the Preferences dialog box which allows you to set the autosave function and operating mode for xgeom or simplify the output of a DXF export.

91

Xgeom User’s Manual You may turn the autosave function on or off by clicking on the Auto Save On checkbox. If you set the function to on, you may also set the frequency of saves by editing the text entry box. The default is 5 minutes between saves. If the autosave function is set to on, xgeom creates a backup of your circuit at the time interval entered in the dialog box. The name of the backup file is the same name as your circuit file but has a “.bck” suffix. For example, if your circuit file is “filter.geo”, the backup file is named “filter.bck”. This file is deleted when exiting xgeom, but will remain in your directory if your system fails. You may choose to have xgeom operate in the version 4.0 style by clicking on the Remain in mode checkbox. If this option is on, xgeom will remain in a given mode until you explicitly exit by pressing the ESC key or selecting another mode. For example, when this option is not set, if you select Add Port, after you add one port, xgeom will automatically exit the Add Port mode. If this option is set, you will remain in Add Port mode until you press the ESC key or select another mode, such as Add Via.

TIP When not in Xgeom 4.0 Style, pressing the shift key while selecting a mode will leave you in the mode until you explicitly exit it.

You may click on the Simplified Write checkbox if you are having trouble reading a DXF file output by xgeom. Export the file with this checkbox set by selecting File ⇒ Export ⇒ DXF; this format should be easily read by most DXF users.

File - Exit Use File ⇒ Exit when you are finished, and wish to exit the xgeom program. If you have not saved your circuit(s) to disk, you are asked for a verification.

92

Chapter 7 Function Reference XGEOM

The Edit Menu The Edit Menu contains functions that help you make changes to your circuit. Most of the items in this menu are unavailable, appearing “faded”, until you select an object to edit. Chapter 2, “Editing Your Circuit”, covers the basic concepts of how to use the Edit menu. This section describes all of the Edit menu items.

Edit - Undo Edit ⇒ Undo undoes the last action taken on the geometry. A history is kept and it is possible to perform multiple undos. The menu item is unavailable if no changes have been made. You may also access this command by using the shortcut key, ^Z.

Edit - Redo If you perform an Edit ⇒ Undo and wish to restore the action, select Edit ⇒ Redo. A history is kept and it is possible to perform multiple redos. The menu item is unavailable if all undos have been reversed. You may also access this command by using the shortcut key, ^Y.

Edit - Cut Edit ⇒ Cut removes the selected objects from your circuit and places them into the clipboard. Any selected points, ports, and vias are also deleted with the polygon and copied into the clipboard. First, select the objects that you wish to put into the buffer. These objects may be on a single layer or multiple layers. Then choose Edit ⇒ Cut. The objects are removed from your circuit and placed into the buffer. The objects remain in the buffer until you replace them by doing another Cut, Copy or Duplicate.You may also access this command by using the shortcut key, ^X.

TIP One use for Edit

⇒ Cut is to move objects from one layer to another.

93

Xgeom User’s Manual Edit - Copy Edit ⇒ Copy copies the selected objects to the clipboard, from where they may be pasted in a different place in the circuit or even in another circuit file. The command leaves the selected objects in their original locations. First, select the objects that you wish to copy into the clipboard. The objects may be on multiple levels. Then choose Edit ⇒ Copy. The objects are copied into the clipboard and remain in the clipboard until you replace them by doing another Cut, Copy or Duplicate. Note that ports or vias attached to a polygon will be copied to the clipboard along with the polygon. You may also access this command by using the shortcut key, ^C.

TIP One use for Edit ⇒ Copy is to copy objects from one layer to another so that you have identical objects on both levels. This is sometimes done to simulate thick metal.

Edit - Paste Edit ⇒ Paste copies the objects that are in the clipboard to your circuit. The objects remain in the buffer until you replace them by doing a Cut, Copy or Duplicate.You may also access this command by using the shortcut key, ^V.

TIP Edit ⇒ Paste is typically used after using Edit ⇒ Copy to copy objects from one layer to another or from one xgeom file to another. Also, when a multilayer copy or cut is performed, the objects in the clipboard will retain their relative spacing.

Edit - Duplicate Edit ⇒ Duplicate is a quick way of combining Edit ⇒ Copy and Edit ⇒ Paste into one command. Edit ⇒ Duplicate copies the selected objects into the clipboard, and places a duplicate of the objects into your circuit. Use this option 94

Chapter 7 Function Reference

Edit - Delete Edit ⇒ Delete removes the selected objects and points from your circuit, but does not place them into the clipboard. The Delete and Backspace keys are shortcuts for this command.

TIP This command is useful when you wish to remove points or objects, but do not wish to overwrite the contents of the clipboard.

Edit - Clip You can use Edit ⇒ Clip if you have a complex circuit already entered into xgeom and you want to analyze just a portion of the circuit. To do this, choose Edit ⇒ Clip and lasso the area that you wish to analyze. All objects or portions of objects are removed that are not inside of your lassoed area. NOTE:

Edit ⇒ Clip is a multilayer operation regardless of the Single Layer Select setting. Any objects, or portions of objects, on another level inside the selected area will not be removed and any objects, or portions of objects, on another level outside the selected area will be removed. Sometimes the selected area will pass through a polygon, i.e., only part of a polygon is clipped. In this case, any ports or edge vias associated with that polygon are removed. This is true even if the port or edge via is inside of the selected area. To maintain the port or edge via, you must lasso the entire polygon associated with the port or edge via.

95

XGEOM

when you want to duplicate part of a circuit on the same layer. One example of this would be if you wanted to create a mirror image of a polygon. You would first duplicate your polygon, and then flip it. You may also access this command by using the shortcut key, ^E.

Xgeom User’s Manual You can use Edit ⇒ Clip if you suspect that there may be objects outside your substrate. You can select Edit ⇒ Clip and lasso the entire substrate. Any objects outside of the substrate are removed.

!

WARNING Save your file before using Edit ⇒ Clip. Clipping an area may delete large portions of your circuit, ports, and vias.

Edit - Select All Edit ⇒ Select All selects all objects on all levels. One use for Select ⇒ All is to change the metal type of all objects at once.You may also access this command by using the shortcut key, ^A.

Edit - Unselect Edit ⇒ Unselect will unselect all selected objects. This menu item is only available whenever objects are selected.

TIP To unselect all selected points and objects, click the left mouse button anywhere on your circuit that is not inside a polygon or close to a point.

Edit - Reselect Edit ⇒ Reselect reselects all points and objects that were unselected in the last unselection action. For example, if you have selected a group of objects and then unselect one object by use of a click of the mouse while pressing the control key, Reselect will once again add that object to those presently selected. This item is unavailable whenever the last action included a selection of an object or point. You may also access this command by using the shortcut key, ^T.

96

Chapter 7 Function Reference XGEOM

Edit - Copy Picture Edit ⇒ Copy Picture allows you to save a bitmap of your current view in the clipboard to allow you to import it to other programs under Windows95 and WindowsNT. NOTE:

This command is only valid on the Win95/98/NT versions of xgeom. To make a copy of your plot on a Unix system, see See “File - Export” on page 89.

Edit - Single Layer Select Edit ⇒ Single Layer Select allows you to select single-layer select mode or multilayer select mode. Clicking on Single Layer Select will toggle the mode. If this item is set, the mode is single-layer select. If it is turned off, the mode defaults to multilayer select. Single-layer select means that only objects on the current layer can be selected. Multilayer select mode allows you to select objects on any level, not just the present level. See “Selecting Objects for Editing,” page 24 for a complete description of selecting. If a layer is complete, and you wish to avoid inadvertent changes on it while in multilayer select mode, you may wish to lock it through use of the View ⇒ Metalization Levels command.

Edit - Select Filter Edit ⇒ Select Filter opens the Selection Filter dialog box that allows you to disable selection of object types. For instance, you want to remove all vias, but leave the rest of your circuit intact. You would click on the Only Objects Checked

97

Xgeom User’s Manual Below radio button in the dialog box and set only the Vias checkbox. Then perform a Edit ⇒ Select All, followed by an Edit ⇒ Delete. This will eliminate all vias in the circuit.

The Selection Filter dialog box set to select only Vias.

Any Object : Clicking on this radio button will make all items available for selection. Only Objects Checked Below: Clicking on this radio button will allow you to choose items for editing selection. To enable selection on a particular item, click on its checkbox so that the box is filled in. To disable selection, click on the checkbox so that it appears “open”.

98

Chapter 7 Function Reference XGEOM

The View Menu The View menu options allow you to change how information is presented on the display.

View − Zoom In You can zoom into a particular area of your circuit by using View ⇒ Zoom In. After selecting the option, the cursor, when in the main window, changes to indicate that xgeom is waiting for you to tell it where you want to zoom. Move the cursor to a corner of the region you want magnified and press the left mouse button. Holding it down, drag the cursor to the desired opposite corner of the desired area and release the mouse button. The specified region is expanded to fill the window. You may also use the Zoom In button on the tool bar, the shortcut key, Space Bar or the middle mouse button. See “Tool Bar,” page 152 for more details.

View − Zoom Out Use View ⇒ Zoom Out to shrink your circuit in your window. This option is quite useful when you accidentally zoom in a little too far, and need to zoom back out. Each time you select View ⇒ Zoom Out the circuit shrinks in size. As a shortcut, you can type ^W to zoom out. Repeatedly typing ^W will eventually shrink your circuit down to just a small dot on the screen. If this happens, use View ⇒ Full View to restore the display to full size. You may also use the Zoom Out button on the tool bar. See “Tool Bar,” page 152 for more details.

View − Previous View View ⇒ Previous View allows you to toggle between the present view and the previous view. You may also use the Previous View button on the tool bar or the shortcut key, ^L. See “Tool Bar,” page 152 for more details.

99

Xgeom User’s Manual View − Full View View ⇒ Full View resets the area displayed so that the entire substrate surface is visible on the available screen. ^F is the shortcut key for this command. You may also use the Full View button on the tool bar or the shortcut key. See “Tool Bar,” page 152 for more details.

View − New View View ⇒ New View opens a new xgeom window with the current circuit displayed. This command is useful if you need to observe the overall circuit while also working on a section of the circuit. You may also use the New View button on the tool bar. See “Tool Bar,” page 152 for more details.

View − Aspect Ratio Select View ⇒ Aspect Ratio to specify the aspect ratio of the displayed plot. The Aspect Ratio dialog box will appear on the display.

100

Chapter 7 Function Reference

TIP When editing long thin circuits, Aspect Ratio may be changed to allow you to view your entire circuit in detail.

View - Up One Level Select View ⇒ Up One Level to move the view in the xgeom window up one level of metalization in your circuit. This can also be accomplished by use of the shortcut key, ^U, the button on the tool bar, the Level menu on the tool bar or the Up Arrow key.

View - Down One Level Select View ⇒ Down One Level to move the view in the xgeom window down one level of metalization in your circuit. This can also be accomplished by use of the shortcut key, ^D, the button on the tool bar, the Level menu on the tool bar or the Down Arrow key.

View - Metalization Levels With xgeom, you can view multiple levels at the same time. Each level may be visible or invisible. It is also possible to lock a level to prevent multilayer selecting commands from effecting objects on it. If a layer is locked, editing commands will act on objects on that layer only if it is the present layer. To accomplish this, select View ⇒ Metalization Levels.

101

XGEOM

To obtain a wider image, enter a greater value in the X text entry box than in the Y text entry box. To obtain a taller image, you enter a greater value in the Y text entry box than in the X test entry box. A value of X=1.0, Y=1.0 displays the circuit in its true proportions on any CRT. This option may be needed when obtaining hardcopy from devices with aspect ratios different from your CRT.

Xgeom User’s Manual The Metalization Levels dialog box appears as shown below. The box indicates the present visibility value for all levels and their lock status. Initially, when xgeom starts up, all levels are visible and unlocked so that multilayer selecting commands will affect objects on all layers of a circuit.

Current Level drop list: This controls the level displayed in the xgeom window. This allows you to view different levels in your circuit without the need to exit the dialog box. Select a level from the drop list or edit the text entry box and click on the Apply button to update the display. Visible button: Clicking on this button makes the highlighted level visible. When a level is visible, and not the present level, polygons and other shapes can be seen on the screen as a dashed outline. Invisible button: Clicking on this button makes the highlighted level invisible. When a level is invisible, its objects are not shown except when it is the present layer viewed. Lock button: Clicking on this button locks the highlighted level. No objects on this layer will be selected on another level while in multilayer select mode. You can use this to prevent inadvertent changes on a completed level. It is important to remember, however, that the lock does not prevent editing if it is the present layer.

102

Chapter 7 Function Reference XGEOM

Unlock: Clicking on this button unlocks the highlighted level for multilayer selecting. All Visible button: Clicking on this button makes all levels visible. All Invisible button: Clicking on this button makes all levels invisible.

View - Object Visibility View ⇒ Object Visibility opens the Object Visibility dialog box which enables you to choose which objects you wish to display.

The Object Visibility dialog box with only ports selected.

All Objects radio button: Clicking on this button will make all objects visible. Only Objects Checked Below radio button: Clicking on this button will allow you to choose which objects to make visible. To enable display on a particular item click on its checkbox so that the checkbox is filled in. To disable display, click on the checkbox so that it appears “open”. The default upon startup is that all objects are visible.

View − Cell Fill View ⇒ Cell Fill toggles the display of the actual metalization, as subsectioned by em, on and off. ^M is the shortcut key for this command. When cell fill is “on”, metal is displayed using the pattern specified in the Parameters ⇒ Metal Types dialog box. All via posts are also displayed. The metalization is never displayed on the ground plane except for the via posts. The edges of the metalization are the actual edges that are analyzed by em. Selection of a finer grid in Parameters ⇒

103

Xgeom User’s Manual Box results in a better approximation to the edge. Other consequences are a better representation of the current distribution, more accurate results and a longer analysis time. Selection of the edge fill mode, by selecting Modify ⇒ Attributes, can also improve the fit of the actual metal to the desired edge. Finally, moving the outline edge slightly can also provide a better fit.

View − Cell Grid This option toggles the grid on and off. It is related to the cell size selected in the Box Parameters dialog box. The distance between two grid points is equal to the cell size. This grid is not to be confused with the snap distance set in the Snap Grid Setup dialog box. By default, the snap coordinates are set to coincide with the cell grid, however, the snap coordinates may be set to any other value and may not necessarily coincide with the cell grid. When View ⇒ Cell Grid is active, one dot is displayed at the corner of every possible cell. Metalization is allowed to cover an entire cell or none of the cell. If the Diagonal edge fill option is active then half a cell, cut diagonally, may also be filled/empty. To modify the cell size, see section “Parameters - Box,” page 129.

View - Tool Bar This option toggles the Tool bar on and off at the top of the xgeom menu.The default setting upon startup is on. For a detailed description of the Tool Bar, see “Tool Bar,” page 152.

View - Status Bar This option toggles the Status bar on and off at the bottom of the xgeom menu.The default setting upon startup is on. The status bar is shown in the figure on page 16.

104

Chapter 7 Function Reference XGEOM

View - Tool Box You select this option to open or close a tool box which allows for quick access to a number of xgeom’s functions. This option toggles between open and closed. If the option is on, the tool box is opened automatically when you open an xgeom file and closed automatically when the last file is closed. For a detailed description of the Tool Box, see “Tool Box,” page 155.

View - Measuring Tool You can use this feature to measure distances in your circuit. The measuring tool consists of an anchor and a readout box. Both will be displayed when you select View ⇒ Measuring Tool. The anchor, which is drawn as a small cross, is the reference point for taking measurements. The readout box provides the coordinates of the current anchor, whose default value is 0,0, the current cursor position, which is given as a delta from the anchor, and the length of the line from the anchor to cursor position.

Anchor Setup button Mouse button Readout box

Anchor Setup dialog box

If you wish to place the anchor in a different position, you may click on the Mouse button, then click on the new anchor position in your circuit. For a more precise positioning of the anchor, click on the Anchor (...) button in the Readout box. The Measuring Tool Setup box is opened. You have three options for entering an anchor position. • •

You may directly enter coordinates by editing the X and Y text entry boxes via keyboard. You may use the origin point (0,0) of the drawing by clicking on 105

Xgeom User’s Manual

•

the Drawing Origin button. You may use the center of the box by clicking on the Box Center button.

Any selection will cause the X and Y display to be updated with the current anchor position. Once the position is correct, click on the OK button to apply this value and close the setup box. The Anchor entry in the readout will now display the new anchor value and all subsequent measurements will be taken from this point. Turning the Allow anchor to be dragged checkbox on allows you to move the anchor directly, by dragging it with the mouse. However, when the option is on, the anchor has precedence over all other objects. If the anchor is at the same location as a point, that point can not be selected by clicking. You are also not permitted to add a new point in the same location as the anchor. If you need to perform an operation at the anchor location, be sure you first turn Allow anchor to be dragged off so that the action is permissible.

TIP When a polygon or set of points are selected, the dimensions of the bounding box of the selected object(s) appears in the cursor section of the status bar.

View - Redraw You select View ⇒ Redraw to redraw the xgeom display window.

106

Chapter 7 Function Reference XGEOM

The Tools Menu The Tools menu is used to access the tools which enable you to add or change objects in a circuit.

Tools - Pointer You select Tools ⇒ Pointer to place xgeom in pointer mode. This is the default operating mode of xgeom, in which objects can be selected and moved by use of the cursor and mouse. The mode is indicated by the word Pointer in the mode field of the status bar at the bottom of the xgeom window. You can also use the shortcut ESC key to place you in this mode.

Tools - Reshape You use Tools ⇒ Reshape to modify the shape of an existing object. For example, if you have a polygon which needs to be larger on one side, select Tools ⇒ Reshape. This places you in reshape mode, indicated by the cursor and a message in the status bar. Lasso the points of interest to select them. Selection is indicated by a black dot on the relevant points. Then drag one of the selected points to the desired location and release the mouse. As you are dragging the selected points a rubber band will appear indicating the new shape of the polygon. When the mouse is released, the polygon will be redrawn with a new shape to fit the new location of the selected points. This operation is illustrated in on the next page.

TIP To move an entire structure, without moving a port attached to the box wall, set the Multilayer select mode, then select Tools ⇒ Reshape. Lasso the entire circuit, with the exception of the port. Dragging one of the highlighted points will cause the entire circuit with the exception of the port to be moved.

107

Xgeom User’s Manual

Polygon before reshaping

Selected points on polygon in reshape mode.

Appearance of polygon after moving the selected points.

108

Chapter 7 Function Reference XGEOM

Tools - Add Points to Polygon The Add Points to Polygon menu option may be used when you want to add additional points to an existing polygon. To add points to a polygon, select Tools ⇒ Add Points to Polygon and click the mouse with the cursor on the line segment to which you want to add points. The new point appears highlighted. Now drag the point to the desired position. A “rubber band” appears connecting the highlighted new point to the existing endpoints on either side. When you release the mouse, the polygon is redrawn with the new point. To add more points, repeat the procedure above. When adding points is completed, exit the add points mode by selecting Tools ⇒ Pointer or pressing the ESC key. The measuring tool can be used to determine the correct position while adding points. Also, the Delete or Backspace key may be used while adding points to delete a point while it is still selected.

Tools - Add Metalization Use Tools ⇒ Add Metalization when you wish to add metal to your circuit. Tools ⇒ Add Metalization ⇒ Draw Rectangle This menu item provides you with a quick way to create rectangular polygons. To add a rectangle, select Tools ⇒ Add Metalization ⇒ Draw Rectangle or use the shortcut key ^R. There is also a button in the Tool box for this task. Now place the cursor where you want one corner of the rectangle to be. Next, drag the cursor to form an outline of the desired rectangle. When the rectangle is the size and dimensions you want, release the mouse button. The rectangle will be drawn and xgeom will return to pointer mode. If you wish to add more than one rectangle at a time, press the shift key while making the menu selection. This will leave you in Add Metalization ⇒ Draw Rectangle until you explicitly exit this mode.

109

Xgeom User’s Manual Tools ⇒ Add Metalization ⇒ Draw Polygon To add a polygon, select Tools ⇒ Add Metalization ⇒ Draw Polygon or use the shortcut key ^P. There is also a button in the Tool box for this task. Now click where you want the first point of the polygon to appear. As you move the cursor a moving rubber band appears stretching from the first point to the cursor location. Click on the next point of the polygon, and a fixed line appears between the first two points and the rubber band now stretches from the last input point. The rubber band will continue to follow the cursor from the last point input until the polygon is complete. This can be accomplished by either clicking again on top of the first point entered or double-clicking on the last point to be entered. Once the polygon is complete, xgeom returns to pointer mode, unless the mode was shift-selected. If so, you can continue to add polygons. Tools ⇒ Add Metalization ⇒ Rectangle To add a rectangle from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ Rectangle. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Rectangle,” page 71. Tools ⇒ Add Metalization ⇒ InterDigCap To add an interdigital capacitor from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ InterDigCap. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Interdigital Capacitor,” page 72. Tools ⇒ Add Metalization ⇒ Donut To add a donut from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ Donut. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Donut,” page 74.

110

Chapter 7 Function Reference XGEOM

Tools ⇒ Add Metalization ⇒ Meander To add a meander from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ Meander. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Meander,” page 76. Tools ⇒ Add Metalization ⇒ Round Spiral To add a round spiral from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ Round Spiral. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Round Spiral,” page 77. Tools ⇒ Add Metalization ⇒ Rectangular Spiral To add a rectangular spiral from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ Rectangular Spiral. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Rectangular Spiral,” page 78. Tools ⇒ Add Metalization ⇒ Parallel Lines To add a set of parallel lines from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ Parallel Lines. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Parallel Lines,” page 79. Tools ⇒ Add Metalization ⇒ Fan Stub To add a fan stub from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ Fan Stub. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Fan Stub,” page 80.

111

Xgeom User’s Manual Tools ⇒ Add Metalization ⇒ Lange To add a Lange coupler from the palette of standard geometries, select Tools ⇒ Add Metalization ⇒ Lange. The Geometry Parameters dialog box appears on your display. For details about this geometry’s parameters, see “Lange Coupler,” page 81.

Tools - Add Dielectric Brick Use Tools ⇒ Add Dielectric Brick when you wish to add dielectric bricks to your circuit. Tools ⇒ Add Dielectric Brick ⇒ Draw Rectangle This menu item provides you with a quick way to create rectangular polygons. To add a rectangle, select Tools ⇒ Add Dielectric Brick ⇒ Draw Rectangle or use the button in the Tool box for this task. Now place the cursor where you want one corner of the rectangle to be. Next, drag the cursor to form an outline of the desired rectangle. When the rectangle is the size and dimensions you want, release the mouse button. The rectangle will be drawn and xgeom will return to pointer mode. If you wish to add more than one rectangle at a time, press the shift key while making the menu selection. This will leave you in Add Dielectric Brick ⇒ Draw Rectangle until you explicitly exit this mode. Tools ⇒ Add Dielectric Brick ⇒ Draw Polygon To add a polygon, select Tools ⇒ Add Dielectric ⇒ Draw Polygon or use the button in the Tool box for this task. Now click where you want the first point of the polygon to appear. As you move the cursor a moving rubber band appears stretching from the first point to the cursor location. Click on the next point of the polygon, and a fixed line appears between the first two points and the rubber band now stretches from the last input point. The rubber band will continue to follow the cursor from the last point input until the polygon is complete. This can be accomplished by either clicking again on top of the first point entered or doubleclicking on the last point to be entered. Once the polygon is complete, xgeom returns to pointer mode, unless the mode was shift-selected. If so, you can continue to add polygons.

112

Chapter 7 Function Reference

Use Tools ⇒ Add Port to add ports to your circuit. See Chapter 3, “Using Ports” for a detailed description of ports. Select Tools ⇒ Add Port to place xgeom in Add Port mode, as indicated by a change in cursor and the messages in the status bar at the bottom of the xgeom window. Then click on the center of the desired edge of the polygon to which you wish to add a port. Ports are nearly always placed on the edge of the substrate so that the port has access to ground. Note that the edge must be on the substrate edge for the port to have access to ground. If this is not the case, em issues an error message and terminates. NOTE:

If you wish to add an auto-grounded port, you must first add the port as described above, then go to the Port Attributes dialog box to change its type to auto-grounded. For details of how to accomplish this, see section "Modify - Attributes" on page 118.

If you wish to add multiple ports, press the shift key while making your menu selection. To delete a port, return to pointer mode by selecting Tools ⇒ Pointer or pressing the ESC key. Click on the port to select it, then use Edit ⇒ Cut to delete it.

Tools - Add Via Use Tools ⇒ Add Via to add edge vias. See Chapter 4, “Using Vias” for a detailed description of vias. To add an edge via, select Tools ⇒Add Via, then click on the polygon edge that you want the via added to. A dark dashed line appears showing where the edge via is. With the metalization turned on, the via subsections are displayed. Each via post has an “up” via symbol which is an “up” triangle, indicating that the via connects this level to the next level above. Move up one level. The subsection vias are shown on this level with a “down” via symbol, a “down” via triangle. Via posts that are too small to be seen individually have a single via symbol representing multiple via posts.

113

XGEOM

Tools - Add Port

Xgeom User’s Manual If you wish to add multiple vias, press the shift key while making your menu selection. To delete a via, return to pointer mode, by selecting Tools ⇒ Pointer or pressing the ESC key. Make sure that you are on the level where the edge via starts, and emanates upward, and not on the level above where the via ends. Click on the triangle to select it, then use Edit ⇒ Delete to delete it. See “Creating an Airbridge: An Example,” page 59 for more details on how to get the via metalization exactly where you want it to go. The length of the via must be small with respect to the wavelength. If not true, subdivide the dielectric layer into multiple layers and put vias on each of these layers.

Tools - Ortho Selecting Tools ⇒ Ortho places xgeom in the orthogonal mode. This will affect you in two ways. First, if you add a polygon while the ortho feature is active, xgeom will only allow horizontal or vertical lines in the object as shown in the figure below.

Ortho feature off

Ortho feature on

TIP Holding down the shift key while adding, moving, or editing a polygon will place you in orthogonal mode temporarily.

114

Chapter 7 Function Reference

Horizontal move

Notice, that in ortho mode, if a move is made in the horizontal plane, then the vertical position is fixed. If the move is in the vertical plane, then the horizontal position is fixed

Vertical move

115

XGEOM

Second, when moving a selected object or point, you may only move in the horizontal plane or vertical plane at any given time. In other words, Tools ⇒ Ortho ensures that you can only change position in one planar coordinate, either x or y, at a time. See the example shown below.

Xgeom User’s Manual Tools - Snap Setup Selecting Tools ⇒ Snap Setup opens the Snap Grid Setup dialog box, shown below. This allows you to specify a new snap distance. All captured coordinates and all ruler measurements are snapped to this grid. For example, with a snap of 0.1 mils, a captured point at 10.05 mils is impossible.

Snap Grid radio buttons

Divisions field Custom text boxes

NOTE:

The snap distance of your circuit affects how the polygons are entered. It does not change the size of your subsections.

You select one of the following four buttons, pictured above, to select your snap type: Cell Size: Sets the snap coordinates to match the cell grid. If the cell size is changed and this option is in effect, the snap coordinates change to match the new cell grid. The current cell size is displayed following the button. Divide Cell: Sets the snap coordinates to match a subdivision of cell size. Enter the number of divisions per cell in the Divisions field. Custom: Any arbitrary snap grid can be specified by entering the desired values in the X and Y Custom text boxes. Use this option if, for example, you want to capture a polygon with all coordinates snapped to the nearest micron. No Snap: Alternatively, one can select the No Snap button, in which case captured coordinates are exactly at the cursor position. This actually means all coordinates snap to the nearest pixel.

116

Chapter 7 Function Reference

Y - No Snap: Will turn off the snap in the y-direction. This option allows you to snap in the x-direction only.

117

XGEOM

X - No Snap : Will turn off the snap in the x-direction. This option allows you to snap in the y-direction only.

Xgeom User’s Manual

The Modify Menu The Modify menu allows you to make changes to existing objects in your circuit. NOTE:

Please note that you must first select a polygon(s) and/or port(s) to access the following modify menu items.

Modify - Attributes To change the parameters associated with a polygon or a port, select the object, then select Modify ⇒ Attributes. If the object selected is a metal polygon, the Metalization Attributes dialog box is opened, as shown in the figure on page 119. This dialog box allows you to set the metal type, fill type and subsection size. If the object selected is a brick polygon, the Brick Attributes dialog box is opened, shown in the figure on page 121. This dialog box allows you to set the brick type and subsection size. If the object selected is a port, the Port Attributes dialog box is opened, as shown in the figure on page 122. This dialog box allows you to set the port type, number, impedance values, and if the port is an autogrounded port, set the reference plane or calibration length. If both polygons and ports are selected, the Metalization Attributes box will be opened, followed by the Brick Attributes box and then the Port Attributes dialog box. These dialog boxes are described below. NOTE:

118

If multiple polygons or ports are selected, then changes made in the Attributes dialog boxes will affect the parameters of all selected polygons or ports.

Chapter 7 Function Reference XGEOM

The Metalization Attributes dialog box This dialog box shows the current metal type, fill type and subsection size associated with the selected polygon(s).

Metalization Attributes dialog box.

Metal field: : Choose a metal type from the drop list in the Metal field. Changing the metalization changes only the metalization loss. The polygon is still on the same physical level; i.e., it is still between the same two layers of dielectric. When xgeom starts, it initially only has one metal type, “Lossless”. If you have not set up additional metal types in the Parameters ⇒ Metal Types menu, then the pull-down menu only presents the one choice of “lossless”. To add new metal types, see section “Parameters - Metal types,” page 138. Fill Type field: The “fill type” of a polygon is defined as the type of subsection used by em at the edges of a polygon. Choose a fill type from the drop list in the Fill Type field. The four options in the drop list are defined as follows: Staircase

Provides the best rectangular fit to the metalization edge. Staircase subsections are used in the middle of a polygon for all choices. Xgeom defaults to “staircase” edge fill for all new metal structures.

Diagonal

Allows triangular as well as rectangular subsections. This provides a better fit to diagonal edges but requires more analysis time. An example appears on page 120. For a more detailed explanation see section 119

Xgeom User’s Manual “A Coupled Open-Miter with Diagonal Fill,” page 163 in the Em User’s Manual. Corner

Provides an improved representation of the current at gap corners. For a more detailed explanation see “A Coupled Open-Miter with Diagonal Fill,” page 163 in the Em User’s Manual.

Both

Includes both Corner and Diagonal fill.

Two identical polygons, on the lef subsectioned with Staircase fill, and on the right subsectioned with Diagonal fill.

Subsection Size: Edit the X Min, X Max, Y Min and Y Max text entry boxes to set the subsectioning for the polygon(s). First time users should leave the numbers set to the default values (X Min 1, Y Min 1, X Max 100, Y Max 100). Refer to Chapter 3, “Subsectioning” in the Em User’s Manual for a description of these parameters.

120

Chapter 7 Function Reference XGEOM

The Brick Attributes dialog box The Brick Attributes dialog box, shown below, shows the current brick material and subsection size associated with the selected polygon(s).

The Brick Attributes dialog box.

Brick field: Choose a brick type from the drop list in the Brick field. Changing the brick changes only the brick material. The brick is still within the same physical layer. When xgeom starts, it initially only has one brick material, “Air”. If you have not set up additional brick material choices in the Parameters ⇒ Brick Materials menu, then the pull-down menu only presents one choice of “air”. To add new brick material choices, see section “Parameters - Brick Materials,” page 136. Subsection Size: Edit the X Min, X Max, Y Min and Y Max text entry boxes to set the subsectioning for the polygon(s). First time users should leave the numbers set to the default values (X Min 1, Y Min 1, X Max 100, Y Max 100). Refer to Chapter 3, “Subsectioning” in the Em User’s Manual for a description of these parameters.

121

Xgeom User’s Manual The Port Attributes dialog box The Port Attributes dialog box shows the number, type and impedance associated

with the selected port(s). It also allows you to set reference planes or calibration lengths for auto-grounded ports. The Port Attributes dialog box.

Port Number: Renumber the port by entering a new value in the Port Number text box. Type: Select the port type from a drop list consisting of Standard and Autogrounded. Impedance: Enter the four impedance values in the text entry boxes: Resistance (in ohms), Reactance (in ohms), Inductance (in nanohenrys), and Capacitance (in picofarads). Autoground Port Data: This section of the dialog box allows you to set a reference plane and calibration length for an autogrounded port by entering values in the text entry boxes or by using the mouse. For a more detailed discussion of ports and explanations for the options available in this dialog box, see Chapter 3, “Using Ports”.

122

Chapter 7 Function Reference

Selecting Modify ⇒ Flip opens the Flip dialog box, shown below, which allows you to flip the selected polygon(s) in the following manner: •

Left-right flip: Flips about a vertical axis so that its left side is on the right.

•

Up-down flip: Flips about a horizontal axis so that its top is on the bottom.

The axis runs through a pivot point which you also set in the dialog box. You can perform a flip on a single polygon or multiple polygons; the flip operation will be done to all the objects selected when Modify ⇒ Flip is performed.

Left-Right button: Click on the left-right button to flip the selected polygon(s) about a vertical axis so that its left is on the right and vice versa. Up-Down button: Click on the up-down button to flip the selected polygon(s) about a horizontal axis so that its top is on the bottom and vice versa. The following choices are available for pivot points: Selection Center: The flip axis runs through the center of the area enclosing all selected objects. Individual Centers:If there are multiple polygons selected, each polygon will flip about its own center.

123

XGEOM

Modify - Flip

Xgeom User’s Manual Pivot Point:You can set a custom pivot point by clicking on the Set Pivot button in the Flip dialog box, and then clicking in the xgeom window to set the pivot point. Once the pivot point is set, the Pivot Point radio button will be available and selected. The polygon will then flip about an axis through this custom point. Meas. Tool Position:This option is only available if the measuring tool is active when Modify ⇒ Flip is selected. The polygon will flip with the measuring tool anchor position as a pivot point. Box Center:This selection uses the box center as the pivot point for the flip.

Modify - Rotate Modify ⇒ Rotate rotates the selected polygon(s) clockwise or counterclockwise for a specified angle about a specified pivot point. Selecting Modify ⇒ Rotate opens the Rotate dialog box, shown below, which allows you to rotate the selected polygon(s) about a pivot point, clockwise or counterclockwise, which you also set in the dialog box. You can perform a rotate on a single polygon or multiple polygons; the rotate operation will be done to all the objects selected when Modify ⇒ Rotate is performed.

Angle: Enter the desired angle of rotation in this field. To determine angle values, refer to the figure on page 157.

124

Chapter 7 Function Reference XGEOM

Counterclockwise button: Click this button to rotate the polygon(s) in the counterclockwise direction. Clockwise button: Click this button to rotate the polygon(s) in the clockwise direction. Set Pivot button: Click on this button to set a pivot point using the mouse. This will enable the Pivot Point selection in the Pivot Point field. The following choices are available for pivot points: Selection Center: Selected objects are rotated about the center of the area enclosing all the selected objects. Individual Centers: If there are multiple polygons selected, each polygon will rotate about its own center. Pivot Point: The polygon will rotate about the custom point set after clicking the Set Pivot button. Meas. Tool Position: This option is only available if the measuring tool is active when Modify ⇒ Rotate is selected. The polygon will rotate with the measuring tool origin position as a pivot point. Box Center:This selection uses the box center as the pivot point for the rotation.

125

Xgeom User’s Manual Modify - Resize Modify ⇒ Resize allows you to resize an object or multiple objects using a scaling factor. When you select the Resize menu item, the Resize dialog box is opened. Enter the desired scaling factor in the Percentage field. Percentages over the value of 100 will enlarge the object; below 100 will shrink the object.

Original polygon

Reduced polygon (%80)

Multiple objects will resize from the selection center, keeping relative, not absolute positions.

126

Chapter 7 Function Reference XGEOM

Modify - Snap to Modify ⇒ Snap to opens the Snap Objects dialog box which allows you to “snap” or align the points of the polygon(s) in one of three useful ways:

Cells: All points on the selected polygons are aligned to a cell edge. You can choose to align points in both the X and Y direction, only the X direction, or only the Y direction. • •

•

X and Y - Both the X and Y coordinates of each point in the selected polygon(s) will correspond to a cell edge. X only - The X coordinate of each point in the selected polygon(s) will correspond to a cell edge. The Y coordinate of the points does not change. Y only - The Y coordinate of each point in the selected polygon(s) will correspond to a cell edge. The X coordinate of the points does not change.

Current Snap Grid: All points on the selected polygons are aligned to the Snap Grid as defined in “Tools - Snap Setup,” page 116. If the current Snap Grid is set to the cell size, then this option is not enabled, since in that case, the Cells option above is the equivalent of this selection. You can choose to align points in both the X and Y direction, only the X direction, or only the Y direction. •

X and Y - Both the X and Y coordinates of each point in the selected polygon(s) will correspond to a grid edge. 127

Xgeom User’s Manual •

•

X only - The X coordinate of each point in the selected polygon(s) will correspond to a grid edge. The Y coordinate of the points does not change. Y only - The Y coordinate of each point in the selected polygon(s) will correspond a a grid edge. The X coordinate of the points does not change.

Box Walls: This option is enabled only if you are in Reshape mode and have selected points on a polygon. You then select a box wall: Left, Right, Top or Bottom. All selected points will be moved to correspond to the Box wall. This can be used to ensure that your metalization extends to the box wall.

Modify - Add Vias to All Modify ⇒ Add Vias to All will make the entire selected polygon(s) a via, by adding vias to all edges of the selected polygon(s).

Modify - Convert to Metal Selecting Modify ⇒ Convert to Metal will convert selected polygon(s) to metal.

Modify - Convert to Brick Selecting Modify ⇒ Convert to Brick will convert selected polygon(s) to dielectric brick.

TIP Use the two commands above if you enter a polygon incorrectly.

128

Chapter 7 Function Reference XGEOM

The Parameters Menu The Parameters menu allows specification of the parameters of the box, length units, substrate and dielectric layers, metalization, ports, reference planes and parallel subsections.

Parameters − Box You select Parameters ⇒ Box to open the Box Parameters dialog box in order to display the parameters of the circuit box, as well as to allow for modification of those parameter values. The default box size for a new file is 160 mils X 160 mils with a cell size of 10 mils X 10 mils.

Metal types

Sizes

Max. Subsection Size text box

Lock check boxes

Symmetry check box

Box Parameters - Sizes The entries in this section define the area of the box, and include Cell Size, Box Size and Num. Cells. Changing one of the size factors will cause the other factors to be updated according to the relationship below. The dialog box allows you to “lock” any of these values while entering changes. Note that due to this relationship, if you lock in two of the size parameters, you have de facto locked in the third. These three values are related in the following manner: Num. Cellsx * Cell Sizex = Box Sizex 129

Xgeom User’s Manual Num. Cellsy * Cell Sizey = Box Sizey where Num. Cells must be an integer. Cell Size:This entry row defines the size of a single cell of the box area. You enter an X dimension (width) and a Y dimension (height) through the keyboard. If you wish the cell size to remain constant when changing either of the other size factors, click on the lock check box at the end of the entry row. Selecting cell size is important. The em analysis automatically subsections the circuit based on the cell size. Box Size:This entry row defines the size of the box area. You enter an X dimension (width) and a Y dimension (height) through the keyboard. Alternatively, if you wish to set the box size and location using the mouse, click on the Set Box Size with Mouse button. Then click in the xgeom window on the location of one corner of the new box, then drag to the opposite corner. If you wish the box size to remain constant when changing either of the other size factors, click on the lock check box at the end of the entry row. Num. Cells:This entry row defines the number of cells in the box area. You enter an X dimension (width) and a Y dimension (height) through the keyboard. If you wish the number of cells to remain constant when changing either of the other size factors, click on the lock check box at the end of the entry row. Note that if the cell size is not locked, xgeom picks the next smallest allowed cell size which allows for an integer number of cells to fit in the box area specified by the Box Size and Num. Cells. Box Parameters - Max. Subsection Size Max. Subsection Size: Em uses a variable subsection size. Small subsections are used where needed, such as around corners, and larger subsections are used elsewhere. This reduces the size of the matrix which must be inverted, often providing a dramatic increase in the speed of an analysis. In no case are the subsections smaller than a single cell whose dimensions are specified above.

130

Chapter 7 Function Reference

See section “Modify - Attributes,” page 118 for information on how to specify the maximum and minimum subsection size on a polygon by polygon basis. Box Parameters -Symmetry Symmetry check box: You specify whether symmetry is turned on or off by clicking on the Symmetry check box. Most circuits should have symmetry off. If a circuit is symmetric about the center line parallel to the X axis and has ports only on this center line on any layer, then activating symmetry results in a faster analysis. When symmetry is turned on, everything below the line of symmetry is ignored, and all metal above the line of symmetry is “reflected” about the symmetry line. Therefore, you may enter in just half of your circuit, making sure that you enter in the half that is above the line of symmetry. It is recommended that first time users capture the entire circuit when using symmetry, not just half the circuit. This way, toggling symmetry on and off will affect only the speed of the analysis, not the final result.

!

WARNING Do not place ports below the line of symmetry.

Box Parameters - Top Metal, Bottom Metal This option allows you to select the metal type for the box top and box bottom. You can choose either a predefined metal type or user defined metal type. The are three predefined metal types: •

Lossless, models a perfect conductor 131

XGEOM

This field allows you to limit the size of the subsection, generated by em, in terms of subsections per wavelength, where the wavelength is approximated at the beginning of the analysis. The highest analysis frequency is used to calculate the wavelength. The default of 20 is fine for most work and means that the maximum size of a subsection is 18 degrees at the highest frequency of analysis. Increasing this number decreases the maximum subsection size until the limit of 1 subsection = 1 cell is reached.

Xgeom User’s Manual • •

WG Load, models a perfect matched waveguide load. Free Space, which removes the top or bottom cover.

WG Load is useful for modeling infinite arrays in em. The top or bottom cover will then be terminated with a perfectly matched waveguide load. Note this is not the same as an open environment and the method for modeling radiating structures in an open environment has not changed. To model an open environment select Free Space, which sets the impedance of the cover to 377 ohms/sq, the impedance of free space. You should also note that the sidewalls are always modeled as perfect conductors. Top Metal:Use the drop list to select the metal type for the top of the box. Choose a metal type from the list which includes the default types of WG Load, Free Space, and Lossless in addition to any user defined metal types. For details on how to define a user metal type, and default metal types, see section "Parameters Metal types" on page 138. Bottom Metal: Use the drop list to select the metal type for the bottom of the box. The metal types are the same as listed above for the box top.

Parameters - Units Parameters ⇒Units allows you to select the default measurement units for your circuit by opening the Units dialog box.

132

Chapter 7 Function Reference

Size Field: If you wish to define nonstandard units, enter the conversion factor to meters in this field. Applying New Units radio buttons: When changing units, you can choose to keep your circuit the same physical size or allow it to change size. If you choose to keep it the same physical size, parameters will be updated to reflect that. For instance if a cell is 10 mils by 10 mils and you change the unit to inches, it now measures 0.01 inches by 0.01 inches. If you choose to allow the circuit to change size, parameters will retain the same amount of the new measurement, i.e., a 10 mil by 10 mil cell will now measure 10 inches by 10 inches. Click the appropriate checkbox before applying the units.

!

WARNING If you change the size, be sure to assign a new name to avoid confusion. For instance, you would not want to define mils as equal to 3 meters.

Parameters - Dielectric Layers Parameters ⇒ Dielectric Layers opens the Dielectric Layers dialog box which allows specification of the dielectric layers in the box including adding or deleting dielectric layers. For your convenience, the xgeom “level” number appears on the

133

XGEOM

Name field: Select a unit of measure from the drop list in the name field. The size field will be updated to reflect the unit chosen. If you wish to define nonstandard units, type the name in this field.

Xgeom User’s Manual left providing you with an approximate “side view” of your circuit. A “level” is defined as the intersection of any two layers and is where your circuit metal is placed. The metal is attached to the bottom of the upper dielectric layer.

Dielectric layer text boxes: Each layer has an entry row in the dialog box with entries available for the parameters of each dielectric layer: • • • • • • •

thickness (Thickness) relative dielectric constant (Erel) dielectric loss tangent (Dielectric Loss Tan) dielectric conductivity (Diel Cond) relative magnetic permeability (Mrel) magnetic loss tangent (Magnetic Loss Tan) Z-Partitioning (Z-Parts. for bricks)

You select the entry to be changed by clicking on its field. Not all parameters can be displayed simultaneously in the dialog box. Use the scroll bars to access the additional parameters.When you click on a text entry box in a dielectric layer’s row of parameters, it becomes the presently selected layer. The dielectric constant and loss of a dielectric layer are defined as follows:

134

Chapter 7 Function Reference Erel: The ratio (ε’/εo’), where ε’ is the real part of the permittivity of the dielectric layer material, and εo’ is the permittivity of free space. The ratio is dimensionless.

•

Loss Tan: The ratio (ε’’/ε’), where ε = ε’ - jε’’, and ε is the complex permittivity of the dielectric layer material. The ratio is dimensionless.

•

Diel. Cond: The quantity σ, where σ is the dielectric conductivity in siemens per meter.

Em uses the above parameters to calculate the total effective tanδ for the dielectric material as follows:

Diel Cond )tan δ = ( Loss Tan ) + (---------------------------ω ( Erel )ε o'

Here, ω is the radian frequency (ω = 2πf, where f is frequency in hertz). Note that tanδ has both a frequency-dependent term and a frequency-independent term. The above equation for tanδ can also be expressed in terms of conductivities as follows:

( Total Effective Cond ) = ( Loss Tan )ω ( Erel )ε o' + ( Diel Cond )

Both equations are equivalent. Each describes how em uses the input dielectric parameters to compute loss in the dielectric layer. Above button: To add a dielectric layer above an existing layer, click on an entry in the existing layer’s row, then click on the Above button. An additional layer will be added above the selected layer. This will be indicated by the appearance of another row of parameter text boxes and a level number to the left of the row.

135

XGEOM

•

Xgeom User’s Manual Below button: To add a dielectric layer below an existing layer, click on an entry in the existing layer’s row, then click on the Below button. An additional layer will be added below the selected layer. This will be indicated by the appearance of another row of parameter text boxes and a level number to the left of the row. Delete button: To delete a dielectric layer, click on an entry in its row, then click on the Delete button. If you attempt to delete a layer on which metalization is present, xgeom will prompt you for confirmation before deleting the dielectric layer, since any metalization present on the layer will also be deleted.

Parameters - Brick Materials You select Parameters ⇒ Brick Materials to open the Brick Materials dialog box which allows you to define the characteristics of brick materials for use in your circuit. The default brick material is Air; this is defined for all xgeom files. All brick materials may be used on all levels and with different polygons on the same level. A brick material specifies the dielectric constant, loss tangent and bulk conductivity used by em, and whether the dielectric is isotropic or anisotropic. Since the dielectric parameters are unrelated to the level on which a polygon is located, changing the brick material of a polygon does not change the physical location of the polygon. It changes only the dielectric parameters.

136

Chapter 7 Function Reference XGEOM

Brick Materials Default for Add Bricks: Choose a brick material from the drop list of defined brick materials. This brick material will be applied to any new dielectric bricks as you add them to the circuit. Default button:This button is a shortcut for selecting a brick material as the default, rather than using the drop list discussed above. Click on the material to be used as the default, then click on this button. New button: Clicking on the New button allows you to define a new brick material for use in your circuit. When you click the New button, a new row appears. You then need to enter parameters to define the brick material. •

Name: To customize the name, edit the text entry box under the name column.

•

Ani: If you wish the brick material to be anisotropic, indicating that its parameters are different in each dimension, click on this checkbox until it is filled in. You can then enter the parameters of the brick material by editing the entries in the row.

If the brick material is isotropic, you only need to fill in the parameters in the X dimension, which are all displayed in the dialog box as follows: • • •

Erel Dielectric Loss Tan Diel. Cond. (S/m)

If the brick material is anisotropic, you will need to define the parameters in all three dimensions, X, Y, and Z. To access all the parameters, use the scroll bars. The parameters are as follows: • • •

Erelx,y,z Dielectric Loss Tanx,y,z Diel. Cond.x,y,z

Duplicate button:Clicking on the Duplicate button will add a new brick material whose parameters are identical to the existing selected brick material.

137

Xgeom User’s Manual Delete button: To delete a brick material, click in its row, then click on the Delete button. If any polygons currently refer to this brick material, the polygon is changed to air. Fill Pattern: This dialog box allows you to choose a fill pattern for the brick material so that types are easily distinguishable while editing your circuit. Click on the arrow buttons in the Pattern column, to advance through the choices available.

Parameters - Metal types You select Parameters ⇒ Metal Types to open the Metal Types dialog box which allows you to define the characteristics of metal types for use in your circuit. The default metal type is Lossless; this is defined for all xgeom files. All metal types may be used on all levels and multiple metal types may be used on the same level. These metal types may also be used for the box top and bottom. A metal type specifies the metalization loss used by em. Since the metalization loss is unrelated to the level on which a polygon is located, changing the metal type of a polygon does not change the physical location of the polygon. It changes only the metalization loss. Metalization loss is discussed in more detail following the description of the dialog box.

138

Chapter 7 Function Reference XGEOM Default for Add Metalization: Choose a metal type from the drop list of defined metal types. This metal type will be applied to any metalization added to the circuit. Default button:This button is a shortcut for selecting a metal type as the default, rather than using the drop list discussed above. Click on the metal type to be used as the default, then click on this button. New button: The New button allows you to define a new metal type for use in your circuit. When you click the New button, a new row appears. To customize the name, edit the text entry box under the name column. You can then enter the parameters of the metal type by editing the entries in the row: Metalization Resistivity •

DC Resistance (RDC)

•

Skin effect (RRF)

Metalization Reactance •

DC Reactance (XDC)

139

Xgeom User’s Manual •

Surface Inductance (Ls)

Duplicate button: Clicking on the Duplicate button will add a new metal type whose parameters are identical to the existing selected metal type. You select a metal type to duplicate by clicking on a text entry box in its row of parameters. Delete button: To delete a metal type, click in its row, then click on the Delete button. If any polygons currently refer to this metalization, the polygon is changed to lossless. Fill Pattern: This dialog box also allows you to chose a fill pattern for the metal type, so that types are easily distinguishable while editing your circuit. Click on the arrow buttons in the Pattern column, to advance through the choices available. Loss Definitions There are two kinds of metalization loss specified. The first is DC, low frequency loss. In this case, the skin depth is much thicker than the metalization. This is also known as the electrically thin case. Thin film resistors at any frequency fall into this case, as well as thin conductors at low frequency. DC resistivity loss is independent of frequency. The units are Ohms/square with the value given by:

R DC = 1 ⁄ σ t where:

σ is bulk conductivity (Mhos/M) t is the metalization thickness(M) In contrast, at higher frequencies, the skin effect surface impedance is a function of frequency. The number that you type into xgeom, RRF, called the skin effect coefficient, is multiplied by the square root of the frequency, in Hertz, to arrive at the surface impedance defined in Ohms/square. The surface impedance is complex with the imaginary part equal to the real part. The skin effect coefficient that you type into xgeom is approximately given by:

140

Chapter 7 Function Reference XGEOM

µ R RF = skin effect coefficient ≈ π --σ where:

µ = 4π × 10

–7

σ is bulk conductivity (Mhos/M) The above equation assumes that all of the current travels on just one side of the conductor. This is a good approximation for microstrip. The equation should be modified for other structures. Stripline, for example, has current of equal amplitude on both the top and bottom of the conductor. In this case, you should divide the RRF value by two, while maintaining RDC. As an example, σ for copper is 5.8x107 Mhos/M, giving RDC = 0.006 Ohms/ square (t = 3 µM) and a microstrip RRF = 2.6x10-7 (enter “2.6e-7” into xgeom). In reality, the bulk conductivity of copper, or any other given metal, may not equal

141

Xgeom User’s Manual the laboratory value, so the figures as calculated above are likely to be lower than actual results. Table 1 provides calculated results of commonly used metals using the equations above.

Table 7-1: Properties of Commonly Used Metals RDC (Ω/sq) t = 1 mil

RRF (ΩHz-1/2/sq) “Skin Effect” (microstrip)

Metal

σ (S/M)

RDC (Ω/sq) t = 1 µM

Aluminum

3.72e7

0.027

1.1e-3

3.3e-7

0

0

Brass

1.57e7

0.070

2.5e-3

5.0e-7

0

0

Copper

5.80e7

0.017

6.8e-4

2.6e-7

0

0

Gold

4.09e7

0.024

9.6e-4

3.1e-7

0

0

Nichrome

1.00e6

1.000

3.9e-2

2.0e-6

0

0

Silver

6.17e7

0.016

6.4e-4

2.5e-7

0

0

Tantalum

6.45e6

0.155

6.1e-3

7.8e-7

0

0

Tin

8.70e6

0.115

4.5e-3

6.7e-7

0

0

XDC (Ω/sq) Ls (pH/sq) “Reactance”

The transition from electrically thin (DC) to electrically thick (RF) is properly modeled in em. Thus em provides accurate results at low, high and transition frequencies. If the skin effect coefficient (RRF) is set to 0.0, then the value of RDC is used over all frequencies. This is the usual case for resistors. The example files “res400.geo” and “res500.geo” show how to set up a 400 ohm and 500 ohm resistor using lossy metal. The top or bottom of the xgeom box may be set to a perfect matched waveguide load or free space. WG Load is useful for modeling infinite arrays in em. The top or bottom cover will then be terminated with a perfectly matched waveguide load. Note this is not the same as an open environment and the method for modeling radiating structures in an open environment has not changed. To model an open

142

Chapter 7 Function Reference

The surface reactance is specified in Ohms/square and is usually set to zero. See section “Surface Reactance,” page 43 in the Em User’s Manual for an explanation of surface reactance.

Parameters − Ports Selecting Parameters ⇒ Ports opens the Port Impedance dialog box. Using this dialog box allows you to view and modify the impedance values of all the ports in your circuit simultaneously. There is an entry row for each port number, in which you can enter the following values in their respective text boxes: • • • •

Resistance (in ohms): R Reactance (in ohms): X Inductance (in nanohenrys): L Capacitance (in picofarads): C

Note that more than one port may have the same port number and they share the same parameters. Also, negative port numbers use the same parameters as positive port numbers, e.g. Port -1 and Port 1 use the same set of parameters. For a further discussion of port numbers, see “Changing Port Numbering,” page 42.

143

XGEOM

environment select Free Space, which sets the impedance of the cover to 377 ohms/sq, the impedance of free space. You should also note that the sidewalls are always modeled as perfect conductors.

Xgeom User’s Manual The bearing of the values on the port are as depicted below.

R + jX

L C

V

If there are too many ports to be displayed all at once, use the scroll bar to access the entry rows for the other ports.

TIP You may also enter the impedance value for a particular port by selecting Modify ⇒ Attributes when the port is selected in the xgeom window.

Parameters − Ref. Planes/Cal. Length Em has an automatic de-embedding capability. When invoked, em removes the port discontinuity and a desired length of transmission line, moving the reference plane into the interior of the box. The reference planes used for de-embedding are set with this option. One reference plane length per box side may be specified. Parameters ⇒ Ref. Planes/Cal. Length also allows you to specify the calibration length used for de-embedding. A box side has a reference plane or a calibration length, but not both. Please note that it is possible for the calibration length to be greater than the box length.

144

Chapter 7 Function Reference

This dialog box is used only for standard ports. Setting Reference planes/Calibration lengths for an auto-grounded port must be done in the Modify Attributes dialog box. See section "Modify - Attributes" on page 118 for details. When you select Parameters ⇒ Ref. Planes/Cal. Length, the Reference Planes/ Calibration Lengths dialog box is opened. Each box wall (Top, Left, Right, or Bottom) has a text box containing the following items:

Type field: Use the drop list to select either a reference plane or a calibration length for the box side. Length field: Enter the desired length for the reference plane or calibration length for the box side. Use Mouse button: This button allows you to set the reference plane length using the mouse. After clicking on this button, a plus (+) cursor will appear in the xgeom window. Click in the desired place in your circuit where the reference plane or 145

XGEOM

NOTE:

Xgeom User’s Manual calibration length should be. Once you click in the xgeom window, the length text box is updated with the corresponding value. Reference planes will be drawn on the circuit. Xgeom may change the reference plane slightly so that it is aligned with the edge of a cell. The actual position is indicated with a short vertical line at the actual reference plane location. To remove a reference plane, enter a value of zero in the Length field or click outside the boxwall on that side. There is only one reference plane or calibration length per box side. The same reference plane is used on all ports on all levels on that side of the box.

Parameters − Parallel Subs. Parameters ⇒ Parallel Subs. opens the Parallel Subsections dialog box, and allows the removal of X or Y subsections for a specified distance from a box wall. This feature is useful for reducing computational time although it may affect the accuracy of the results. It is referred to as “Parallel Subs” because all subsections in the specified region that are parallel to the selected box wall are removed. Refer to the Em User’s Manual for more information about removing parallel subsections before using this function.

Parallel Subsections dialog box.

146

Chapter 7 Function Reference

Length field: Enter the desired length for the parallel subsection from the box wall. Use Mouse button: Click on this button to set the length using the mouse. Once you click in the xgeom window, the length text box is updated with the corresponding value. To remove a parallel subsection, enter a value of zero in the Length field or click outside the boxwall on that side. Em will remove the appropriate subsections (Y for the left and right sides, X for the top and bottom) up to the specified distance, shown as a grayed-out area. An example is shown below. The Parallel Subsections option removes either X or Y subsections over a specified distance from the box wall.

}

}

Only X subsections (Y Subsections removed)

147

XGEOM

When you select Parameters ⇒ Parallel Subsections the Parallel Subsections dialog box is opened, as shown in the figure on page 146. Each box wall (Top, Left, Right, or Bottom) has a text box containing the following items:

Xgeom User’s Manual

The Help Menu The Help menu provides you with basic operating information such as the xgeom version and licensing.

Help - License Info For platforms which support floating licenses, the License dialog box will allow you to obtain or release a license.

The Users with Licenses scroll list will display a list of users who are presently in possession of a license. If you do not have a license, the Get command button will be enabled. Click on this button to obtain a license, if any are available. If you have a license, the Release command button will be enabled. Click on this button if you wish to give up your license.

Help - System Info This command opens a dialog box to provide system information, such as license ID, display type, and memory available.

148

Chapter 7 Function Reference XGEOM

Help - Debug This command opens a dialog box which allows you to set the parameters of a trace for debug purposes. This will be used to provide information to technical support in case of a problem.

Help - About This command opens a dialog box which provides the version number for xgeom and licensing type information.

149

Xgeom User’s Manual

Keyboard Shortcuts Table 2 describes keyboard shortcuts used in xgeom. All keys are shown here. Shortcut keys which also have a button on the tool bar are detailed in Table 3, “Tool Bar,” on page 152. All of these functions may also be implemented by the menu. Most shortcut keys are “control” keys, abbreviated with a “^” symbol. As an example, to type a Control-P (^P), hold down the Control key and press the P key. The DEL key is the Delete and/or Backspace key and the ESC is the Escape key.

Table 2 Keyboard Shortcuts Key

150

Meaning

^N

New File

^O

Open File

^S

Save File

^Z

Undo

^Y

Redo

^X

Cut

^C

Copy

^V

Paste

^E

Duplicate

^A

Select All

^T

Reselect

^W

Zoom Out

Space Bar Middle Mouse button

Zoom In

^L

Previous View

Chapter 7 Function Reference XGEOM

Table 2 Keyboard Shortcuts Key

NOTE:

Meaning

^F

Full View

^U

Up One Level

^D

Down One Level

^M

Toggle Cell Fill

^P

Add Metalization- Draw Polygons

^R

Add Metalization-Draw Rectangle

DEL

Delete

ESC

Pointer mode (Cancel current mode)

The ESC key may also be used to cancel actions. For example, if you are in the middle of adding a polygon, pressing the ESC key will exit the Add mode and the polygon will not be added to the circuit.

151

Xgeom User’s Manual

Tool Bar The tool bar consists of a row of buttons which appears just under the main menu in the xgeom window. The tool bar allows you to access frequently used functions quickly, without going through pull down menus. Each button, its use, and a keyboard shortcut to access it, is described below.

Table 3 Tool Bar Tool Bar button

Name

Keyboard Shortcut

File ⇒New

^N

File ⇒ Open

^O

File ⇒ Save

^S

Cut

Edit ⇒ Cut

^X

Copy

Edit ⇒ Copy

^C

Paste

Edit ⇒ Paste

^V

New Document

Open Document

Save Document

152

Menu Equivalent

Chapter 7 Function Reference XGEOM

Table 3 Tool Bar Tool Bar button

Name New Window (View) of Same Document

Menu Equivalent View ⇒ New View

Keyboard Shortcut None

Zoom In

View ⇒ Zoom In

Zoom Out

View ⇒ Zoom Out

^W

Full View

View ⇒Full View

^F

Previous View

View ⇒ Previous View

^Y

Up One Level

View ⇒ Up One Level

^U

Down One Level

View ⇒ Down One ^D Level

Space Bar Middle Mouse Button

153

Xgeom User’s Manual Table 3 Tool Bar Tool Bar button

Name Ortho

Toggle Measuring Tool Analyze File

Print

Menu Equivalent Tools

None

View ⇒ Measuring None Tool File ⇒ Analyze

None

File ⇒ Print

None

Level Drop None List

154

⇒ Ortho

Keyboard Shortcut

None

Chapter 7 Function Reference XGEOM

Tool Box The Tool Box consists of a window of buttons which appears next to the xgeom window on your display upon opening a circuit file. The Tool Box allows you to access frequently used tools quickly, without going to the tools menu. The table below identifies each button in the tool box and gives a brief description of its functions. For more detailed explanations, see the corresponding menu entry. As you move your cursor over each button, a brief description of its function appears in the Messages box in the status bar at the bottom of the xgeom window. Note that the Metal Mode/Brick Mode button toggles between the two modes, changing each time you click on the button. The Tool Box is closed automatically when the last circuit file is closed. The tool box can also be turned on or off by selecting View ⇒ Tool Box.

TIP As with menu selects, pressing the shift key while clicking on the button will leave you in the selected mode until you exit it explicitly. See “Shift Selecting Modes,” page 18.

155

Xgeom User’s Manual .

Table 4 Tool Box Button

156

Name

Description

Pointer

Places xgeom in pointer mode, allowing you to select objects.

Reshape Polygons

Places xgeom in reshape mode, allowing you to move points in polygons.

Add Points to Polygons

Allows you to add points to a selected polygon.

Add Port to Polygon Edge

Places xgeom in Add Port mode allowing you to add ports to metal polygons.

Add Via to Polygon Edge

Places xgeom in Add Via mode allowing you to add vias to metal polygons.

Add a Polygon

Adds a polygon of metal or dielectric brick according to the selections below.

Add a Rectangle

Adds a rectangle of metal or dielectric brick according to the selections below.

Metalization Mode

When Add a Polygon or Rectangle is selected, objects added will be metal.

Dielectric Brick Mode

When Add a Polygon or Rectangle is selected, objects added will be dielectric brick.

Chapter 7 Function Reference XGEOM

Typing in Coordinates From the Keyboard Polygons and points may be entered and moved by using the mouse or their coordinates may be typed in directly. You can type in polar or Cartesian coordinates either as absolute coordinates or relative coordinates. Relative coordinates are relative to the last point entered. The following section shows you how to enter in a rectangle by typing in the coordinates. The coordinate system used is shown below. Use View ⇒ Measuring Tool to change the origin used for Cartesian coordinates.

+Y

90° 45°

135°

180°

+X (0,0)

0°

225° 270°

Cartesian Coordinates

315° ( – 45° )

Polar Coordinates

An Example of Keying in Coordinates For this example, we will draw a rectangular polygon that is 60 mils wide and 30 mils high. First, start with a new xgeom window by selecting File ⇒ New from the main menu. If during this example you type in the wrong numbers, just press the Delete or Backspace key and the last entered point will be removed. Now select Tools ⇒ Add Metalization ⇒ Draw Polygon and type in the X and Y coordinates of the first point of the polygon. For this example, type 50,60

157

Xgeom User’s Manual and press the Return key. The point you are entering will appear in the left-hand field of the status bar at the bottom of the xgeom screen. Moving the mouse stretches a rubber band from the coordinates X=50 mils and Y=60 mils, just as if you had clicked the left mouse button on that point. See the figure on page 159. Note that coordinates are snapped to the nearest snap grid point; i.e., typing in “50.01,59.98” would have the same result as “50,60”. To enter the next point 60 mils to the right of this point, type @60<0 A line appears starting from the first point you picked and ending 60 mils to the right of it. The at sign, “@”, tells xgeom that the number you specify is a distance from the last point that you entered. The 60 is the distance to move, and the less than symbol (“<“) tells xgeom that the next number is the angle at which the point is to be placed. Angles are defined in degrees as shown on page 157. For the next point, let’s use Cartesian coordinates. Type @0,30

158

Chapter 7 Function Reference

Keyboa rd

50,90

50,60 (Starting Point)

@0,30

@60<0

Now type 50,90 At this point, your screen should look like the picture above. This time the number was entered as an absolute coordinate just to illustrate that you may mix absolute coordinates with relative coordinates. Mixing absolute coordinates with relative coordinates on the same line is not allowed; e.g., “50,@0” is not allowed. You may finish your polygon using the same technique as the mouse: by clicking exactly on the last point entered, or clicking on the first entered point. You may also type in the equivalent coordinates to finish. The easiest way to complete a polygon is to type: @

159

XGEOM

Another line of our rectangle appears. Once again we used the “@” sign to tell xgeom that the coordinates we entered are relative to the last point. The first number (0) is always the X value and the second number after the comma is the y value. In this case the Y value is positive, so the new point is 30 mils above the previous point.

Xgeom User’s Manual This completes the polygon. Xgeom interprets the “@” sign as relative coordinates. When none are given, zero (0) is assumed. In this case “@” is equivalent to “@0,0” which is the same as clicking the mouse with the cursor on the last point. This completes the polygon. Note also that typing in “@5” is equivalent to “@5,0” and adds a point 5 units to the right of the last point. You may mix mouse and keyboard entry of points. If the “@” sign is used for the first point when the measuring tool anchor display is visible, the values are relative to the anchor. Otherwise, a relative first point is treated as an absolute value. The above example used Tools ⇒ Add Metalization ⇒ Draw Polygon. Keyboard entry is also supported for Tools ⇒ Add Metalization ⇒ Draw Rectangle and Tools ⇒ Add Points to Polygon. As soon as you type a number or an at sign, “@”, while any of these functions are active, xgeom starts reading from the keyboard. Note that for entering a rectangle, you only enter two points: the origin of the rectangle and the location of the opposite corner. Direct keyboard entry of coordinates can be used in place of mouse actions in most circumstances, including while adding rectangles, moving objects and zooming.

Panning Panning is accomplished through the use of scroll bars which appear on the edges of the xgeom window. To cause the display to move in increments, click on the arrow button at either end of a scroll bar. Alternately, place the cursor on the bar at the center and drag the bar to the desired position.

160

Index XGEOM

Index

A ^A 86, 140 accuracy 136 add dielectric bricks 102, 146 default brick material 127 polygon 102 rectangle 102 metalization 99–102 default metal type 129 donut 64, 100 fan stub 70, 101 interdigital capacitor 62, 100 lange coupler 71, 102 meander 66, 101 parallel lines 69, 101 polygon 100, 102 rectangle 61, 99, 100 rectangular spiral 68, 101 round spiral 67, 101 points 19, 150 making an error 20, 99 to polygon 99 polygons 104 ports 103 auto-grounded 31 standard 30 via-ports 30 via 103

air-bridge 49 example 49–52 analyze 79 anisotropic dielectric bricks 53 aspect ratio 90 @ (at) sign 148–150 attributes 16 dielectric bricks 108, 111 modify metalization 109 ports 112 auto-grounded ports 29, 103 calibration lengths 112 reference planes 112 autosave 82

B backup 82 backup file 25, 82 balanced ports 33 bottom metal 122 box parameters 119 bottom metal 122 maximum subsection size 120 sizes 119 symmetry 121 top metal 122 Box Parameters dialog box 94, 119 box size 119, 120 Brick Attributes dialog box 111 161

Xgeom User’s Manual brick materials 127 parameters 126 Brick Materials dialog box 126 bricks See dielectric bricks bulk conductivity 130, 131

C ^C 84, 140, 142 calibration lengths 34, 35, 36, 112, 134 capacitance 133 cartesian coordinates 147 cell fill 93, 109 patterns 130 cell grid 94, 106 cell size 94, 106, 119, 120 cells/lambda 120 clicking 14 clip 85 clipboard 21, 83, 85 delete 85 duplicate 84 paste 84 close 78 comments 81 conductivity 124 control key 8, 18 conventions 5 status bar 6 tool bar 6 tool box 6 convert 118 coordinates 147 default origin 148 entry from keyboard 147–150 polar coordinates 148 copy 15, 22, 84 edit 84 points 22 162

polygons 22 across layers 22, 84 copy picture 87 corner mode 110 cut 15, 21, 47, 83 polygons 21

D ^D 91, 141, 143 de-embedding 134 de-embedding ports via-ports 28 DEL 85, 141 delete 15, 20, 83, 85, 88, 141 clipboard 85 objects 21 parallel subsections 137 points 20 polygons 21 reference planes 136 delete key 20, 99 deleting ports 31, 103 vias 47, 104 diagonal mode 109 dialog boxes Box Parameters 94, 119 Brick Attributes 111 Brick Materials 126 Dielectric Brick Attributes 16 Dielectric Layers 123 Flip 22, 113 Levels 44 License 138 Metal Types 128 Metalization Attributes 16, 109 Metalization Levels 92 Object Visibility 93

Index XGEOM

Parallel Subsections 136 Port Attributes 16, 112 Port Impedance 34, 133 Print Setup 80 Reference Planes/Calibration Lengths 135 Resize 116 Rotate 114 Save As 78, 79 Selection Filter 88 Snap Grid Setup 94, 106 Snap Objects 117 Unit 122 Dielectric Brick Attributes dialog box 16 dielectric bricks 16, 53–58, 102, 126, 127, 146 add 102 anisotropic 53 application 53 attributes 108, 111 convert 118 defining new types 127 fill patterns 128 materials 55, 56, 126 polygon 108 visibility 55 dielectric conductivity 124 dielectric constant 124 dielectric layers 123 auto-grounded ports 30 dielectric conductivity 124 magnetic loss tangent 124 magnetic permeability 124 Dielectric Layers dialog box 123 dielectric loss tangent 124 dielectric thickness 124 dielectrics 123 direct entry 147 discontinuities 134 donut 64, 100

double clicking 8 down one level 91 duplicate 15, 22, 84 clipboard 84 DXF simplified write 82 DXF export file 79, 82

E ^E 85, 140 edge-vias 42 edit 83–88 clip 85 copy 15, 22, 84 copy picture 87 cut 15, 21, 47, 83 delete 15, 20, 85, 88 duplicate 15, 22, 84 multi-layer 87 paste 22, 84 points 18 reselect 86 resize 23 select all 86, 88 select filter 87 single layer 87 single layer select 18 undo 83 unselect 17, 86 em accuracy 136 invoking 79 low frequencies 132 subsections 1, 93, 120 emergency backup 25 ESC 97, 141 escape key 97, 141 163

Xgeom User’s Manual example files patch.geo 28, 48 exit xgeom 78, 82 export 79 DXF file 79 GDS file 80 picture 79

F ^F 90, 141, 143 fan stub 70, 101 feed-structure 29 file 74–75, 76–82 analyze 79 close 78 comments 81 exit 75, 82 export 79 DXF 79, 82 GDS 80 picture 79 new 74, 76 open 75, 76 page setup 80 preferences 81 print 80 revert to saved 79 save 78 save as 78 fill patterns 128, 130 flip 22, 113 modify 22, 23 pivot point 113 Flip dialog box 22, 113 full view 90

G .geo 1, 78 164

gaps 110 GDS export file 80 grid 106

H ^H 140 help menu 75 hot keys 8, 18, 97, 99, 100, 102, 140 shift key 18

I impedance 33, 34, 112 inductance 133 interdigital capacitor 62, 100 internal ports 26, 48 invisible 91 invoking em 79

K keyboard control key 8 coordinate entry 147 ctrl-A 86, 140 ctrl-C 84, 140, 142 ctrl-D 91, 141, 143 ctrl-E 85, 140 ctrl-F 90, 141, 143 ctrl-H 140 ctrl-M 93, 141 ctrl-N 76, 140, 142 ctrl-O 77, 140, 142 ctrl-P 100, 141 ctrl-R 99, 141 ctrl-S 78, 140, 142 ctrl-T 86, 89, 140, 143 ctrl-U 91, 141, 143 ctrl-V 84, 140, 142

Index

L lange coupler 71, 102 lasso 16 Levels dialog box 44 License dialog box 138 lock 87, 91, 92, 119, 120 loss 109 definitions 130 formula 130 magnetic loss tangent 124 microstrip 131 stripline 131 tangent 124 vias 47 loss tangent 124 lossless 109, 128

M ^M 93, 141 magnetic loss tangent 124 magnetic permeability 124 main menu 76–139 accesses 7 maximum cell size 120 meander 66, 101 measuring tool 95, 114 anchor 95–96 menu

XGEOM

ctrl-W 140, 143 ctrl-X 83, 140, 142 ctrl-Y 140 ctrl-Z 140 DEL 85, 141 ESC 97, 141 shortcut keys 8 space bar 140 table of keys 140

edit See edit file See file help 75 modify See modify parameters See parameters startup 74 tools See tools view See view menu bar accesses 7 metal types 93, 128 defining new types 129 fill patterns 130 Metal Types dialog box 128 metalization commonly used metals 132 convert 118 fill patterns 130 fill type 109 loss 109, 130 modify attributes 109 reactance 129 resistivity 129 thickness 130 type 109 Metalization Attributes dialog box 16, 109 metalization levels 91 invisible 92, 93 locked 92 unlocked 93 visibility 44 visible 92, 93 Metalization Levels dialog box 92 metalization loss vias 126 metals via-posts 93 microstrip

165

Xgeom User’s Manual loss 131 mode add points to polygons 19 orthogonal mode 104 pointer 13, 14, 21, 97 reshape 14, 18, 97 shift selecting 8 modify 16, 108–118 attributes 16, 108 metalization 109 ports 31, 33, 36, 112 convert to brick 118 convert to metal 118 flip 22, 23, 113 rotate 23, 114 snap to grid 117 mouse clicking 14 move points 18 polygons 21 polygons between layers 22 move layer 84 multi-layer 87 editing 84 multi-layer select 17 multiple selects 15

N ^N 76, 140, 142 new 76 new features 2 new file 76 new view 90 number of cells 119, 120

O ^O 77, 140, 142 166

Object Visibility dialog box 93 objects across layers 84 copy 84 delete 21, 83, 85 editing 14 flip 113 moving 105 multiple selects 15 orthogonal 104 points add 150 ports 25–30, 103 reference planes 134, 136 rotate 114 selecting 14 snap 117 unselect 17 vias 41–48, 103, 104 visibility 93 open File 76 ortho 104

P ^P 100, 102, 141 page setup 80 palette 59–71 donut 64 fan stub 70 interdigital capacitor 62 lange coupler 71 meander 66 parallel lines 69 rectangle 61 rectangular spiral 68 round spiral 67 palette of standard geometries see palette or standard geometries

Index XGEOM

panning 150 parallel lines 69, 101 parallel subsections delete 137 Parallel Subsections dialog box 136 parameters 119–137 box 119 bottom metal 122 box size 120 cell size 94, 120 maximum subsection 120 number of cells 120 symmetry 121 top metal 122 brick materials 126 calibration lengths 134 dielectric layers 123 metal types 93, 128 metalization 109, 111 parallel subsections 136 ports 34, 133 reference planes 134 units 122 partitioning 124 paste 21, 22, 84 clipboard 84 polygons 21 patch.geo 28, 48 picture export 79 pivot point flip 113 rotate 115 pointer 13, 97 pointer mode 13, 14, 21, 97, 141 points add 19, 150 making an error 20

copy 22 delete 20 edit 18 move 18 select 14 polar coordinates 147 polygons 20 add 100, 102 add points 19, 99 brick 108 copy 22, 84 across layers 22 cut 21 delete 21, 83, 85 flip 22, 113 modify 108 move 21 moving between layers 22 paste 21 resize 22, 23, 116 rotate 22, 23, 114 select 14 snap 117 Port Attributes dialog box 16, 112 Port Impedance dialog box 34, 133 ports 20, 25–36, 133 add 103 auto-grounded 31 standard 30 via-ports 30 attached to polygons 16 auto-grounded 29, 103, 112 calibration lengths 36 example 36–39 reference planes 36 calibration length 34 de-embedding 27, 30 deleting 31, 103

167

Xgeom User’s Manual discontinuity 134 impedance 33, 34, 112, 133 internal 26, 48 manipulating 30 modify attributes 31, 36, 112 non-sequential 33 numbering 30, 32, 112 push-pull 33 reference planes 34 standard calibration lengths 35, 135 reference planes 35, 135 type 112 via-ports 28, 48 preferences 81 autosave 82 remain in mode 82 simplified write 82 previous view 89 print 80 Print Setup dialog box 80 push-pull ports 33

R ^R 99, 141 reactance 129, 133 rectangle 60, 61 rectangular spiral 68, 101 redo 140 redraw 96 reference planes 34, 35, 36, 134 auto-grounded ports 30 delete 136 Reference Planes/Calibration Lengths dialog box 135 remain in mode 82 reselect 86 reshape 14, 97 168

reshape mode 14, 18, 20, 97 resistance 133 resistivity 129 resistors 132 resize 22, 23, 116 scaling factor 116 Resize dialog box 116 revert to saved 79 rotate 22, 23, 114 angle 114 pivot point 115 Rotate dialog box 114 round spiral 67, 101 rubber banding 99

S ^S 78, 140, 142 save 78 revert to saved 79 save as 78 Save As dialog box 78, 79 scaling factor 116 scroll bars 150 select 87 all 86 control key 18 lasso 16 multi-layer 17 multiple 15 objects 14 point 14 polygons 14 reselect 86 select all 87 select filter 87 shift key 18 unselect 86 select all 88

Index XGEOM

Selection Filter dialog box 88 shift key 8, 18 shift selecting 145 shift selecting modes 8 shortcut keys 8, 18, 99, 100, 102 ctrl-A 86, 140 ctrl-C 84, 140, 142 ctrl-D 91, 141, 143 ctrl-E 85, 140 ctrl-F 90, 141, 143 ctrl-H 140 ctrl-M 93, 141 ctrl-N 76, 140, 142 ctrl-O 77, 140, 142 ctrl-P 100, 141 ctrl-R 99, 141 ctrl-S 78, 140, 142 ctrl-T 86, 89, 140, 143 ctrl-U 91, 141, 143 ctrl-V 84, 140, 142 ctrl-W 140, 143 ctrl-X 83, 140, 142 ctrl-Y 140 ctrl-Z 140 DEL 85, 141 ESC 97, 141 space bar 140 table of keys 140 shorted vias 48 simplified write 82 single layer 87 single layer select 18 skin effect 130 snap 19, 21, 94, 106, 117 no x snap 107 no y snap 107 none 106 snap distance 106

snap grid 106, 117 Snap Grid Setup dialog box 94, 106 Snap Objects dialog box 117 snap setup 106 space bar 140 S-Parameters 1 spiral rectangular 68, 101 round 67, 101 staircase mode 109 standard geometries 59–71 donut 64 fan stub 70 interdigital capacitor 62 lange coupler 71 meander 66 parallel lines 69 rectangle 60, 61, 100 rectangular spiral 68 round spiral 67 startup menu 74–75 file 74–75 exit 75 new 74 open 75 status 97 status bar 2, 3, 4, 6, 7, 94, 97, 103, 145, 148 stripline loss 131 subsectional vias 43 subsections 110 maximum size 120 parallel 136 size 110, 111 triangular 109 vias 43 X Min 110 Y Max 110

169

Xgeom User’s Manual Y Min 110 surface impedance 130 surface reactance 133 symmetry 121

T ^T 86, 89, 140, 143 thick metal 32 tool bar 3, 6, 8, 31, 43, 89, 94 table of icons 142 tool box 3, 6, 8, 30, 95, 145 table of icons 146 tools 97–107 add dielectric bricks 102 add metalization 99–102 donut 64, 100 fan stub 70, 101 interdigital capacitor 62, 100 lange coupler 71, 102 meander 66, 101 parallel lines 69, 101 rectangle 61, 100 rectangular spiral 68, 101 round spiral 67, 101 add points to polygon 99 add ports 103 add via 103 add vias 42 ortho 104 pointer 13, 97 reshape 14, 97 snap setup 106 top metal 122 triangle subsections 109

U ^U 91, 141, 143 undo 83, 140 170

units setting 122 Units dialog box 122 unlock 93 unselect 17, 86 control key 18 up one level 91

V ^V 84, 140, 142 via 103 via-fence 43 via-ports 28, 48 via-posts 43, 93 vias 20, 41–48 add 42 deleting 47, 104 edge-vias 42 example 49–52 loss 47 metalization loss 126 restrictions 104 shorted 48 subsections 43 summary 52 symbol 43, 103 to ground 44 via-posts 43 view 89–96 aspect ratio 90 cell fill 93 cell grid 94 down one level 91 full 90 measuring tool 95 metalization levels 44, 91 new 90 object visibility 93

Index XGEOM

previous 89 redraw 96 status bar 94 tool bar 94 tool box 95 up one level 91 zoom in 89 zoom out 89 visible 91

Z ^Z 140 zoom 89 in 89 out 89 Z-partitioning 124

W ^W 140, 143 WG load 128 window status bar 94 tool bar 94, 142 tool box 95, 145

X ^X 83, 140, 142 X Max 110 X Min 110 xgeom 4.0 style 82 autosave 82 clipboard 21, 85 cut 83 exit 78, 82 menu bar 7 overview 1 window 94, 95, 145

Y ^Y 140 Y Max 110 Y Min 110

171

Xgeom User’s Manual

172

Em User’s Manual

Em User’s Manual

Table of Contents

Table of Contents Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 A Simple Outline of the Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Subsectioning the Circuit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Zero Voltage Across a Conductor . . . . . . . . . . . . . . . . . . . . . . 4 The Input and Output Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Em Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Em Origins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Enhancements in Release 6.0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 New Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Em User’s Manual Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Describing Menu Bar Accesses . . . . . . . . . . . . . . . . . . . . . . . 11 Invoking Sonnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 UNIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2

Important Options and Concepts . . . . . . . . . . . . . . . . . . . . . . . 17 De-embedding the Port Discontinuity . . . . . . . . . . . . . . . . . . . . . . . . . 18 Using Multi-Frequency Caching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Using Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Invoking Single Precision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Adjusting the Subsectioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Removing Parallel Subsections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

i

EM

1

Em User’s Manual

3

Subsectioning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Selecting Cell Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Setting the Maximum Subsection Size Parameter . . . . . . . . . . . . . . . 31 Defining the Subsectioning Frequency . . . . . . . . . . . . . . . . . . . . . . . . 31 Changing the Subsectioning of a Polygon . . . . . . . . . . . . . . . . . . . . . 32 Default Subsectioning of a Polygon . . . . . . . . . . . . . . . . . . . X Min and Y Min for a Manhattan Polygon. . . . . . . . . . . . . X Min and Y Min for a Non-Manhattan Polygon . . . . . . . . Using X Max and Y Max for any Polygon . . . . . . . . . . . . . .

32 34 36 37

Using the Edge Mesh Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4

Metalization and Dielectric Loss . . . . . . . . . . . . . . . . . . . . . . . . 41 Metalization Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Surface Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Surface Reactance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Dielectric Layer Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

5

Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Port Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Box-Wall Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ungrounded-Internal Ports . . . . . . . . . . . . . . . . . . . . . . . . . . Via Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Automatic-Grounded Ports . . . . . . . . . . . . . . . . . . . . . . . . . . Special Considerations for Auto-Grounded Ports . . . . . . . .

50 51 52 53 54

Specifying Port Normalizing Impedances. . . . . . . . . . . . . . . . . . . . . . 55 Special Port Numbering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Ports with Duplicate Numbers . . . . . . . . . . . . . . . . . . . . . . . 58 Ports with Negative Numbers . . . . . . . . . . . . . . . . . . . . . . . . 58 6

De-embedding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 Enabling the De-embedding Algorithm . . . . . . . . . . . . . . . . . . . . . . . 62

ii

Table of Contents De-embedding Port Discontinuities . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Box-Wall Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Ungrounded-Internal Ports . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Auto-Grounded Ports. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Shifting Reference Planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

De-embedding Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 De-embedding Error Codes . . . . . . . . . . . . . . . . . . . . . . . . . . 78 7

De-embedding Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Defining Reference Planes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 De-embedding Without Reference Planes. . . . . . . . . . . . . . . 82 Reference Plane Length Minimums . . . . . . . . . . . . . . . . . . . 82 Reference Plane Lengths at Multiples of a Half-Wavelength 84 Reference Plane Lengths Greater than One Wavelength . . . 84 Non-Physical S-Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 85 Box Resonances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Higher Order Transmission Line Modes . . . . . . . . . . . . . . . . . . . . . . . 88

8

Network File Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 Network File Analyses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Cascading S-, Y- and Z-Parameter Data Files . . . . . . . . . . . . . . . . . . . 90 Thin-Film Resistor Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 The Network File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 A Network File Invoking a Geometry File Analysis . . . . . . . . . . . . . . 97 Inserting Lumped Elements into a Circuit . . . . . . . . . . . . . . . . . . . . . 102 Using Ungrounded-Internal Ports . . . . . . . . . . . . . . . . . . . . 107

9

Circuit Subdivision - A Filter Example . . . . . . . . . . . . . . . . . 111 Example Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 iii

EM

Box-Wall Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Coupled Transmission Lines . . . . . . . . . . . . . . . . . . . . . . . . . 74 Auto-Grounded Ports. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Em User’s Manual Dividing the Circuit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Creating the Geometry Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Analyzing the Geometry Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Creating the Network File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Analysis of the Network File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 Alternate Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 10

Intelligent Frequency Selection . . . . . . . . . . . . . . . . . . . . . . . . 125 Automatic Frequency Selection Example . . . . . . . . . . . . . . . . . . . . . 126 Using FINDMIN and FINDMAX. . . . . . . . . . . . . . . . . . . . . . . . . . . 130

11

The em Network File. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 Format of the em Network File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 Header Line. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 Data Blocks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 The DIM Data Block. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rules for the DIM Data Block . . . . . . . . . . . . . . . . . . . . . . Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

137 138 139 139

The VAR Data Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 name . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 Referencing Variables in the CKT Data Block . . . . . . . . . 141 The CKT Data Block. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . defNp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . netname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . filename . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

143 145 145 145 145 145 146

Table of Contents Using Data Tags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 Geometry File Consistency . . . . . . . . . . . . . . . . . . . . . . . . . 148 High Precision em Output Files. . . . . . . . . . . . . . . . . . . . . . 148 The FILEOUT Data Block. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

The FREQ Data Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 keyword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 Analysis Control Keywords For em . . . . . . . . . . . . . . . . . . 154 Sorted Frequency Sweeps . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Using END to Control the Order of Frequency Sweeps . . . 156 Frequency Interpolation of em Output Data . . . . . . . . . . . . 156 AUTO, FINDMIN and FINDMAX for Basic Analyses . . . 157 Overriding the FREQ Block . . . . . . . . . . . . . . . . . . . . . . . . 157 The OUT Data Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 netname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 meas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 filename . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 delim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 com . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 12

Using Diagonal Fill . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 A Coupled Open-Miter with Diagonal Fill . . . . . . . . . . . . . . . . . . . . 163

13

Vias and 3-D Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 Adding Vias to the Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

v

EM

netname . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 format. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 filename . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 param . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 outtype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 SPICE and PSPICE keywords. . . . . . . . . . . . . . . . . . . . . . . 152 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

Em User’s Manual Restrictions on Vias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 Simple Via Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 A Conical Via . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 14

Dielectric Bricks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 Applications of Dielectric Bricks . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 Dielectric Brick Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 Guidelines for Using Dielectric Bricks . . . . . . . . . . . . . . . . . . . . . . . 173 Subsectioning Dielectric Bricks . . . . . . . . . . . . . . . . . . . . . Using Vias Inside a Dielectric Brick . . . . . . . . . . . . . . . . . De-embedding and Dielectric Bricks . . . . . . . . . . . . . . . . . Air Dielectric Bricks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

173 174 174 175

Limitations of Dielectric Bricks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Diagonal Fill . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Antennas and Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Ebridge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 15

Antennas and Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 Modeling Infinite Arrays. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 Modeling an Open Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 Validation Example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

16

SPICE Lumped Model Synthesis . . . . . . . . . . . . . . . . . . . . . . 187 Class of Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 Using The SPICE Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 PSpice Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 N-Coupled Line Option. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 A Simple Microwave Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 Topology Used for SPICE Output . . . . . . . . . . . . . . . . . . . . . . . . . . 194 A High Speed Digital Example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

vi

Table of Contents Other Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 17

Coplanar Waveguide Discontinuities and Balanced Ports . . 203 The Coplanar Short . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 The Coplanar Cross Junction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206

Thick Metal with Arbitrary Cross-Section . . . . . . . . . . . . . . 209

19

Package Resonances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 Unwanted Box Resonances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 Verifying the Box Resonance Problem . . . . . . . . . . . . . . . . 218 Removing Box Resonances . . . . . . . . . . . . . . . . . . . . . . . . . 220

20

Viewing Tangential Electric Fields . . . . . . . . . . . . . . . . . . . . . 223

21

Accuracy Benchmarking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 An Exact Benchmark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Residual Error Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 Using the Error Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

22

Range of Analysis Validity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Subsection Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 Metalization and Dielectric Thickness. . . . . . . . . . . . . . . . . . . . . . . . 234 Numerical Precision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235

23

Time Required for Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 The “Wall” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240 Detailed Parameter Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . 241

vii

EM

18

Em User’s Manual

24

Em Interface: Analysis of a Geometry File . . . . . . . . . . . . . . 245 Invoking em. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 Selecting a Geometry File Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 246 Analysis Input Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 Selecting a Geometry File. . . . . . . . . . . . . . . . . . . . . . . . . . 247 Editing a Geometry File . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 Specifying Frequency Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 Setting up a Simple Sweep . . . . . . . . . . . . . . . . . . . . . . . . . 248 Setting Up a Complex Sweep . . . . . . . . . . . . . . . . . . . . . . . 249 Using an Analysis Control File. . . . . . . . . . . . . . . . . . . . . . 250 Selecting Run Options. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 Selecting Job Window Options. . . . . . . . . . . . . . . . . . . . . . 251 Selecting Additional Options . . . . . . . . . . . . . . . . . . . . . . . 252

25

Em Interface: Analysis of a Network File. . . . . . . . . . . . . . . . 257 Selecting a Network File Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 258 Analysis Input Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 Selecting a Network File. . . . . . . . . . . . . . . . . . . . . . . . . . . 258 Editing a Network File . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 Specifying Frequency Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 Selecting a Simple Sweep. . . . . . . . . . . . . . . . . . . . . . . . . . 260 Selecting Internal Sweep . . . . . . . . . . . . . . . . . . . . . . . . . . 260 Selecting Run Options. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 Selecting the Verbose Option . . . . . . . . . . . . . . . . . . . . . . . 261 Selecting Additional Options for a Network File . . . . . . . . 261

26

Em Interface: Run Controls. . . . . . . . . . . . . . . . . . . . . . . . . . . 263 Editing Analysis Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 Adding Frequency Controls . . . . . . . . . . . . . . . . . . . . . . . . Entering Intelligent Frequency Controls. . . . . . . . . . . . . . . Entering a Separator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Editing a Frequency Control Entry . . . . . . . . . . . . . . . . . . .

viii

265 267 268 268

Table of Contents Specifying SPICE Parameters . . . . . . . . . . . . . . . . . . . . . . . 268 Adding Comments to the Analysis Control File . . . . . . . . . 269 Specifying the Subsectioning Frequency . . . . . . . . . . . . . . 270 Saving Frequency Controls . . . . . . . . . . . . . . . . . . . . . . . . . 270 Specifying Output Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 Running an em Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272

Using the em Output Window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 Closing the Output Window . . . . . . . . . . . . . . . . . . . . . . . . 277 Re-Opening the Output Window . . . . . . . . . . . . . . . . . . . . . 277 Saving the Contents of the Output Window . . . . . . . . . . . . 278 Invoking emgraph to Plot Response Data. . . . . . . . . . . . . . 278 Invoking emvu to View Current Density . . . . . . . . . . . . . . 278 Invoking patvu to View the Far-Field Radiation Patterns. . 279 Invoking a Text Editor to View Response Data . . . . . . . . . 279 Job Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 Creating a New Job File . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 Opening an Existing Job File. . . . . . . . . . . . . . . . . . . . . . . . 280 Loading an Existing Job File . . . . . . . . . . . . . . . . . . . . . . . . 282 Saving a Job File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 Reverting to a Saved Job File . . . . . . . . . . . . . . . . . . . . . . . 283 Em Control Preferences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 Setting Multi-Frequency Caching Parameters. . . . . . . . . . . 283 Selecting Startup Run Options . . . . . . . . . . . . . . . . . . . . . . 285 Setting Up a Default Simple Sweep for Analyses . . . . . . . . 285 Appendix I

The em Command Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 Input/Output Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294

ix

EM

Viewing the Run List. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 Editing the Run List. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 Starting an em Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 Pausing an em Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 Continuing an em Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 277 Stopping an em Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 277

Em User’s Manual

Appendix II

The Analysis Control File Format. . . . . . . . . . . . . . . . . . . . . . 297

Appendix III

LEVEL1 and LEVEL1plus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 LEVEL1 Suite. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 LEVEL1plus Suite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304

Appendix IV

Warning and Error Messages . . . . . . . . . . . . . . . . . . . . . . . . . 305 Warning Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306 Error Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 De-embedding Error Codes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310

Appendix V

Sonnet Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 Sonnet Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 Sonnet Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321

Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347

x

Chapter 1 Introduction

EM

Chapter 1

Introduction

Em performs electromagnetic analysis [75, 76, 78] for arbitrary 3-D planar [50] (e.g., microstrip, coplanar, stripline, etc.) geometries, maintaining full accuracy at all frequencies. Em is a “full-wave” analysis in that it takes into account all possible coupling mechanisms. The analysis inherently includes dispersion, stray coupling, discontinuities, surface waves, moding, metalization loss, dielectric loss and radiation loss. In short, it is a complete electromagnetic analysis. Since em uses a surface meshing technique, i.e. it meshes only the surface of the circuit metalization, em can analyze predominately planar circuits much faster than volume meshing techniques. Em does a full three dimensional analysis that includes both 3-D fields and 3-D currents. This is in contrast to 2.5-D analyses which, while including full 3-D fields, allow only 2-D currents. Thus, a 2.5-D analysis does not allow vias or any other vertical current.

1

Em User’s Manual Em analyzes 3-D structures embedded in planar multilayered dielectric on an underlying fixed grid. For this class of circuits, em can use the FFT (Fast Fourier Transform) analysis technique to efficiently calculate the electromagnetic coupling on and between each dielectric surface. This provides em with its several orders of magnitude of speed increase over volume meshing and other non-FFT based surface meshing techniques.

A Simple Outline of the Theory Em performs an electromagnetic analysis of a microstrip, stripline, coplanar waveguide, or any other 3-D planar circuit by solving for the current distribution on the circuit metalization using the Method of Moments.

Em analyzes planar structures inside a shielding box. Port connections are usually made at the box sidewalls.

Subsectioning the Circuit The analysis starts by subdividing the circuit metalization into small rectangular subsections. In an actual subsectioning, see , small subsections are used only where needed. Otherwise, larger subsections are used since the analysis time is

2

Chapter 1 Introduction directly related to the number of subsections. Triangular subsections, which can be larger without sacrificing accuracy, can be used to fill in the diagonal “staircase” at the user’s discretion.

Em calculates the tangential electric field on all subsections, given current on one subsection. This figure shows the actual subsectioning for an example circuit.

3

EM

Em evaluates the electric field everywhere due to the current in a single subsection. Em then repeats the calculation for every subsection in the circuit, one at a time. In so doing, em effectively calculates the “coupling” between each possible pair of subsections in the circuit.

Em User’s Manual

Zero Voltage Across a Conductor Each subsection generates an electric field everywhere on the surface of the substrate, but we know that the total tangential electric field must be zero on the surface of any lossless conductor. This is the boundary condition: no voltage allowed across a perfect conductor. The problem is solved by assuming current on all subsections simultaneously. Em adjusts these currents so that the total tangential electric field, which is the sum of all the individual electric fields just calculated, goes to zero everywhere that there is a conductor. The currents that do this form the current distribution on the metalization. Once we have the currents, the S-parameters (or Y- or Z-) follow immediately. If there is metalization loss, we modify the boundary condition. Rather than zero tangential electric field (zero voltage), we make the tangential electric field (the voltage on each subsection) proportional to the current in the subsection. The constant of proportionality is the metalization surface resistivity (in ohms/square). In short, Ohm’s Law.

The Input and Output Files Em executes in the following manner: 1

Reads an ASCII geometry file describing the circuit or a network file defining the circuit analysis.

2

Reads an ASCII “.an” file or accepts direct input from the job window specifying the analysis frequencies.

3

Number crunches.

4

Outputs one or more ASCII files with the results.

There are three types of files associated with an em geometry file analysis: a file which controls the analysis, a file which specifies the circuit geometry and circuit response file(s). More than one circuit response file may be specified for an em

4

Chapter 1 Introduction analysis. Which circuit response files are generated depends on the options selected. In addition, em can generate a current density file and/or a SPICE lumped element equivalent netlist file.

EM

There are two other types of files associated with an em network file analysis: a file which defines the network to be analyzed and a circuit response file. More than one output response file may be specified for a network file analysis. The network file may cite circuit response files, execute a geometry file analysis, or obtain external analysis frequency control from a control file. The file extensions used in em are listed in Table 1.

Table 1 Em File Extensions Extension

I/O Type

File Type

.geo

Input

Circuit geometry file (Required Extension)

.an

Input

Analysis control file (Required Extension)

.net

Input

Network File (Required Extension)

.d

Output

Circuit response file (De-embedding applied)

.nd

Output

Circuit response file (No de-embedding)

.pd

Output

Circuit response file (Precision de-embedded)

.pnd

Output

Circuit response file (Precision no de-embedding)

.sp (.s)

Output

Touchstone format frequency sorted response file. n = number of ports < 10. nn= number of ports > 10.

.jxy

Output

Current density file

.lc

Output

SPICE lumped element equivalent netlist file

.lct

Output

SPICE RLCG matrix sets file

.psp

Output

PSPICE output file

.csv

Output

Comma separated value data for Excel. See -f option.

5

Em User’s Manual The analysis control file determines the frequencies for analysis. Its other uses are detailed in Appendix II, “The Analysis Control File Format.” An example file is given in the next section. If a name other than “ctl.an” is selected, a name ending with “.an” is required. The analysis control file is not mandatory. Analysis frequencies may also be controlled through options selected in the em job window or by an internal command in the case of a network file. The circuit geometry file is generated with a geometry capture program such as xgeom. Geometry files can also be generated by the interface provided by Cadence or Barnard Microsystems and by the Sonnet programs gds, which translates GDSII Stream files, dxfgeo, which translates DXF files, and ebridge, which provides an interface between HP-EEsof’s Series IV or ADS layout module and Sonnet. The file name is required to have a “.geo” extension. The geometry file specifies the size of the box containing the circuit, the number of and material properties of the dielectric layers as well as a polygonal description of the metalization outline. Performing circuit analysis with the em circuit network capability requires a network file. This file is not required to run an electromagnetic analysis. This file consists of a header line, optional comment lines and several data blocks which define the circuit analysis to be performed. The format of the file is specified in Chapter 11, “The em Network File.” A network file name ending with “.net” is required. The circuit response file format can be selected in the Output File dialog box for compatibility with popular circuit analysis programs. The circuit response file contains automatically generated comment lines as well as any comment lines from the analysis control and geometry files. The various types of circuit response files are cited in Table 1. The Touchstone format frequency sorted response file provides Touchstone format S-Parameter data sorted by frequency. If the number of ports in the circuit is less than 10, em uses the .sp extension, where is the number of ports. For example, if a circuit has three ports, the file extension is .s3p. If the number of ports is 10 or greater, em uses the .s extension where is the number of

6

Chapter 1 Introduction ports. For example, if the number of ports is 14, the file extension is .s14. For more details about this file type, please see the -SNP command option in Appendix I, "The em Command Line" The output for a network file analysis has a file name ending with “.rsp.” The format of the data is determined by a command within the network file. EM

Em Applications Em is appropriate for a wide range of 3-D planar structures. The via capability allows analysis of airbridges, wire bonds, spiral inductors, wafer probes and internal ports as well as for simple grounding. It is appropriate to use em for: •

Evaluation of specific discontinuities or groups of interacting discontinuities to assist in the design of a 3-D planar circuit. Em provides ultra-precise S-parameters for discontinuities allowing designers to work with confidence. Em can also quickly synthesize an equivalent lumped model for discontinuities. The lumped model can be used directly in circuit theory programs.

•

Design validation. Using em for design validation effectively eliminates expensive design iterations (i.e., “tweak”, refabricate, etc.) of the passive, planar portion of a circuit. At present, because em makes no compromise in accuracy, performing an analysis of, say, an entire amplifier with a single electromagnetic analysis on a mid-range workstation or PC is difficult. However, validation of large portions of an amplifier design is reasonable. If a circuit is designed from the start with electromagnetic analysis in mind, much larger circuits can be done. For example, the analysis works best with tightly packed, rectangular circuits, designed on a common dimension grid.

•

Microwave package evaluation. It is important to assess how a circuit will operate in the package environment. Em analyzes a circuit inside a conducting box. If the box (acting as a dielectric loaded resonator) is resonant at a fre-

7

Em User’s Manual quency where the circuits still have gain, poor performance results. Em provides an analysis of a package prior to fabrication. Resonances can then be dealt with on the computer rather than on the test bench. •

Microstrip antennas. The “top” of em’s box can be effectively removed. While radiation is outside of em’s primary thrust, a wide variety of microstrip antennas and radiating discontinuities can be evaluated.

•

High speed digital interconnect. When an approximate model is not good enough, em can synthesize a SPICE lumped model including all delays and couplings. The lumped model is synthesized directly from electromagnetic data.

Em is not appropriate for doing an initial design. Rather, the faster circuit theory simulators (which do not typically include stray coupling) should be used for the first cut. Em can then enhance the simulator performance by providing custom, ultra-precise discontinuity data and by validating large portions of the final circuit, including all stray interactions. Em is designed to work with your existing CAE software. Since the output data is in Touchstone or Compact format (at your discretion), em provides a seamless interface to your CAE tool.

Em Origins The technique used in em was developed at Syracuse University in 1986 by Rautio and Harrington [75, 76, 78]. It was originally developed as an extension of an analysis of planar waveguide probes [81]. The technique expresses the fields inside the box as a sum of waveguide modes and is thus closely related to the spectral domain approach. The complete theory has been published in detail in peer reviewed journals. A full bibliography of relevant papers is presented in Appendix V.

8

Chapter 1 Introduction

Enhancements in Release 6.0 This section summarizes new capabilities and changes in 6.0. If you are not yet familiar with em, you may want to just skim this section, skipping any terms that are unfamiliar. If you are an experienced user, this section merits detailed reading.

New Features Multi-Frequency Caching: Multi-frequency caching (MFC) is a technique that can dramatically reduce the em computation time when analyzing large circuits over many frequencies without any loss in accuracy. MFC pre-computes frequency independent data and stores it on your hard disk. This data is later recalled while the simulation is performed. For more information about MultiFrequency Caching, see “Using Multi-Frequency Caching,” page 20. Em Control interface: The em Control interface provides an interactive interface to em. This interface consists of menus and dialog boxes which allow you to select run options and execute em analyses. You may save the settings of the control interface in a job file. Run List: The em interface allows you to edit a list of command lines which form a batch file for running analyses. The command lines are generated interactively from the settings in the various em dialog boxes. This list is saved as part of the job file. The job file contains the settings of the interface and must have a “.job” extension. For more information about the Run List, please see “Running an em Analysis,” page 272. Direct Frequency Control: The em interface allows you to control analysis frequencies directly through use of the Simple Sweep and Complex Sweep options in the job window. Em analyses do not require an analysis control file, although it is still possible to use one to control the analysis frequencies. For information about specifying frequency control information, please see “Specifying Frequency Control,” page 248 and page 260.

9

EM

The major changes incorporated in Version 6.0 is the new UNIX interface which is compatible with WindowsNT 4.0 and Windows95 and the combining of em and emgen into one program. This will allow you to run both electromagnetic and circuit network analyses from an interactive window whose settings can be saved in a file. Listed below are the new features and changes introduced in Version 6.0.

Em User’s Manual Output Data Format: Em outputs Magnitude/dB data as well as Magnitude/ Angle and Real/Imaginary data. See “Selecting Additional Options,” page 252. Invoking emgraph: You now have the ability to invoke the Sonnet plotting program, emgraph, directly from the em interface to observe your analysis results. For information about invoking emgraph, please see “Invoking emgraph to Plot Response Data,” page 278. For details about emgraph’s operation, refer to the Emgraph User’s Manual in your 6.0 manual set. Invoking xgeom: You now have the ability to invoke the Sonnet circuit geometry input program, xgeom, directly from the em interface. This provides efficient access to your circuit to make modifications while running em analyses. For information about invoking xgeom, please see “Editing a Geometry File,” page 248. For details about xgeom’s operation, refer to the Xgeom User’s Manual in your 6.0 manual set.

Changes The combination of em and emgen: Em and emgen have been combined into one program. Both are accessible through a windows interface activated by selecting Em Control from the Sonnet menu in the Windows desktop Start menu. In this documentation, emgen is referred to as the em circuit network capability and is accessed by running a Network file analysis. The circuit network capability must be purchased separately as an add-on to the em program. Intelligent Frequency Control: This feature, previously only available in emgen, automatically determines where to place frequency points for an analysis. Intelligent frequency control is now available with em. For details, see “Entering Intelligent Frequency Controls,” page 267.

Em User’s Manual Layout Chapter 2 through Chapter 7 discuss important design considerations while using em. Many of the discussions contain examples to illustrate the point under consideration. Chapter 8 through Chapter 11 discuss the em circuit network capability. This is followed by Chapter 12 through Chapter 23 in which advanced design topics are discussed. The last chapters, Chapter 24 through Chapter 26 provide a user’s guide to the em interactive interface, em Control.

10

Chapter 1 Introduction

Describing Menu Bar Accesses In this manual, we describe accessing the menu bar of em Control using a “pointer” description to illustrate selecting the desired menu buttons.

Invoking Sonnet You use the Sonnet task bar, shown below, to access all the modules in the em Suite. Opening the Sonnet task bar, for both Windows and UNIX systems is detailed below.

11

EM

For example, View ⇒ Open Graph means to move the cursor to View on the menu bar, press and hold down the left mouse button, drag the cursor down the menu which appears until Open Graph is highlighted. Release the mouse button. Emgraph is invoked to display the response data of the file being analyzed.

Em User’s Manual

UNIX 1

Open a terminal. If you do not know how to do this, please see your system administrator.

2

Enter “sonnet” at the prompt. The Sonnet task bar appears on your display.

12

Chapter 1 Introduction

Windows 1

Select Start ⇒ Programs ⇒ Sonnet ⇒ Sonnet from the Windows desktop Start menu.

EM

The Sonnet task bar appears on your display.

13

Em User’s Manual Once the Sonnet task bar is open, for UNIX or Windows systems, clicking on any given button opens the appropriate module. The table below shows which modules are invoked by each button.

Table 2 Sonnet Task Bar Buttons Button

14

Button Name

Sonnet Program

Edit Circuit

xgeom

Analyze Circuit

em Control

View Response

emgraph

View Current

emvu

View Far Field

patvu

Online Manuals

Adobe Acrobat

Chapter 1 Introduction The translation programs, dxfgeo and gds, are accessed through the Sonnet task bar main menu, as shown below. Select Convert Dxf for dxfgeo and select Convert Gds for gds.

EM

For details on each program, please refer to the appropriate user’s manual.

15

Em User’s Manual

16

Chapter 2 Important Options and Concepts

EM

Chapter 2

Important Options and Concepts

In the previous chapter, we described the basics of running em: what input files are required, how you set up and save a job file, and what the output results mean. In this chapter we discuss the following techniques which are frequently used to obtain more accurate results and/or shorter analysis times: •

De-embedding the port discontinuity

•

Using multi-frequency caching (MFC)

•

Using symmetry

•

Invoking single precision

•

Adjusting the subsectioning

•

Removing parallel subsections.

17

Em User’s Manual

De-embedding the Port Discontinuity Each port in a circuit analyzed by em introduces a discontinuity into the analysis results. In addition, any feed transmission lines that might be present introduce phase shift, and possibly, impedance mismatch and loss. Depending upon the nature of your analysis, this may or may not be desirable. De-embedding is the process by which the port discontinuity and transmission line effects are removed from the analysis results. The em de-embedding algorithm is described in detail in Chapter 6 and Chapter 7. To summarize, this algorithm performs the following analysis steps when enabled:

18

1

Calculates port discontinuities.

2

Removes effects of port discontinuities from analysis results.

3

Optionally shifts reference planes (removes effects of feed or port transmission lines from analysis results).

4

Calculates transmission line parameters Eeff and Z0.

Chapter 2 Important Options and Concepts Run em with de-embedding enabled whenever you do not want to include the effects of port discontinuities in your analysis results. In fact, the De-embed option is selected by default whenever a new job file is opened. To enable de-embedding, set the De-embed check box in the em job window to on, or specify an output file with a “.d” file extension in the Output Files dialog box.

EM The circuit “steps.geo” is shown above. The de-embedded and non-de-embedded analysis results are shown below. The circuit is available in the Sonnet example files.

De-embedded 50 Ohm S-Params. Mag/Ang. Touchstone Format (S11 S21 S12 S22): 10.0000000 0.513865 -141.5 0.857871 -53.90 0.857871 -53.90 0.513865 -146.3 P1 F=10.000 Eeff=(6.6763 0.0000) Z0=(44.73370 0.000000) R=0.00000 C=0.117753 P2 F=10.000 Eeff=(6.2013 0.0000) Z0=(73.08549 0.000000) R=0.00000 C=0.044139

The de-embedded results of a “steps.geo” analysis.

Primary 50 Ohm S-Params. Mag/Ang. Touchstone Format (S11 S21 S12 S22): 10.0000000 0.515668 152.64 0.856789 -163.5 0.856789 -163.5 0.515668 60.400

The non-de-embedded results of a “steps.geo” analysis.

19

Em User’s Manual

Using Multi-Frequency Caching Multi-frequency caching (MFC) is a technique that can dramatically reduce the em computation time when analyzing large circuits over many frequencies without any loss in accuracy. This technique is enabled by checking the “MultiFrequency Caching” run option in the em Control job window. MFC precomputes frequency independent data and stores it on your hard disk. This data is later recalled while the simulation is performed. By storing and reusing the data in this manner, MFC significantly reduces the required matrix fill time. In standard em analyses, matrix fill is completely recomputed at each frequency. Thus, if you are analyzing at a large number of frequencies (greater than four), and if matrix fill time is appreciable for your circuit, enabling MFC will speed up your analysis. Since an analysis executes in either the same or less time, this option should always be used except when cache disk space is unavailable. The MFC technique does require some time to pre-compute the frequency independent data. Because of this pre-computation time, MFC does not provide any advantage when you simulate at four or fewer frequencies. If you set up an analysis to run at four or fewer frequencies, and you check the “Multi-Frequency Caching” run option, em will automatically disable MFC for the simulation. Circuits for which the matrix fill time is large will benefit the most from MFC. Characteristics of such circuits include: • • • • •

Circuits with a large number of cells in the X and Y dimensions. Circuits with a large number of metallization layers. Circuits with vias. Circuits with polygons using diagonal fill. Circuits with polygons for which XMIN/YMIN is greater than 1.

It is important to remember that MFC only reduces the matrix fill time. MFC has no affect on the matrix solve time. MFC saves time by writing data to your hard disk and reusing that data later. Therefore, you need to have hard disk space available in order to use MFC. When em begins a simulation with MFC enabled, it displays the amount of disk space 20

Chapter 2 Important Options and Concepts required, and where the data will be stored. In general, if you have disk space available, it is recommended that you always run with MFC enabled. MFC will not slow the simulation down.

When MFC is enabled, em displays a message similar to the following: Multi-frequency caching enabled below GHz. This frequency is the maximum frequency at which the pre-computed data is valid and is used as a cutoff frequency. Above the cutoff frequency, it is not usually possible to compute frequency independent data. If your frequency analysis band is entirely below , em runs with MFC enabled over your entire band. If your band is above , em disables MFC for the entire simulation. If your band has frequencies both above and below , em automatically uses MFC up to and disables MFC for all frequencies greater than . It is possible to disable the automatic cutoff. For details, see the -F option in the Appendix I, "The em Command Line".

Using Symmetry The microstrip circuit in “steps.geo” (see the circuit on page 19) is symmetric about the horizontal center line. Em can take advantage of this, provided all ports are also on the center line. This second condition is important. If there is a port off the center line, em creates the port’s image on the other side of the center line and shorts the two together, a result usually not desired.

21

EM

If you wish to change the location where em will store the data, select File ⇒ Preferences from the em Control main menu, and enter a new disk location in the Cache Directory dialog box. You may also set a Cache Limit in the same preferences window. This limit sets the maximum amount of disk space that you will allow em to use for MFC. If a particular simulation exceeds this limit, em will automatically disable MFC for that simulation. For details on how to set the location and limit the amount of memory, see “Setting Multi-Frequency Caching Parameters,” page 283.

Em User’s Manual If you want to short two symmetrical ports together you should identically number the ports above and below the line of symmetry. Ports with the same number are electrically connected together. This may be useful for designers interested in the even-mode of a pair of coupled lines, as shown below.

This circuit cannot be run using symmetry.

This circuit may be run with symmetry, because the ports above and below the line of symmetry are the same number.

For the "steps.geo" example file, your em analysis may take advantage of the symmetry setting. Run xgeom and modify “steps.geo” using Parameters ⇒ Box to open the Box Parameters dialog box. When you click on the Symmetry checkbox, it toggles from Off to On and a dashed center line is drawn on the circuit. Visually check to make sure the circuit really is centered on the center line. If not, move the circuit until it is centered. In electromagnetic terms, the symmetry option places a magnetic wall along the center line. In terms of performance, the number of subsections is reduced by nearly half, reducing the size of the matrix to one quarter of its original memory requirement, and decreasing matrix solve time by about a factor of eight. Thus, symmetry is a powerful tool in reducing both analysis time and hardware requirements. You will always get identical analysis results when using symmetry. When finished, save the file under the new name “steps_sy.geo”. You may want to add a comment or two indicating the modification as you store the file. You may also obtain a copy of this file from the Sonnet examples by using Sonnet ⇒ Copy Examples. If you are not familiar with Sonnet ⇒ Copy Examples, see “Obtaining the Example Files,” page 7 in the Sonnet Tutorial. Run an analysis in the following manner:

22

Chapter 2 Important Options and Concepts Invoke the em program by selecting Em Control from the Sonnet task bar.

2

Click on the Geometry File radio button under File Type to select the analysis of a geometry file.

3

Check to make sure the default directory is correct. Then, enter “steps_sy.geo” in the Geometry File text entry box.

4

Click on the Simple Sweep radio button and enter a value of 10.0 in the Start text entry box.

5

The Verbose and De-embed options are already on; therefore, click on the Run command button to execute the analysis.

The output to the screen should be the same as the previous section, with one possible exception. If a non-zero user time was displayed in the previous example, the time should be about half that in this example. Note that the number of subsections is cut by almost half. While not particularly important here, the improvement in speed and memory use is very important for large circuits.

Invoking Single Precision Em normally works with double precision. In many cases, a single precision matrix reduction is all that is needed. Keep in mind that the matrix elements are still calculated in double precision because of potential numerical difficulties. Once calculated, they are then truncated and stored in single precision format.To invoke a double precision matrix using the file “steps_sy.geo”, do the following: 1

Invoke the em program.

2

Click on the Geometry File radio button under File Type to select the analysis of a geometry file.

3

Enter “step_sy.geo” in the Geometry File text entry box.

4

Click on the Simple Sweep radio button and enter a value of 10.0 in the Start text entry box.

5

Click on the Memory Save checkbox to turn the option off and use a double precision matrix.

23

EM

1

Em User’s Manual 6

The Verbose and De-embed options are already on; therefore, click on the Run command button to execute the analysis.

The results are virtually identical to the previous results. By invoking single precision, you cut the memory required by a factor of two. This has a significant effect if your problem is large enough to cause em to swap out to disk. For these big problems, use of single precision can result in an order of magnitude increase in speed. On some machines, the floating point operation speed is also increased because single precision calculations can be done faster than double precision. The Sun SPARCstation is such a machine. Unfortunately, compilers on certain other computers perform single precision arithmetic by first converting to double precision, performing the operation and then converting back to single precision. If such a machine is not memory swapping in double precision, there is no increase in speed in single precision. It may even result in a slightly slower analysis. In general, use the memory saver option on large circuits. It is not needed on small circuits. On occasion, the memory saver option can result in slightly different results. At extremely low frequencies, when the subsections size is on the order of 0.0001 wavelength or smaller, single precision may not be enough to allow the matrix to be accurately inverted. In this case, em issues a warning as corrupted results may be generated. If you are using the memory save option at very low frequencies and you notice non-physical results, eliminating the option and re-analyzing may remove the problem.

Adjusting the Subsectioning The most direct way to reduce analysis time is to reduce the number of subsections. A reduction in number of subsections by half can result in one, or even two, orders of magnitude increase in speed.

24

Chapter 2 Important Options and Concepts The easiest way to reduce subsection count is to reduce the amount of metal to be analyzed. Any metal which is unlikely to carry significant current should be removed. Distant ground plane regions in coplanar waveguide, for example. Also, long, unneeded lengths of connecting transmission line provide no additional information in an analysis and should be removed.

Be careful not to reject viable options because of the dimensions being used. For example, if you want to analyze a line 7.0 microns wide, the obvious potential cell dimensions are 1 micron (for a seven cell wide line) and 7 microns (for a one cell wide line). There are other options, also. For example, the cell size can be specified to be 7/3 = 2.33333 microns resulting in a three cell wide line. Next, use your engineering judgement in determining the dimensions of a structure. If a 7.1 micron wide line is called for, and you use a 7 micron cell size, the residual error introduced by making the line 7 microns wide is probably less than the manufacturing tolerances. However, there are circuits which have very fine geometries combined with very large overall structures. The fine geometry determines the cell size, requiring the overall circuit to use a large number of cells. To address this problem, you may specify a minimum and maximum subsection size, in terms of cells, for each polygon. You use the parameters X Min, Y Min, X Max and Y Max to do this. Chapter 3, “Subsectioning,” discusses this in more detail. A third way to reduce the number of subsections is to cut the circuit into pieces and analyze each piece separately. For example, instead of analyzing an entire bandpass filter, analyze half of it. Then, if both halves are identical, just use the em circuit network capability, or a circuit theory program to cascade the half filter analysis with itself.

25

EM

A second way to reduce subsection count is to simply use a larger cell size. It can be surprising how little a larger cell size changes results on many circuits.

Em User’s Manual Most filters, and other structures as well, can be broken down into a number of small circuits which can be easily analyzed. In fact, a pure circuit theory analysis takes this “breaking down” to the limit. An example of this approach is discussed in the beginning of Chapter 8.

Removing Parallel Subsections A transmission line connected to a port is subsectioned with both x and y directed subsections included in order to determine both x and y directed currents. However, in most cases, a majority of the current flows in the longitudinal direction and there is very little current flowing in the transverse direction. For example, in a horizontal, x directed transmission line, most of the current is flowing in the x direction. There is very little transverse, y directed, current flowing. Without parallel subsection removal, all of the y directed subsections are still included in the analysis, though their current is almost zero. This wastes subsections, memory and time. To remove the transverse subsections, in xgeom, select Parameters ⇒ Parallel Subsections. Then select a side and either specify a length in the text entry box or use the mouse to specify the length for which parallel subsections are to be excluded. The transverse subsections are parallel to the side of the box being referenced. This option should be used carefully. Here are a few rules:

26

•

Never remove subsections for lengths greater than 1/4 to 1/2 wavelength.

•

Remove subsections only from transmission lines with very small transverse currents such as feed lines. For example, do not remove subsections from discontinuities, lines that change in width or regions where structures approach one another.

•

Since this option affects all xgeom drawing levels, check all levels for possible problems.

Chapter 3 Subsectioning

EM

Chapter 3

Subsectioning

The Sonnet subsectioning is based on a uniform mesh indicated by the small dots in the xgeom screen. The small dots are placed at the corners of a “cell”. One or more cells are automatically combined together to create subsections. Cells may be square or rectangular (any aspect ratio), but must be the same over your entire circuit. The cell size is specified in xgeom in the Box Parameters dialog box which is opened by selecting Parameters ⇒ Box. The analysis solves for the current on each subsection. Since multiple cells are combined together into a single subsection, the number of subsections is usually considerably smaller than the number of cells. This is important because the analysis solves an N x N matrix where N is the number of subsections. A small reduction in the value of N results in a large reduction in analysis time and memory. Care must be taken in combining the cells into subsections so that accuracy is not sacrificed. Em automatically places small subsections in critical areas where current density is changing rapidly, but allows larger subsections in less critical areas, where current density is smooth or changing slowly.

27

Em User’s Manual However, in some cases you may wish to modify the automatic algorithm because you want a faster, less accurate solution, or a slower, more accurate solution, than is provided by the automatic algorithm. Also, in some cases, you may have knowledge about your circuit that the software does not. For example, you may know that there is very little current on a certain area of your metal. Or you may have chosen a small cell size because you have a small dimension in your circuit, but do not need the accuracy of a small cell size in larger structures within your circuit. In these cases, you can change the method by which em combines cells into subsections. This chapter explains how em combines cells into subsections and how you can control this process to obtain an analysis time or the level of accuracy you require. There is also a discussion of selecting the cell size and how that may affect the em analysis. The following methods may be used to control the way in which em combines cells into subsections: •

Selecting cell size.

•

Setting the subs/lambda parameter.

•

Defining the subsectioning frequency.

•

Changing the subsectioning of a polygon.

•

Using the edge mesh option.

Selecting Cell Size As you know, em subdivides the circuit into subsections which are made up of “cells,” the building block in xgeom. The following discussion describes how to select a cell size.

TIP Select a cell size that is smaller than 1/20 of a wavelength.

28

Chapter 3 Subsectioning Before calculating a cell size, is it important to calculate the wavelength at your highest frequency of analysis. An exact number is not important. If you know the approximate effective dielectric constant of your circuit, use this in the wavelength calculation; otherwise, use the highest dielectric constant in your structure.

TIP When possible, round off dimensions of your circuit so that they have a larger common multiple.

Since your circuit geometry is snapped to the nearest cell, you must find a cell size such that all of the dimensions of the circuit are a multiple of this cell size. For example, if your circuit has dimensions of 30 microns, 40 microns and 60 microns, possible cell sizes are 10 microns, 5 microns, 2.5 microns, 2 microns, etc. Large cell sizes result in more efficient analyses, so choosing 10 microns is probably best.

TIP Calculate the X cell size and the Y cell size independently.

The X cell size and Y cell size do not have to be the same number. Calculate the X cell size based on just your dimensions in the X direction, and your Y cell size based on just your dimensions in the Y direction.

29

EM

Most circuits require that your cell size be smaller than 1/20 of a wavelength. Larger cell sizes usually result in unacceptable errors due to incorrect modeling of the distributed effects across the cell. Cell sizes smaller than λ/20 may increase the accuracy slightly but usually increases the total number of subsections, which increases the analysis time and memory requirements.

Em User’s Manual For example, if you have a spiral inductor with widths of 10 microns and spacings of 11 microns, modify the 11 microns to 10 microns. You may now use a cell size of 2, 2.5, 3.333, 5 or 10 microns instead of 1 micron, speeding up the analysis by several orders of magnitude with little impact on circuit performance. This concept is illustrated below.

Circuit 1: Requires 80 cells Runs slow, uses more memory More accurate

8 µm

3 µm 1 µm cell size 1 cell

1 cell

Circuit 2: Requires only 6 cells Runs fast, uses less memory Less accurate

8

4 µm 4 µm cell size

Circuit 1 takes more time and memory to analyze than circuit 2 even though they have approximately the same amount of metal. This is because the dimensions in circuit 2 are divisible by 4, so a 4 um cell size may be used. Circuit 1 requires a 1 um cell size. Think about the sensitivity of your circuit to these dimensions and

30

Chapter 3 Subsectioning your fabrication tolerances. If your circuit is not sensitive to a 1 micron change or can be made with only a +/- 1 micron tolerance, you can easily round off the 3 micron dimension in circuit 1 to the 4 micron dimension in circuit 2.

Setting the Maximum Subsection Size Parameter

The default of 20 subsections/λ is fine for most work. This means that the maximum size of a subsection is 18 degrees at the highest frequency of analysis. Increasing this number decreases the maximum subsection size until the limit of 1 subsection = 1 cell is reached. You might want to increase this parameter for a more accurate solution. For example, changing it from 20 to 40 decreases the size of the largest subsections by a factor of 2, resulting in a more accurate (but slower) solution. Keep in mind that using smaller subsections in non-critical areas may have very little effect on the accuracy of the analysis while increasing analysis time. Another reason for using this parameter is when you require extremely smooth current distributions using emvu. With the default value of 20, large interior subsections may make the current distribution look “choppy”. Setting this value to a large number forces all subsections to be only 1 cell wide, providing smooth current distribution. Again, analysis time is impacted significantly. The Max. Subsection Size parameter is specified in xgeom in the Box Parameters dialog box which is opened by selecting Parameters ⇒ Box.

Defining the Subsectioning Frequency The subsectioning parameter Max. Subsection Size defined as subsections per wavelength normally uses the highest analysis frequency to determine the wavelength. However, this may be changed by using the keyword “FMAX” 31

EM

The parameter Max. Subsection Size allows the specification of a maximum subsection size, in terms of subsections per wavelength, where the wavelength is approximated at the beginning of the analysis. The highest analysis frequency is used in the calculation of the wavelength.

Em User’s Manual followed by a frequency in the ctl.an file or by entering a frequency in the Subsectioning Frequency text entry box in the Analysis Control dialog box when setting up frequency control information in em. That frequency is now used for the wavelength determination instead of the highest frequency of analysis. Thus, the same subsectioning can be used for several analyses which differ in the highest frequency being analyzed. NOTE:

The Subsectioning Frequency must be greater than the highest analysis frequency to be used since em uses the higher value of the two.

Changing the Subsectioning of a Polygon Em allows you to control how cells are combined into subsections for each polygon. This is done using the parameters “X Min”, “Y Min”, “X Max” and “Y Max”. These parameters may be changed for each polygon, allowing you to have coarser resolution for some polygons and finer resolution for others. See “The Metalization Attributes dialog box,” page 119 in the Xgeom User’s Manual, for information on how to change these parameters. Before discussing how to make use of these parameters, we need to first understand em’s automatic subsectioning for a polygon when the parameters are set to their default settings.

Default Subsectioning of a Polygon The default values for the subsectioning parameters are X Min = 1, Y Min = 1, X Max = 100 and Y Max = 100. These numbers specify the smallest and largest allowed dimensions of the subsections in a polygon. With X Min = 1, the smallest subsection in the X dimension is one cell. With X Max = 100, subsections are not allowed to go over 100 cells in length. shows how these default subsectioning parameters are used. Notice in the corner, the subsection size is just one cell. The current density changes most rapidly here, thus, the smallest possible subsection size is used.

32

Chapter 3 Subsectioning

Subsection size is 1 cell by 1 cell on corner Subsection size is 1 cell wide along edge EM

Subsections are wide and long away from edge

Cell Size = A portion of circuit metal showing how em combines cells into subsections. In this case the subsectioning parameters are set to their default values: X Min = 1, Y Min = 1, X Max = 100 and Y Max = 100.

As we go away from the corner, along the edge, the subsections become longer. For example, the next subsection is two cells long, the next one is four cells long, etc. If the edge is long enough, the subsection length increases until it reaches X Max (100) cells, or the maximum subsection size parameter, whichever comes first, and then remains at that length until it gets close to another corner, discontinuity, etc. Notice, however, that no matter how long the edge subsection is, it is always one cell wide. This is because the current density changes very rapidly as we move from the edge toward the interior of the metal (this is called the edge singularity).

33

Em User’s Manual In order to allow an accurate representation of the very high edge current, the edge subsections are allowed to be only one cell wide. However, the current density becomes smooth as we approach the interior of the metal. Thus, wider subsections are allowed there. As before, the width goes from one cell to two cells, then four, etc.

X Min and Y Min for a Manhattan Polygon On occasion, you may wish to change the default subsectioning for a given polygon. You can do this using the subsectioning parameters X Min, Y Min, X Max and Y Max. X Min and Y Min affect Manhattan polygons differently than non-Manhattan polygons, where a Manhattan polygon is one that has no diagonal edges. For Manhattan polygons, X Min and Y Min set the size of the edge subsections. By default, X Min and Y Min are 1. This means the edge subsections are 1 cell wide. When X Min is set to 2, the subsections along vertical edges are now 2 cells wide in the X direction (see the figure on page 35). This reduces the number of subsections and reduces the matrix size for a faster analysis. However, accuracy may also be reduced due to the coarser modeling of the current density near the structure edge or a discontinuity.

34

Chapter 3 Subsectioning

1 Cell Wide (Y Min = 1)

{

{

2 Cells Wide (X Min = 2)

4 Cells

8 Cells

EM

2 Cells

4 Cells

Y

Cell Size =

X

A portion of circuit metal showing how em combines cells into subsections for Manhattan polygons when X Min = 2 and Y Min = 1.

If X min or Y min are greater than your polygon size, em uses subsections as large as possible to fill the polygon. NOTE:

The subsection parameters, X Min, Y Min, X Max and Y Max are specified in cells (not mils, mm, microns, etc.). For example, X Min = 5 means that the minimum subsection size is 5 cells.

Although the X Min and Y Min parameters are very useful options, it is not a substitute for using a larger cell size. For example, a circuit with a cell size of 10 microns by 10 microns with X Min = 1 and Y Min =1 runs faster than the same circuit with a cell size of 5 microns by 5 microns with X Min = 2 and Y Min = 2. 35

Em User’s Manual Even though the total number of subsections for each circuit may be the same, em must spend extra time calculating the value for each subsection for the circuit with the smaller cell size.

X Min and Y Min for a Non-Manhattan Polygon For non-Manhattan polygons, when X Min is increased, the smallest subsection size is still one cell. All edge subsections are still one cell wide. This is so the diagonal edges can be represented. However, the next subsections in become longer and/or wider more quickly than before (see the figure below). This reduces the number of subsections and reduces the matrix size for a faster analysis.

1 Cell by 1 Cell on corner

4 Cells (2 * X Min)

Y

Cell Size =

X

A portion of circuit metal showing how em combines cells into subsections for non-Manhattan polygons, with X Min = 2 and Y Min = 1. Non-Manhattan polygons always have 1 cell wide edge subsections. The next subsection in is 2*X Min = 4 cells.

36

Chapter 3 Subsectioning

Using X Max and Y Max for any Polygon As mentioned before, the Max. Subsection Size parameter can be used to control the maximum subsection size for your circuit. You may also control the maximum subsection size of individual polygons by using the X Max and Y Max parameters.

NOTE:

If the maximum subsection size specified by X Max or Y Max is larger than the size calculated by the Max. Subsection Size parameter, the Max. Subsection Size parameter takes priority.

Using the Edge Mesh Option When using the Edge Mesh option, all Manhattan polygons are treated as if they were non-Manhattan polygons. In other words, the edge subsections are always one cell wide regardless of X Min or Y Min. When used in conjunction with large

37

EM

For example, if X Max and Y Max are decreased to 1, then all subsections will be one cell. This results in a much larger number of subsections and a very large matrix which are the cause of increased analysis time. Thus, this should be done only on small to medium size circuits where high accuracy is required.

Em User’s Manual X Min or Y Min values, this option can be very useful in reducing the number of subsections but still maintaining the edge singularity, as shown below. This is very often a good compromise between accuracy and speed.

A Manhattan polygon showing how em combines cells into subsections using the Edge Mesh option in conjunction with a large X Min and Y Min.

In the case pictured above, X Min and Y Min are set to be very large, and the frequency is low enough so that the Max. Subsection size parameter corresponds to a subsection size that is larger than the polygon.

38

Chapter 3 Subsectioning To invoke the edge mesh option, click on the Edge Mesh checkbox in the Additional Options dialog box.

39

EM

The Additional Options dialog box showing the Edge Mesh feature invoked (box checked).

Em User’s Manual

40

Chapter 4 Metalization and Dielectric Loss

EM

Chapter 4

Metalization and Dielectric Loss

Metalization Loss Metalization loss is specified in xgeom in the Metal Types dialog box which is opened by selecting Parameters ⇒ Metal Types. Losses may be assigned to circuit metal, top cover and ground plane. Sidewalls are always assumed to be perfect conductors. A common misconception is that only one type of metalization is allowed on any given level. In fact, different metalizations (i.e., different losses) can be mixed together on any and all levels. It is possible to have, say, a thin film resistor next to a gold trace on the same level.

41

Em User’s Manual

Surface Resistance The metalization loss is based on surface resistance. Two numbers are required. The first parameter, RDC, determines loss at low frequency (where the conductor is much thinner than the skin depth). Surprisingly, electromagnetic analyses often predict zero loss at low frequency because they assume RDC is zero. RDC is useful for thin film resistors as well as low frequency metal loss. The second parameter is the skin effect coefficient, RRF. Em multiplies this number by the square root of the frequency (in Hertz) to yield the ohms/square value at high frequency. See the “Parameters - Metal types,” page 138 in the Xgeom User’s Manual, for examples of RDC and RRF. The equations for RDC and RRF are repeated here:

R DC = 1 ⁄ ( σt )

R RF = Skin effect coefficient =

( πµ ) ⁄ σ

where σ is the bulk conductivity in mhos/meter, t is the metalization thickness in meters, and µ = 4π x 10-7 H/m. Typical values for RDC and RRF are 0.004 and 3e7. If you start getting very strange loss results, check RRF, paying special attention to the exponent. Em also properly models the transition between electrically thin (low frequency) and electrically thick (high frequency) conductors. The transition frequency from RDC to RRF is the square of RDC/RRF. At this frequency, and a relatively narrow band around it, both coefficients are important. Em’s modeling of loss is very precise if accurate values of RDC and RRF are used. Expect the final result to be just as accurate as the values of RDC and RRF specified.

42

Chapter 4 Metalization and Dielectric Loss If an extremely precise evaluation of loss is needed, build (or find) and measure a simple structure of the desired metalization. Then adjust the values of RDC (low frequency) and RRF (high frequency) until the calculated loss matches the measured value. Then you can use these values for any circuit which uses the same metalization. You are now effectively using measured values for R DC and RRF. The typical values given above work well as a first approximation.

Another aspect of loss is that the surface impedance of a good conductor has an imaginary part which is equal to the real part. This reactive surface impedance is physically due to the increased surface inductance caused by the current being confined closer to the surface of the conductor. This surface reactance is included in RRF. The effect is small, but potentially significant in certain cases. Some electromagnetic analyses use a “perturbational” approach for loss. This means that they assume the current flowing everywhere is the same as the lossless case. This approximation works for low loss (good conductor). However for thin film resistors (high loss), the lossless (short circuit) current is not the same as the lossy current and a perturbational approach fails. Em’s loss analysis is not perturbational. It works just as well for 100 ohms/square resistor material as it does for 0.004 ohms/square good conductor.

Surface Reactance Surface reactance, Xdc, is specified, in ohms/square, in xgeom in the Metal Types dialog box accessed by selecting Parameters ⇒ Metal Types. Em uses the same reactance at all frequencies.

43

EM

The alternative to using measured loss value to determine RRF is to develop a physical model. Such a model would have parameters such as bulk conductivity, bottom side surface roughness, top side surface roughness, metal porosity and edge roughness as at least a few of its parameters. Developing such a model accurately is expected to be very difficult and is beyond the scope of this analysis software. Even if developed, such a model only converts the problem to that of measuring the physical parameters needed for the model. Measuring the required physical parameters (like metal porosity) is likely to be even more difficult than measuring the loss and determining RRF in the first place.

Em User’s Manual Until recently, the only surface resistivities of practical interest were pure real, i. e., pure loss. With the growing application of superconductors in high frequency work, surface reactance reaches significant levels. A superconductive effect known as “kinetic inductance” slows the velocity of the electrons with no loss of energy. This can be modeled as a surface inductance. The effect of surface inductance is to make εeff larger, or the velocity of propagation slower. For normal conductors, εeff can never be larger than εrel. In a superconductor, this is no longer true. This unusual effect becomes significant for very thin substrates. Surface inductance, Ls, is specified, in xgeom in the Metal Types dialog box accessed by selecting Parameters ⇒ Metal Types. This parameter takes into account the surface reactance at higher frequencies. There are three recommended approaches to obtaining a value for Ls. A first order approximation is to assume the metal is a perfect conductor.

R DC = 0

R rf = 0

Ls = 0

This model works well for moderate frequencies (less then 150 GHz) and moderate circuit dimensions which are much greater than the London depth of penetration. The second approach is a model which is still valid at moderate frequencies, but includes effects due to kinetic inductance. The kinetic inductance is a function of temperature and can be approximated in the following manner:

R DC = 0

44

R rf = 0

Ls = µ0 λL ( T )

Chapter 4 Metalization and Dielectric Loss where –7

µ 0 = 4π ( 10 ) H/m 4

λ L ( T ) = λ0 ⁄ ( 1 – ( T ⁄ T c ) )

London depth of penetration at temp. EM

λ 0 = London depth at T = 0 meters T c = Critical (Transition) Temperature in degrees Kelvin The third model should be used to account for high frequency effects or effects due to small circuit dimensions. In these cases, the surface resistances proportionality to ω2 begins to dominate and the following model is suggested. The resistivity is a function of frequency-squared, and Sonnet presently does not have a method to do this. Therefore, if you are analyzing over a broad band, you need to have a separate geo file for each frequency, using the following equations.1

R rf = 0 1 2 2 R DC = --- ω µ 0 ( λ L ( T ) ) 3 σ N ( η n ⁄ ( η n + η s ) ) 2 Ls = µ0 λL( T ) where

ω = 2πf radians/sec σ N = Conductivity of the superconductor in its normal state (Mhos/m3) η n = Normal state carrier density (1/m3) η s = Superconducting state carrier density (1/m3) µ 0 and

λL ( T )

are as defined above

1. Shen, Z. Y., “High-Temperature Superconducting Microwave Circuits,” Boston, 1994, Artech House. 45

Em User’s Manual

Dielectric Layer Parameters You can set the dielectric constant and loss of a dielectric layer by changing the following parameters in xgeom by selecting Parameters ⇒ Dielectric Layers. •

Erel: The relative dielectric constant (εr). The ratio (ε’/εo’), where ε’ is the real part of the permittivity of the dielectric layer material, and εo’ is the permittivity of free space. The ratio is dimensionless.

•

Dielectric Loss Tan: The dielectric loss tangent. The ratio (ε’’/ε’), where ε = ε’ - jε’’, and ε is the complex permittivity of the dielectric layer material. The ratio is dimensionless.

•

Diel. Cond: The dielectric conductivity, σ, where σ is the bulk conductivity in siemens per meter.

•

Mrel: The relative magnetic permeability (µr) of the dielectric layer material.

• Magnetic Loss Tan: The magnetic loss tangent of the dielectric layer material. •

Z-Partitioning: The z-partitioning parameter for the dielectric layer. Note that the number of Z partitions only affects dielectric bricks. Changing this value for a particular layer will have absolutely no affect on the analysis if there are no bricks on the layer. If there are multiple bricks on the layer, the Z subsectioning for all of those bricks will be identical. The more partitions (better resolution) used in the Z-dimension, the more accurate the analysis; however, analysis time and memory requirements also increase.

Em uses the above parameters to calculate the total effective tanδ for the dielectric material as follows:

Diel Cond )tan δ = ( Loss Tan ) + (----------------------------ω ( Erel )ε o'

46

Chapter 4 Metalization and Dielectric Loss Here, ω is the radian frequency (ω = 2πf, where f is frequency in hertz). Note that tanδ has both a frequency-dependent term and a frequency-independent term. The above equation for tanδ can also be expressed in terms of conductivities as follows:

Both equations are equivalent. Each describes how em uses the input dielectric parameters to compute loss in the dielectric material. See “Parameters - Dielectric Layers,” page 133 of the Xgeom User’s Manual, for information on setting these parameters.

47

EM

( Total Effective Cond ) = ( Loss Tan )ω ( Erel )ε o' + ( Diel Cond )

Em User’s Manual

48

Chapter 5 Ports

EM

Chapter 5

Ports

Port Types All ports in em are two-terminal devices. In most applications, the first terminal is attached to a metal polygon and the second terminal is attached to ground. Such ports are referred to as grounded ports. Occasionally, however, it is useful to attach the two terminals of a port between two adjacent polygons. These ports are referred to as ungrounded ports. When analyzing multi-port circuits to find S-, Y- or Z-parameters, all of the ports in the circuit should be grounded. An ungrounded port can have a different ground reference from other ports in the circuit, which, in turn, can corrupt the results. In addition to being either grounded or ungrounded, ports can be further characterized by their location in a circuit and by whether or not em can de-embed them. Each port type is described in the sections that follow.

49

Em User’s Manual

Box-Wall Ports A standard box-wall port is a grounded port, with one terminal attached to a polygon edge coincident with a box wall and the second terminal attached to ground. An example of a standard box-wall port is shown below. Standard boxwall ports can be de-embedded.

-

1

+ Box wall port on page 46.

50

Chapter 5 Ports

Ungrounded-Internal Ports A standard ungrounded-internal port is located in the interior of a circuit and has its two terminals connected between abutted metal polygons. An ungroundedinternal port is illustrated below. Ungrounded-internal ports can be de-embedded by em.

a)

EM

-

+ 1

b)

In part a of the figure, the ungrounded-internal port is attached between two polygons which have equal widths. This is not a necessary condition for ungrounded-internal ports. These ports can also be attached between polygons which are abutted, but have unequal widths, as shown in part b. The only difference between the two conditions is that de-embedding requires the use of more standards (and therefore more time) when the polygons have unequal widths.

51

Em User’s Manual

Via Ports A via port has one terminal connected to a polygon on a given circuit level and the other terminal connected to a second polygon on a circuit level above or below the first polygon. An example of this port type is shown in below. See “Adding Via Ports,” page 40 of the Xgeom User’s Manual, for information on how to create a via port. Upper Polygon

+ 1

Via Port Lower Polygon An example of a circuit with a standard via port. A side view of the enclosed area on the circuit is shown on the right.

Em cannot de-embed via ports. However, in a circuit which contains a combination of via ports and other port types, the other port types can still be deembedded. Em will automatically identify all of the other ports present in the circuit and de-embed them, but leave the via ports un-de-embedded. The example file “patch.geo” has an example of a via port used in a patch antenna. The example file “viaports.geo” shows conceptually how via ports are used in an amplifier. The designer would analyze the four port in “viaports.geo” and then use a circuit theory program to attach a transistor between the via ports, ports 3 and 4 in “viaports.geo”. Both example files can be obtained with the Sonnet ⇒ Copy Examples command.

52

Chapter 5 Ports In most cases where you need grounded ports, your first choice would be to use auto-grounded ports (as discussed in the next section), especially since it is possible to de-embed an autogrounded port. The two most common cases where a via port would be used is when you wish to attach a port between two adjacent levels in your circuit or when you want a port to go up to the box cover rather than down to ground. EM

Automatic-Grounded Ports An automatic-grounded port is a special type of port used in the interior of a circuit. This port type has one terminal attached to the edge of a metal polygon located inside the box and the other terminal attached to the ground plane through all intervening dielectric layers. An auto-grounded port with a reference plane shift is shown below.

In many circuits, the addition of auto-grounded ports has little influence on the total analysis time of the em job. However for some circuits, auto-grounded ports may require some extra overhead calculations, thus increasing the total analysis time. Therefore, they should be used only when they provide an advantage over standard box-wall ports. Auto-grounded ports provide advantages over standard box-wall ports when: •

the layout of your circuit does not allow a direct path for a feed line to be connected between the port and the box wall (as in the figure above), or

•

your circuit requires a large feed structure to reach the box wall. If all or part of your feed structure can be eliminated, using an auto-

53

Em User’s Manual grounded port could reduce the total number of subsections in your circuit, thus decreasing the analysis time and/or memory requirements. Auto-grounded ports are similar to via ports with the exception of the following characteristics: •

Via ports require you to manually create vias that extend upward through the dielectric to the edge of a metal polygon. This is not the case with auto-grounded ports. You simply place auto-grounded ports anywhere a grounded port is needed. Em automatically detects the presence of auto-grounded ports in the circuit and connects the port terminals appropriately.

•

Auto-grounded ports connect directly through all dielectric layers to the ground plane. Via-ports allow the flexibility of connecting between any two adjacent dielectric layers.

•

Auto-grounded ports are de-embedded when the de-embedding option is used, while via ports are not.

•

Reference planes may be set with auto-grounded ports but cannot be set for via ports.

Special Considerations for Auto-Grounded Ports Metal Under Auto-Grounded Ports You cannot have metal directly beneath an auto-grounded port in a multi-layer circuit. Auto-grounded ports are two-terminal devices with one terminal connected to an edge of a metal polygon and the second terminal connected to the ground plane. When em detects the presence of an auto-grounded port, it automatically connects the two terminals in this manner. This includes circuits which have multiple dielectric layers between the polygon and the ground plane. However, in order for em to accomplish this, there must be a direct path from the edge of the metal polygon to the ground plane. When an auto-grounded port is used in a circuit where there is more than one dielectric layer between the port and the ground plane, em checks to make sure that there is no metal directly beneath the auto-grounded port. If metal is found, em prints an error message and stops.

54

Chapter 5 Ports Edge of Metal Polygon is Lossless

Auto-Grounded Ports on Box-Wall Auto-grounded ports are designed to be used in the interior of a circuit. If an autogrounded port is placed on a box-wall, it is treated as if it were a standard box-wall port.

Specifying Port Normalizing Impedances Whenever results are to be incorporated into a circuit theory based analysis program, the normalizing impedance for each port should be 50 ohms. In rare cases, S-parameters normalized to some other impedance is desired. For example, you may want to see what the reflection coefficient of a structure is when port 2 is connected to a 1.0 pF capacitor in parallel with a 10 ohm resistor (a power FET input model), while the input is being driven with a source which has a 35 ohm internal impedance. In this case, normalize port 1 to 35 ohms and port 2 to 10 ohms plus 1.0 pF. In another example, pure electromagnetics is frequently carried out using Sparameters normalized to the characteristic impedance of each connecting transmission line. These are often called “generalized” S-parameters, in spite of the fact that the line impedance is usually not specified, thus precluding precise conversion to 50 ohms for use in circuit theory software. To understand the physical meaning of normalized S-parameters, recall that our standard 50 ohm S-parameters are measured by terminating all ports in 50 ohms and then measuring ratios of incident to reflected (or transmitted) wave

55

EM

Auto-grounded ports can attach to the edge of any metal polygon in the interior of a circuit. There are no restrictions on the loss parameters of the metal used in the polygon. However, along the edge of the metal polygon where the port is attached, em does force the cells to be lossless. For most circuits, this should have little or no effect on the results. If, however, the port is attached to a highly lossy metal polygon, such as a thin-film resistor, the edge cell(s) of that polygon will be made lossless, and the output results may be affected.

Em User’s Manual amplitudes. If we were to use 60 ohm terminations, instead of 50 ohm terminations, the resulting measurements of traveling wave ratios would yield Sparameters normalized to 60 ohms. If we were then to take the S-parameters which had been measured using 60 ohm port terminations and use it in a circuit theory program (which expects you to use 50 ohm terminations), we would get incorrect results. However, 60 ohm S-Parameters do have uses. Say you want to know what percentage of power is absorbed by a 60 ohm load terminating port 2 of a two port circuit. That is simply one minus the magnitude squared of S11, using 60 ohm Sparameters. If you are interested in a different load, use a different normalizing impedance. S-Parameters can be renormalized to any real impedance by using the circuit network capability of em, or a circuit theory based program, to cascade appropriate transformers on each port. This technique can not be used with a complex impedance, as is the case with a lossy line or complex load. In em, the default normalizing impedance is 50 ohms. If you would like a different normalization, refer to the section “Parameters - Ports,” page 143 of the Xgeom User’s Manual, on how to specify the normalizing impedance for each port. The normalizing impedance is represented by four numbers. First is the real part in ohms. Next comes the reactive part in ohms. Third is the inductive part in nanohenries (nH). The last number is the capacitive part in picofarads (pF). The inductive and capacitive part modify only the reactive portion of the load, they are included so you do not have to manually re-calculate the reactive part at each frequency.

56

Chapter 5 Ports The resistance, reactance and inductance are all connected in series when the specific normalizing impedance is calculated at each frequency of analysis as illustrated below. The capacitance is connected in parallel with the result and then the final normalizing impedance at the frequency of analysis is calculated.

L

EM

R + jX C V

Equivalent circuit of an em port.

NOTE:

The normalizing impedances are ignored if Y- or Z-parameters are specified for output. Y- and Z-parameters are always normalized to one ohm.

This capability should be used only by the most advanced users. If the above discussion is not clear, seek assistance. Under no circumstances should this capability be used on data which is to be incorporated in a circuit in a standard circuit theory program other than Sonnet. Many such programs assume Sparameters normalized to exactly 50 ohms. For example, the 50 ohm S-parameters of a microstrip step junction look like a very small shunt capacitance and a very small series inductance. In other words, the reflection coefficient is almost zero and the transmission coefficient is almost unity with a couple degrees of phase. If, instead of 50 ohm S-parameters, we were to use S-parameters normalized to the characteristic impedances of the lines connecting to each of the two ports (this is common in electromagnetics), the S-

57

Em User’s Manual parameters would look like a transformer between the two impedances. This causes many circuit theory programs, which are expecting 50 ohm S-parameters, to give grossly incorrect results.

Special Port Numbering All ports are assigned a number at the time they are created in xgeom. By default, the ports are numbered by the order in which they are created (i.e. first port created is assigned the number 1, second port created is assigned the number 2, etc.). With this default method, all ports are positive and unique. However, there are some applications that require the ports to have duplicate, or even negative, numbers.

Ports with Duplicate Numbers All ports with the same number, as pictured below, are electrically connected together. As many physical ports as desired may be given the same numeric label. Such ports are sometimes called “even-mode” or “push-push” ports and have many uses, including simulating thick metal or the even-mode response of a circuit. See Chapter 18, “Thick Metal with Arbitrary Cross-Section,” for an example of using “push-push” ports.

Ports with identical port numbers are electrically connected together.

Ports with Negative Numbers Ports may also have negative numbers as shown in the figure on page 59. This feature can be used to redefine ground. Strictly speaking, em sums the total current going into all the positive ports with the same port number and sets that equal to 58

Chapter 5 Ports the total current going out of all the ports with that same negative port number. For example, for a circuit with a +1 port and a -1 port, em sets current flowing into port +1 to be equal to the current flowing out of port -1. Thus, they are sometimes called “balanced”, “push-pull” or “odd-mode” ports. Coplanar lines can be represented with balanced ports. See Chapter 17, “Coplanar Waveguide Discontinuities and Balanced Ports,” for an example of push-pull ports. EM

An example of pushpull ports.

59

Em User’s Manual

60

Chapter 6 De-embedding

EM

Chapter 6

De-embedding

Each port in a circuit analyzed by em introduces a discontinuity into the analysis results. In addition, any transmission lines that might be present introduce phase shift, and possibly, impedance mismatch and loss. Depending upon the nature of your analysis, this may or may not be desirable. De-embedding is the process by which the port discontinuity and transmission line effects are removed from the analysis results. The figure on page 62 illustrates the general layout of a circuit to be analyzed with em. The device under test (DUT), shown as a box in the figure, is the circuitry for which we wish to obtain analysis results. The DUT is located inside the metal box and is connected to one or more ports. The ports may be located on box walls, as in the figure, or in the interior of the metal box (see Chapter 5 for a description of port types available in em). Typically, transmission lines are necessary to connect the ports to the DUT. When de-embedding is enabled, em performs the following sequence of steps: 1

Calculates port discontinuities. 61

Em User’s Manual 2

Removes effects of port discontinuities from analysis results.

3

Optionally shifts reference planes (removes effects of feed transmission lines from analysis results).

4

Calculates transmission line parameters Z0 and Eeff.

Metal Box Walls Port

Transmission Line

Device Under Test (DUT)

Transmission Line Port

General layout of a circuit to be analyzed with em.

Upon completion of the de-embedding process, em outputs de-embedded Sparameter results, Z0, Eeff and the calculated port discontinuities. An abbreviated summary of the de-embedding algorithm used is presented in [66] and the complete theory is presented in [67].

Enabling the De-embedding Algorithm To demonstrate de-embedding with em, we will analyze the filter shown in the next figure on page 63. This circuit consists of five sections making up the filter metalization, two ports and two transmission lines connecting the ports to the filter 62

Chapter 6 De-embedding metalization. Reference planes have been defined for port 1 and port 2 at the left and right edges of the filter metalization, respectively. These reference planes instruct em to remove the effects of the transmission lines up to the filter metalization when de-embedding is enabled.

EM

Port 1

Transmission Line

Filter Metalization (DUT)

Transmission Line

Port 2

Port discontinuities and transmission lines at the upper left and lower right are removed from the em analysis results by enabling de-embedding.

63

Em User’s Manual The geometry file shown in the figure above, “filter.geo,” is available in the Sonnet examples directory. You may use Sonnet ⇒ Copy Examples to obtain a copy of this file. NOTE:

Adding reference planes to a circuit in xgeom does not automatically enable de-embedding in em. De-embedding must be enabled at the time em is executed. The procedure for enabling de-embedding is described below.

There are two methods for enabling de-embedding in em. The first method is to set the De-embed option in the job window to “on”. The second method is to specify an output file with a “.d” file extension in the Output Files dialog box. In this case, the De-embed checkbox is automatically set to “on”. To demonstrate de-embedding with the example “filter.geo” do the following: 1

Select em Control from the Sonnet task bar. The em program window will appear with a new untitled job file.

2

Click on the Geometry File radio button under File Type to select the analysis of a geometry file.

3

Check that the default directory is your project directory. If the directory is not correct either click on the Browse button to chose the correct directory and file, or enter the directory in the Start In : text entry box.

4

Enter “filter.geo” in the Geometry File text entry box.

5

Click on the Simple Sweep radio button in the Frequency Control section of the job window. Then enter “10” in the Start text entry box. The Frequency Unit is set to GHz, so no action need be taken.

6

The Verbose and De-embed options are already set, as defaults, so you need take no action for these items.

7

Click on the Run command button to execute the em analysis.

Below is the output written to the output window by em.

64

Chapter 6 De-embedding

ELECTROMAGNETIC ANALYSIS OF 3-D PLANAR CIRCUITS Version 6.0 (C) Copyright 1987 - 1999 Sonnet Software, Inc. All rights reserved. filter.geo ewa36361 filter.d

EM

Circuit geometry file: Analysis control file: De-embedded response file:

Starting subsectioning...done. Circuit uses 455 subsections, 2 Mbytes. Analyzing 10.000000 GHz. Waveguide mode calculation time = 0 seconds. Loading matrix, source level 0, field level 0. Analyzing entire circuit (DUT and transmission lines) Matrix fill time = 1 seconds. Reducing matrix. Matrix solve time = 0 seconds. De-embedding ports on left box wall. First standard, 28 subs, 28.125 mils -- Analyzing 10.000000 GHz. Loading matrix, source level 0, field level 0. De-embedding Reducing matrix. Second standard, 53 subs, 56.25 mils -- Analyzing 10.000000 GHz. Port 1 Loading matrix, source level 0, field level 0. Reducing matrix. De-embedding ports on right box wall. First standard, 28 subs, 28.125 mils -- Analyzing 10.000000 GHz. Loading matrix, source level 0, field level 0. De-embedding Reducing matrix. Port 2 Second standard, 53 subs, 56.25 mils -- Analyzing 10.000000 GHz. Loading matrix, source level 0, field level 0. Reducing matrix. Pre-computational time (seconds) -- Subsectioning: 0, Caching: 0 Analysis time per frequency (mm:ss) -- 0:02 user, 0:00 system, 0:02 real Analysis time per function (seconds) -- Modes: 0, Fill: 1, Solve: 0 De-embedded 50 Ohm S-Params. Mag/Ang. Touchstone Format (S11 S21 S12 S22): 10.0000000 1.000000 -72.59 0.000641 17.050 0.000641 17.050 1.000000 -73.31 P1 F=10.000 Eeff=(6.4556 0.0000) Z0=(51.78808 0.000000) R=0.00000 C=0.041639 P2 F=10.000 Eeff=(6.4762 0.0000) Z0=(51.88224 0.000000) R=0.00000 C=0.041650

De-embedded S-parameter, transmission parameter, and port discontinuity results em: finished. Total time for 1 freq (mm:ss) -- 0:02 user, 0:00 system, 0:02 real

65

Em User’s Manual Listed below are some comments concerning the output information shown above: •

Em begins the analysis by subsectioning the entire circuit contained inside the metal box. This includes the DUT, port discontinuities and all transmission lines.

•

Em then analyzes the entire structure at the first frequency (10 GHz). S-parameter results for the entire structure are obtained. If de-embedding had not been enabled, these S-parameters would be output by em as the non-de-embedded results and the analysis would be complete.

•

After analyzing the entire structure, em de-embeds the port (#1) on the left box wall. To do this, em creates and analyzes two calibration standards. The listing shows the number of subsections (28, 53) and length (28.125 mils, 56.25 mils) for each standard.

•

Em then de-embeds the port (#2) on the right box wall in the same manner.

•

In the last block of information, em displays the de-embedded Sparameter results along with the feed transmission line parameters (Z0 and Eeff) and calculated discontinuity (R and C) for each deembedded port. “P1” and “P2” stand for “port 1” and “port 2”, respectively. A detailed discussion concerning the port discontinuities (R and C) is presented in the next section.

De-embedding Port Discontinuities All ports in em introduce a discontinuity into the analysis results. Sometimes, this is desirable. For example, when analyzing a circuit fabricated with box walls, the effects introduced by a box-wall port discontinuity are real. Under this circumstance, the discontinuity should not be removed. However, in analyses where only the behavior of the DUT is of interest, all port discontinuities should be removed by de-embedding.

66

Chapter 6 De-embedding When enabled, em’s de-embedding algorithm automatically removes the discontinuities for box-wall, ungrounded-internal and auto-grounded ports (see Chapter 5 for a description of port types available in em). A via port is the only type of port that cannot be de-embedded by em. The port discontinuities for the port types which can be de-embedded are described in the sections that follow.

EM

Box-Wall Ports Box-wall ports have one port terminal connected to a polygon inside the metal box, and the second port terminal connected to ground (see the figure on page 50). The port discontinuity is modeled as a series resistor, R, and capacitor, C, shunted to ground as shown below. If the circuit being analyzed is completely lossless, the resistor value, R, is zero. Even with loss in the circuit, the capacitive reactance is normally very large compared to the resistance. S-parameters from em without de-embedding Device Under Test

R Box-wall port discontinuity

C

Port discontinuity associated with a box-wall port.

When enabled, em’s de-embedding feature automatically calculates the values of R and C for each box-wall port present in the circuit. The port discontinuities are then removed by cascading a negative R and C as illustrated above.

67

Em User’s Manual

S-Parameters from em with de-embedding.

S-parameters from em without de-embedding

Device Under Test

Block cascaded to negate box-wall port discontinuity

-R

R

-C

C

Box-wall port discontinuity

De-embedding automatically cancels the discontinuity associated with a box-wall port.

68

Chapter 6 De-embedding

Ungrounded-Internal Ports Ungrounded-internal ports do not have access to ground. For these ports, the two port terminals are connected between abutting polygons. The port discontinuity is again modeled as a series resistor, R, and capacitor, C, but is now connected between polygons as shown in . EM

Ungrounded-internal port discontinuity

-

+

R Metallized polygon

C

S-Parameters from em without deembedding

Metallized polygon

Port discontinuity associated with an ungrounded-internal port.

69

Em User’s Manual When enabled, the de-embedding feature automatically calculates the values of R and C for each ungrounded-internal port in the circuit. The port discontinuities are then removed by cascading a negative R and C as illustrated below. Block cascaded to negate port discontinuity

-

+ -R -C

Metallized polygon

R

C

Ungrounded-internal port discontinuity

S-parameters from em with de-embedding S-parameters from em without de-embedding

Metallized polygon

De-embedding automatically cancels the discontinuity associated with an ungrounded-internal port.

Auto-Grounded Ports Auto-grounded ports have one port terminal connected to a polygon edge in the interior of the metal box and the second port terminal connected to the ground plane. The model for the port discontinuity is shown in the figure on page 71. This discontinuity includes three components: a resistance, R, a capacitance, C, and an impedance, Zvia. Zvia is the impedance associated with the via used to construct

70

Chapter 6 De-embedding the auto-grounded port (see section “Automatic-Grounded Ports,” page 53, for a description of auto-grounded ports). R and C are the open-end resistance and capacitance of the polygon to which the auto-grounded port is attached.

Auto-grounded port discontinuity

EM

Device Under Test

Zvia

R C

S-parameters from em without deembedding

Port discontinuity associated with an auto-grounded port.

When de-embedding is enabled, em automatically calculates the values of R, C and Zvia. The port discontinuity is then removed by cascading negative values of these parameters as illustrated on page 72.

71

Em User’s Manual

.

D ev ice U n d er Test

A u to -g ro u n d ed p o rt d isco n tin u ity

Z v ia

R

-R

C

-C

-Z via

S -p aram eters fro m em w ith o u t d e-em b ed d in g

S -p aram eters fro m em w ith d e-em b ed d in g

B lo ck cascad ed to n eg ate p o rt d isco n tin u ity

De-embedding automatically cancels the discontinuity associated with an auto-grounded port.

Shifting Reference Planes Transmission lines are required in many circuits to connect ports to the device under test (DUT). If the length of a transmission line is more than a few degrees relative to a wavelength, unwanted phase (and possibly loss) will be added to the S-parameter results. If the impedance of the transmission line differs from the normalizing impedance of the S-parameters (usually 50 ohms), an additional error in the S-parameters results. Thus, if we are only interested in the behavior of the DUT, any “long” transmission lines connecting the ports to the DUT should be removed during de-embedding. The process of removing lengths of transmission line during de-embedding is known as “shifting reference planes”.

72

Chapter 6 De-embedding Reference planes may be specified in xgeom for box-wall and auto-grounded ports, but not ungrounded-internal ports. When em detects a reference plane, and de-embedding is enabled, it automatically builds and analyzes the calibration standards necessary to de-embed the port and shift the reference plane by the specified length. Reference planes are not necessary for de-embedding. If you do not specify a reference plane in xgeom for a box-wall or auto-grounded port, the reference plane length defaults to zero. This means that em will not shift the reference plane for that port when de-embedding is enabled. However, em will remove the discontinuity for that port.

Box-Wall Ports The figure below shows a circuit with a length of transmission line, TRL, inserted between a box-wall port and the device under test.

Transmission line

TRL S-parameters from em without de-embedding

R C

Device Under Test

Box-wall port discontinuity

Port discontinuity and transmission line associated with a box-wall port.

73

EM

NOTE:

Em User’s Manual When de-embedding is enabled, em removes the transmission line in a manner similar to that used to remove the port discontinuity. Em calculates S-parameters for the TRL alone, and then cascades a “negative” TRL along with negative R and C as illustrated in the next figure. Block cascaded to negate transmission line S-parameters from em with de-embedding

S-parameters from em without de-embedding Transmission line

-TRL Block cascaded to negate port discontinuity

TRL -R

R

-C

C

Device Under Test

Box-wall port discontinuity

Illustration of how de-embedding removes the port discontinuity and transmission line associated with a box-wall port.

Coupled Transmission Lines The two previous figures illustrated how the reference plane for a single transmission line attached to a box-wall port is shifted during de-embedding. In general, there may be multiple transmission lines on a given box wall on one or more circuit levels. This is illustrated in the next figure. In this situation, em shifts the reference plane an equal distance for all transmission lines on the given box wall. All coupling between the transmission lines is accounted for and removed.

74

Chapter 6 De-embedding

NOTE:

When shifting a reference plane for coupled lines, em assumes the following: a) all coupled lines are either horizontal or vertical b) the width of each coupled line is constant c) the spacing between coupled lines is constant. EM

1

Ports

2 . . .

N-coupled transmission lines

N Box Wall De-embedding shifts the reference plane an equal distance for all Ncoupled transmission lines on a given box wall. The coupling between transmission lines is removed by the de-embedding process.

75

Em User’s Manual

Auto-Grounded Ports Below is a circuit with a length of transmission line, TRL, inserted between a metallized polygon and an auto-grounded port.

Device Under Test

TRL

Transmission line S-parameters from em without de-embedding

Zvia

R C

Auto-grounded port discontinuity

Port discontinuity and transmission line associated with an auto-grounded port.

76

Chapter 6 De-embedding When de-embedding is enabled, em removes the transmission line in a manner similar to that used to remove the port discontinuity. Em calculates S-parameters for TRL alone, and then cascades a “negative” TRL along with negative R, C and Zvia as shown below.

TRL

EM

Device Under Test

Block cascaded to negate transmission line

-TRL

Transmission line S-parameters from em without de-embedding

Zvia

R C

-R -Z via -C

Auto-grounded port discontinuity S-parameters from em with de-embedding

Block cascaded to negate port discontinuity

Illustration of how de-embedding removes the port discontinuity and transmission line associated with an auto-grounded port.

De-embedding Results The listing below shows the de-embedded results obtained earlier in the chapter from the analysis of the example filter circuit (see page 63). This example illustrates the format of the de-embedded data that is written both to your output window and to the de-embedded results file (“.d” file extension).

77

Em User’s Manual

De-embedded 50 Ohm S-Params. Mag/Ang. Touchstone Format (S11 S21 S12 S22): 10.0000000 1.000000 -75.45 0.000757 14.090 0.000757 14.090 1.000000 -76.37 P1 F=10.000 Eeff=(6.4556 0.0000) Z0=(51.78808 0.000000) R=0.00000 C=0.041639 P2 F=10.000 Eeff=(6.4762 0.0000) Z0=(51.88224 0.000000) R=0.00000 C=0.041650

Example showing format of results obtained when de-embedding is enabled in em.

You should notice the following about the results in above: •

The first line is a comment line which describes the analysis results on the line below. In this example, the results are de-embedded 50 ohm S-parameters in Touchstone magnitude/angle format.

•

The second line gives the analysis results.

•

The remaining two lines give de-embedding information for each port in the circuit. The various fields are defined as follows:

P#:

Port number.

F:

Frequency in units defined earlier in the results file.

Eeff:

Effective dielectric constant of the transmission line connected to the port.

Z0:

TEM equivalent characteristic impedance of the transmission line, in ohms.

R:

Equivalent series resistance of port discontinuity, in ohms.

C:

Equivalent series capacitance of port discontinuity, in pF.

De-embedding Error Codes There are certain situations, discussed in detail in Chapter 7, “De-embedding Guidelines,” for which em is unable to obtain accurate de-embedded results. Em will usually, but not always, detect these situations and replace any suspect results with an error message. The format of the error message is “undefined: ”, where is a code which indicates the reason that em is unable to determine the de-embedded results. Table 2 describes the various error codes which may be displayed by em.

78

Chapter 6 De-embedding

Table 3 Codes displayed for indeterminate de-embedded results Code

De-embedded S-Parameters

nd

N/A

Port was not de-embedded. No data is available.

mp

Valid

Multiple ports on same box wall.

sl

Caution

Length of first de-embedding standard is too short.

nl

Valid

Length of first standard is multiple of half wavelength.

mv

Valid

Multiple values of Eeff or Z0 for a single port number.

bd

Caution

Bad Eeff or Z0 data due to unknown reason.

Description EM

The second column of Table 2, labeled “De-embedded S-Parameters”, gives the status of the de-embedded S-parameters corresponding to each error code. Error code “nd” indicates that the port was not de-embedded, therefore the status is not applicable. Error codes “mp”, “nl” and “mv” have a status of “Valid”. This indicates that while em was not able to determine Eeff or Z0, the de-embedded Sparameter results are completely valid. Error codes “sl” and “bd” have a status of “Caution”. This indicates that you should be cautious about using the deembedded S-parameter results as they may be corrupt. The “nd” error code indicates that the port cannot be de-embedded. Via ports are the only port type available in em that cannot be de-embedded. Thus, you will get this error code only when de-embedding circuits which contain via ports.

79

Em User’s Manual The “mp” error code indicates that em is unable to determine Eeff and Z0 because the circuit has multiple ports on the same side of the box. The reason for this is that more than one value is required to describe the multiple modes associated with coupled transmission lines. The “sl” code indicates that the length of the first de-embedding standard is too short. We recommend that the length be at least one substrate thickness. See the section “Reference Plane Length Minimums,” page 82, for details. The “nl” code indicates that the length of the first de-embedding standard is a multiple of a half wavelength. In this case, em is unable to determine Eeff and Z0, but the de-embedded S-parameter results are completely valid. See the section “Reference Plane Lengths at Multiples of a Half-Wavelength,” page 84, for details. The “mv” code indicates that a single port number is used for multiple ports in the circuit, and that the Eeff and Z0 values vary for the different ports. Finally, the “bd” error code indicates that em is unable to determine Eeff and/or Z0 for an unknown reason. Low precision and box resonances in the calibration standards are sources of error that occasionally lead to the “bd” code.

80

Chapter 7 De-embedding Guidelines

EM

Chapter 7

De-embedding Guidelines

The previous chapter describes the basics of de-embedding: what it is, how it is enabled, and what it does when enabled. This chapter presents guidelines for obtaining good de-embedded results.

Defining Reference Planes Xgeom and em place very few restrictions on the reference planes which may be defined for a given circuit. This is done intentionally so as to provide maximum flexibility for all users. However, there are some basic guidelines concerning reference planes that should almost always be followed. These guidelines are discussed below.

81

Em User’s Manual

De-embedding Without Reference Planes De-embedding does not require reference planes. Reference planes are optional for all box-wall and auto-grounded ports. If you do not specify a reference plane for a particular port in xgeom, em will assume a zero-length reference plane for that port. This means that de-embedding will remove the discontinuity associated with that particular port, but will not shift the reference plane for it. As discussed in the next section, em may generate bad de-embedded results if you attempt to remove a very short (but greater than zero) reference plane length. However, if you de-embed without a reference plane, em will not attempt to remove any length of transmission line at all. As a result, de-embedding without a reference plane does not lead to any error. Therefore, we recommend that you de-embed without reference planes rather than specify very short, non-zero, reference plane lengths.

Reference Plane Length Minimums The only explicit restriction on the minimum reference plane length that you may specify is that the reference plane must be at least three cell lengths long. If the reference plane is less than three cell lengths long, em displays an error message and stops. Otherwise, em performs the analysis with whatever reference plane length you have specified.

82

Chapter 7 De-embedding Guidelines If, however, the reference plane is very short relative to the substrate thickness or the width of the transmission line, em may generate poor de-embedded results. This is due to one or both of the following reasons which are illustrated below Fringing fields from DUT interact with fringing fields from port.

1

Box Wall

DUT

1

2

Metal Box

Poor de-embedding results may be obtained when very short (but nonzero) reference plane lengths are used.

•

The port is too close to the device under test (DUT). There are fringing fields associated with the port and separate fringing fields associated with the DUT. If the port and DUT are too close, the fringing fields interact. The deembedding algorithm (which is virtually identical to algorithms used in deembedding measured data) is based on circuit theory and cannot handle fringing field interaction. See [56] for a detailed description of the problem.

•

The first calibration standard is too short. In this situation, the discontinuity associated with port #1 interacts with the discontinuity associated with port #2. As a result, the first calibration standard does not “behave” like a transmission line and its S-parameters are invalid.

There is no precise rule as to how long a reference plane must be made in order to prevent the above effects from corrupting the de-embedded results. The required reference plane length is dependent upon the circuit geometry and the nature of

83

EM

First calibration standard is too short. Port #1 interacts with port #2.

Em User’s Manual the analysis. However, we recommend that you use reference plane lengths equal to or greater than one substrate thickness. This is sufficient for most types of analysis.

Reference Plane Lengths at Multiples of a Half-Wavelength Eeff and Z0 cannot be calculated when the length of the reference plane is an integral multiple of a half wavelength. For example, at an extremely low frequency the electrical length of the reference plane may be a fraction of a degree (i.e., zero half-wavelengths). In this case, the analysis is unable to accurately evaluate the electrical length and, especially, the characteristic impedance. At some point as the length of the reference plane approaches a multiple of a halfwavelength, em is able to determine that the calculated values of Eeff and Z0 are becoming corrupt. When this occurs, em outputs the error message “undefined: nl” in place of the Eeff and Z0 values (see the section “De-embedding Error Codes,” page 78). Note, however, that while em is unable to determine Eeff and Z0, the de-embedded S-parameter results are still perfectly valid.

Reference Plane Lengths Greater than One Wavelength If the length of the reference plane is more than one wavelength, incorrect Eeff results might be seen. However, the S-parameters are still completely valid. Em’s calculation of Eeff is based on phase length. If the reference plane is, say, 365 degrees long, em first calculates Eeff based on a phase length of 5 degrees. However, em has some “smarts” built in. If a non-physical result is seen, em increases the calculated phase length by 360 degrees at a time until physical (i.e., Eeff Š 1.0) results are obtained. This usually corrects the problem. Thus, it takes a particularly long reference plane before the Eeff calculation fails. When it does fail, it suddenly jumps down to a value just above 1.0. Z0 and the deembedded S-parameter data still have full validity. This failure mode is rarely seen.

84

Chapter 7 De-embedding Guidelines

Non-Physical S-Parameters Generally, reference planes should not be set in xgeom such that they extend beyond a discontinuity in the circuit. Doing so may result in non-physical Sparameters.

L1 L2 W1

W2

Now, consider the figure on page 86. This circuit is identical to the circuit shown above except that the length of the reference plane originating on the left box wall has been increased. If em is run with de-embedding enabled on this circuit, it “removes” a length of transmission line equal to the specified reference plane

85

EM

To illustrate this problem, consider the circuit shown below. In this circuit, the reference planes do not extend beyond any discontinuities. When de-embedding is enabled, the port #1 discontinuity is removed along with a transmission line of width W1 and length L1. Similarly, the port #2 discontinuity is removed along with a transmission line of width W2 and length L2. The de-embedded result is a set of 2-port S-parameters for the block in the middle of the circuit.

Em User’s Manual length. This occurs even though the actual port transmission line is shorter than the reference plane length. As a result, the de-embedded S-parameters are nonphysical. Aragorn

Discontinuity begins here

L2

W1

W2

L1

Example circuit for which non-physical S-parameters will be obtained when em is run with de-embedding enabled.

A second de-embedding example leading to non-physical S-parameter results is shown in the next figure. In this example, the circuit has two via pads on each side of the port transmission line. The via pads are grounded to the box wall.

86

Chapter 7 De-embedding Guidelines When em is run with de-embedding enabled on this circuit, it “removes” three coupled transmission lines with a length equal to the reference plane length. Since the reference plane extends from the box wall beyond the vias, the de-embedded S-parameters are again non-physical.

EM

W1 L2 W2

W4

W3 L

Box Resonances Because em’s de-embedding algorithm is based on circuit theory, it is unable to de-embed a structure contained inside a resonant cavity; a limit it shares with all de-embedding algorithms. Thus, whenever you wish to de-embed a circuit with box resonances, you must take the necessary steps to remove those box resonances. (See Chapter 19 for a detailed description on identifying and removing box resonances.) Note that if you do de-embed a circuit with box resonances, em may generate a “bd” de-embedding error code: see section “Deembedding Error Codes,” page 78. This error code indicates that em has detected bad values for Eeff and Z0.

87

Em User’s Manual

Higher Order Transmission Line Modes De-embedding removes the port discontinuity and the connecting length of transmission line. The de-embedding assumes that there is only one mode propagating on the connecting transmission line, usually the fundamental quasiTEM mode. If higher order modes are propagating, the de-embedded results are not valid. (The same is true for actual, physical, measurements.) If this is the case, we strongly recommend using a thinner substrate, unless, for some reason, multimode operation is desired. Even when higher order microstrip modes are evanescent, there can still be problems. If the port is so close to the discontinuity of interest that their fringing (evanescent) fields interact, the de-embedding looses validity. Again, this is a problem which also arises in an actual physical measurement if the device to be de-embedded is too close to the fixture connector.

88

Chapter 8 Network File Analysis

EM

Chapter 8

Network File Analysis

The em network file provides you with a powerful circuit analysis tool. Examples of ways in which the circuit network capability may be used include: •

Cascading S-, Y- and Z- parameter data files. You can read and combine multiple sets of S-, Y- and Z-parameter data files, including results from previous em runs. This is particularly useful when analyzing large, complex circuits which require subdivision for an em analysis. When analyzing a network file, em will automatically interpolate between frequencies if there are differences between the data files.

•

Inserting lumped elements into a circuit. Lumped elements, such as resistors, capacitors, inductors and ideal transmission lines, can be combined with S-, Y- and Z-parameter data files.

•

Intelligent frequency selection. Em, when analyzing a network file, may be set up to automatically select frequency points for an analysis. Frequency points are spaced close together in regions where the circuit response varies

89

Em User’s Manual rapidly and are spaced farther apart in regions where the circuit response is essentially constant. Em may also be set up to automatically find the frequencies where the minimum and maximum responses of a circuit occur.

Network File Analyses The sequence of steps for a network file analysis may be summarized as follows: 1

Em reads the network file which contains circuit and analysis control information. This includes S-, Y- and Z-parameter data files, lumped elements, references to geometry files and intelligent frequency selection specifications. The format of the network file is similar to the format used in other netlist programs.

2

Em uses an analysis control file and a geometry file to run each electromagnetic analysis invoked by the network file.

3

Em performs the circuit analysis specified in the network file.

4

Em combines the electromagnetic results with the circuit results to obtain the desired output results. This may include sorted S-, Y- and Zparameters, and the frequencies at which minimum or maximum circuit responses occur (Fmin, Fmax).

Note that the above sequence of steps is generalized for analyses which include both electromagnetic and circuit analysis. In cases where the overall analysis is restricted to either electromagnetic analysis or circuit analysis, some of the steps are omitted.

Cascading S-, Y- and Z-Parameter Data Files A particularly useful feature provided by a network file is the ability to cascade multiple S-, Y- and Z-parameter data files. There are no restrictions on the file formats which may be cascaded. For example, you can cascade em Z-parameter data in Touchstone format with measured S-parameter data in Super-Compact format. In addition, em can analyze at frequencies which are not included in the data files. Em automatically interpolates if there are any differences between the requested frequency points and those in the data files. 90

Chapter 8 Network File Analysis To demonstrate the cascading operation, we will analyze the two-port circuit shown below. This circuit consists of two identical thin film resistors connected in series. We will use the S-parameters from the geometry file analysis on the thinfilm resistor as input to the network file analysis. The desired output result is an overall set of two-port S-parameters for the series combination of resistors.

S -param eter file “res16.d”

N ode 2

S -param eter file “res16.d”

EM

N ode 1

1

N ode 3

2

The two-port S-parameters contained in file “res16.d” are cascaded to obtain an overall set of two-port S-parameters.

91

Em User’s Manual

Thin-Film Resistor Example Before cascading the resistors “.d” files together, we need to obtain the results of the geometry file analysis of the thin-film resistor shown in the next figure. Use Sonnet ⇒ Copy Examples to obtain the response file, “res16.d”, from the Sonnet example files.

Resistive Material

Transmission Lines

The geometry file for the thin-film resistor.

The output file “res16.d” appears below.

92

Chapter 8 Network File Analysis

EM

! ************************************************** !< FTYP RSP !< PROG EM 6.0, id leslie1b.2986, on PC/DOS 95050501. !< CMD em -dv -rres16.d res16.geo -- Sat Mar 27 12:24:46 1999 !< CKDATE Wed Mar 1 11:00:11 1996 ! Last time res16.geo was updated. ! Maximum subsection size is lambda/20 at 400 MHz. Estimated Eeff = 5.4. ! All dimensions are in mils. ! A = 200.000000(16), B = 200.000000(16), C = 275.000000, with 2 layers. ! Lay H( mils) Erel Etan(d) Esigma Murel Mtan(d) Nz Subs ! 0 250.0000 1.000000 0.000000 0.000000 1.000000 0.000000 1 17 ! 1 25.00000 9.800000 0.000000 0.000000 1.000000 0.000000 1 0 ! Estimated Memory: 1 Mbytes Subsection Total: 17 ! Ports: 2 Box-Wall ! Circuit has loss. ! Loss parameters of metals used in circuit: ! Thin Film -- Rdc: 16.77 Rrf: 0 Xdc: 0 Ls: 0 ! Circuit and excitation are symmetric about X axis. ! De-embedded 50 Ohm S-Params. Mag/Ang. Touchstone Format (S11 S21 S12 S22): ! Comments following s-parameters give port data with following syntax: ! P# F=x MHz Eeff=(x+jy) Z0=(x+jy) Ohms R=x Ohms C=x pF ! All box-wall reference planes are zero length. # MHZ S MA R 50 ! Pre-computational time (seconds) -- Subsectioning: 0, Caching: 0 ! Analysis time per frequency (mm:ss) -- 0:00 user, 0:00 system, 0:00 real ! Analysis time per function (seconds) -- Modes: 0, Fill: 0, Solve: 0 ! 200.000000 0.143614 -2.626 0.856388 -3.076 0.856388 -3.076 0.143614 -2.626 !< P1 F=200.00 Eeff=(6.3966 0.0000) Z0=(51.23505 0.000000) R=0.00000 C=0.094942 !< P2 F=200.00 Eeff=(6.3966 0.0000) Z0=(51.23505 0.000000) R=0.00000 C=0.094942 ! 300.000000 0.143613 -3.939 0.856393 -4.614 0.856393 -4.614 0.143613 -3.939 !< P1 F=300.00 Eeff=(6.3968 0.0000) Z0=(51.23438 0.000000) R=0.00000 C=0.094938 !< P2 F=300.00 Eeff=(6.3968 0.0000) Z0=(51.23438 0.000000) R=0.00000 C=0.094938 ! 400.000000 0.143611 -5.254 0.856400 -6.152 0.856400 -6.152 0.143611 -5.254 !< P1 F=400.00 Eeff=(6.3970 0.0000) Z0=(51.23348 0.000000) R=0.00000 C=0.094934 !< P2 F=400.00 Eeff=(6.3970 0.0000) Z0=(51.23348 0.000000) R=0.00000 C=0.094934 ! Total time for 3 freqs (mm:ss) -- 0:01 user, 0:00 system, 0:01 real

The output file “res16.d” is created by em when the thin-film resistor example shown on page 92 is analyzed.

93

Em User’s Manual

The Network File The network file consists of a header line, optional comment lines and several data blocks which define the circuit analysis to be performed. The network file that we will use in our example is shown below.

!< FTYP NET ! File: cascade.net ! Date: March 27, 1999 ! Cascade two s-parameter files. DIM FREQ

MHZ

! Use MHZ freq units

S2P S2P DEF2P

1 2 2 3 1 3

res16.d res16.d RESNET

Touch

cascade.d

CKT

FILEOUT RESNET

! Input resistor net ! Input resistor net ! Overall network

S MA R 50

! Define output data

FREQ SWEEP

200.0 400.0 100.0

! 200 - 400 MHZ

The network file “cascade.net” cascades S-parameter data files. Em reads S-parameter data from the input file “res16.d”, and writes the resulting S-parameter data to the output file “cascade.rsp”.

You can use Sonnet ⇒ Copy Examples to obtain a copy of the network file, “cascade.net”. Presented below is a brief explanation of the above network file. For a detailed description of the network file, see Chapter 11, “The em Network File.”

94

•

The first line in the above, “!< FTYP NET”, is the header line. This is a required line which identifies the file as an em network file.

•

Following the header line in is a block of comment lines. All comments begin with an exclamation point (!). Note that the

Chapter 8 Network File Analysis exclamation point does not need to be in the first column of the file. Everything following an exclamation point on a given line is considered a comment. There are no restrictions on the number or location of comment lines which may be placed in an em network file. The DIM data block defines units for circuit parameters specified later in the network file. In our example, the frequency units are defined as MHZ.

•

The CKT data block defines the network(s) to be analyzed. In the above example, a two-port element, S2P, is placed between nodes 1 and 2 with S-parameters from input file “res16.d”. A second 2-port element, S2P, is placed between nodes 2 and 3 with S-parameters from the same file, “res16.d”. Finally, a 2-port network, “RESNET”, is defined between nodes 1 and 3.

•

The FILEOUT data block specifies the desired output results. In this example, this block specifies that results for the network “RESNET” should be stored using Touchstone (Touch) format in the output file “cascade.rsp”. The final four fields, “S MA R 50,” specify S-parameters, magnitude-angle format and a real normalizing impedance of 50 ohms for all ports.

•

The FREQ data block specifies the analysis frequencies. In this case, the frequency is swept from 200 MHZ to 400 MHZ in steps of 100 MHZ. Note that the MHZ unit comes from the DIM block.

To analyze the example circuit with em, do the following: 1

Select em Control from the Sonnet task bar to open the em program window with a new job file.

95

EM

•

Em User’s Manual 2

Click on the Network radio button in the File Type section of the job window. The appearance of the job window will be updated and appear as shown below.

3

Check that the default directory is correct. If not, use the Browse button to select the correct directory and the file, “cascade.net” or edit the Start In text entry box to enter the correct directory.

4

Enter “cascade.net” in the Network File text entry box.

5

The Internal Sweep radio button is already selected under the Frequency Control section.

6

Click on the Run command button to execute the em analysis.

The listing on page 97 shows the resulting S-parameters stored in the output file “cascade.d”.

96

Chapter 8 Network File Analysis

EM

!< FTYP RSP ! S-Parameter Data File !< VER 6.0 !< PROG em !< DATE Sat Mar 27 12:34:38 1999 ! ! Network : RESNET with 2 Ports # MHZ S MA R 50 200.000000 0.250782 -5.307 0.748778 -6.262 0.748778 -6.262 0.250782 -5.307 300.000000 0.250310 -7.959 0.748702 -9.393 0.748702 -9.393 0.250310 -7.959 400.000000 0.249650 -10.61 0.748596 -12.52 0.748596 -12.52 0.249650 -10.61

The output file “cascade.d” is generated when the network file “cascade.net” is analyzed with em.

A Network File Invoking a Geometry File Analysis The preceding described how em may be used to perform strictly circuit analyses. The next example demonstrates a network file analysis which invokes a geometry file analysis in conjunction with using previously generated data. The example builds on results generated in the previous chapters. To demonstrate a combined network file/geometry file analysis, the two-port Tattenuator shown in the next figure on page 98 will be analyzed. Em will be set up to perform the following steps: 1

Read S-parameter data from the file “res16.d”, copied in Chapter 8.

2

Perform an electromagnetic analysis of the geometry file “res67.geo”, a 67 ohm thin-film resistor.

3

Combine the S-parameter results from the electromagnetic analysis with the S-parameter results from “res16.d” to obtain an overall set of Sparameters for the T-attenuator.

97

Em User’s Manual

1

Node 1

S-parameter file “res16.d”

Node 2

S-parameter file “res16.d”

Node 3

geometry file “res67.geo”

The two-port T-attenuator will be analyzed with em to demonstrate a combined electromagnetic/circuit analysis.

98

2

Chapter 8 Network File Analysis Pictured below is the geometry file “res67.geo”, which is a 67 ohm thin-film resistor. This file is read by em and used for the geometry file portion of the analysis.

EM

Transmission Line

67 ohm Thin-Film Resistor

Transmission Line

The network file shown on page 100 will be input to em. Since this file is similar to the network file described on page 94, we will only highlight the differences between the two files here.

99

Em User’s Manual

!< FTYP NET ! File: combine.net ! Date: July 1, 1996 ! Perform combined electromagnetic and circuit analysis. DIM FREQ

MHZ

S2P S2P GEO DEF2P

1 2 2 1

CKT

FILEOUT ATTEN

2 3 0 3

Touch

res16.d res16.d res67.geo ATTEN

combine.rsp

OPT=vd

CTL=internal

S MA R 50

FREQ SWEEP

200.0 400.0 100.0

The network file “combine.net” illustrates a combined electromagnetic and circuit analysis.

The primary distinction between the network file shown above and the file on page 94 is that this file contains an instruction to perform a geometry file analysis. The GEO keyword in the CKT data block instructs em to use the geometry file “res67.geo” and command options “vd” to run an electromagnetic analysis. The CTL keyword at the end of the GEO line specifies how em acquires the analysis frequencies. CTL can be set equal to an external file, i.e., CTL=ctl.an, or it can be set equal to the word “internal” as it is in the above example. When it is set to “internal”, em automatically creates a temporary analysis control file using the frequencies specified in the FREQ data block.

100

Chapter 8 Network File Analysis After first performing the electromagnetic analysis on the geometry file, em then performs all required circuit analysis before outputting the requested results.

TIP

To obtain copies of the geometry file, “res67.geo”, and the network file, “combine.net”, use Sonnet ⇒ Copy Examples. In addition, if you wish to perform the analysis and you do not have file “res16.d” available, you can obtain it also by using Sonnet ⇒ Copy Examples. To analyze the example circuit with em, do the following: 1

Select em Control from the Sonnet task bar to open the em program window with a new job file.

2

Then click on the Network radio button in the File Type section of the job window.

3

Check that the default directory is correct. If not, use the Browse button or edit the Start In text entry box to enter the correct directory.

4

Enter “combine.net” in the Network File text entry box.

5

The Internal Sweep radio button is already selected under the Frequency Control section.

6

Click on the Run command button to execute the em analysis.

The listing on page 102 shows the output file “combine.rsp”. This file contains the overall set of S-parameters for the T-attenuator.

101

EM

Before executing a GEO statement, em checks for the existence of data at the specified control frequencies. If the data already exists, and the geometry file has not changed since the data was generated, em does not execute an electromagnetic analysis, but uses the available data.

Em User’s Manual

!< FTYP RSP ! S-Parameter Data File !< VER 6.0 !< PROG em !< DATE Thu Jul 1 12:00:00 1996 ! ! Network : ATTEN with 2 Ports # MHZ S MA R 50 200.000000 0.008927 67.709 0.500515 -5.759 0.500515 -5.759 0.008927 67.709 300.000000 0.013072 68.924 0.501159 -8.646 0.501159 -8.646 0.013072 68.924 400.000000 0.017178 67.772 0.502054 -11.54 0.502054 -11.54 0.017178 67.772

The output file “combine.rsp” is generated when input file “combine.net” is analyzed with em. This file contains the overall set of S-parameters for the T-attenuator.

NOTE:

When the above example is run, em generates two output files; one has a “.d” extension and the other has a “.pd” extension. The file with the “.d” extension is the standard em de-embedded results file. The file with the “.pd” extension is a high precision data file used by the circuit network capability. See section “High Precision em Output Files,” page 148 for details.

Inserting Lumped Elements into a Circuit Another very useful feature that the em circuit network capability provides is the ability to insert lumped elements into a circuit after an electromagnetic analysis has been performed on that circuit. To demonstrate the use of lumped elements, we will again analyze the T attenuator. In this chapter, however, the three resistors will not be analyzed as part of the geometry file, but will be inserted as lumped elements in the network file analysis. The figure below shows the circuit layout with the lumped resistor elements. The transmission line structures geometry file will be analyzed with em first. A network file will then be used to insert the three resistors and calculate two-port S-parameters for the overall circuit.

102

Chapter 8 Network File Analysis

16.77 Ω

16.77 Ω

1

2

Lumped Elements

Lumped Element 67.11 Ω

Geometry File metalization

The two-port T attenuator will be re-analyzed to demonstrate the use of lumped elements.

103

EM

To accomplish this task, it is necessary to create a geometry file with the transmission line structure and three “holes” where lumped elements will eventually be inserted. The figure on page 104 shows such a geometry file. Here, pairs of auto-grounded ports have been placed on the edges of each lumped element “hole”. When the lumped elements are inserted later on, each is connected across the corresponding pair of auto-grounded ports. Note that under certain conditions, ungrounded-internal ports can be used instead of auto-grounded ports. See “Using Ungrounded-Internal Ports,” page 107, for details.

Em User’s Manual

The geometry file “lumped.geo” contains three sets of auto-grounded ports placed at locations where lumped elements will eventually be inserted.

104

Chapter 8 Network File Analysis Below is the network file that will be used for this example. !< FTYP NET ! File: lumped.net ! Date: Sept 1, 1996 ! Analyze T attenuator using lumped elements.

FREQ RES

EM

DIM MHZ OH

CKT GEO RES RES RES DEF2P FILEOUT ATTEN

1 2 3 4 5 6 7 8 3 4 5 6 7 8 1 2

lumped.geo R=16.77 R=16.77 R=67.11 ATTEN

Touch

lumped.rsp

OPT=vd

CTL=internal

S MA R 50

FREQ SWEEP

200.0 400.0 100.0

The network file “lumped.net” is used to insert the three T-attenuator resistors as lumped elements.

The network file above instructs em to perform the following steps: 1

Perform an electromagnetic analysis on the geometry file “lumped.geo” using the verbose and de-embed options. Frequency control is set to internal; therefore, the values used are those specified in the SWEEP command which appears later in the file.

2

Insert a 16.77 ohm resistor between nodes 3 and 4.

3

Insert a 16.77 ohm resistor between nodes 5 and 6.

4

Insert a 67.11 ohm resistor between nodes 7 and 8.

5

Calculate an overall set of S-parameters for the T attenuator.

6

Write the results to output file “lumped.rsp”.

105

Em User’s Manual Use Sonnet ⇒ Copy Examples to obtain copies of the geometry file, “lumped.geo”, and the network file, “lumped.net”. To perform the analysis: 1

Select em Control from the Sonnet task bar to open the em program window with a new job file.

2

Then click on the Network radio button in the File Type section of the job window.

3

Check that the default directory is correct. If not, use the Browse button or edit the Start In text entry box to enter the correct directory.

4

Enter “lumped.net” in the Network File text entry box.

5

The Internal Sweep radio button is already selected under the Frequency Control section.

6

Click on the Run command button to execute the em analysis.

The listing below is the output file “lumped.rsp”. Note that these results are similar to the results given in for distributed elements.

!< FTYP RSP ! S-Parameter Data File !< VER 6.0 !< PROG emgen !< DATE Thu Sep 1 12:00:00 1996 ! ! Network : ATTEN with 2 Ports # MHZ S MA R 50 200.000000 0.007892 66.627 0.500390 -4.890 0.500390 -4.890 0.007892 66.627 300.000000 0.011500 69.402 0.500788 -7.338 0.500788 -7.338 0.011500 69.402 400.000000 0.015127 69.447 0.501343 -9.790 0.501343 -9.790 0.015127 69.447

The output file “lumped.rsp” shows the resulting S-parameters obtained when the T attenuator circuit is analyzed using lumped elements.

106

Chapter 8 Network File Analysis

Using Ungrounded-Internal Ports

The figure below shows a geometry file for the T attenuator with ungroundedinternal ports at each lumped element location. Note that the gaps between polygons at these locations have been removed. This is because you must attach ungrounded-internal ports between two abutted polygons. This slightly impacts the overall performance of the attenuator.

Z3

The geometry file “lumped2.geo” uses ungrounded-internal ports at locations where lumped elements will eventually be inserted.

107

EM

In the example presented above, a pair of auto-grounded ports was placed at each location in the em circuit layout where a lumped element would eventually be inserted (see ). It is also possible to perform the same analysis using ungroundedinternal ports, because each resistor in this example is a series lumped element without access to ground (see ). Any time access to ground is not required for a lumped element, you can replace the pair of auto-grounded ports with a single ungrounded-internal port.

Em User’s Manual The network file shown on page 109 connects the desired resistors across the ungrounded-internal ports of the network shown on page 107. Since ungroundedinternal ports do not have access to ground, only a single node is specified when connecting an element across them. See the RES statements in the CKT data block.

!

WARNING Ungrounded-internal ports have one terminal connected to an edge of a polygon and the second terminal connected to an abutted edge of a second polygon. Ungrounded-internal ports do not have access to ground. Therefore, only 1-port elements or 1-port networks may be connected across ungrounded-internal ports.

108

Chapter 8 Network File Analysis

!< FTYP NET ! File: lumped2.net ! Date: Sept 1, 1996 ! Analyze T-attenuator using ungrounded-internal ports.

FREQ RES

EM

DIM MHZ OH

VAR Z3 = 16.77 Z4 = 16.77 Z5 = 67.11 CKT GEO RES RES RES DEF2P FILEOUT ATTEN

1 2 3 4 5 3 4 5 1 2

TOUCH

lumped2.geo R^Z3 R^Z4 R^Z5 ATTEN

OPT=vd

lumped2.rsp

CTL=internal

S MA R 50

FREQ SWEEP

200.0 400.0 100.0

The network file “lumped2.net” connects the three T attenuator resistors across the corresponding ungrounded-internal ports.

You can use Sonnet ⇒ Copy Examples to obtain copies of the geometry file, “lumped2.geo”, and the network file, “lumped2.net”. To initiate the analysis once the input files are in place, do the following: 1

Select em Control from the Sonnet task bar to open the em program window with a new job file.

2

Then click on the Network radio button in the File Type section of the job window. 109

Em User’s Manual 3

Check that the default directory is correct. If not, use the Browse button or edit the Start In text entry box to enter the correct directory.

4

Enter “lumped2.net” in the Network File text entry box.

5

The Internal Sweep radio button is already selected under the Frequency Control section.

6

Click on the Run command button to execute the em analysis.

The listing below shows the S-parameter results obtained from the analysis with ungrounded-internal ports. These results are very similar, but not identical, to the results in for auto-grounded ports. The differences are primarily due to the change in the gap size between polygons at the points where lumped elements are inserted.

!< FTYP RSP ! S-Parameter Data File !< VER 6.0 !< PROG emgen !< DATE Thu Sep 1 12:00:00 1996 ! ! Network : ATTEN with 2 Ports # MHZ S MA R 50 200.000000 0.009217 68.496 0.500482 -5.785 0.500482 -5.785 0.009217 68.496 300.000000 0.013510 70.114 0.500994 -8.683 0.500994 -8.683 0.013510 70.114 400.000000 0.017788 69.364 0.501707 -11.59 0.501707 -11.59 0.017788 69.364

The output file “lumped2.rsp” contains the S-parameter results for the T attenuator analyzed with lumped elements and ungrounded-internal ports.

110

Chapter 9 Circuit Subdivision - A Filter Example

EM

Chapter 9

Circuit Subdivision - A Filter Example

This chapter will provide an in-depth example of a filter in which a network file analysis is used to make the circuit more manageable for your processing resources. The circuit is a seven-section edge-coupled microstrip bandpass filter (courtesy of Kaman Sciences, Inc.). The filter as a whole presents a difficult analysis problem in that it could require more memory than is available and/or excessive CPU usage; therefore, the circuit will be broken down into smaller pieces for electromagnetic analysis. For this entire structure, the response changes rapidly with respect to frequency. This is not true, however, for the response of certain sections of the circuit. These sections will become the subdivisions used to analyze the structure.

111

Em User’s Manual Subdividing not only reduces the size of circuit to be analyzed, but allows us to take advantage of the interpolation feature of a network analysis. Interpolation requires fewer frequencies to be analyzed while providing the same level of accuracy; therefore, analyses of the subdivisions are performed at a limited number of frequencies. Then a network file is used to connect the response data of the subdivisions to simulate the full filter. Interpolation is automatically performed between frequencies in the data files. This provides accurate response data for the filter at a larger number of frequencies using fewer resources than would be required if the circuit was approached as a whole. Performing user-guided subdivision as a method of analysis should, in general, be done as follows: 1

Decide how the circuit is to be split up. This step often requires expertise and experience to avoid splitting the circuit at a junction where significant coupling or rapidly varying response is present.

2

Create the individual geometry files in xgeom which make up the circuit.

3

Use em to analyze the individual geometry files at a limited number of frequencies.

4

Create a network file which connects the individual response data files in a network equivalent to the circuit as a whole and defines the frequencies of analysis. This allows em to interpolate between the frequency points used in Step 3.

5

Use em to run an analysis of the network file and output response data for the circuit.

Example Files All of the files associated with this example are contained in one directory and may be obtained using Sonnet ⇒ Copy Examples. To copy all the files using only one command, type the following in the Sonnet ⇒ Copy Examples Command text entry box when it appears:

112

Chapter 9 Circuit Subdivision - A Filter Example copyex bpfilter This will copy a folder which contains all the example files related to this example. The filter to be analyzed is shown below, and is available in the file “bpf_w.geo.” EM

The filter circuit, “bpf_w.geo,” with an aspect ratio of 1:4. The dashed line shows the first place to split the circuit.

113

Em User’s Manual

Dividing the Circuit The first step, as mentioned above, is to decide how to split the circuit. The main consideration is to divide the circuit in places where the coupling mechanism is at a minimum. Other considerations, such as symmetry and wavelength, may need to be addressed. Notice that the filter is symmetrical; therefore the first split should be to divide the circuit in half and connect the halves after analysis.

!

WARNING This circuit is symmetrical about the y-axis and not the x-axis, nor do the ports lie on the plane of symmetry; therefore, it would be incorrect to set the Symmetry checkbox in the Box Parameters dialog box.

114

Chapter 9 Circuit Subdivision - A Filter Example Half of the filter still presents a difficult analysis problem; therefore, the circuit will be broken down even further into eight geometry files. This breakdown is shown below.

bpf_p5.geo bpf_p8.geo bpf_p1.geo

bpf_p2.geo

bpf_p2.geo bpf_p3.geo bpf_p6.geo

bpf_p6.geo

Half of the filter circuit divided into eight separate “.geo” files.

Note that the breaks in the circuit are made where coupling is not a significant factor. Places on the circuit where a high degree of coupling is present are kept within an individual geometry file so that the coupling may be accurately accounted for. You may notice that some of the geometry files, for example, “bpf_p2.geo,” are used twice instead of one file with a circuit that is twice the length. The division was placed to optimize interpolation results. In order to prevent interpolation from skewing the results each piece must be non-resonant and its length under λ/2. Each circuit is 1/4 wavelength long in keeping with this stricture.

115

EM

bpf_p7.geo

bpf_p4.geo bpf_p4.geo

Em User’s Manual

Creating the Geometry Files A geometry file must be created in xgeom for each piece of the circuit. When creating the individual files, you should ensure that the parameters of the circuit, such as the metal loss and dielectric are the same. The physical dimensions of the circuit elements should also be kept intact. You should insert ports in the individual files to connect them to the pieces on either side. The figure below shows the first two individual files which make up the filter as shown on page 113. These files, along with the other individual geometry files, are available in the “bpfilter” directory that you copied at the beginning of this chapter.

bpf_p1.geo

bpf_p2.geo

TIP Cell size may vary from geometry file to geometry file as long as the physical dimensions of the circuit are preserved. Therefore, choose a cell size for each circuit piece which provides the most efficient analysis without sacrificing accuracy.

116

Chapter 9 Circuit Subdivision - A Filter Example

Analyzing the Geometry Files

In this case, each of the individual geometry files were analyzed at five frequencies: 4.0, 7.75, 11.5, 15.25 and 19.0 GHz. The individual files are analyzed at the same first and last frequency as the overall analysis and at enough points in between to provide for reasonable interpolation of data at frequencies which fall between these values. When the network file analysis is performed, em will interpolate to provide simulation data at other frequencies. The figure on page 118 shows the Smith charts for the file “bpf_p2.d” analyzed at the five frequencies cited above and the same geometry file analyzed only at 4.0 GHz and 19.0 GHz. As you can see from the Smith chart on the right using only two data points would yield erroneous data when interpolating. On the other hand, five frequencies yields a fairly smooth curve and will provide acceptable interpolated data.

117

EM

Once the individual files have been created, each should be analyzed using em. When importing response data, em will interpolate data if the imported analysis results are not at exact evaluation frequencies. This ability to interpolate between frequencies in the input data files can provide a great deal of efficiency by eliminating the need to simulate at each frequency.

Em User’s Manual

The Smith charts for file “bpf_p2.d” and a subset of only two frequencies. Em analyses need to be performed at enough frequencies to allow accurate interpolation of data.

Creating the Network File The next step after analyzing the individual files is to create a network file which uses the response data from the analyses as input for a network which is equivalent to the whole circuit. The listing on page 119 shows the network file, “bpf_p.net”, that will be used for this example. The network file instructs em to perform the following steps:

118

1

Import the “.d” data files resulting from the analysis of the individual geometry files.

2

Define the network HALF which defines half the bandpass filter.

3

Define a circuit FILTER comprised of two HALF networks connected together.

4

Analyze FILTER from 4.0 GHz to 19.0 GHz in steps of 0.05 GHz.

Chapter 9 Circuit Subdivision - A Filter Example

!< FTYP NET ! Net file for Bandpass filter DIM FREQ GHZ

CKT S3P 1 2 3 bpf_p1.d S4P 2 3 4 5 bpf_p2.d S4P 4 5 6 7 bpf_p2.d S4P 6 7 8 9 bpf_p3.d S4P 8 9 10 11 bpf_p4.d S4P 10 11 12 13 bpf_p4.d S4P 12 13 14 15 bpf_p5.d S4P 14 15 16 17 bpf_p6.d S4P 16 17 18 19 bpf_p6.d S4P 18 19 20 21 bpf_p7.d S4P 20 21 22 23 bpf_p8.d DEF3P 1 22 23 HALF HALF 1 22 23 HALF 2 23 22 DEF2P 1 2 FILTER

EM

VAR

The network file, “bpfilter.net.” !Insert !Insert !Insert !Insert !Insert !Insert !Insert !Insert !Insert !Insert !Insert !Define

data file bpf_p1.d data file bpf_p2.d data file bpf_p2.d data file bpf_p3.d data file bpf_p4.d data file bpf_p4.d data file bpf_p5.d data file bpf_p6.d data file bpf_p6.d data file bpf_p7.d data file bpf_p8.d network HALF

!Insert network HALF !Insert network HALF !Define network FILTER

FILEOUT FILTER TOUCH bpf_p.rsp S DB R 50

!Define output filter as “bpf_p.rsp”

FREQ SWEEP 4.0 19.0 0.05 !Analyze FILTER from 4.0 GHZ to 19.0 GHZ in 0.05 GHZ step

The listing for the network file, “bpf_p.net”, used for this example.

119

Em User’s Manual The nodal network HALF is illustrated below. The network HALF is equivalent to half the filter in “bpf_w.geo.” The network file then connects the two HALF circuits, in the network FILTER, substituting the response data in place of the circuit. bpf_p1.d 1

2

bpf_p2.d

2

1

3

3

2

4

1

…

20

bpf_p7.d 1

3

2

4

bpf_p8.d

22

1

3

2

4

21

3

23

Once half the filter is defined, the two halves are connected to make a whole circuit equivalent to that found in “bpf_w.geo.” Note that the circuit HALF on the right in the figure below is reversed in both the vertical and horizontal directions.

1

1 HALF

2

22

3

HALF

1

2

3

2

23 Graphical representation of the nodal network, FILTER, as defined in bpf_p.net.

Em is then instructed to run an analysis of the whole structure, from 4.0 GHz to 19.0 GHz in 0.05 GHz steps, placing the output in the file “bpf_p.rsp.” The “.d” data files contain results of analyses at a subset of the frequencies just cited above. The network file analysis includes interpolation at frequencies for which data does not exist.

Analysis of the Network File The last step to complete the analysis of the filter is to analyze the network file using em. The analysis completes very quickly. Even with the analysis time of the individual geometry files, there is a considerable difference in the amount of time and computer resources used to obtain an answer for the bandpass filter as a sum of its parts, rather than approaching it as a whole. In this case, subdivision yielded approximately a 500x improvement in processing time and a 25x improvement in 120

Chapter 9 Circuit Subdivision - A Filter Example memory space required. This improvement comes as a result of reducing the number of subsections for any given analysis since both computation time and memory requirements rise sharply as the subsections go up, as shown on the chart below. For this example, the entire filter circuit used approximately 3500 subsections while any given individual piece only required a few hundred. Full Filter

EM

Time & Memory

Piecewise Analysis

Number of Subsections

Another contributing factor to the efficiency of the subdivision method comes from taking advantage of the interpolation performed when analyzing a network file. Interpolation reduces the number of analysis frequencies, thereby saving considerable computational time. In this instance, 301 frequency points were required. The need to analyze even more frequency points would increase the efficiency of using this method. Conversely, you would derive less benefit from this method if less frequency points were needed.

121

Em User’s Manual The emgraph plot shown below depicts the measured response data, “bpf_meas.rsp,” of the bandpass filter circuit as compared to the geometry file analysis of the whole filter, “bpf_w.d,” and the network file analysis of the subdivided circuit, “bpf_p.rsp.”

The measured data from the bandpass filter compared with the geometry file analysis and the network file analysis.

Alternate Approach In the example in this chapter, the analysis of the individual files was performed separate from the network file analysis. Using the GEO command in the CKT block of the network file, it is possible to perform the electromagnetic analysis as part of the circuit analysis. An example of a network file, “bpf_g.net” using this

122

Chapter 9 Circuit Subdivision - A Filter Example method is shown below. The analysis control file, “bpf_g.an” is available in the “bpfilter” directory. Each of the circuits is analyzed from 4.0 GHz to 19.0 GHz in steps of 3.75 GHz. !< FTYP NET ! Net file for Bandpass filter

EM

DIM FREQ GHZ VAR CKT GEO 1 2 3 bpf_p1.geo OPT=vmd CTL=bpf_g.an GEO 2 3 4 5 bpf_p2.geo OPT=vmd CTL=bpf_g.an GEO 4 5 6 7 bpf_p2.geo OPT=vmd CTL=bpf_g.an GEO 6 7 8 9 bpf_p3.geo OPT=vmd CTL=bpf_g.an GEO 8 9 10 11 bpf_p4.geo OPT=vmd CTL=bpf_g.an GEO 10 11 12 13 bpf_p4.geo OPT=vmd CTL=bpf_g.an GEO 12 13 14 15 bpf_p5.geo OPT=vmd CTL=bpf_g.an GEO 14 15 16 17 bpf_p6.geo OPT=vmd CTL=bpf_g.an GEO 16 17 18 19 bpf_p6.geo OPT=vmd CTL=bpf_g.an GEO 18 19 20 21 bpf_p7.geo OPT=vmd CTL=bpf_g.an GEO 20 21 22 23 bpf_p8.geo OPT=vmd CTL=bpf_g.an DEF3P 1 22 23 HALF !Define network HALF

!Analyze !Analyze !Analyze !Analyze !Analyze !Analyze !Analyze !Analyze !Analyze !Analyze !Analyze

bpf_p1.geo bpf_p2.geo bpf_p2.geo bpf_p3.geo bpf_p4.geo bpf_p4.geo bpf_p5.geo bpf_p6.geo bpf_p6.geo bpf_p7.geo bpf_p8.geo

HALF 1 22 23 !Define network HALF HALF 2 23 22 !Define network HALF DEF2P 1 2 FILTER!Define network FILTER FILEOUT FILTER TOUCH bpf_g.rsp S DB R 50!Save output in bpf_g.rsp FREQ SWEEP 4.0 19.0 0.05!Set up control frequencies for analysis

The network file, “bpf_g.net” which shows an example of performing geometry file analysis within a network file analysis.

The results using this method are identical to those obtained earlier in the chapter. The only real change is a reduction in the amount of user intervention required.

123

Em User’s Manual Note that an external analysis control file, analyzing at less frequencies than the network file, was used in the GEO statements to take advantage of the interpolation ability of the network analysis. Before executing a GEO statement, em checks for the existence of data at the specified control frequencies. If the data already exists, and the geometry file has not changed since the data was generated, em does not execute an electromagnetic analysis, but uses the available data. This check saves having to run all eleven analyses over, when only one of the geometry files has been changed.

124

Chapter 10 Intelligent Frequency Selection

EM

Chapter 10

Intelligent Frequency Selection

The frequency response of many circuits varies slowly in some frequency regions and rapidly in others. For efficiency reasons, it is often desirable to analyze such circuits with coarse frequency resolution in the slowly varying regions and fine frequency resolution in the rapidly varying regions. Typically, however, you do not know where the slowly and rapidly varying regions lie before the analysis is performed. This makes it necessary to either analyze over the entire frequency band with the fine resolution or to manually adjust the frequencies as the analysis is being performed. Em’s automatic frequency selection feature alleviates this difficulty. When enabled, this feature automatically determines where to place frequency points. In the rapidly varying regions, the frequency points will be spaced close together. In the slowly varying regions, frequency points will be spaced farther apart.

125

Em User’s Manual

Automatic Frequency Selection Example To demonstrate the automatic frequency selection feature, we will analyze the amplifier circuit shown below. This circuit consists of a pair of matching networks and a pair of auto-grounded ports at the location where the transistor S-parameters

126

Chapter 10 Intelligent Frequency Selection will be inserted.

EM

The geometry file “amp.geo” shows an amplifier with matching networks and a pair of auto-grounded ports at the location where transistor Sparameters will be inserted.

You can use Sonnet ⇒ Copy Examples to obtain the geometry file shown above, “amp.geo”.

127

Em User’s Manual Listed below is the network file for this example. It is the AUTO keyword in the FREQ data block that enables the intelligent frequency selection feature. Specifically, the AUTO line in the listing below tells em to analyze nodal network “amp” at 20 frequency points between 2.0 GHz and 20.0 GHz, with a precision of 0.010 GHz. The precision value specifies the finest frequency resolution allowed for the analysis. Thus, for this example, all frequency points chosen by em will be spaced by at least 0.010 GHz. !< FTYP NET ! File: amp.net ! Date: Jan 1, 1997 ! Amplifier example. DIM FREQ

GHZ

GEO S2P DEF2P

1 2 3 4 3 4 1 2

amp.geo amp_dev.s2p amp

OPT=vd

TOUCH

amp.rsp

S MA R 50

NET=amp

N=20

CKT

FILEOUT amp

CTL=internal

FREQ AUTO

2.0 20.0 0.010

The AUTO keyword in the FREQ block instructs em to analyze at 20 frequency points between 2.0 GHz and 20.0 GHz, with a precision of 0.010 GHz

!

WARNING The smaller the frequency “precision”, the longer the required computation time. Be careful when you choose the precision value for a particular analysis. You may use Sonnet ⇒ Copy Examples to obtain a copy of the network file, “amp.net”, along with the transistor S-parameter input file, “amp_dev.s2p”.

128

Chapter 10 Intelligent Frequency Selection Then click on the Run button or press the Return key to copy the two files to the default directory. Select em Control from the Sonnet task bar to open the em program window with a new job file.

2

Then click on the Network radio button in the File Type section of the job window.

3

Enter “amp.net” in the Network File text entry box.

4

The Internal Sweep radio button is already selected under the Frequency Control section.

5

Click on the Run command button to execute the em analysis.

EM

1

Note that this analysis may take a few minutes to complete. If you do not wish to wait for the analysis to complete, you can obtain the output response file, “amp.rsp” by using Sonnet ⇒ Copy Examples. The figure on page 130 shows the frequency response of |S11| and |S21| obtained from the analysis. Notice that the frequency points are not evenly spaced. The points are more concentrated near “corners” of the response and less concentrated along “straight edges” of the response. Emgraph may be invoked from em by clicking on the Open Graph button in the Output window or by selecting View ⇒ Open Graph from the main menu.

TIP If you wish to use the intelligent frequency selection features of em for a geometry file analysis, you may use the IFS command button in the Control Analysis dialog box when editing a Complex Sweep or an Analysis Control File.

129

Em User’s Manual

The magnitude of S11 and S21 is plotted versus frequency, using emgraph, for the example amplifier circuit.

Using FINDMIN and FINDMAX The previous example described the AUTO keyword and how it enables automatic selection of frequency points within a specified frequency band. In this section, the FINDMIN and FINDMAX keywords are described. FINDMIN and FINDMAX determine the frequencies where the circuit response reaches a minimum and maximum, respectively. To demonstrate FINDMIN and FINDMAX, we will again analyze the amplifier shown on page 127. This time, however, we will determine the frequency at which |S21| reaches a maximum. The plot above shows that the maximum occurs somewhere between 16.0 and 20.0 GHz. Thus, the analysis will be constrained to this frequency band.

130

Chapter 10 Intelligent Frequency Selection The figure below shows a network file that will accomplish the task described above. Here, the FINDMAX keyword has been specified in the FREQ data block. This statement instructs em to determine the frequency of maximum |S21| for network “AMP”, between 16.0 and 20.0 GHz, with a precision of 0.0010 GHz. The precision value specifies the finest frequency resolution allowed for the analysis. Thus, in this example, all frequency points chosen by em while searching for maximum |S21| will be spaced by at least 0.0010 GHz. EM

!< FTYP NET ! File: findmax.net ! Date: Sept 1, 1996 ! Determine frequency of maximum circuit response. DIM FREQ

GHZ

CKT GEO S2P DEF2P FILEOUT AMP

1 2 3 4 3 4 1 2

amp.geo amp_dev.s2p AMP

OPT=vd

CTL=internal

Touch

findmax.rsp

S MA R 50

NET=AMP

S2_1

16.0 20.0 0.0010

FREQ FINDMAX

The FINDMAX keyword in the FREQ block instructs em to determine the frequency of maximum |S21| between 16.0 and 20.0 GHz. A precision of 0.0010 GHz is specified.

You can use Sonnet ⇒ Copy Examples to obtain a copy of the network file shown above, “findmax.net”. The analysis can then be performed as follows: 1

Select em Control from the Sonnet task bar to open the em program window with a new job file.

2

Then click on the Network radio button in the File Type section of the job window.

131

Em User’s Manual 3

Enter “findmax.net” in the Network File text entry box.

4

The Internal Sweep radio button is already selected under the Frequency Control section.

5

Click on the Run command button to execute the em analysis.

Shown below is the sorted response file “findmax.rsp”. The maximum value of |S21| is 2.128671 at 17.847 GHz.

!< FTYP RSP ! S-Parameter Data File !< VER 6.0 !< PROG emgen !< DATE Thu Sep 1 12:00:00 1996 ! ! Network : AMP with 2 Ports # GHZ S MA R 50 16.0000000 0.770403 4.0070 1.616757 -127.5 17.6390000 0.284598 -4.083 2.107314 170.36 17.8330000 0.221455 17.305 2.128562 159.30 17.8460000 0.219825 19.365 2.128670 158.52 17.8470000 0.219719 19.526 2.128671 158.46 17.8480000 0.219615 19.687 2.128670 158.40 17.8490000 0.219515 19.849 2.128669 158.34 17.8530000 0.219140 20.497 2.128650 158.10 17.8620000 0.218464 21.972 2.128540 157.56 17.8870000 0.217841 26.150 2.127732 156.03 17.9610000 0.227358 38.411 2.120736 151.44 18.0000000 0.238855 44.146 2.114385 148.93 18.2300000 0.372999 62.464 2.028162 133.69 20.0000000 0.993288 7.9480 0.039507 162.26

0.123244 0.181064 0.185135 0.185291 0.185303 0.185314 0.185325 0.185368 0.185460 0.185668 0.185869 0.185731 0.180400 0.049682

-112.2 -162.2 -171.5 -172.2 -172.2 -172.3 -172.4 -172.6 -173.0 -174.3 -178.2 179.69 166.91 169.51

0.359779 0.446832 0.532143 0.538237 0.538707 0.539177 0.539648 0.541532 0.545782 0.557658 0.593240 0.611908 0.719460 0.737714

|S21| of the amplifier response reaches a maximum of 2.128671 at 17.847 GHz.

132

-1.676 50.421 50.882 50.809 50.803 50.797 50.791 50.765 50.703 50.501 49.658 49.027 43.969 3.8837

Chapter 11 The em Network File

EM

Chapter 11

The em Network File

The em network file consists of several data blocks. These data blocks define the circuit to be analyzed, designate how em should perform the analysis and determine what types of output the analysis is to produce. It is not necessary to create an em network file when the overall analysis is restricted to a geometry file. In that case, you only need the geometry file, “.geo.” However, when circuit analysis is to be performed, you must specify an em network file in the job window. Below you will find a detailed description of the em network file. This section is primarily intended to serve as a reference. Thus, you may wish to skim it initially and return later for specific details.

133

Em User’s Manual

Format of the em Network File The figure below shows the general format of the em network file. This file consists of the following: a header line, optional comment lines and data blocks which provide the information to perform the analysis. Everything in the network file is case insensitive. !< FTYP NET

Header

! File: example.net ! Date: Sept 1, 1996 ! General format of network file.

Comment Lines

DIM [Define units used to specify circuit parameters]

VAR [Define variables for use in CKT block]

CKT [Define circuit to be analyzed]

Data Blocks FILEOUT [Define output data files]

FREQ [Define analysis frequencies]

OUT [Define tabular output]

This is the general format of the em network file. This file consists of a header, optional comment lines and several data blocks which provide the information to perform the analysis.

134

Chapter 11 The em Network File

Header Line All network files must begin with the header line shown in , “!
Comments You may insert a comment on a line in the em network file by entering an exclamation point (!) followed by the comment. Whenever em detects an exclamation point, it considers everything to the right of the exclamation point to be a comment. There are no restrictions on the number or location of comments that may be placed in an em network file.

TIP The string “!< ” (exclamation point - “less than” sign - space) is a special anticomment symbol recognized by em. When em detects a line beginning with this string, it interprets the line as an uncommented line. For example, em would interpret the line “!< input data” as “input data”. This anti-comment symbol may be useful if you read em network files with other netlist programs.

135

Em User’s Manual

Data Blocks Data blocks are sections in the em network file which specify particular types of analysis information. Each data block begins with a special keyword to indicate the type of information that it contains. Following the keyword are appropriate parameter values for that section. Table 4 lists the data blocks which may be included in em network files.

Table 4 Em Network File Data Blocks Order

Data Block

Description

1

DIM

Define units used to specify circuit parameters.

2

VAR

Define variables for use in CKT data block.

3

CKT

Define circuit to be analyzed.

4

FILEOUT

Define output data files.

5

FREQ

Define analysis frequencies.

6

OUT

Define tabular output.

Table 4 shows the required order for the data blocks if they are present in the network file. The only data block required to be present is the CKT block. All other data blocks are optional.

136

Chapter 11 The em Network File

The DIM Data Block The DIM data block is an optional block which is used to override the default units for the em network file. The listing below shows the syntax for the DIM data block. DIM

unit1 unit2 . . . unitN

EM

parameter1 parameter2 . . . parameterN VAR ... CKT ... FILEOUT ... FREQ ... OUT ...

The DIM data block is used to override the default units for em.

parameter The parameter field designates a circuit parameter which can be specified in various units. For example, frequency is a circuit parameter which can be specified in HZ, KHZ, MHZ, GHZ, THZ, or PHZ. See Table 5 for a list of the circuit parameters which can be specified in the DIM data block.

137

Em User’s Manual

unit The unit field designates the particular unit which is used to specify a circuit parameter. For example, HZ is a particular unit which can be used to specify the frequency parameter. See Table 5 for a list of the units which can be specified in the DIM data block.

Table 5 DIM Data Block Parameters and Units

138

paramet er

Descriptio n

FREQ

Frequency

HZ (Hertz) MHZ (1e+6 Hertz) THZ (1e+12 Hertz) Hertz)

KHZ (1e+3 Hertz) GHZ (1e+9 Hertz) PHZ (1e+15

GHZ

RES

Resistance

OH (Ohms) MOH (1e+6 Ohms)

KOH (1e+3 Ohms)

OH

IND

Inductance

FH (1e-15 Henries) ries) NH (1e-9 Henries) MH (1e-3 Henries)

PH (1e-12 Hen-

NH

UH (1e-6 Henries) H (Henries)

units

Defaul t

CAP

Capacitance

FF (1e-15 Farads) NF (1e-9 Farads) MF (1e-3 Farads)

PF (1e-12 Farads) UF (1e-6 Farads) F (Farads)

PF

LNG

Length

MIL (mils) UM (1e-6 meters) CM (1e-2 meters)

IN (inches) MM (1e-3 meters) M (meters)

MIL

ANG

Angle

DEG (degrees)

RAD (radians)

DEG

Chapter 11 The em Network File

Rules for the DIM Data Block The parameters given in Table 5 may be included in a DIM block using any combination and order. If more than one unit is specified for a given parameter, em will use the final unit specified for that parameter. EM

Example The listing below shows an example of how the DIM data block may be used in an em network file. DIM FREQ IND CAP

MHZ PH PF

! ! ! !

DIM keyword indicates start of block Frequency in megahertz Inductance in picohenries Capacitance in picofarads

139

Em User’s Manual

The VAR Data Block The VAR data block is an optional block which defines variables for subsequent use in the CKT data block. The listing below shows the syntax for the VAR data block. DIM ... VAR

name1 = value1 name2 = value2 . . . . . . nameN = valueN CKT ... FILEOUT ... FREQ ... OUT ...

name The name field specifies the alphanumeric name of the variable being defined.

value The value field specifies the number which will be substituted for name everywhere name is referenced in the CKT data block.

140

Chapter 11 The em Network File

Referencing Variables in the CKT Data Block Variables that have been defined in the VAR data block may be referenced in the CKT data block by using a “^” symbol as follows: parameter^name

VAR char_impedance = 50.0 electrical_length = 90.0 reference_freq = 1.0 CKT TLIN

1 2

Z^char_impedance E^electrical_length F^reference_freq

141

EM

When em detects a “^” symbol in the CKT block, it will set parameter equal to the value which corresponds to name. As an example, consider the VAR and CKT data blocks listed below. In this example, em sets the Z-parameter equal to 50.0, the E-parameter equal to 90.0, and the F-parameter equal to 1.0.

Em User’s Manual

The CKT Data Block The CKT data block is a required block which defines the circuit to be analyzed. The figure below shows the syntax of the CKT data block. DIM ... VAR ... CKT

element1 element2 element3 aa. aa. aa. elementN defNp [filename1] FILEOUT ... FREQ ... OUT

142

nodes nodes nodes . . . nodes nodes

parameters parameters parameters aaa. aaa. aaa. parameters netname1

Chapter 11 The em Network File

element The element field specifies a type of circuit element. Table 6 lists the circuit elements which may be included in the CKT data block.

Table 6 Nodes

Parameters

Description

RES

n1 [n2]

R=res resistance in RES units.

Resistor

CAP

n1 [n2]

C=cap capacitance in CAP units.

Capacitor

IND

n1 [n2]

L=ind inductance in IND units.

Inductor

TLIN

n1 n2 [n3]

Z=Z0 characteristic impedance in RES units. E=len electrical length in ANG units. F=freq reference frequency for len in FREQ units.

Transmission line

143

EM

Element

CKT Data Block Elements

Em User’s Manual Table 6 Element TLINP

Nodes n1 n2 [n3]

CKT Data Block Elements Parameters Z=Z0 characteristic impedance in RES units. L=len physical length in LNG units. K=Eeff effective dielectric constant. A=atten attenuation in dB per unit length in LNG units. F=freq frequency for scaling atten in FREQ units.

Description Physical Transmission Line

Attenuation versus analysis frequency (f): A(f) = attenfreq = 0 A(f) = atten * sqrt(f/freq)freq > 0 SNP

n1 n2 ... nN [n(N+1)]

file [tag] alphanumeric name of input data file. optional field which references a particular data tag located in the input data file. The tag has the following format inside the file: “!< DATA_TAG tag”.

Input data file

GEO

n1 n2 ... nN [n(N+1)]

file alphanumeric name of the geometry file. OPT=opt list of em command line options. CTL=ctl em analysis frequencies: ctl may be an analysis control file or it may be the string “internal”. If “internal”, the em analysis is performed at frequencies specified in the FREQ data block.

Geometry file for electromagnetic analysis

144

Chapter 11 The em Network File

nodes

parameters The parameters field specifies the required parameter values for a given element. The parameters must be specified in the order given by Table 6.

defNp The defNp field defines a nodal network of one or more elements in the CKT block. The “n” in defNp is equal to the number of ports in the network (all of which are referenced to node 0). The nodes listed after defNp become ports in ascending order. Networks may be included as elements in subsequently defined networks.

netname The netname field assigns names to nodal networks defined with defNp. All assigned names must be unique; only alphanumeric characters are allowed; and the first character cannot be a number.

filename The filename field specifies the alphanumeric name of an output file in which 50 ohm S-parameter results will be stored. This field is optional. Note that an output file can also be specified with the FILEOUT data block.

145

EM

The nodes field specifies reference points to which circuit elements are connected. Any positive integer value may be used. Node 0 is reserved for ground. Bracketed nodes ([n]) are optional. If a bracketed node is not specified, it defaults to 0 (ground). For example, if you specify “RES 1 R=50.0”, a 50 ohm resistor is connected between node 1 and ground. Conversely, if you specify “RES 1 2 R=50.0”, a 50 ohm resistor is connected between node 1 and node 2. At each location in the CKT block where a defNp statement is used, the node numbers are reset.

Em User’s Manual You may include the string “$BASE” in filename. Upon detecting this string, em substitutes the network file name, minus the “.net” extension, for “$BASE”. For example, if the network file is named “filter.net”, and you specify “$BASE_new.rsp” for file name, em will create an output file named “filter_new.rsp”.

Example The listing below shows how the CKT data block may be used in an em network file. CKT S2P RES GEO GEO DEF2P

1 2 2 3 1

2 3 0 4 4

S3P Net1 DEF3P

1 2 3 3 4 1 2 4

filter.nd R=100.0 via.geo feed.geo Net1 device.d Net2

! ! OPT=vd CTL=internal ! CTL=feed.an ! $BASE_Net1.rsp!

Read data in filter.nd Insert 100 ohm resistor Analyze via.geo with em Use control file feed.an Define network Net1

! Read data in device.d ! Insert network Net1 ! Define network Net2

Using Data Tags Input data files may contain more than one “data set”. A “data set” consists of a Touchstone or Super-Compact header line, and a block of S-, Y- or Z-parameter data. The S-, Y- or Z-parameter block may contain one or more frequency points. The figure on page 147 shows the general format of an input data file with three data sets. Note that in addition to the header lines and parameter blocks, the figure also has three DATA_TAG lines. DATA_TAG lines are optional lines which you

146

Chapter 11 The em Network File can insert into an input file to distinguish one data set from another. In the example below, the first set has the tag “transistor,” the second set has the tag “feedline” and the third set also has the tag “transistor.” !< DATA_TAG transistor [Touchstone or Super-Compact header line] [S, Y or Z parameter block]

EM

!< DATA_TAG feedline [Touchstone or Super-Compact header line] [S, Y or Z parameter block] !< DATA_TAG transistor [Touchstone or Super-Compact header line] [S, Y or Z parameter block]

Input data files may contain more than one data set. To distinguish one data set from another, optional DATA_TAG lines are inserted into the data file.

When you use an SNP statement to read an input data file, em determines whether or not you have specified the tag field (see Table 6 on page 143). If you have not specified a tag, em simply ignores any DATA_TAG lines and reads all data present in the input file. Conversely, if you have specified a tag, em looks for a matching DATA_TAG line in the input file. If em does not find at least one matching DATA_TAG line, it issues an error message and stops. If em finds one or more matching DATA_TAG lines, it reads the data from each matching set. The “Use last data sets only” option gives you additional control over the reading of an input data file. This option instructs em to exclude all data sets in the input file except the final set with a tag which matches the tag specified in the SNP statement. If no tag is specified, the last data set in the file is read. This option is available in the Advanced Options dialog box, which is accessed by the Additional Options command button in the job window when a network file is selected.

147

Em User’s Manual

Geometry File Consistency The GEO statement instructs em to perform an electromagnetic analysis on a specified geometry file. Before starting the electromagnetic analysis, however, em checks the specified output file for previously generated results to determine whether or not any “consistent” results already exist. If so, em simply reads those results from the output file rather than perform the electromagnetic analysis. Previously generated em results are considered “consistent” if they satisfy two criteria: •

The geometry file has not been modified since the em data was created. This is determined by comparing the date and time the geometry file was last updated to the date and time listed on the CKDATE line in the em output file.

•

The batch command options specified in the GEO statement match the options listed on the CMD line in the em output file.

The circuit network capability advanced option, Do not check for consistency, may be used to override the geometry file consistency checks described above. When enabled, this option instructs the circuit network capability to read all data present in the em output file, regardless of consistency. This option may be selected in the Advanced Options dialog box, which is accessed by the Additional Options command button when a Network file is selected in the job window.

High Precision em Output Files The GEO statement instructs em to perform an electromagnetic analysis using the specified geometry file, analysis control file and batch command options (see Table 6). When doing so, em automatically creates one or two high precision output files in addition to the standard em output files. If only non-de-embedded data is requested, em creates one file with the extension “.pnd”. If only deembedded data is requested, em creates one file with the extension “.pd”. If both non-de-embedded and de-embedded data is requested, em creates both “.pnd” and “.pd” files.

148

Chapter 11 The em Network File The high precision output files contain S-parameter data in real/imaginary format. They are primarily intended for internal use by em. Generally, you do not need to be concerned with these files. However, it is recommended that you save the “.pnd” and “.pd” files whenever you save corresponding “.nd” and “.d” files. This will ensure that the high precision data is still available should you need to reanalyze in the future. EM

The FILEOUT Data Block The FILEOUT data block is an optional block which defines the output data files. The listing below shows the syntax for the FILEOUT data block. DIM ... VAR ... CKT ... FILEOUT netname1 netname1 netname2 aa. aa. aa. netnameN

format SPICE format aa. aa. aa. format

filename1 filename2 filename3 aaa. aaa. aaa. filenameN

[param][outtype] [impedance] [SPICE keywords] [param][outtype] [impedance] aaaa. aaaa. . aaaa. aaaa. . aaaa. aaaa. . [param][outtype] [impedance]

FREQ ... OUT

netname The netname field references a nodal network defined in the CKT block. 149

Em User’s Manual

format The format field indicates the format type to use for the output data. Valid options are: TOUCH

Touchstone format.

SPICE

Spice Lumped Model Synthesis (generic format)

PSPICE

Spice Lumped Model Synthesis (PSpice format)

SC

Super-Compact format.

CSV

Comma separated values. For use with commonly available spread sheet programs such as Excel.

filename The filename field specifies the name of the file which will store output results. If no extension is specified, the default extension of “.rsp” is used. The same filename may not be used in multiple FILEOUT statements. You may include the string “$BASE” in filename. Upon detecting this string, em substitutes the network file name, minus the “.net” extension, for “$BASE”. For example, if the network file is named “filter.net”, and you specify “$BASE_new.rsp” for filename, em will create an output file named “filter_new.rsp”.

param The param field is an optional field which is used to specify the desired output parameter type. S (S-parameters), Y (Y-parameters) and Z (Z-parameters) are available. If param is not specified, it defaults to S. Note that you must specify the param field if you have specified the outtype field. Not used with the SPICE format.

150

Chapter 11 The em Network File

outtype

impedance The impedance field specifies the normalizing impedances for each port in the network referenced by netname. Table 7 describes the available options for the impedance field. If you do not specify impedance, default settings are “R 50” for S-parameter data and “R 1” for Y- and Z-parameter data. Not used with the SPICE format.

Table 7 FILEOUT Data Block Impedance Keywords Argument

Description

R nr

All ports in netname are normalized by the same real value, nr. Valid for S-, Y- and Z-parameter data.

Z nr ni

All ports in netname are normalized by the same complex value, nr + jni. Valid only for S-parameter data.

TERM nr1 ni1 nr2 ni2 ... nrN niN

First port in netname is normalized by the complex value, nr1 + jni1. Second port in netname is normalized by the complex value, nr2 + jni2, etc. Valid only for S-parameter data.

151

EM

The outtype field is an optional field which specifies the format of the output data. This field can be set to MA (magnitude-angle in degrees), MR (magnitude-angle in radians), RI (real-imaginary) and DB (magnitude in dB-angle in degrees). If outtype is not specified, it defaults to MA. Note that you must specify the outtype field if you have specified the impedance field. Not used with the SPICE format.

Em User’s Manual

SPICE and PSPICE keywords These keywords are optional when using the SPICE or PSPICE format and may be placed in any order at the end of the line. If these values are not specified by the user, then the default value is used. CMIN: Minimum allowed capacitance (pF). The default value is 0.1 pF. LMAX: Maximum allowed inductance (nH). The default value is 100.0 nH. RMAX: Maximum allowed resistance (ohms). The default value is 1000.0 ohms. KMIN: Minimum allowed mutual inductance (dimensionless ratio). The default value is 0.01. RZERO: Resistor to go in series with all lossless inductors (resistance in ohms). Needed for some versions of SPICE. The default value is 0.0

Example The figure below shows an example of how the FILEOUT data block may be used in an em network file. FILEOUT Net1 Net2 Net3 Net4 Net5

152

Touch SC SPICE Touch Touch

file1.rsp file2.rsp file2.rsp file3.rsp file4.rsp

S CMIN=1.0 Y S

MA

TERM 0 80 50 0 50 0

LMAX=10.0 RI R 100 MA

Chapter 11 The em Network File

The FREQ Data Block The FREQ data block is an optional block which specifies analysis control parameters. The syntax of the FREQ data block is shown below. DIM ...

EM

VAR ... CKT ... FILEOUT ... FREQ keyword1 keyword2 aaa. aaa. aaa. keywordN

parameters1 parameters2 aaaaa. aaaaa. aaaaa. parametersN

OUT ...

keyword The keyword field specifies an analysis control keyword. Below are listed analysis control keywords that are recognized by em.

parameters The parameters field specifies all required information for the given keyword. The keyword descriptions below detail the required parameters for each keyword.

153

Em User’s Manual

Analysis Control Keywords For em ANN: ANN comment Write comment following ANN keyword to output response file. If ANN appears inside the FREQ block of a em network file, em will write comment to the specified response files. END: END Sort and analyze all frequencies (not yet analyzed) which precede the END keyword. SWEEP: SWEEP f1 f2 fstep Linear frequency sweep from f1 to f2 with a step size of fstep. ESWEEP: ESWEEP f1 f2 Nfreq Exponential frequency sweep from f1 to f2 with a common ratio between the Nfreq frequency points. LSWEEP: LSWEEP f1 f2 Nfreq Linear frequency sweep from f1 to f2. Step size is equal to(f2-f1)/(Nfreq-1). STEP: STEP f1 f2 … fN Discrete frequencies at f1, f2, …, fN. AUTO: AUTO NET=network N=Nfreq f1 f2 prec Automatic frequency selection using network as the basis. Em begins by analyzing at f1 and f2. It then analyzes at Nfreq frequencies between f1 and f2. The prec field specifies the frequency grid upon which frequencies are selected. For example, if prec = 0.10, f1 = 1.0 and f2 = 2.0, the algorithm is constrained to the following frequencies: 1.00, 1.10, 1.20, 1.30, 1.40, 1.50, 1.60, 1.70, 1.80, 1.90 and 2.00. Note that while network is used as the basis for selecting frequencies, all networks in the circuit are analyzed at each selected frequency.

154

Chapter 11 The em Network File FINDMIN: FINDMIN NET=network param [MAX=Nfreq] f1 f2 prec

The search for the minimum is constrained to frequencies which fall on a grid controlled by prec, f1, and f2 (see description of AUTO). If Nfreq is specified, the total number of frequency points analyzed is limited to the endpoints f1 and f2, plus Nfreq points between f1 and f2. Note that while frequencies are selected to determine the minimum frequency response of network, all networks in the circuit are analyzed at each selected frequency. FINDMAX: FINDMAX NET=network param [MAX=Nfreq] f1 f2 prec FINDMAX is identical to FINDMIN except that it finds the frequency at which the maximum frequency response of network occurs.

Sorted Frequency Sweeps Multiple frequency sweep statements may be specified in a single FREQ data block. For example, the FREQ block shown below has one STEP and two SWEEP statements. FREQ STEP SWEEP SWEEP

2.0 13.0 10.0 30.0 10.0 5 25 10

First frequency example.

Notice that the frequencies for the SWEEP statements overlap. By default, em sorts all analysis frequencies in ascending order before performing an analysis. Thus, if the above FREQ block were used in em, the analysis would be performed with the following frequency order: 2.0, 5.0, 10.0, 13.0, 15.0, 20.0, 25.0 30.0.

155

EM

FINDMIN finds the frequency at which the minimum frequency response of network occurs. The param field specifies a basis S-, Y- or Z-parameter using one of the following formats: pxy or px_y, where p is S-, Y- or Z, and x,y are a pair of port indices. The px_y format must be used when a port index with two or more digits is referenced. For example, S[port 1 - port 2] may be specified as S12 or S1_2, but S[port 15 - port 1] may only be specified as S15_1.

Em User’s Manual

Using END to Control the Order of Frequency Sweeps You could also set up the example shown on page 155 so that frequencies from the two SWEEP statements are not sorted. This is done in the example below. FREQ STEP SWEEP END SWEEP

2.0 13.0 10.0 30.0 10.0 5 25 10

END statements may be used in the FREQ block to control the order of frequency sweeps.

Here, an END statement has been inserted between the SWEEP statements. This statement tells em to analyze at all frequencies before END first, and then analyze the frequencies following END. For the example in , the analysis would be performed with the following frequency order: 2.0, 10.0, 13.0, 20.0, 30.0, 5.0, 15.0, 25.0. You could also include multiple END statements in the FREQ block. Then, all frequencies above the first END statement would be analyzed first, all frequencies between the first and second END statements would be analyzed second, etc.

Frequency Interpolation of em Output Data You may specify one set of analysis frequencies in the FREQ block of a network file and a second set of analysis frequencies in an external analysis control file referenced by a GEO statement (see “The CKT Data Block,” page 142). When you do, em performs the electromagnetic analysis only at the set of frequencies specified in the external analysis control file. Em then interpolates the electromagnetic analysis output data to obtain results at the set of frequencies specified in the FREQ block.

156

Chapter 11 The em Network File

AUTO, FINDMIN and FINDMAX for Basic Analyses

AUTO NET=GEO N=20 1.0 20.0 0.10

Overriding the FREQ Block You may specify an analysis control file (“.an”) along with a network file (“.net”) in the em job window. When you do, the information contained in the analysis control file overrides any information contained in the FREQ block of the network file. You may include any of the keywords and parameters described above in the analysis control file. Note that specifying an analysis control file along with a network file overrides only the information contained in the FREQ block of the network file. If the network file contains a GEO statement (see “The CKT Data Block,” page 142) which references an external analysis control file, the external analysis control file is not overridden.

157

EM

You may include AUTO, FINDMIN and FINDMAX statements in an analysis control file when using em to perform basic electromagnetic analyses. In this case, the network parameter must be set equal to “GEO” to identify the circuit in the geometry file as the “network” whose response is analyzed. For example, an AUTO statement might appear as follows:

Em User’s Manual

The OUT Data Block The OUT data block is an optional block which is used to specify tabular output data. The syntax of the OUT data block is shown below. DIM ... VAR ... CKT ... FILEOUT ... FREQ ... OUT

netname1 [COM=com] netname2 [COM=com] aaa. aaa.

meas1

[meas2 ...]

filename1 [DELIM=delim]

meas3

[meas4 ...]

filename2 [DELIM=delim]

aa. aa.

aaaaa. aaaaa.

aaaa. aaaa.

. aaaaaaa. . aaaaaaa.

netname The netname field references a nodal network defined in the CKT data block.

meas The meas field specifies the type of measurement to output. Table 8 shows the measurement types which may be specified in the OUT block. A single line may include multiple measurement types.

158

Chapter 11 The em Network File

Table 8 OUT Data Block Measurement Types Description

MAG[px_y] or MAG[pxy]

Magnitude of px_y or pxy, where “p” is S, Y or Z, “x” is port x and “y” is port y. Note that pxy cannot be used if “x” or “y” is greater than 9. Examples: MAG[S1_1], MAG[Y11], MAG[Z10_20].

ANG[px_y] or ANG[pxy]

Phase angle of px_y or pxy in degrees (See MAG for details).

RE[px_y] or RE[pxy]

Real part of px_y or pxy (See MAG for details).

IM[px_y] or IM[pxy]

Imaginary part of px_y or pxy (See MAG for details).

DB[Sx_y] or DB[Sxy]

Magnitude in dB of Sx_y or Sxy. See MAG above for details.

filename The filename field specifies the alphanumeric name of the file in which the tabular data will be stored. You can send multiple measurements to the same filename. In this case, the measurement data will appear in multiple columns separated by the specified delim. You may include the string “$BASE” in filename. Upon detecting this string, em substitutes the network file name, minus the “.net” extension, for “$BASE”. For example, if the network file is named “filter.net”, and you specify “$BASE_new.out” for filename, em will create an output file named “filter_new.out”.

159

EM

meas

Em User’s Manual

delim The delim field is an optional field used to specify the delimiter between columns in filename. SPACE, TAB and COMMA delimiters are available. If the delim field is not specified, em defaults to TAB.

com The com field is an optional field used to specify a comment string. This string will precede all comments in filename. The comment string may be one or more characters in length. If the com field is not specified, em defaults to “!”.

Example Below is an example of how the OUT data block may be used in an em network file. Notice that the same output file may be specified on multiple lines. OUT Net1 Net2

160

MAG[S1_1]ANG[S1_1] file1.out RE[Z2_1]IM[Z2_1] file1.out DELIM=SPACE COM=##

Chapter 12 Using Diagonal Fill

EM

Chapter 12

Using Diagonal Fill

This chapter discusses the use of the diagonal fill option for a metal polygon. When a polygon is first added to a circuit, the default fill type is staircase. The metal will be “filled” in using whole cells, thus a diagonal edge would resemble a staircase. An example of this is shown in the figure on page 162.

161

Em User’s Manual

Metal “off”

Metal “on”

An example of staircase fill for a metal polygon.

On the left is the outline of the polygon with the metal fill turned “off”. On the right is the polygon with the metal turned “on” with a staircase fill. As can be seen, the cell fill makes a staircase approximating the diagonal edge of the polygon. Note that the error caused by such an approximation decreases as the X and Y cell sizes are decreased. Thus, it is possible to make this error arbitrarily small by choosing sufficiently small X and Y cell sizes. However, for a coarser cell size, the diagonal fill option, which uses triangular subsections on the edge, may provide a better fit.

162

Chapter 12 Using Diagonal Fill This fill option involves additional computation time. However, depending on the circuit, using this option may allow a given level of precision to be achieved more quickly. The circuit on page 162 is shown below with diagonal fill.

EM

A Coupled Open-Miter with Diagonal Fill As an example, a right angle mitered bend which is closely interacting with an open end will be analyzed. Attempts to model this discontinuity using a circuit theory based program do not include the fringing field interaction between the two discontinuities. The circuit is contained in “openmite.geo,” available using the Sonnet ⇒ Copy Examples command and is shown on the left in the figure on page 164. For details on how to use diagonal fill for a metal polygon, see “The Metalization Attributes dialog box,” page 119 in the Xgeom User’s Manual.

163

Em User’s Manual Notice that there are “gussets” where the miter polygon attaches to the transmission line shown on the left in the figure below. Due to the nature of the triangle subsections used in diagonal fill, acute angles (less than 90 degrees) cannot be modeled. Em automatically cuts acute angles off, as illustrated on the right. Thus, the tabs on the miter polygon.

On the left, “openmite.geo,” with a closely interacting mitered bend and open end discontinuity. If the mitered region is captured as a simple triangle, without tabs, as shown on the right, the vertices with acute angles are cut off.

This analysis required more time due to the use of diagonal fill to model the mitre accurately. Once included, however, the em analysis time is relatively insensitive to the amount of diagonal fill. Analyzing the file yields the following results: 10.0000000 1.000000 -150.6

Circuits with diagonal edges may benefit from using diagonal fill. The edges of such structures are frequently much better approximated with the diagonal edges allowed by diagonal fill. Include it only on polygons with diagonal edges that carry significant current. See references [42] and [68] in the Sonnet Bibliography for a detailed description of diagonal fill (triangle subsections).

164

Chapter 13 Vias and 3-D Structures

EM

Chapter 13

Vias and 3-D Structures

Up to this point, most of the consideration has been with two-dimensional (2.5-D) structures, i. e., only X and Y current has been needed. However, em can handle full 3-D current as well. The third (Z) dimension of current is handled by a special kind of subsection called a via.

Adding Vias to the Circuit Vias, as used in em, are actually more general than the vias usually used in circuit design. In circuit design, vias connect metal on the substrate surface to the groundplane beneath the substrate, a ground via. In em, vias can connect metalization between any substrate or dielectric layer, not just bottom layer to ground. We call these “level-to-level” vias: Thus, em’s vias can be used in modeling airbridges, spiral inductors, wire bonds and probes as well as the standard ground via.

165

Em User’s Manual Em’s vias use a uniform distribution of current along their length and thus are not intended to be used to model resonant length vertical structures. Keep the via lengths small with respect to a wavelength. To create vias, use xgeom to specify one or more polygon edges as edge-vias (see Chapter 4, “Using Vias,” in the Xgeom User’s Manual for details). Em places subsectional vias (called “via-posts”) along the entire length of all edge-vias. Vias always go up from the selected polygon edge with the length of the via equal to exactly the thickness of the dielectric layer. The via-posts are rectangular cylinders with a horizontal cross-sectional area equal to one cell. If you make the cell size smaller, the vias become smaller with more of them along the edge-via. Of course, the length of the edge-via is unchanged. Current in a subsectional via is uniform through out the body of the via and is Z directed. Via loss is determined by the polygon from which the via originates.

Restrictions on Vias The height of the via should be a small fraction of a wavelength. The via height is the same as the thickness of the substrate or dielectric layer it penetrates. So, for ground vias, this is usually no problem. If a microstrip substrate is a significant fraction of a wavelength thick, overmoding also becomes a major problem. If vias are used to form, for example, a septum, or an interior wall, keep an eye on potential problems.

Simple Via Example A simple via is stored in the file “via.geo” and is shown in the figure on page 167. This file was generated by adding a via to ground at the end of the stub in “open_120.geo”, converting an open circuited stub to a short circuited stub. The via subsections are indicated by small triangles pointing down, indicating metal going into the screen. Triangles pointing up indicate metal coming out of the screen. The file “via.geo” is also available using Sonnet ⇒ Copy Examples.

166

Chapter 13 Vias and 3-D Structures The vias were captured by placing a polygon on the ground level indicating the periphery of the ground via. A square was used to model a via that is actually round. Such a model gives surprisingly accurate results. If an even more accurate model is needed, an octagon can be used.

Note that the top end of the via, shown below, is a “hat” which is larger than the via itself. There are no restrictions on the polygons at the top of a via. Em’s subsectioning algorithm handles the subsectioning accurately.

A simple via to ground. On the top, as it would appear in xgeom (top view). On the bottom, a view in perspective.

An analysis of the above via at 10 GHz with de-embedding gives an S11 phase of 148.32 degrees. A perfect short circuit is 180 degrees. This means the via has an inductive phase of about 32 degrees.

167

EM

After the polygon is in place at the base of the via, each side of the base polygon is converted into vias going up. See “Making Entire Polygons Vias,” page 57 in the Xgeom User’s Manual for details.

Em User’s Manual

A Conical Via One may simulate a conical ground via with a staircase approximation. Simply divide, say, a 100 µM GaAs substrate into four 25 µM substrates. Then put polygons (and specify edge-vias) at appropriate places to form a step approximation to the via sides. For an example, see the file “cvia.geo” in the examples directory. This circuit is a conical via to ground placed in the center of a through line, the purpose being to measure the via inductance. You may use Sonnet ⇒ Copy Examples to obtain the file “cvia.geo”. Then select xgeom from the Sonnet task bar and open the file “cvia.geo”. The “cvia.geo” file is a very detailed model of a conical via. If you are modeling a large circuit (say, an inter-stage matching network) with multiple vias, you may want to use a simpler model for faster analysis. A more complex structure using simple vias is the 20 µM wide shunt stub off a 50 ohm transmission line shown below (courtesy Raytheon Research Division). The stub is near enough to the via/capacitor combination that there is significant coupling, causing the error seen in the circuit theory calculation as compared to the measured and simulated em results shown on page 169. PORT 1

PORT 2

VIA MIM CAP

The coupling between the 20 µM wide stub and the via/capacitor combination requires an electromagnetic analysis.

168

Chapter 13 Vias and 3-D Structures The em analysis accurately models the coupling and would have eliminated a redesign and re-fabrication cycle. The circuit theory analysis requires only a fraction of a second per frequency while em requires 13 seconds per frequency (HP-710). But, in this case, since the circuit theory analysis yields the wrong answer, taking the time to perform an electromagnetic analysis has a very high return. 1.0

EM

Circuit Theory analysis chart 0.9 Mag S21 Em Data Measured

0.8

0.7 10

15

20

25

Frequency (GHz)

If the circuit had been spread out (the stub straightened out and kept at least 100 µM from any other components), circuit theory would probably have provided an accurate analysis and electromagnetic analysis would be unneeded. This circuit, “raystub.geo,” can be obtained using Sonnet ⇒ Copy Examples.

169

Em User’s Manual

170

Chapter 14 Dielectric Bricks

EM

Chapter 14

Dielectric Bricks

Although em is primarily a planar electromagnetic simulator, it also has the capability to add “dielectric brick” material anywhere in your circuit. A dielectric brick is a solid volume of dielectric material embedded within a circuit layer. See the illustration below. Dielectric bricks can be made from any dielectric material (including air) and can be placed in circuit layers made from any other dielectric material (including air). For example, dielectric bricks can be used to simulate structures such as a dielectric resonator block in an “air” circuit layer, or an “air hole” in a dielectric substrate circuit layer.

171

Em User’s Manual

Level 1

Level 0 Dielectric Brick

Dielectric Layer Level 0 Metal Dielectric Layer Level 1 Metal Side View of Circuit shown above. All realizable values for the dielectric constant, loss tangent and bulk conductivity can be used. Furthermore, it is possible to set these parameters independently in each dimension to create anisotropic dielectric bricks. Em is appropriate for simple structures using dielectric bricks; however, for more complicated circuits you may need a full 3-D electromagnetic analysis tool like Micro-Stripes from KCC Ltd. Micro-Stripes is a powerful full 3-D EM analysis

172

Chapter 14 Dielectric Bricks tool based on the Transmission Line Matrix (TLM) technique, and is ideal for the analysis of waveguide components, non-planar circuit structures, transitions and antennas. Micro-Stripes is available in North America through Sonnet Software.

Applications of Dielectric Bricks

Dielectric Brick Parameters Loss in the dielectric bricks is calculated in the same fashion as loss for the dielectric layer. For a detailed discussion of these parameters, see “Dielectric Layer Parameters,” page 46. See “Defining Dielectric Brick Materials,” page 65 of the Xgeom User’s Manual for information on setting these parameters.

Guidelines for Using Dielectric Bricks Subsectioning Dielectric Bricks A dielectric brick simulates a volume of dielectric material. Because a brick simulates a volume, it must be subsectioned in the X, Y and Z dimensions. The more subsections (better resolution) used in each dimension, the more accurate the analysis. X/Y subsectioning of dielectric bricks is identical to X/Y subsectioning of metal polygons. You can control the X/Y subsectioning of both through your choice of grid size, XMIN, YMIN, XMAX, YMAX and subsections-per-wavelength. See Chapter 3, “Subsectioning,” for details.

173

EM

The use of dielectric bricks is appropriate for applications where the effects of dielectric discontinuities or anisotropic dielectric materials are important. Examples of such applications include dielectric resonators, dielectric overlays, airbridges, microstrip-to-stripline transitions, dielectric bridges and crossovers, microslab transmission lines, capacitors and module walls.

Em User’s Manual Z subsectioning of dielectric bricks is controlled by the “number of Z-partitions” parameter. This parameter specifies the number of Z subsections for all dielectric bricks on a particular dielectric layer. See “Z-Partitioning,” page 67 of the Xgeom User’s Manual for information on setting this parameter.

Using Vias Inside a Dielectric Brick Vias through dielectric bricks are treated the same as vias through the standard dielectric layers. Note that via ports inside dielectric bricks are not allowed.

De-embedding and Dielectric Bricks The effects of port discontinuities and interconnecting transmission lines are removed from the simulated data by using de-embedding. In general, the deembedding procedure is not changed when dielectric bricks are present in the circuit. However, there are two important considerations. First, a dielectric brick located in the interior of a circuit (not touching a box wall) represents a discontinuity in that circuit. Thus, no reference plane should be set so that it extends from the box wall beyond the leading edge of the dielectric brick. This is similar to the restriction that reference planes should not extend beyond metal discontinuities in the circuit. The second consideration concerns the creation of the two “thru-line” standards used to perform the de-embedding. To create the standards for a given box wall, em identifies all metal polygons in the circuit which have an edge in common with the wall, and then creates one thru-line extending from that edge for every such polygon. If there are also dielectric bricks with an edge in common with the given box wall, em also creates “dielectric brick thru-lines” which extend from that edge. In short, this makes it possible to de-embed ports on box walls where one or more dielectric bricks have an edge in common with that box wall.

174

Chapter 14 Dielectric Bricks

Air Dielectric Bricks

Limitations of Dielectric Bricks Diagonal Fill Diagonal fill is not allowed for dielectric bricks. All dielectric bricks must use “staircase fill”. Thus, dielectric bricks with curved or rounded edges must be stairstep approximated. Note that the error caused by such an approximation decreases as the X and Y cell sizes are decreased. Thus, it is possible to make this error arbitrarily small by choosing sufficiently small X and Y cell sizes.

Antennas and Radiation Patvu does not support dielectric bricks. Circuits containing dielectric bricks can be analyzed with patvu, but the radiation effects of the dielectric bricks are not accounted for in the analysis.

Ebridge The ebridge interface to the HP-EEsof Series IV or ADS circuit analysis program does not create dielectric bricks.

175

EM

Dielectric bricks can be made of any dielectric material (dielectric constant, loss tangent and bulk conductivity) and can be placed in any circuit layer. This allows, for instance, “alumina” bricks to be created in an “air” circuit layer. However, it is also possible to reverse this scenario. Dielectric bricks made of “air” can also be created in alumina circuit layers. This is an important consideration to remember. Depending upon the circuit geometry for a given application, this ability to reverse the dielectric characteristics may simplify the circuit and make it faster to analyze.

Em User’s Manual

176

Chapter 15 Antennas and Radiation

EM

Chapter 15

Antennas and Radiation

To this point, this manual has been focused on using em for the analysis of high frequency circuits and transmission structures. However, there is a large class of radiating structures for which em has proven very useful. This chapter describes how to use em to analyze 3-D planar radiating structures, such as microstrip patch arrays and microstrip discontinuities, using the “Open Waveguide Simulator” technique. The underlying assumptions of this technique are described in detail. Common modeling mistakes are also pointed out. Examples are provided to illustrate the correct use of the modeling technique. If you find that these modeling techniques are not sufficient to handle your design, then you may need a full 3-D electromagnetic analysis tool like Micro-Stripes from KCC Ltd. Micro-Stripes is a powerful full 3-D EM analysis tool based on the Transmission Line Matrix (TLM) technique, and is ideal for the analysis of waveguide components, non-planar circuit structures, transitions and antennas. Micro-Stripes is available in North America through Sonnet Software.

177

Em User’s Manual

Background Since em is an analysis of 3-D planar circuits in a completely enclosing, shielding, rectangular box, the analysis of radiating structures is not an application which immediately comes to mind. However, em can be used to simulate infinite arrays using a waveguide simulator. In this technique, as shown in on page 179, a portion of the array is placed within a waveguide. The waveguide tube is vertical, connecting the radiating patches to the termination, which is a matched load. The images formed by the waveguide walls properly model the entire infinite array scanned to a specific angle. The waveguide simulator inspired what we now call the Open Waveguide Simulator Technique described in the next section.

Modeling Infinite Arrays The sidewalls of the shielding box in the em analysis easily represent the sidewalls of the waveguide in the infinite array waveguide simulator. A side view is shown in the figure on page 179. Providing a termination for the end of the waveguide requires a little more thought. Any waveguide mode can be perfectly terminated by making the top cover resistivity in em equal to the waveguide mode impedance. This can be done in xgeom automatically at all frequencies and all modes by selecting “WGLOAD” from the metals in the Top Metal drop list in the Box Parameters dialog box.

178

Chapter 15 Antennas and Radiation

Waveguide Walls

fc 2  Z TM = η 1 – ----  f

f > fc Waveguide Termination

f > fc

EM

η Z TE = ------------------------fc 2 1 –  ----  f

Array Patches

v c mπ 2 nπ 2 f c = ------  ------- +  ------  B 2π  A 

Substrate The waveguide simulator for infinite arrays inspired the technique described here. In this side view, the waveguide walls form images of the array of microstrip patches, simulating an infinite array. vc is the velocity of light in the medium filling the waveguide.

In a phased array with the array scanned to a specific direction, a single waveguide mode is generated. The em software can model the waveguide simulator of that infinite array just by setting the top cover impedance to the impedance of the excited waveguide mode.

Modeling an Open Environment If we can use a closed (i.e., terminated) waveguide to model an infinite array, we can also model radiation from a finite array; although, it must be done under certain conditions. It is important to keep in mind that, unless the analysis is

179

Em User’s Manual carefully prepared, these conditions are easily violated, yielding incorrect results. When the conditions are met, useful results can be obtained, as shall be demonstrated. First Condition: Make both of the lateral substrate dimensions greater than one or two wavelengths. When using the Open Waveguide Simulator, we view the sidewalls of the shielding box as forming a waveguide whose tube extends in the vertical direction, propagating energy from the antenna toward the “Termination” in . Radiation is then approximated as a sum of many waveguide modes. If the tube is too small, there are few, if any, propagating modes, violating the First Condition. There is an easily made mistake when modeling radiation from small discontinuities. Discontinuities are usually small with respect to wavelength. For a discontinuity analysis, the sidewalls are usually placed one or two substrate thicknesses from the discontinuity. In this case, the substrate dimensions are unlikely to meet the First Condition. If the sidewalls form below a cut-off waveguide, there is no radiation. Second Condition: Make sure the sidewalls are far enough from the radiating structure that the sidewalls have no affect. Another way to look at this condition is to consider the image of the structure (discontinuity or antenna) created by the sidewall. Position the sidewall so that the image it forms has no significant coupling with the desired structure. Usually two to three wavelengths from the sidewall is sufficient for discontinuities. For single patch antennas, one to three wavelengths is suggested. Requirements for specific structures can easily be greater than these guidelines. If the First Condition requires a larger substrate dimension than the Second Condition, it is very important that the larger dimension is used. If you are using patvu, the larger the box the better. Patvu assumes that Sparameters from em are from a perfect open environment. If some of the power is reflected due to a box that is too small, the input power calculated by patvu will be slightly incorrect. Patvu then calculates antenna efficiencies greater then 100%. If this occurs, the box size should be increased.

180

Chapter 15 Antennas and Radiation Third Condition: Place the top cover outside the fringing fields (i.e., near field) of the radiating structure, preferably a half wavelength. If this condition is violated, the resistive top cover becomes involved in the reactive fringing fields which form the near field of the radiator. This changes what would have been reactive input impedance into resistive input impedance, overestimating the radiation loss. EM

Do not place the top cover thousands of wavelengths away from the radiator. Extreme aspect ratios of the box should be avoided. Empirical data for patch antennas has shown that a distance of about 1/2 wavelength works best. Fourth Condition: Set the top cover to Free Space. This value is a compromise. As shown by the equations on the previous page, all TE modes have a characteristic impedance larger than 377 ohms (Ω), while all TM modes are lower. Thus, while a 377 ohms/square top cover does not perfectly terminate any mode, it forms an excellent compromise termination for many modes. This approximates removing the top cover of the box. If the box is large, it, in turn, approximates radiation, as shall be demonstrated. Fifth Condition: The radiating structure can not generate a significant surface wave. If there is a significant, compared to required accuracy, surface wave, it is reflected by the sidewalls of the box. Unless this is the actual situation, such antennas are inappropriate for this technique. Actually, the Fifth Condition is a special case of the Second Condition, since if there is significant surface wave, the Second Condition cannot be met. This condition is stated explicitly because of its importance. In general, any surface wave is both reflected and refracted when it encounters the edge of the substrate. This boundary condition is different from either the conducting wall of Sonnet or the infinite substrate provided by a true open space analysis.

181

Em User’s Manual A dual patch antenna is illustrated conceptually below. Free Space Top Cover

Double Patch Antenna

Feed point

Radiation can be simulated by including a lossy top cover, a lossy dielectric layer (optional) and by placing the sidewalls far from the radiator (drawing not to scale). Place the top cover one half wavelengths from the radiator.

The feed point is created in xgeom by creating a via to ground at the feed point. Then the ground end of that via is specified as a port, just as one would specify a more typical port on the edge of the substrate at a box sidewall. A file showing an antenna similar to this one is named “patch.geo” and is available using the Sonnet ⇒ Copy Examples command.

182

Chapter 15 Antennas and Radiation

Validation Example

Top view of a triple patch antenna (courtesy of Matra Defense). The central patch is fed with a coaxial probe (indicated by a down pointing triangle). Each patch is resonant at a different frequency to increase the overall antenna bandwidth.

Good results are also regularly obtained on single microstrip patch antennas. We cite this example as one of the more sophisticated antennas analyzed using the Open Waveguide Simulator technique. In this antenna, each patch has a slightly different resonant frequency, resulting in an increased bandwidth. The antenna is fed from below with a coax probe attached to the central patch. The feed point is indicated with a triangle. The substrate is 3.04 mm thick with a dielectric constant of 2.94. The drawing is to scale with substrate dimensions of 200 mm x 100 mm. The top cover is 200 mm above the substrate surface. Cell size is 0.78125 mm square. A loss tangent of 0.001 is used in both air and substrate. The small air loss helps terminate the propagating modes. The antenna geometry file, “tripat.geo,” is available using the Sonnet ⇒ Copy Examples command. 183

EM

For validation, we offer work performed by E. Ongareau of Matra Defense, Antennas & Stealthness Dept., France, as presented at the 1993 EEsof User’s Group meeting at HYPER in Paris. (Reprinted with permission.) The antenna is a triple patch structure, with a top view shown below. The antenna is a test realization intended only for validation. It is not designed for optimum VSWR.

Em User’s Manual The chart below shows the result. We see that the resonant frequencies of each patch (i.e., the low VSWR points) have differences between measured and calculated of about 1%. This is typical of most analyses of patch antennas using this technique. The differences in resonant frequency (i.e., the reflection zeros) then determine the differences in the rest of the VSWR plot. The degree to which these differences are due to analysis error, fabrication error and measurement error cannot be determined from this data.

7

Measured Calculated

6

VSWR

5 4 3 2 1 1% 2.0

2.1

2.2 2.3 2.4 Frequency (GHz)

2.5

2.6

The measured and calculated data for the triple patch antenna were obtained completely separately, so there was no chance to “tweak” the model for agreement.

184

Chapter 15 Antennas and Radiation If the typical differences between measured and calculated data shown on page 184 are acceptable, given the specific requirements for a particular project, then the Open Waveguide Simulator technique can provide useful results.

EM 185

Em User’s Manual

186

Chapter 16 SPICE Lumped Model Synthesis

EM

Chapter 16

SPICE Lumped Model Synthesis

This chapter describes how to use em to automatically synthesize SPICE files. This capability is useful for circuits which are small with respect to the wavelength of the highest frequency of interest. This includes structures such as discontinuities like step, tee and cross junctions. Other applications include modeling cross-talk and propagation delay in digital interconnect circuits and multiple spectrum circuits that combine digital, analog and RF functions. This option automatically takes the results of the electromagnetic analysis of a circuit and synthesizes a lumped model of inductors, capacitors, resistors and mutual inductors. This information is then formatted and saved as an ASCII SPICE file.

187

Em User’s Manual A second SPICE related capability of em is the generation of L, C, R and G matrices for N-coupled transmission lines. These matrices represent the distributed parameters of the transmission lines (for example, inductance per unit length).

Class of Problems The SPICE generation capability is intended for any circuit which is small with respect to the wavelength of the highest frequency of excitation. Typically, 1/20th wavelength is an appropriate limit. (If a circuit is too large, split it into two or more circuits and analyze each separately.) This limitation is due to the circuit theory limitations of modeling a circuit with lumped elements. The Sonnet electromagnetic analysis is not intrinsically limited in this fashion. The model generated by the analysis includes any lumped elements (including mutual inductors) between any ports of the circuit layout. Lumped elements from any port to ground are also included. The synthesis capability does not allow internal nodes (nodes which are not connected to a port in the layout) with the single exception of the internal node required to specify a resistor in series with an inductor. Any circuit which requires internal nodes for an accurate model should be split into several parts so that the required points become nodes. Internal ports without ground reference give incorrect results. Any internal ports should be carefully specified and checked for reasonable results. The SPICE file generation capability is usually not appropriate for large microwave circuits because such circuits are usually larger than a small fraction of a wavelength. The SPICE model synthesis capability is fast enough that it can be used on circuits with hundreds of ports.

188

Chapter 16 SPICE Lumped Model Synthesis

Using The SPICE Option

Make sure the frequency is not too low. In circuits with vias, when the subsection size is less than 10-5 wavelength, numerical precision can be a problem. For example, if the subsection size is 1 mm, it would be unwise to analyze below 1 MHz. If the frequency specified is low enough, em warns you that you may need quadruple precision. The quadruple precision option can be enabled by clicking on the Quad Precision checkbox in the Additional Options dialog box. Be aware, however, that Quad Precision can slow the analysis down substantially; use the option only when necessary. Analyzing the circuit at a higher frequency is often a better solution. After completing the analysis, always do a “reality check” for reasonable values. If you have bad data, the frequency may be too high or too low. If the frequency is too low, the solution may have unity S-parameters, causing a strange SPICE model. To be absolutely sure your results are good, select a second pair of frequencies, different from the first pair by, say, a factor of two, and re-analyze the circuit. You should obtain similar results between the two analyses. To use the SPICE generation option, click on the Output Files button in the job window, which will open the Select Output Files dialog box. Set the .lc name checkbox to “on” and enter a file name in the corresponding text entry box. This will select the SPICE option specifying the default value of 1 for the displayed precision for the lumped element values. This results in output like “C1 1 3 12.1pf”.

189

EM

First, select two frequencies, with an approximate 10% difference, for analysis and specify them in the analysis control file or in the Simple Sweep option in em. The SPICE synthesis needs electromagnetic results at two frequencies to accomplish its work. If four frequencies are specified, two lumped models are generated. If an odd number of frequencies is specified, em terminates with an appropriate error message.

Em User’s Manual If you wish a higher number of digits, you must open the Additional Options dialog box and enter -xn in the Advanced text entry box, where n is an integer value from 0 to 7. When working with very small structures, specifying, for example, “-x4” may be desired. This results in output like “C1 1 3 0.0312pf”. Values from 0 to 7 may be used. The digits 8 or 9 generate the same output as 7. Any capacitors with a value of 0 are not included in the model. Any inductors or resistors which are essentially open circuits are also excluded. To reduce the number of lumped elements in the model, open circuit limits can be specified in the Spice Control dialog box, shown below. This dialog box is used when editing Complex Frequency controls or an Analysis Control file.

The values are defined as follows: CMIN: Minimum allowed capacitance (pF). The default value is 0.1 pF. LMAX: Maximum allowed inductance (nH). The default value is 100.0 nH. RMAX: Maximum allowed resistance (ohms). The default value is 1000.0 ohms. KMIN: Minimum allowed mutual inductance (dimensionless ratio). The default value is 0.01. RZERO: Resistor to go in series with all lossless inductors (resistance in ohms). Needed for some versions of SPICE. The default value is 0.0 All calculated component values which fall outside the allowed range specified by the user in the frequency controls are excluded from the resulting lumped model. The RZERO entry is provided for those versions of SPICE which need inductors to have some small loss to avoid numerical difficulties. The default value of 0.0 disables this capability. Output is sent to the file ending in “.lc” specified in the Output Files dialog box.

190

Chapter 16 SPICE Lumped Model Synthesis In most cases, the de-embedding option should also be used. Otherwise the small shunt capacitance from the port discontinuity is also included in the lumped model.

PSpice Option

N-Coupled Line Option If the structure being modeled is an N-coupled line, the SPICE option described above can be applied to a short length of line to generate a single section of the LC-L-C. lumped model of a transmission line. This approach has two disadvantages. First, the lumped LC model of a transmission line is approximate. Second, the LC model can be very time consuming to analyze. There is a better alternative. That alternative is to use distributed LC parameters of a transmission line, specifically, inductance per unit length and capacitance per unit length. Analyses which use such data are much faster than those which use simple lumped models. In addition, accuracy is maintained at all frequencies for which TEM mode propagation is an adequate approximation. For a single line, the L and C distributed parameters are each a single number. For N-coupled lines, L and C become N by N matrices. When metal loss is included, we now also have an R matrix. The resistance is in series with the inductance. When there is dielectric loss, a G matrix is also calculated. The conductance is in parallel with the capacitance. The synthesis determines whether a G or R matrix are needed only from the calculated S-parameters. The circuit geometry is not referenced at any time.

191

EM

A SPICE lumped model can also be generated using PSpice output format. All of the rules and assumptions described in the previous section for .lc SPICE files still apply. To generate a SPICE lumped model in PSpice format, specify "-pspice" under Advanced Options in em control. See "SAN-104B: Generating PSpice Files Using Electromagnetic Analysis" on page 565 in the Sonnet Application Notes for an illustration of the use of PSpice output format. The output file name extension is ".psp".

Em User’s Manual To generate RLCG matrices, select a “.lct” file in the Output Files dialog box and enable the De-embed option. De-embedding is often required as described in the previous section. The “ctl.an” file needs only one frequency specified. If two frequencies are specified, two RLCG matrix sets are generated. Your “.geo” file must be an N-coupled line with ports 1 through N as input and ports N+1 through 2N as output. The input of line M should be port M and its output should be port M+N. The software does not check for this condition, but issues a warning message if the number of ports is not an even number. This restriction does not apply to generating “.lc” files, only generating “lct” files. There is no limit on N. The results are per unit length, where a unit length is the de-embedded length of the N-coupled line. The length must be short compared to the wavelength at the frequency of analysis.

A Simple Microwave Example Let’s say you want a lumped model for the steps discontinuity used in the beginning of this manual. A copy of the file “steps.geo” can be obtained by using Sonnet ⇒ Copy Examples. In order to synthesize a lumped model, two frequencies need to be analyzed in the following manner:

192

1

Select em Control from the Sonnet task bar. The job window will appear on your display.

2

Enter “steps.geo” in the Geometry File text entry box.

3

Click on the Simple Sweep option in the Frequency Control section. Enter 1.0 in the Start text entry box and 1.1 in the Stop text entry box.

4

The Verbose and De-embed options are already selected, so you need take no action on these items.

Chapter 16 SPICE Lumped Model Synthesis Click on the Output Files command button to open the Select Output Files dialog box.

6

Click on the .lc name checkbox to set it to “on.” The default name of “steps.lc” will appear in the corresponding text entry box. Click on the OK button to close the dialog box.

7

Click on the Run command button to execute the analysis.

EM

5

The lumped model is stored in the file specified in the Select Output Files dialog box, “steps.lc”:

* Limits: C>0.1 pF, L<100 nH, R<1000 Ohms, K>0.01. * Analysis time per frequency (mm:ss) -- 0:01 user, 0:00 system, 0:01 real * Analysis time per function (seconds) -- Modes: 0 Fill: 0 Solve: 0 * Analysis frequencies: 1000.000000, 1100.000000 MHz .subckt 1 2 C1 1 0 0.3pf C2 2 0 0.2pf L1 1 2 0.3nh .ends

193

Em User’s Manual The first line indicates the open circuit limits. For example all capacitors less than 0.1 pF are excluded from the model. These limits can be modified as discussed in “Using The SPICE Option,” page 189. The next two lines detail the processing time. The next line documents the analysis frequencies. This is followed by the SPICE lumped model. In this case we have two capacitors to ground (node 0) and one inductor connected between port 1 and port 2. Ground is node 0 and all the remaining node numbers correspond to ports of the same number. If two decimal places of precision are desired, open the Additional Options dialog box by clicking on the Additional Options command button in the em job window. Enter -x2 in the Advanced text entry box.

Topology Used for SPICE Output The topology of the lumped element model generated by em depends on the circuit being analyzed. In general, the model contains an inductor (in series with a resistor if using loss), a capacitor and a resistor (when using loss) connected in parallel from each port to ground. A similar parallel RLC network is also connected between each port. Therefore, a four-port circuit can contain more elements than a two-port circuit. Each inductor may also have a mutual inductance to any other inductor in the network. The figure on page 195 shows the most complex equivalent circuit possible for a two-port (mutual inductances not shown). Any values that are outside of the open circuit limits are not included.

194

Chapter 16 SPICE Lumped Model Synthesis

1

2

EM

Equivalent circuit of a two-port structure generated by em’s SPICE option. Mutual inductances also exist between all inductors, but are not shown. Any component whose value is outside of the open circuit limits are not printed in the SPICE output file.

195

Em User’s Manual

A High Speed Digital Example The figure below shows the top level of an example circuit. There are 32 input ports and 32 output ports for a total of 64 ports. The first eight bits (ports 1-8) go into the circuit, down to the second level, underneath all the other lines, and come up on the right hand side as the last eight bits. Thus, this is a byte-reversal network.

Top Level

The top level of the byte-reversal network. The byte order on input (left side) is reversed on output (right side). This is a top view.The arrow heads indicate connections between levels (down vias).

196

Chapter 16 SPICE Lumped Model Synthesis

Bottom Level EM

The bottom level of the byte-reversal network. The triangles indicate connections between levels (up vias).

A copy of the above file, “br32.geo” can be obtained using Sonnet ⇒ Copy Examples. The circuit was analyzed at 10 and 15 MHz, where each line is about 1 degree long. If you wish to perform the analysis yourself, do the following: 1

Select em Control from the Sonnet task bar. The main window with a new job window will appear on your display.

2

Enter “br32.geo” in the Geometry File text entry box.

3

Click on the Simple Sweep option in the Frequency Control section. Enter 10.0 in the Start text entry box and 15 in the Stop text entry box.

4

Select MHz from the Frequency Unit drop list at the end of the Simple Sweep.

197

Em User’s Manual 5

The Verbose and De-embed options are already selected, so you need take no action on these items.

6

Click on the Output Files command button to open the Select Output Files dialog box.

7

Click on the .lc name checkbox to set it to “on.” The file name “br32.lc” is automatically input in the corresponding text entry box. Click on the OK button to close the dialog box.

8

Click on the Additional Options button. The Advanced Options dialog box will appear on your display.

9

Enter “-e” in the Manual Options text entry box. This option reduces the number of subsections at cross-over points. Click on the OK button to close the dialog box and apply the option.

10

Click on the Run command button to execute the analysis.

The lumped model is stored in the file specified in the Select Output Files dialog box, “br32.lc”.

198

Chapter 16 SPICE Lumped Model Synthesis The following is a portion of the resulting SPICE model (… indicates information left out):

EM

.subckt bytervrs 1 2 3 4 . . . 63 64 C1 a1 0 a14.5pf C2 a1 2 a 1.6pf C3 a1 25 a0.7pf C4 a1 55 a0.1pf . . . C57 1 64 0 9.1pf L1 1 33 77.2nh L2 2 34 77.3nh . . . L32 32 64 79.6nh K1 L1 L2 0.2 K2 L1 L3 0.1 . . . K70 L31 L32 0.2 .ends

Nodes 1 - 64 correspond to the ports of the same number in the circuit layout, . Node 0 is ground. For example, C1 represents the capacitance from port 1 to ground. L1 represents the inductance from port 1 to port 33 (i.e., port 1 is connected to port 33). There is also a capacitance (not listed above) from port 33 to ground. C2 is a stray capacitance coupling ports 1 and 2, generating cross-talk. The capacitive coupling causes cross-talk whenever there is a time varying voltage difference between ports 1 and 2. Mutual inductance K1 inductively couples the port 1 to port 33 line (L1) to the port 2 to port 34 line (L2), also generating cross-talk. The mutual inductance causes cross-talk whenever either the port 1 or port 2 line carries a time varying current. A quick inspection of this file reveals the worst cases for cross-talk (i.e., largest mutual inductors and capacitors). 199

Em User’s Manual The sub-circuit was placed in a complete SPICE file with an example analysis shown below.

0.15

Crosstalk Voltage (Volts)

0.10

Port 25 (Reverse)

0.05 0.00 -0.05

Port 57 (Forward)

-0.10 -0.15

0

2

4

Time (nS)

6

8

10

A SPICE analysis was performed using the em synthesized model of the byte-reversal network. Shown here is the cross-talk to the port 25 - port 57 line caused by a 1 Volt signal with a 10 pS rise time on port 1. All ports are terminated in 50 ohms. Analysis courtesy of CONTEC Microelectronics USA.

Other Techniques Classical techniques use, for example, just an electrostatic or just a magnetostatic analysis to derive a model. This is adequate for uniform transmission lines embedded in homogenous dielectric (no different layers). In an arbitrary predominantly planar circuit, as we have here, a single static analysis provides only half a circuit model, just the capacitors or just the inductors.

200

Chapter 16 SPICE Lumped Model Synthesis Since em is a full dynamic analysis, both the inductive and capacitive portion of the model are obtained with one analysis (at two frequencies). In addition, the techniques usually used for the static analyses are of a volume gridding variety (e.g., finite elements, finite difference). Even under the simplifications allowed by static analysis, the circuit shown in is well beyond the capability of such software tools.

The model which results has 1057 capacitors, 32 inductors and 70 mutual inductors. If loss is included, the model would also include resistors. Many of the capacitors in the lumped model are 0.1 and 0.2 pF, just over the default minimum capacitance of 0.1 pF. A much simpler model (with a little less accuracy) is possible by including the command “CMIN 0.3” in the ctl.an file. Nearly all of the 1057 capacitors are then excluded from the model because they are too small. Note that the experimental approach to modeling this circuit involves building the circuit, measuring a 64 port structure (requiring 2080 separate complex measurements at each frequency), developing and entering an appropriate 1000+ element model in a circuit simulator, and optimizing each of the 1000+ variables for a best fit. Such a task is well beyond the state-of-the-art. In contrast, by using the em analysis, the total, end-to-end time is a few minutes. This includes the time required for manual circuit layout capture and inspection of the final results.

201

EM

However, because em is a surface meshing analysis, it can analyze the circuit of in under 1 minute on an HP-710 using about 1 Mbyte of memory. The SPICE file is generated after analyzing two frequencies.

Em User’s Manual

202

Chapter 17 Coplanar Waveguide Discontinuities and Balanced Ports

EM

Chapter 17

Coplanar Waveguide Discontinuities and Balanced Ports

Em also handles coplanar waveguide well because ground planes and dielectric layer thicknesses can be made any value while still maintaining full accuracy and without compromise in speed. This section describes a very simple, yet important, coplanar waveguide discontinuity, the short circuit. A second example is the coplanar cross junction. The cross junction also illustrates the use of level-to-level vias to form airbridges. Airbridges are needed to suppress the slot line mode as well as provide a ground current return path in some cases.

203

Em User’s Manual

The Coplanar Short The coplanar short, see below, finds wide use as a measurement calibration standard.

A co-planar waveguide short circuit discontinuity is easily handled by em.

Notice that both ground ports have been labeled “-1”, while the signal port is labeled “1”. This represents a balanced port. Any number of ports can have the same positive or negative labels. This is done in xgeom by selecting the ports in question and opening the Port Attributes dialog box by selecting Modify ⇒ Attributes. Em sums the total current going into all the positive ports with the same port number and sets that equal to the total current going out of all the ports with that same negative port number. Thus the name “balanced”, or “push-pull,” port. Although this circuit is symmetric, there is another issue to consider before invoking symmetry. This is addressed at the end of this section. Slot line can be represented by removing one of the “-1” ports, above. Be careful, however, the sign of the slot line port can be reversed by swapping the remaining “-1” port with the “1” port. It is possible, in fact, to get an S21 phase of 180 degrees for a zero length line. This happens if the ports on either end of a slot line have opposite sign.

204

Chapter 17 Coplanar Waveguide Discontinuities and Balanced Ports Note that we end the width of the ground lines just before they would have touched the sidewall. If the side of the ground line touches the side wall, it shorts out to the sidewall, thus allowing ground current to return via the sidewall instead of through the ground line. This defeats the purpose of the balanced port. Be sure that your ground lines touch the sidewall only at the location of negative port numbers when using balanced ports.

You can get a copy of the coplanar short, “cosht.geo”, using the Sonnet ⇒ Copy Examples command. To analyze the circuit, perform the following: 1

Select em Control from the Sonnet task bar. The main window with a new job window will appear on your display.

2

Enter “cosht.geo” in the Geometry File text entry box.

3

Click on the Simple Sweep option in the Frequency Control section. Enter 10.0 in the Start text entry box.

4

The Verbose and De-embed options are already selected, so you need take no action on these items.

5

Click on the Run command button to execute the analysis.

The result is 10.0000000 1.000000 169.31

A perfect short is 180 degrees; therefore, the short is about 11 degrees inductive.

205

EM

For large coplanar (or slot line) structures, there may be significant current on the outside edges of the ground strip. This can be verified with emvu. If this current is undesired, it can be eliminated by connecting the outside edge of the ground strips to the sidewalls of the box about a quarter wavelength from each port. This forms a quarter wavelength shorted (slot line) stub. At the port, the stub presents an open circuit to the current which, otherwise, would have started flowing along the outside edge of the ground strip.

Em User’s Manual It is possible to get the same result in less time. Notice that the circuit is physically symmetric. To add in symmetry, you would select Parameters ⇒ Box in xgeom. If this were the only change you made before analyzing the circuit, it would not work. The reason for this is that when a circuit is symmetric, em does not subsection any part of the circuit below the center line. After subsectioning the circuit, em checks to make sure all ports have at least one subsection. But notice that the “-1” port below the center line is still there. This port does not have any subsections and em prints an error message to that effect and terminates. So also remove the lower of the two “-1” ports and save the circuit under “cosht_sy.geo”. An analysis of this circuit provides almost identical data in much less time. Whenever you see an error message stating that there are no subsections for a given port, check for ports below the axis of symmetry, if symmetry is turned on.

The Coplanar Cross Junction The coplanar cross junction (see the figure on page 207) illustrates the analysis of coplanar waveguide structures including the effect of airbridges as an integral part of the coplanar discontinuity. The airbridges are needed in order to suppress the slot line mode which would otherwise propagate in the coplanar transmission line. In the case of the cross junction, the airbridges are needed for a second purpose, to provide a return path for ground current. The ground conductor on both sides

206

Chapter 17 Coplanar Waveguide Discontinuities and Balanced Ports must be continuous between all ports. Otherwise the ground current on one side is interrupted. If this happens, very strange results are seen, both in the em analysis and in actual measurements.

EM

A coplanar cross junction with airbridges (dotted lines) to short out slot line modes and to provide a return path for current.

The airbridges are indicated by dashed lines. Both ends of each airbridge are supported by a via (indicated by the up triangle) which also provides electrical connection to the ground conductors. The airbridges are 1.5 µm above the GaAs. This circuit is stored in the file “cocross.geo” in the examples directory and is accessible with the Sonnet ⇒ Copy Examples command.

207

Em User’s Manual

208

Chapter 18 Thick Metal with Arbitrary Cross-Section

EM

Chapter 18

Thick Metal with Arbitrary CrossSection

The previous chapter discussed balanced, or “push-pull,” ports. In this section, we use “push-push” ports to create thick metal lines where the vertical cross-section has an arbitrary geometry. To demonstrate this capability, we use a simple trapezoidal geometry, the cross section shown in the figure on page 211. To analyze the thick metal, set up the dielectrics so that there is one layer of dielectric with the same thickness as the metal. Then, place a polygon representing the top side of the thick metal on the top side of that dielectric layer. Also place a polygon representing the bottom side of the thick metal on the bottom side of that dielectric layer.

209

Em User’s Manual Next, place ports on both top and bottom sides. To connect the top side port to the bottom side port, give both ports the same number. This is done in xgeom by selecting the ports in question and opening the Port Attributes dialog box by selecting Modify ⇒ Attributes. As many physical ports as desired may be given the same numeric label, and all ports with the same label are automatically connected together. Such ports can be called “push-push”, in contrast with the “push-pull” ports of the previous chapter. A circuit implementing the above transmission line is stored in “thkthru.geo”. A copy can be obtained by using Sonnet ⇒ Copy Examples. An analysis, using a Simple Sweep frequency control of 10 GHz run with the Verbose and De-embed options set, yields:

10.0000000 0.316130 -126.0 0.948716 144.01 0.948716 144.01 0.316130 -126.0 P1 F=10.000 Eeff=(7.6412 0.0000) Z0=(26.49793 0.000000) R=0.00000 C=0.241587 P2 F=10.000 Eeff=(7.6412 0.0000) Z0=(26.49793 0.000000) R=0.00000 C=0.241587

For extremely thick lines, where the assumption that there is no current wrapping around the edge might not be valid, vias must be included along the edges of the line so that current flowing there is included. Note that vias, representing Z-

210

Chapter 18 Thick Metal with Arbitrary Cross-Section directed current, are not needed to represent current flowing along the edge of the line. They are only needed for current which flows over and around the edge of the line, a rare situation in most planar high frequency circuit designs.

Top Current

Bottom Current A trapezoidal cross-section transmission line viewed in perspective. If the line has no current going around the edge, it can be modeled, as shown, as two infinitely thin sheets of current, one at the top and the other at the bottom of the actual metal.

If a more detailed cross-section is desired, one can use more than two levels of infinitely thin metal to model the cross-section. Do not place metal on the interior (i.e., metal which is not on the top, bottom or edge) of the thick line. There is no current inside a good conductor and there is no need to waste subsections there. When a discontinuity is encountered in thick metal, vias probably should be included between the top and bottom metal as current may need to flow up or down at that point. Also if a line is longer than about an eighth wavelength, it would be a good idea to “tack” the top and bottom together with vias periodically to prevent unexpected resonances. Do not place vias at the edge of the substrate next to a sidewall on a port. In this situation, the vias short out to the sidewall. This shorts out the port. Such a situation can be identified by the port refection coefficient magnitude close to unity and the phase close to 180 degrees, a short circuit. The example directory contains an example of a thick step junction in a file called “thkstep.geo”, which can be obtained by using Sonnet ⇒ Copy Examples.

211

EM

No Current Around Edge

Em User’s Manual

212

Chapter 19 Package Resonances

EM

Chapter 19

Package Resonances

The circuit is designed, fabricated and meets all specifications. But before you are ready to ship you need to consider the package around the circuit. Once the lid goes on, it must be tested for package resonances and chip-to-chip coupling. Em can help with the package design. Recall that em analyzes a circuit in a conducting 6-sided metal box. The sidewalls of the box are always perfect conductors. Thus, you can find out on the computer early in the design cycle if the package is going to have resonances or evanescent waves that are of any concern. As an example, we analyzed a model of an amplifier in a box, shown in the figure on page 214. The circuit is stored in the file “package.geo” in the examples directory, and can be obtained using the Sonnet ⇒ Copy Examples command. We have left the transistor location open in this circuit for two reasons. First, the electromagnetic analysis performed by em does not support active devices. If you want to use the circuit network capability of em to include a transistor, place an

213

Em User’s Manual auto-grounded port here. Second, if there is a resonance, it shows up as a value of S21 approaching unity. This indicates a strong, package resonance induced, coupling between input and output. We have also included the bias lines, ports 3 and 4, to make sure that a package resonance does not kill the bias isolation. Feedback through bias leads is a significant cause of low frequency oscillations and smoked devices.

The file “package.geo” is a model of an amplifier used to check for package resonances. The entire width of the box is not shown.

The amplifier model is only approximate, we do not need exact line widths and precise layout. The purpose of this analysis is only to check for resonances. The detail used in this case is probably more than is needed. A simpler circuit should be sufficient for a package resonance search. The lowest resonance found was at 31.7625 GHz, shown on the plot on page 215. Given the approximate nature of the way we captured the circuit, we can conclude that there is a package resonance somewhere around 31 -32 GHz. There may be other resonances; the search was not exhaustive. Coupling between bias ports, which is not shown, also becomes large at resonance, making instability a virtual certainty. In addition to searching for resonances by looking at the S-parameters, as we have done here, we can also search for resonances by invoking the Detect Box Resonance option. The Detect Box Resonance option is described in “Verifying the Box Resonance Problem,” page 218. 214

Chapter 19 Package Resonances This package has big problems at resonance. Fortunately, with the em analysis in hand, modifications can be made and tested before costly fabrication. A second analysis was performed from 20 GHz to 60 GHz with the top cover removed; i.e., selecting Free Space as the Top Metal in Box Parameters. This analysis showed several low Q (wide bandwidth) resonances, none of which exceeded 35 dB down. The plot is shown on page 216.

Results of a search for package resonances shows strong coupling between input and output at 31.7625 GHz.

215

EM

A related application of this analysis is to use the circuit package as the resonant, frequency determining element in an oscillator design.

Em User’s Manual

The package resonances disappear when the top cover is removed.

Unwanted Box Resonances A problem which affects actual measurements, as well as the em analysis, is box resonances. If present, resonances put glitches into the simulated data. Em has the ability to detect box resonances and remove them, if desired. This section shows the effects of unwanted box resonances and their resolution. Box resonances can also corrupt de-embedding results. Because em’s deembedding feature is based on circuit theory, it possesses the same limitation that all de-embedding algorithms share. It is unable to de-embed a structure contained inside a resonant cavity (box). This means that if a box resonance exists for a deembedding calibration standard, the final S-parameters should be suspect.

216

Chapter 19 Package Resonances To obtain an example of data with a box resonance use Sonnet ⇒ Copy Examples to get the file “open_120.geo”. To analyze the circuit, perform the following: Select em Control from the Sonnet task bar. The em main window with a new job window will appear.

2

Enter “open_120.geo” in the Geometry File text entry box in the job window.

3

Click on the Simple Sweep radio button under Frequency Control. Then enter 24.0 in the Start box, 24.6 in the Stop box and 0.1 in the Step box.

4

The default value of GHz is already set, as well as the default options, Verbose and De-embed so no action need be taken on these items.

5

Click on the Run command button to execute the em analysis.

The S-parameter results are shown below. 24.0000000 1.000000 -54.96 !< P1 F=24.000 Eeff=(8.1759 ! 24.1000000 1.000000 -56.91 !< P1 F=24.100 Eeff=(8.1667 ! 24.2000000 1.000000 -61.60 !< P1 F=24.200 Eeff=(8.1394 ! 24.3000000 1.000000 -172.2 !< P1 F=24.300 Eeff=(7.7543 ! 24.4000000 1.000000 -43.44 !< P1 F=24.400 Eeff=(8.2908 ! 24.5000000 1.000000 -49.36 !< P1 F=24.500 Eeff=(8.2456 ! 24.6000000 1.000000 -51.58 !< P1 F=24.600 Eeff=(8.2336

0.0000) Z0=(36.47273 0.000000) R=0.00000 C=0.211362

0.0000) Z0=(36.22324 0.000000) R=0.00000 C=0.214332

0.0000) Z0=(35.53605 0.000000) R=0.00000 C=0.221539

0.0000) Z0=(29.14039 0.000000) R=0.00000 C=0.311480

0.0000) Z0=(39.71460 0.000000) R=0.00000 C=0.188056

0.0000) Z0=(38.25533 0.000000) R=0.00000 C=0.199493

0.0000) Z0=(37.85267 0.000000) R=0.00000 C=0.203249

De-embedded results showing the effects of a box-resonance.

217

EM

1

Em User’s Manual Here we see the effects of a box resonance with the jump in phase, at 24.3 GHz, of the de-embedded open end. The sudden steep phase change in S11 is a sure clue of a box resonance. Note that Eeff and Z0 are also affected. This is because there is also a resonance in at least one of the standards that em creates for deembedding. In some cases, the standards do not have a box resonance and Z0 and Eeff are unaffected. To see an example of this, reduce the top layer thickness from 1000 mils to 50 mils. The box resonance moves to 24.85 GHz while the Z0 and Eeff calculations remain smooth.

Verifying the Box Resonance Problem When you see a “glitch” in the S-parameter data or in the characteristic impedance, you can verify whether or not it is a box resonance by using the detect box resonance option in em. To do this, proceed as follows: 1

Select em Control from the Sonnet task bar. The em main window with a new job window will appear on your display.

2

Click on the Additional Options command button. This will open the Additional Options dialog box.

3

Click on the Detect Box Resonance checkbox. Then click on the OK command button to close the dialog box and apply the option.

4

Enter “open_120.geo” in the Geometry File text entry box in the job window.

5

Click on the Simple Sweep radio button under Frequency Control. Then enter 24.0 in the Start box, 24.6 in the Stop box and 0.1 in the Step box.

6

The default value of GHz is already set, as well as the default options Verbose and De-embed

7

Click on the Run command button to execute the em analysis.

The output at 24.3 GHz is shown on page 219.

218

Chapter 19 Package Resonances

De-embedded results showing a box-resonance at 24.3 GHz.

Notice that the resonance detection information is printed out before, not after, the frequency of analysis. Also note that the S-parameters are the same; the resonance has not been removed. We see box resonances listed for the primary structure (the open end) and for both standards. The box resonance information tells us that the resonance is Transverse Electric (TE), i.e., there is no Z directed electric field which is perpendicular to the substrate surface. Several of the modes are (0,1,*) indicating that there is no variation (first digit) of field in the X direction, which is the direction of the open stub, and a half sine wave of variation (second digit) in the Y direction. The “*” in the mode number indicates that the variation in the Z direction is unknown. Since we generally have several different dielectrics as we go along the Z axis, a mode number here is not always clear. The digit in front of the parentheses is the mode significance. The mode significance can be used to compare one resonance with another, but has no physical meaning. For example, the first mode in the primary structure has a significance of 8. This means that we are closer to exciting this mode than we are to the second mode, which has a significance of 1. The most important mode is the (1,0,*) mode in the first standard, with a significance of 29.

219

EM

! Box resonances, primary structure. ! TE: 8(0,1,*) 1(1,1,*) ! Box resonances, first standard, side 0. ! TE: 1(1,1,*) 29(1,0,*) 1(0,3,*) 0(1,1,*) ! Box resonances, second standard, side 0. ! TE: 0(1,1,*) 19(1,0,*) 1(0,1,*) 0(0,1,*) ! 24.3000000 1.000000 -172.2 !< P1 F=24.300 Eeff=(7.7543 0.0000) Z0=(29.14039 0.000000) R=0.00000 C=0.311480

Em User’s Manual

Removing Box Resonances One way to remove a box resonance is to change the size of the box, either larger or smaller, to move the resonant frequency out of band. If the problem occurs in de-embedding, you may be able to change the length of the calibration standard in xgeom to move the box resonance out of the band of interest. The box is a resonator because it is completely enclosed. Another way to keep it from resonating is to just take off the top cover. We can use an approximation of this by setting the top cover resistivity to 377 ohms/square, the impedance of free space. You do this by selecting Free Space as the Top Metal in the Box Parameter dialog box in xgeom. This is an accurate approximation provided the cover is not so close that it interacts with the evanescent fringing fields surrounding the circuit. If you wish to model a sheet of resistive material in the box, rather than removal of the top cover, setting the top cover to that resistivity is accurate no matter how close the cover is to the circuit. It is an approximation only when used to model the removal of the top cover. If you wish to analyze the circuit with the top cover removed, perform the following:

220

1

Select em Control from the Sonnet task bar. The em main window with a new job window will appear on your display.

2

Enter “open_120.geo” in the Geometry File text entry box in the job window.

3

Click on the Editor command button to the right of the Geometry File text entry box to invoke the xgeom program with the “open_120.geo” file open.

4

Select Parameters ⇒ Box to open the Box parameters dialog box. Click on the Top Metal drop list and select Free Space from the metals available in the drop list. Click on the OK command button to close the dialog box and apply the changes.

5

Select File ⇒ Save As to open the Save As dialog box. Enter “openloss.geo” in the file name text entry box and click on the OK button to save the file.

Chapter 19 Package Resonances To analyze the file, enter “openloss.geo” in the Geometry File text entry box in the em job window.

7

Click on the Simple Sweep radio button under Frequency Control. Then enter 24.0 in the Start box, 24.6 in the Stop box and 0.1 in the Step box.

8

The default value of GHz is already set, as well as the default options, Verbose and De-embed

9

Click on the Run command button to execute the em analysis.

EM

6

The results are shown below. 24.0000000 0.989757 -54.86 !< P1 F=24.000 Eeff=(8.1768 ! 24.1000000 0.975815 -56.41 !< P1 F=24.100 Eeff=(8.1707 ! 24.2000000 0.931200 -58.15 !< P1 F=24.200 Eeff=(8.1638 ! 24.3000000 0.855048 -54.81 !< P1 F=24.300 Eeff=(8.1921 ! 24.4000000 0.906994 -50.38 !< P1 F=24.400 Eeff=(8.2331 ! 24.5000000 0.951370 -51.13 !< P1 F=24.500 Eeff=(8.2327 ! 24.6000000 0.969205 -52.31 !< P1 F=24.600 Eeff=(8.2288

5.2e-3) Z0=(36.48887 0.096410) R=0.15584 C=0.211187

0.0115) Z0=(36.30066 0.231696) R=0.34909 C=0.213525

0.0309) Z0=(36.03638 0.674346) R=0.94949 C=0.216409

0.0669) Z0=(36.50667 1.710846) R=2.25883 C=0.212178

0.0422) Z0=(37.70634 1.320674) R=1.64077 C=0.202681

0.0198) Z0=(37.77502 0.697364) R=0.81424 C=0.202991

0.0113) Z0=(37.65657 0.442684) R=0.48646 C=0.204634

De-embedded results with most box-resonances removed.

Most of the resonance has been removed. To remove the resonance more completely, the top cover can be moved closer to the substrate surface. In fact, if the cover is moved to within 50 mils, there is absolutely no indication of a resonance left.

221

Em User’s Manual With the top cover off, em is also including radiation loss in the de-embedded discontinuity (see the S11 magnitude). The loss has a secondary effect in making Eeff and Z0 complex as well. Taking the top cover off works, provided the sidewalls of the box are large enough to form a propagating waveguide up to the top cover, or you can place the top cover close enough to the substrate surface to catch the fields in the box mode. High order “box” modes tend to be confined primarily to the substrate and are difficult to remove in this manner. As you make the box bigger by increasing the substrate surface area, the modes “loosen up” so that they can propagate to the top cover and become absorbed. To completely absorb any single waveguide mode, set the surface impedance of the top cover equal to the impedance of the waveguide mode. If you have multiple modes to absorb, setting the impedance to 377 ohms/square (Free Space Top Metal) is a nice compromise because some modes are above 377 ohms and others are below.

222

Chapter 20 Viewing Tangential Electric Fields

EM

Chapter 20

Viewing Tangential Electric Fields

One reason em is so fast is that all of the electric and magnetic fields are solved for analytically, with “pencil,” “paper” and many equations. The computer need only do an FFT and solve for the current distribution. However, on occasion, you want to view the fields, not the current. You do this with what is called a “sense layer”. The sense layer is a rectangular patch of conductor placed where you want to see the tangential electric field. Actually, describing the sense layer as a conductor is misleading. This is because you set the surface resistance and/or reactance of the conductor to some large value, say 1,000,000 ohms per square. (We suggest setting the reactance to a large value instead of the resistance because it is a little more efficient in em.)

223

Em User’s Manual You set the reactance to such a large value so that the sense layer has little influence on the original fields. An intuitive analogy is to view a sense layer like inserting a sheet of paper (very high reactance) into the fields. Because the reactance of the sense layer is high, the currents are very small. The sheet of paper does not change the fields. When capturing the sense layer, it is best to set X Max and Y Max to 1 for the best image. See the “Adjusting the Subsectioning,” page 24. But even though they are small, what are the currents? The current density is proportional to the tangential electric fields over the area of the sense layer. This is just a two dimensional version of Ohms Law: Current is proportional to Voltage.

224

Chapter 20 Viewing Tangential Electric Fields An example is shown below as viewed by emvu. You may use the Sonnet ⇒ Copy Examples command to get a copy of the geometry file, “tane.geo.” The “tane.geo” file is based on the “gap20.geo” file.

EM The tangential electric fields just above a gap discontinuity. The input voltage comes from the left. Strong fields are present across the gap, especially at the corners. This analysis was performed at 1 GHz.

225

Em User’s Manual

226

Chapter 21 Accuracy Benchmarking

EM

Chapter 21

Accuracy Benchmarking

Electromagnetic analyses are often described as providing what is called “Good Agreement Between Measured And Calculated” (GABMAC). However, in the past, there has been little effort to decide just what “good” means. The more useful result is the “Difference Between Measured And Calculated” (DMAC). In this chapter, we describe a precise benchmark, based on [21], [22] and [24], which allows the evaluation of DMAC for any 2.5-D or 3-D electromagnetic analysis down to the 1 x 10-8 level of accuracy.

An Exact Benchmark What we need to calculate DMAC is an exact benchmark. One source of an exact benchmark is stripline. The characteristic impedance of a stripline has an exact theoretical expression K(k) is the complete elliptical integral of the first kind. For

227

Em User’s Manual

evaluation on a computer, a polynomial for K(k) is available in Abramowitz and Stegun, Handbook of Mathematical Functions, pp. 590 - 592. (Be sure to note the errata, m1 = 1m2, not 1-m.):

b w η 0 K ( k' ) Z 0 ε r = ------ -----------4 K( k) πw k = tanh  --- ----  2 b

k' =

1–k

2

η 0 = 376.7303136 The expression for K(k) cited above provides an accuracy of about 1 x 10-8. When programmed on a computer, the following values are obtained for three different transmission line impedances (unity dielectric constant):

Table 9 Stripline Benchmark Dimensions Z0 (ohms)

228

w/b

25.0

3.3260319

50.0

1.4423896

100.0

0.50396767

Chapter 21 Accuracy Benchmarking For a length of stripline, there are two parameters of interest: characteristic impedance and propagation velocity. With the w/b given above, we know the exact answer (to within 1 x 10-8) for Z0. With a dielectric constant of 1.0, we also know the exact answer for the propagation velocity. It is the speed of light, known to about 1 x 10-9. Any difference from these values is error, or, DMAC.

The “b” dimension is exactly 1.0 mm, the “w” dimension is given by the above table and the length of each line is 4.99654097 mm with a dielectric constant of 1.0. Each of these lines is precisely 0.25 wavelengths long at 15.0 GHz. The geometry files have the subsectioning set so the lines are 16 cells wide and 128 cells long. Analysis time is a few seconds on an HP-710. To evaluate DMAC, do an analysis of the line at 15 GHz, with de-embedding enabled. For the error in characteristic impedance take the percent difference between the calculated value and the exact value, above. For the error in propagation velocity, take the percent difference between the calculated S21 phase and -90 degrees. Total error, in percent, is the sum of the two errors. Some types of analyses do not calculate characteristic impedance. A detailed error analysis shows that, to first order for a 1/4 wavelength long 50 ohm line, the value of |S11| is equal to the error in characteristic impedance. For example, an |S11| = 0.02 means that there is about 2% error in characteristic impedance. To use this approximation for, say, a 25 ohm line, the S-parameters must be converted to 25 ohm S-parameters. This may be done by adding transformers in a circuit theory program.

Residual Error Evaluation We have performed a detailed analysis of the relationship between subsectioning and residual error (DMAC). The simplest way to subsection a line is to use subsections the same width as the line. In Sonnet, and in many other analyses, this results in a uniform current distribution across the width of the line. In reality, the current distribution is singular at the edges of the line. 229

EM

Each of the above three benchmarks are available via the Sonnet ⇒ Copy Examples command. To get the 50 ohm line, to get the file “s50.geo.” The other benchmark circuits are in “s25.geo” and “s100.geo.”

Em User’s Manual Since the current distribution is symmetrical about the center line, using either one or two subsections across the width of the line gives the same amount of error. We find that a one or two subsection wide line gives 5% to 6% error. If there is not much stray coupling, circuit theory can often give a better result. When the line is 16 cells wide, we see about 1% error, much more reasonable. We have found (and you can verify) that convergence is very strong: Double the number of cells per line width and the error is cut in half. When we vary the number of cells per wavelength, along the length of the line, we see an inverse square relationship. Double the number of cells per wavelength along the length of the line and the percent error decreases by a factor of four. An equation which expresses the error as a function of subsectioning is:

16 2 16 E T ≅ -------- + 2  ------- N  NW L

NW ≥ 3

N L ≥ 16

where NW = Number of cells per line width, NL = Number of cells per wavelength along line length, ET = Total Error (DMAC) (%). This equation estimates subsectioning error only. For example, any de-embedding errors are added to the above error. This error estimate should be valid for any electromagnetic analysis which uses roof-top subsectioning. Notice that the quantities used for the error estimate are in terms of cells, not subsections. Cells are the smallest possible subsections size. In Sonnet, subsections in the corners of polygons are one cell on a side. Subsections along the edge of polygons are one cell wide and can be many cells long. Interior subsections can be many cells in both dimensions.

230

Chapter 21 Accuracy Benchmarking We have found that, for most cases, the cell size is the important parameter in determining error. Or in other words, the smallest subsection size is important. For example, the stripline benchmark geometry files, mentioned before, are set to make the lines 16 cells wide, even though those 16 cells may be merged into only 4 or 5 subsections. It is the 16 cells which determine the level of error, not the 4 or 5 subsections.

The above equation can be very accurate in evaluating error. With this precise knowledge of the error, we can now do something about it!

Using the Error Estimates The above error estimate can be used to estimate the error for an overall circuit. Let’s say that a cell size is used that makes some high impedance transmission lines only 1 cell wide. Other, low impedance transmission lines, are, say, 30 cells wide. The 1 cell wide lines give us about 5% error. The 30 cell wide lines give about 0.5% error. In non-resonant situations, you can expect the total error to be somewhere between 5% and 0.5%. If most of the circuit is the low impedance line, the error is closer to 0.5%, etc. However, let’s say that our circuit has resonant structures. Let’s say it is a low pass filter. It is easy to verify by means of circuit theory that the low pass filter is very sensitive to the high impedance lines. This means we can expect about 5% error, even though the high impedance lines only make up half the filter. Given this information, there are several courses of action. First, if 5% error is acceptable, no further effort is needed.

231

EM

In performing this error evaluation, we also found that the error in characteristic impedance due to NW is always high, never low. Also, there is very little variation in the error for different impedance lines. The above equation can be very accurate in evaluating error. And, finally, for NL above about 40 cells per wavelength, all the error is in the characteristic impedance. The error in velocity of propagation is essentially zero.

Em User’s Manual More likely, we wish to analyze the filter with less error. Since we now know the error in the characteristic impedance is 5%, we can physically widen the line so that the characteristic impedance is 5% lower to compensate for the known increase in characteristic impedance due to subsectioning the line only one cell wide. Very precise analyses are possible using this compensation technique.

232

Chapter 22 Range of Analysis Validity

EM

Chapter 22

Range of Analysis Validity

Em is a complete electromagnetic analysis; all electromagnetic effects, such as dispersion, loss, stray coupling, etc., are included. There are only two approximations used by em. First, the finite numerical precision inherent in digital computers. Second, em subdivides the metalization into small subsections. The cell size is important factor in determining the accuracy. By using a smaller cell size, metal edges can be more accurately defined and the current distribution is better represented. The trade-off is increased execution time. A quantitative description of accuracy versus cell size is given in Chapter 21, “Accuracy Benchmarking.” The trade-off between execution time and accuracy increases the degrees of freedom available to the design engineer.

233

Em User’s Manual

Subsection Size There are actually two limits on subsection size, a minimum and a maximum. When we refer to a cell, we have in mind the minimum limit. For example, a circuit may cover 10,000 cells. Em can use variable size cells, which we refer to as subsections. That same 10,000 cell circuit could be subsectioned into only 200 variable sized subsections. The maximum limit determines the size of the largest subsection (which may cover many cells). With a maximum frequency in mind, the cell size should be set so that a single cell is no more than a fraction of a wavelength at the maximum frequency of analysis. The parameter Max. Subsection Size allows the specification of a maximum subsection size, in terms of subsections per wavelength, where the wavelength is approximated at the beginning of the analysis. The highest analysis frequency is used in the calculation of the wavelength. The default of 20 subsections/λ is fine for most work. This means that the maximum size of a subsection is 18 degrees at the highest frequency of analysis. Increasing this number decreases the maximum subsection size until the limit of 1 subsection = 1 cell is reached. As the limits of subsection size, both maximum and minimum, are made smaller, the em analysis is asymptotically exact. Given sufficient computer resources, an arbitrarily accurate answer may be achieved. Chapter 3 contains a detailed explanation of subsectioning.

Metalization and Dielectric Thickness The analysis assumes zero metalization thickness. This could result in some discrepancy for thick lines which are tightly coupled. An approximation for metalization thickness is to have two zero thickness metalization levels, one at the top and the other at the bottom of the actual metalization. Layer to layer vias may be used to form the sides of the thick lines as described in Chapter 18, “Thick Metal with Arbitrary Cross-Section.”

234

Chapter 22 Range of Analysis Validity Dielectric layer thickness and dielectric constant have no impact on the accuracy as long as the layers are greater than 0.05 microns thick. For layers thinner than 0.05 microns, numerical precision may not allow accurate answers. The substrate area has no impact on accuracy. However, it does have a secondary impact on analysis time.

Numerical Precision Em experiences a numerical precision problem (difference between two large numbers) if the subsection size gets too small. For example, if the subsection size is several microns and the frequency is a few kilohertz, there are precision problems. The results quickly become obviously bad. A higher frequency of analysis, or use of quadruple precision can solve the problem. At a higher frequency, the subsection becomes larger, in terms of wavelengths. The precision problem also arises with vias. This is especially true if, in addition to the small cell size, the via is in a thin layer, say a micron or less. Unreasonable results are seen as high as several hundred MHz, or even 1 GHz. The net effect of the loss of precision makes the via appear to have a very long phase length. These problems become worse if the Memory Save option is used. Again, use of the quadruple precision option can solve the problem in exchange for increased analysis time.

235

EM

All coupling between all subsections is calculated and included in the analysis, even coupling between patches of metalization which are not connected to anything.

Em User’s Manual

236

Chapter 23 Time Required for Analysis

EM

Chapter 23

Time Required for Analysis

Em is a memory and computation intensive program. Small circuits are analyzed quickly while large circuits can require a considerable amount of time. In some cases, it may be desirable to run the program in the background. There is no simple rule for calculating the time required for a particular analysis although there are guidelines, presented below, which will afford you some measure of control over that time. The amount of time required is closely related to the number of subsections, which is printed out when em is run with the Verbose option. After a few trials, you will have a good idea of whether an analysis is a few seconds, a few hours, or just totally hopeless by looking at the number of subsections.

237

Em User’s Manual The most important factor in the execution time required is the amount of memory required by em compared with the amount of memory you have. This is also printed out by em if you use the Verbose option. If em says you need 36 Mbytes and you have 16 Mbytes installed, then it is a good idea to kill the run quickly. You can reduce the memory required by using the Memory Save option, which makes the analysis single precision. You can also try to reduce the area of metalization in the circuit. Try to eliminate any metal that is not carrying current and make connecting lines as short as possible (but not too short, see Chapters 6 and 7 on de-embedding). Another approach, if your memory requirement is right on the edge, is to free up some of your computer’s memory. Make sure no one else is also running a big number cruncher at the same time. The estimate of required memory printed out by em is just an estimate. It is usually within 1 Mbyte or so, but could be off by much more. To get both the memory estimate and the number of subsections without going on to actually analyze the circuit, use the Calculate Memory Usage option, available in the Additional Options dialog box. For most circuits, the following equation can be used to estimate the amount of memory that will be used by em: B = K*N2 where B is the number of bytes, and N is the number of subsections. K is equal to 8 if running with double precision and loss. This can be circuit metal loss, top or bottom cover loss or dielectric loss. K is equal to 4 if you are running with loss but using Memory Save or running lossless and using double precision. K is equal to 2 if you are running a lossless circuit and using Memory Save. This equation should be used only as an estimate as it only includes the memory used by the final matrix in em. Circuits with large boxes (in terms of number of cells) or many layers require more memory. You should use this equation to calculate an upper limit on the number of subsections for your computer. For UNIX systems, you can check the memory actually used by typing the “ps” (process status) command. Consult your system administrator or UNIX manuals for details on the “ps” command.

238

Chapter 23 Time Required for Analysis To check how much of your system’s memory is actually available for your use, select Help ⇒ System Info from the em Control main menu. The System Information dialog box appears on your display and contains the information on system memory use.

EM 239

Em User’s Manual

The “Wall” When using circuit theory analysis, an increase in circuit complexity gradually produces an increase in analysis time. With an electromagnetic analysis, the increase happens suddenly. A mere doubling of circuit complexity (say, by using a smaller subsection size) can result in one, or even two, orders of magnitude longer analysis. We call this the “Wall” (see the charts below).

Time

Circuit Theory

Time

Complexity Electromagnetics

Complexity

240

Chapter 23 Time Required for Analysis The Wall is frequently encountered when em runs out of real memory, as described above, and is forced to start swapping out to disk. Execution time can quickly go from a few minutes to a few hours. Either get more memory or modify the circuit so that there are fewer subsections.

The main factor in analysis time is the number of subsections. Em prints out the number of subsections if the Verbose option is used. As you gain experience with em, you will get a good feel for what can be tolerated. For example, on a Sun SPARCstation 1 with 16 Mbytes of memory, up to 1700 subsections (lossless) or 1200 subsections (with loss) can be calculated in an hour or so. At this point the computer runs out of memory and starts swapping to disk, resulting in huge increases in time. To avoid the frustrations of getting on the slow side of the wall, start lossless with big subsections. You may find that big subsections provide all the accuracy you need!

Detailed Parameter Dependencies How do changes in the various input parameters affect the analysis time? First, keep in mind that em has two stages in the analysis. In the first stage, em fills a large matrix. The matrix has one row/column for each subsection. This is where em is calculating the coupling between every possible pair of subsections. In the second stage, em is solving the matrix. Here, em is performing matrix inversionlike functions and is calculating the currents which allow all boundary conditions to be met. The different parameters affect each stage of the analysis differently.

241

EM

To avoid the wall, start with no loss (metal or dielectric) and use a large subsection size. Perform the first analysis at a single frequency to evaluate how long an analysis takes. Then, provided you get fast results, try adding loss or making the subsection size smaller. Keep going until the analysis is as long as you can tolerate and then let it run over a full range of frequencies, perhaps overnight.

Em User’s Manual During analysis of the first frequency in a run, em prints out the amount of time spent on various portions of the analysis. If the time spent in any particular section of the analysis is less than one second, it is not printed out. Sections which are timed include the waveguide mode calculation (prior to matrix fill), the matrix fill and the matrix inversion. Parameters which have no effect on analysis time include the substrate thickness, cover height and number of ports. Each of these parameters is unlimited and have no impact on speed while still maintaining complete accuracy. Including metalization loss increases the matrix solve time only. And this is only if the rest of the structure is lossless. If there is any dielectric loss or ground plane (or top cover) loss, there is virtually no additional impact from also including metalization loss (the whole calculation is already fully complex). The matrix solution time is increased by about a factor of four (if it becomes complex). Metalization loss has no impact on any other segment of the analysis. Including dielectric loss, or ground plane or top cover loss, makes the entire calculation complex. The matrix solve time is increased by about a factor of four, while the matrix fill time is increased by about a factor of two. In calculating the values for the matrix elements during the matrix fill, several two dimensional Fourier Transforms must be calculated. The size of the Fourier Transforms is the same size as the substrate in terms of cells. If a substrate is 128 x 64 cell, each Fourier Transform is 128 x 64 elements. Memory storage is required for only one Fourier Transform at a time and this is usually much smaller than the matrix being filled. For large substrate dimensions (as measured in terms of cells), it is best to use a power of two. With a substrate dimension of 32 cells, there is little difference when a power of two is used. However if the substrate dimension is 512 cells, noticeably slower execution results if one were to use 500 cells. With a power of two, a FFT algorithm is automatically used to evaluate the Fourier Transform which significantly improves processing time. For non-binary substrate sizes, the analysis is speeded up by the use of chirp Ztransforms in place of discrete Fourier Transforms for the matrix fill calculations.

242

Chapter 23 Time Required for Analysis For all the above detail, keep in mind that if the substrate is small (less than 256x256 cells), the Fourier Transform time is of little consequence. If the Quad Precision option is used, all matrix fill operations are performed in quadruple precision. Depending on the computer and specific problem being solved, this can add substantial time to the analysis. The precision used in the matrix solve is not affected by the Quad Precision option.

The number of subsections, for a given cell size, can be reduced by minimizing the number of vertices and the number of diagonal lines in the polygonal description of the circuit. If the circuit is symmetric with no more than two ports, with both ports on the axis of symmetry, invoke the symmetry option for a significant memory and time savings. The matrix solve time is proportional to the number of subsections cubed and is the main limitation on the analysis at this time.

243

EM

Matrix fill time is proportional to the number of subsections squared for large circuits.

Em User’s Manual

244

Chapter 24 Em Interface: Analysis of a Geometry File

EM

Chapter 24

Em Interface: Analysis of a Geometry File

There are two types of analysis that may be done in em: Geometry File or Network File. This chapter describes how to invoke em and set up the analysis of a geometry file. Chapter 25 will describe the analysis of a network file. A geometry file, created by xgeom, provides geometry information for a single 3D planar circuit. The file name must have the extension “.geo”. Em will execute an electromagnetic analysis of this circuit at the frequencies specified directly in the em interface or through use of an analysis control file. The desired output is specified by selecting run options and output file types.

245

Em User’s Manual

Invoking em You start the em program by selecting em Control from the Sonnet task bar. If you do not know how to invoke the Sonnet task bar, refer to “Invoking Sonnet,” page 11. When you select em Control, the main window with a new job window will appear on your display as shown below. File text entry box

File drop down list

Directory drop down list Directory text entry box Em job window.

When em is started, the main window opens with a new job file already open. How to manipulate job files is discussed in “Job Files,” page 279.

Selecting a Geometry File Analysis You must select the type of analysis you wish to run: Geometry File or Network File.

246

Chapter 24 Em Interface: Analysis of a Geometry File •

To analyze a geometry file, click on the Geometry radio button under File Type in the job window. The job window will appear as shown in the figure on page 246. The Frequency Controls and Options now available in the job window are those appropriate for analyzing a geometry file. This window’s appearance will vary slightly for an analysis of a network file. EM

Analysis Input Files This section will discuss specifying, and editing, the geometry file.

Selecting a Geometry File To select the geometry file for analysis, you must specify the directory and the file name. The directory is entered in the box labeled “Start In :” and the file name is entered in the File text entry box just above the Directory text entry box. •

Select the directory and file name. You may enter these several different ways: •

Use the Browse button to browse the file system for the file name; this will set both the directory and the file name.

OR 1

Set the directory by either: • •

2

Editing the Directory text entry box Selecting a previously used directory from the Directory drop down list.

Set the file name by either • •

Editing the File text entry box Selecting a previously used file name from the file name drop down list.

247

Em User’s Manual

Editing a Geometry File You may wish to make changes to a geometry file before running an analysis by editing the circuit in the xgeom program. •

Click on the Edit command button, which appears after the File text entry box, to invoke the xgeom program with the geometry file open. For details on how to use xgeom to modify your circuit, please see the Xgeom User’s Manual.

!

WARNING You must save the circuit geometry file in xgeom for the changes made in the editing session to be available before running an analysis.

Specifying Frequency Control This section of the em job window, highlighted in the figure below, allows you to control the frequencies used in analyzing the geometry file.

Frequency Control section of the em job window.

Setting up a Simple Sweep A simple sweep is used to execute an analysis using only one or multiple frequencies evenly spaced in an ascending order.

248

Chapter 24 Em Interface: Analysis of a Geometry File 1

Click on the Simple Sweep radio button in the Frequency Control section of the job window.

2

Enter the Start, Stop and Step values in the appropriate text entry boxes.

TIP If you omit the step value in a simple sweep, the circuit is analyzed at two frequencies, the start and stop values.

3

Select the frequency units. Select the units for the analysis frequencies from the Frequency drop list. The choices are Hz, KHz, MHz, GHz, THz and PHz. The default setting upon opening a new em interface is GHz.

Setting Up a Complex Sweep Complex sweeps are used to run an analysis using multiple sets of frequencies in sorted or unsorted order, and may also enable intelligent frequency selection (IFS), such as finding the maximum or minimum value for a parameter. Complex sweeps are not available for a network file. Network file frequency control is provided in one of three ways: a simple sweep, internal to the network file or by an analysis control file. 1

Click on the Complex Sweep radio button under Frequency Control in the em job window.

249

EM

If you wish to analyze at only one frequency, enter that frequency in the Start text entry box. Otherwise, Start provides the beginning frequency, Stop the ending frequency and Step the spacing. For instance if the Frequency Units were set to GHz, then the values of 2, 10, 2 would start an analysis at 2 GHz and end at 10 GHz with steps of 2 GHz (e.g. 2, 4, 6, 8 and 10 GHz)

Em User’s Manual This will enable the Edit button to allow you to set up a complex sweep. 2

Click on the Edit command button which appears next to the Complex Sweep radio button. The Analysis Control dialog box will appear as shown below. This dialog box allows you to edit frequency control information. For details on how to edit the complex sweep, see “Editing Analysis Controls,” page 264.

The Analysis Control - Internal dialog box.

Using an Analysis Control File There may be times where you will wish to analyze multiple circuits at the same set of frequencies. Rather than specifying frequency control information on an individual basis, you set the up the information once in an analysis control file. You then specify the analysis control file as input to control each analysis. An analysis control file may be used to control the analysis frequencies for either a geometry or network file. You may specify an analysis control file and edit the contents in the same manner as a Complex Sweep. For details about the contents and syntax of an analysis control file, see Appendix II, “The Analysis Control File Format.”

250

Chapter 24 Em Interface: Analysis of a Geometry File •

Click on the Analysis File radio button under Frequency Control in the em job window. Enter the file name of the desired analysis control file.

NOTE:

You may edit the specified analysis control file by clicking on the Edit command button. This will open the Analysis Control dialog box. For details on using the dialog box to edit your file, see section “Editing Analysis Controls,” page 264.

Selecting Run Options You may select various options for the analysis of the geometry file including advanced options.

Selecting Job Window Options There are four options available in the job window for an geometry file analysis, which may be turned on when you wish to use the option and set to off if the option is not desired. The Verbose and De-embed options are set to on by default for a new job run. Verbose Option: Causes em to display messages in the output window during program execution describing the current state of the analysis. De-embed Option: The circuit is automatically de-embedded to the specified reference planes, or the box edge if no reference planes are specified in the geometry file. For a detailed discussion of de-embedding refer to Chapters 6 and 7.

251

EM

The default directory for the analysis control file will be the same as the input file specified in the top of the job window. If you wish to use a control file from a different directory use the Browse command or type the complete or relative pathname of the file.

Em User’s Manual If this option is on, an output file containing the de-embedded response data is produced. The name defaults to the circuit geometry file basename with a “.d” extension. For example, if the input file is “steps.geo” then the response file is automatically named “steps.d”. If you wish to name the output file differently see “Specifying Output Files,” page 270. Memory Save Option: The system matrix is filled in single precision which reduces memory requirements for storing the matrix to one half of that of double precision. If this option is not used, the matrix is stored in double precision. This option affects only the matrix storage and the matrix solution. The precision of the matrix fill calculations is not affected. Make emvu File Option: Outputs current density information for the entire circuit which can be viewed using emvu. For details on using the emvu program, see the Emvu User’s Manual. The file name for the current density information defaults to the input file basename with the extension “.jxy”. For example, if the input file is “steps.geo”, then the current density file will be named “steps.jxy”. If you wish to name the output file differently, see “Specifying Output Files,” page 270. If the “.jxy” file already exists, it is renamed “.jxb” and output is sent to the new, empty, “.jxy” file. In this case, any previous “.jxb” file is lost.

Selecting Additional Options There are more options available for a geometry file analysis in the Additional Options dialog box. These options provide advanced control over em’s execution.

252

Chapter 24 Em Interface: Analysis of a Geometry File 1

Click on the Additional Options command button in the job window to open the Additional Options dialog box, shown below.

EM The dialog box contains sections for selecting run mode, output parameter type, output data format, output file format, and various options as well as a text entry box for entering advanced options. 2

Select the Run Mode. To execute em in the normal manner and perform a full analysis, you click on the Full Analysis radio button. If you click on the Calculate Memory Usage radio button, em will calculate the number of subsections followed by an estimate of the number of Mbytes of memory required. The actual analysis is not performed. To see how a particular circuit is subsectioned, click on the Generate Subsections Only radio button. Em will output a current density file containing subsectioning information only, which you must view using emvu. The data file will have an extension of “.jxy.” For a detailed discussion of subsectioning, see Chapter 3, “Subsectioning.”

253

Em User’s Manual 3

Select the output parameter type. You select the parameter type by clicking on the S, Y or Z radio button to pick SParameters, Y-Parameters or Z-Parameters, respectively.

4

Select the output data format. You may choose between Touchstone or Compact by clicking on the appropriate radio button.

5

Select the output file format. You may choose between Magnitude/Angle, Real/Imaginary or Magnitude/dB by clicking on the appropriate radio button.

6

Select various options as needed. High Precision Output: Outputs higher precision numbers (more significant figures) to the response file. This file is useful if you plan on using the data later in a network analysis. Detect Box Resonance: Detects box resonance. For a detailed discussion of box resonance, please see Chapter 19, “Package Resonances.” Edge Mesh: Edge subsections are always one cell wide regardless of X Min or Y Min. When used in conjunction with large X Min or Y Min values, this option can be very useful in reducing the number of subsections while still maintaining the edge singularity. Quad Precision: Quadruple precision option. If cell size is less than about 10-5 wavelengths and vias are used in the circuit, numerical precision can cause numerical error. This option switches the matrix fill calculations to quadruple precision so that very low frequency analysis is accurate. For example, for a cell size of 1 micron and a frequency of 1 GHz, errors may be seen, and this option should be used. Be aware, however, that Quad Precision can slow the analysis down substantially; use the option only when necessary. Analyzing the circuit at a higher frequency is often a better solution. This option affects only the matrix fill calculations. This option and Memory Save are completely independent.

254

Chapter 24 Em Interface: Analysis of a Geometry File 7

Enter any desired advanced options in the Advanced Options text entry box. This text entry box allows you to enter advanced command options that are not otherwise available. For details on the options and their specifications see Appendix I, “The em Command Line.”

EM 255

Em User’s Manual

256

Chapter 25 Em Interface: Analysis of a Network File

EM

Chapter 25

Em Interface: Analysis of a Network File

There are two types of analysis that may be done in em: Geometry File or Network File. This chapter describes how to set up the analysis of a network file in em. Chapter 24 describes the analysis of a geometry file. A network file defines a circuit made of components which may consist of multiple geometry files, previously existing data files, resistors, capacitors, inductors and transmission lines. The file name must have the extension “.net.” Em will execute an electromagnetic analysis of this circuit at the frequencies specified within the network file or through use of an external analysis control file. The desired output is specified by selecting run options and output file types. For details on the network file, see Chapter 11, “The em Network File.” For details about invoking em, see “Invoking em,” page 246.

257

Em User’s Manual

Selecting a Network File Analysis You must select the type of analysis you wish to run: Geometry File or Network File. •

To analyze a network file, click on the Network radio button under File Type in the job window. The job window will appear as shown below. The Frequency Controls and Options now available in the job window are those appropriate for analyzing a network file. This window’s appearance will vary slightly for an analysis of a geometry file.

The em job window for a network file analysis.

Analysis Input Files This section will discuss specifying the network file and editing the network file.

Selecting a Network File This allows you to select the network file which you wish to analyze.

258

Chapter 25 Em Interface: Analysis of a Network File •

Enter the file name in the File text entry box. You may enter the file name in one of three ways: edit the text entry box, select a file name from the droplist or click on the Browse command button to open the Browse dialog box which will allow you to change directories if needed.

Editing a Network File You may wish to make changes to a network file before running an analysis by editing the contents of the file. •

Click on the Edit command button, which appears after the File text entry box, to invoke Notepad, or Vi, with the netlist file open. Notepad is an ASCII text editor, available on all Windows platforms. For details on how to use Notepad please see the appropriate documentation in your Windows package. For details on the network file, see Chapter 11, “The em Network File.”

!

WARNING You must save the network file for the changes made in the editing session to be available before running an analysis.

259

EM

You may also change directories by picking one from the drop list attached to the Start In text entry box.

Em User’s Manual

Specifying Frequency Control This section of the em job window, shown below, allows you to control the frequencies used in analyzing the network file. You may select a simple sweep, an internal sweep or an analysis control file. For details on the analysis control file option, see “Using an Analysis Control File,” page 250.

The Frequency Control section of the job window for a network file analysis.

Selecting a Simple Sweep A simple sweep functions the same when used to analyze a network file or a geometry file. For details on the simple sweep, see “Setting up a Simple Sweep,” page 248.

Selecting Internal Sweep An internal sweep uses the frequency control information provided in the FREQ block of the network file as opposed to an external control file. This option is only available for a network file. •

Click on the Internal Sweep radio button under Frequency Control in the em job window

Selecting Run Options You may select various options for the analysis of the network file including advanced options.

260

Chapter 25 Em Interface: Analysis of a Network File

Selecting the Verbose Option The Verbose option is set to on by default in a new job run. This option will cause em to display messages in the output window during program execution describing the present state of the analysis. EM

Selecting Additional Options for a Network File There are additional options available for a network file analysis in the Additional Options dialog box. These options provide advanced control over em’s execution. •

Click on the Additional Options command button in the job window to open the Additional Options dialog box.

The Additional Options dialog box for a network file.

The dialog box contains three options and an Advanced Option text entry box. To select an individual option, click on the checkbox. The option is on if a check appears inside the checkbox. The options and their use are described below. Use last data sets only Option: This will exclude all data sets present in an existing em response file except the final set with a tag which matches the tag specified in the SNP statement. See “Using Data Tags,” page 146. Do not check for consistency Option: Em will not check for geometry file consistency. All data sets in an existing em response file are read regardless of consistency. See “Geometry File Consistency,” page 148 for details. Force Running Option: This will ignore any existing analysis data. All em analyses are run at all frequencies.

261

Em User’s Manual Advanced Options: This text entry box allows you to enter advanced command line options that are not otherwise available. For details on the options and their specifications, see Appendix I, “The em Command Line.”

262

Chapter 26 Em Interface: Run Controls

EM

Chapter 26

Em Interface: Run Controls

This chapter details the em interface by describing how to accomplish various tasks while running the em program. The following will be discussed: •

How to specify frequency control information.

•

How to execute an analysis run.

•

How to view your analysis results.

263

Em User’s Manual

Editing Analysis Controls When editing a complex sweep or an analysis control file, input is provided through the Analysis Control dialog box, shown below.

You add frequency controls by clicking on the Add, Add IFS... or Add Separator command buttons in the Analysis Control dialog box. Entries will appear in the Frequency Control Entries list box.

264

Chapter 26 Em Interface: Run Controls

Adding Frequency Controls •

Click on the Add command button in the Analysis Control dialog box to open the Frequency Control dialog box.

EM This dialog box allows you to define one of four types of sweeps: a simple sweep, single frequency, exponential sweep or linear sweep. You select the type of sweep by clicking on the respective radio button. You then enter the required specifications for that sweep type in the text entry boxes. The sweep types and associated data are described below. An entry line will appear in the Frequency Control Entries list in the Analysis Control dialog box when you click on the OK command button. Sweep: An entry line starting with SWEEP specifies a list of analysis frequencies. Three numbers specify a start, stop and step. Step provides the increment between frequencies. For example, SWEEP 2.0 10.0 2.0 will analyze at 2, 4, 6, 8 and 10 GHz. The start, stop and step are all checked for error conditions. As many SWEEP lines may be used as is needed. When complete, the frequency list is sorted. There is no limit on the number of frequencies. Single Sweep: An entry line starting with STEP followed by up to three discrete frequency points as desired. The text entry boxes are FREQ1, FREQ2 and FREQ3. You may enter up to 3 individual frequencies if desired. For example, 265

Em User’s Manual STEP 3.0 17.5 28.0 will analyze at 3, 17.5 and 28 GHz. Exponential Sweep: An entry line starting with ESWEEP specifies an exponential frequency sweep from the starting frequency to the end frequency with a common ratio between the desired number of frequency points. The text entry boxes are Start, Stop and # of Points. The # of Points is the number of frequency points used for the analysis. For example, ESWEEP 8.0 64.0 4.0 will analyze at 8, 16, 32 and 64 GHz. Linear Sweep: An entry line starting with LSWEEP followed by three integers specifying the start, stop and number of frequencies. The text entry boxes are Start, Stop and # of Points. The # of Points is the number of frequency points used for the analysis. For example, LSWEEP 5.0 40.0 8.0 will analyze at 5, 10, 15, 20, 25, 30, 35 and 40 GHz.

266

Chapter 26 Em Interface: Run Controls

Entering Intelligent Frequency Controls •

Click on the Add IFS command button in the Analysis Control dialog box to open the Intelligent Frequency Control dialog box.

EM This dialog box allows you to define one of three types of sweeps: an auto sweep, a find minimum sweep and a find maximum sweep.You select the type of sweep by clicking on the respective radio button. You then enter the required specifications for that sweep type in the text entry boxes which appear. The sweep types and associated data are described below. The command will appear in the Frequency Control Entries list in the Analysis Control dialog box when you click on the OK command button. For a detailed discussion of intelligent frequency controls, see Chapter 10, “Intelligent Frequency Selection.” Auto: This feature automatically determines where to place frequency points. In the rapidly varying regions, the frequency points will be spaced close together. In the slowly varying regions, frequency points will be spaced farther apart. The text entry boxes are as follows: Start, Stop, Precision and Number of Points. The Number of Points text entry box is the number of frequencies between the start and stop frequency at which em will analyze. The Precision text entry box specifies the finest frequency resolution allowed for the analysis. All frequency points chosen by em will be spaced by at least the precision value. Find Min: Find Min determines the frequency where the circuit response reaches a minimum. The text entry boxes are as follows: Start, Stop, Precision, Parameter and Number of Points. The Precision text entry box specifies the finest frequency

267

Em User’s Manual resolution allowed for the analysis. All frequency points chosen by em will be spaced by at least the precision value. The Parameter text entry box defines the parameter for which you wish to determine the minimum value. For instance, “S2_1” would be S21. The Number of Points text entry box is the number of frequencies between the start and stop frequency at which em will analyze. This optional value can be used to set a reasonable limit on the number of iterations in the analysis. Find Max: Find Max functions in the same manner as Find Min except that it determines the frequency where the circuit response reaches a maximum.

Entering a Separator •

Click on the Add Separator command button to add an END statement to the Frequency Control Entries. The END command causes em to sort and analyze all frequencies not yet analyzed which precede the END keyword. A separator is used to force a particular order of frequencies.

Editing a Frequency Control Entry After you have entered a frequency control, you may wish to change it. •

Click on the entry in the Frequency Control Entries text box. The entry will be highlighted in reverse video. Click on the Edit command button to open the appropriate dialog box to edit the command. The same dialog box that was used to enter the command will now appear. Click on the Delete command button to delete the entry.

Specifying SPICE Parameters The SPICE dialog box allows you to set up parameters for a SPICE lumped element synthesis. These values are used in the analysis control file when an “.lct” output file is specified. For a discussion of SPICE formatted output, see Chapter 16, “SPICE Lumped Model Synthesis.”

268

Chapter 26 Em Interface: Run Controls •

Click on the SPICE command button in the Analysis Control dialog box to open the SPICE dialog box.

EM The SPICE dialog box allows you to enter the parameters necessary for producing a SPICE lumped model. The parameters are defined below. To enter a value, edit the respective text entry box. Rmax: This value specifies the largest resistor allowed for inclusion in the SPICE lumped model, in ohms. The default value is 1000.0 ohms. Cmin: This values specifies the smallest capacitor allowed for inclusion in a SPICE lumped model, in pF. The default value is 0.1 pF. Lmax: This values specifies the largest inductor allowed for inclusion in a SPICE lumped model, in nH. The default value is 100.0 nH. Kmin: This values specifies the smallest mutual inductance allowed for inclusion in a SPICE lumped model; it is a dimensionless ratio. The default value is 0.01. Rzero: This value specifies the resistor to go in series with all lossless inductors, in ohms. This parameter is provided for those versions of SPICE which require inductors with some small loss, to avoid numerical difficulties. The default value of 0.0 disables this capability.

Adding Comments to the Analysis Control File •

Click on the Comments command button in the Analysis Control dialog box to open the Comments text entry box. Edit the text entry box with any comments pertaining to the circuit. These comments will appear in the em response file.

269

Em User’s Manual

Specifying the Subsectioning Frequency Normally, the highest frequency of analysis is used as the subsectioning frequency. This option allows you to specify another subsectioning frequency. •

NOTE:

Enter the desired frequency in the Subsectioning Frequency text entry box in the Analysis Control text entry box. The higher of the two values, the entered subsectioning frequency or the highest frequency specified in the analysis, is used as the subsectioning frequency.

Saving Frequency Controls When you are done editing a Complex Sweep, click on the OK button to save the information and close the Analysis Control dialog box. If you wish to export the complex sweep to a file, click on the Export command button. When you are done editing an analysis control file, you may save the file under the same name, or a different name by clicking on the Save command button and the Save As command button, respectively.

Specifying Output Files There are seven different types of response data files which em can output.

270

Chapter 26 Em Interface: Run Controls •

Click on the Output Files command button in the job window to open the Select Output Files dialog box.

EM To select an output file, click on the checkbox and edit the accompanying text entry box with the desired file name, if the default name is not acceptable. The file name defaults to the basename of the input geometry file with the appropriate extension. For example for a non-de-embedded response file the extension is “.nd”. If “steps.geo” is the input file, then the output file would default to “steps.nd”. The type of output files are described below. .nd name: This file type contains response parameters with no de-embedding done. The file name must end with a “.nd” extension. This is the default output file if the De-embed option is not set. .d name: This file type contains response parameters with de-embedding applied.The file name must end with a “.d” extension. Selecting this output file will also set the De-embed option in the job window. This output file is selected with the default file name if the De-embed option is set. .jxy name: This file type contains current density data for use with the emvu program.The file name must end with a “.jxy” extension. Selecting this output file will also set the Make Emvu File option in the job window. This output file is selected with the default file name if the Make Emvu File option is set. .pnd name: This file type is a high precision circuit response file with no deembedding. The high precision output files contain S-parameter data in real/ imaginary format with more precision than a “.d” or “.nd” file. They are primarily 271

Em User’s Manual intended for internal use when executing a GEO line in a network file. Generally, you do not need to be concerned with these files. However, you may wish to create a high precision file if you plan on re-using the data in a network analysis. .pd name: This file type is the same as the “.pnd” file except the response data is de-embedded. .lc name: This file type contains a SPICE lumped model suitable for incorporating as a “.subckt” directly in a SPICE deck. For a detailed discussion of the SPICE file options, see Chapter 16, “SPICE Lumped Model Synthesis.” The file name must end with a “.lc” extension. .lct name: This file type contains a SPICE distributed N-coupled line RLCG matrix in SPICE format. For a detailed discussion of the SPICE file options, see Chapter 16, “SPICE Lumped Model Synthesis.” The file name must end with a “.lct” extension.

Running an em Analysis This section discusses running an em analysis: starting, stopping, pausing and viewing run lists.

Viewing the Run List The run list allows you to set up a sequence of em jobs, to be run one after another. The run list contains a list of command line equivalents of the options and files specified in each job. Multiple entries in the run list may be used in a batch like manner to execute multiple analysis runs.

272

Chapter 26 Em Interface: Run Controls •

To view the run list, click on the Show Run List toggle button, or select View ⇒ Run List from the main menu. The run list will appear at the bottom of the job window as shown below.

EM

The em job window with the run list displayed.

To hide the run list, click on the Hide Run List toggle button, or select View ⇒ Run List from the main menu.

Editing the Run List The initial command line appears by default when a new em window is opened and appears as follows: Not Run: em -dv

273

Em User’s Manual “Not Run” means the command line has not yet been executed. The -dv is for the De-embed and Verbose options which are set by default at initialization. As you use the interface to select files, frequency controls and run options, this command line will be modified to reflect those choices. For instance, if you select the geometry file, “steps.geo” and the analysis control file, “ctl.an”, then the entry would appear as follows: Not Run: em -dv ctl.an steps.geo steps.d Once the entry has been executed by clicking on the Run command button, the entry will start with “Finished:” in place of “Not Run.” It is possible to have multiple entries in the run list. When the Run command button is clicked, all the entries which have not yet run will be executed. This allows you to set up multiple analyses of the same or different circuits and execute with one action.

Adding a New Entry to the Run List •

To add a new entry to the run list, click on the New command button. The default entry of Not Run: em -dv will be highlighted. The highlighted entry will be affected by actions taken in the job window.

Copying an Entry in the Run List You may wish to add an additional entry to the run list which is similar to an existing entry in the run list. To do so, you may copy an entry and add it to the run list as described below. 1

Click on the entry you wish to copy. This entry will be highlighted, indicating it is the active entry.

274

Chapter 26 Em Interface: Run Controls 2

Click on the Copy command button to the right of the run list. This will enable the Paste Above and Paste Below command buttons.

3

If you wish the new entry to appear above the current entry, click on the Paste Above command button.

Deleting an Entry from the Run List If you wish to delete an entry from the run list, perform the following: 1

Click on the entry to make it the active entry.

2

Click on the Cut command button to delete the entry. The entry deleted by the Cut command is now available for the Paste Above and Paste Below commands.

275

EM

The new entry is highlighted, as the active entry, above the original command. The Paste Below command button operates in the same manner, except that the new entry appears below the original.

Em User’s Manual

Starting an em Analysis •

Click on the Run command button in the job window or select Run ⇒ Start Run from the em main menu. Em will execute the analysis, and the Output window, shown below, will appear on your display. If the verbose option is set, em will output status messages as it executes.

Em output window.

Pausing an em Analysis •

Click on the Pause command button at the bottom of the Output window or select Run ⇒ Pause Run from the em main menu. This will stop the execution of an em analysis until the Continue command button is clicked.

276

Chapter 26 Em Interface: Run Controls

Continuing an em Analysis •

Click on the Continue command button at the bottom of the Output window or select Run ⇒ Continue Run from the em main menu. This will start execution of an em analysis which has been paused.

•

EM

Stopping an em Analysis Click on the Stop command button at the bottom of the Output window or select Run ⇒ Stop Run from the em main menu. This will abort the em execution. Data from frequencies that have been completed will be saved, but data from the frequency currently being processed is lost.

Using the em Output Window The em output window, shown in the figure on page 276, displays the program status as the analysis proceeds when the Verbose option is on. The output window is automatically opened when the Run command button in the job window is selected.

Closing the Output Window •

Click on the Close command button at the bottom of the Output window. This will close the Output window on your display but will not affect the contents of the window.

Re-Opening the Output Window •

Select View ⇒ Output from the em main menu to open the Output window without running an analysis. The scroll bars to the right and bottom of the Output window allow you to move about the contents of the window.

277

Em User’s Manual

Saving the Contents of the Output Window You may save the contents of the Output window in a text file. •

Select File ⇒ Save As from the main menu. This opens the Save As dialog box which allows you to save the contents of the Output window, as a text file, to the specified file name.

Invoking emgraph to Plot Response Data You may observe your response data using the plotting program, emgraph, by invoking the program directly from the output window. •

Click on the Open Graph button at the bottom of the Output window. This will invoke the emgraph program with a Cartesian graph. For information on using emgraph, see the Emgraph User’s Manual. Emgraph may also be invoked by selecting View ⇒ Graph from the em main menu. Note that emgraph will not load a file when analyzing a network; open the desired file from within the emgraph program.

Invoking emvu to View Current Density You may observe your current density using the visualization tool, emvu, by invoking the program directly from the output window. •

Click on the Open Emvu button at the bottom of the Output window. This will invoke the emvu program with a current plot of the JXY Magnitude at the first analysis frequency. For information on using emvu, see the Emvu User’s Manual.

278

Chapter 26 Em Interface: Run Controls

Invoking patvu to View the Far-Field Radiation Patterns You may observe the far-field radiation patterns of your analysis data using the visualization tool, patvu, by invoking the program directly from the output window. •

Select View ⇒ Open Patvu from the menu of the output window.

Invoking a Text Editor to View Response Data You may observe your response data file using the ASCII text editor, Notepad, on Windows systems and the Vi editor on UNIX systems, by invoking the program directly from the output window. •

Select View ⇒ Open Data from the menu of the output window. This will invoke Notepad on Windows, or Vi on UNIX, with the response data file open. For information on using these editors, please see the appropriate documentation for the program. Notepad, or Vi, may also be invoked by selecting View ⇒ Data from the em main menu. Note that em will not load a file when analyzing a network; open the desired file from within the text editor program.

Job Files A job file is used to store all the run options set in the em window that control an analysis. This section will discuss opening job files, creating new job files and saving job files.

279

EM

This will invoke the patvu program with a far-field plot for the default data. For information on using patvu, see the Patvu User’s Manual.

Em User’s Manual

Creating a New Job File A job file allows you to save to disk all the run options and input and output files that have been specified for a particular analysis run. •

Select File ⇒ New from the main menu. A new job window will appear in your display. You may use this window and its dialog boxes to specify input files, output files and run options to control an analysis run. You may then execute the analysis.

Opening an Existing Job File You can open an existing job file in em. The file extension must be “.job”.

280

Chapter 26 Em Interface: Run Controls 1

Select File ⇒ Open from the main menu. The Open File dialog box, shown below for both Windows and UNIX systems, appears on your display.

Open File dialog box for UNIX

281

EM

Open File dialog box for Windows

Em User’s Manual 2

Select the name of the job file you want to open. To open the file in one step, double-click the document name in the scroll list. If the file you want to open is not in the scroll list, change to another directory by double-clicking a directory name in the scroll list or by typing a different name in the text box above the scroll list and clicking Open. This will open a task window containing the specified job file.

3

Click Open. A job window will be opened containing the specified job file.

Loading an Existing Job File You can open an existing job file in em without opening another job window. The file extension must be “.job”. The file you select will replace the active job. 1

Select File ⇒ Load from the main menu. If the active job contains unsaved changes, a query window will appear saying: “The File has been changed, do you want to save it? Click on Save, Discard, or Cancel. Once the file in the active job window is closed, the Open File dialog box, shown in the figure on page 281, will appear on your display.

2

Select the name of the job file you want to open. To open the file in one step, double-click the document name in the scroll list. If the file you want to open is not in the scroll list, change to another directory by double-clicking a directory name in the scroll list or by typing a different name in the text box above the scroll list. The job file will replace the old job in the active job window.

282

Chapter 26 Em Interface: Run Controls

Saving a Job File Saving the Current Job File To save the current job file, select File ⇒ Save from the main menu.

When you save a job you can change its name or location. When you change a job file’s name or location you make a copy of the job file. •

Select File ⇒ Save As from the main menu. This opens the Save As dialog box, similar to the Open File dialog box shown in the figure on page 281, which allows you to change the name or location of your job file. Enter the directory and file name under which you want to save the file. Then click on the Save command button or press the return key.

Reverting to a Saved Job File •

To close the current job file without saving it and open the most recently saved version, select File ⇒ Revert from the main menu. This command is useful if you have made changes to a job file which you wish to discard.

Em Control Preferences The preferences dialog box, accessed by selecting File ⇒ Preferences from the main menu, allows you to control program settings for em Control.

Setting Multi-Frequency Caching Parameters To set the default cache directory and a limit on cache memory perform the following:

283

EM

To Change a Job File’s Name or Location

Em User’s Manual 1

Select File ⇒ Preferences from the main menu. The Preferences dialog box appears on your display.

references dialog box.

2

Enter the desired directory for Cache memory in the Cache Directory text entry box. You must enter the complete path of the directory name. A directory called "sonnet_cache" is created in the specified directory. Any value entered in this text box overrides the global value, if any, entered in the sonnet.ini file. For details about the sonnet.ini file, see Chapter 8, "Initialization File" in the Sonnet Installation Manual.

3

Enter the desired limit, in Megabytes, in the Cache Limit text entry box. Enter the maximum amount of disk space available to use for the cache data. If you do not wish to limit the amount of disk space, leave “None” in the text entry box. If a particular simulation exceeds this limit, em will automatically disable MFC for that simulation.

4

Click on the OK command button. This closes the dialog box and applies your entries. These preferences are used for multi-frequency caching in all subsequent em jobs, including the present setup.

284

Chapter 26 Em Interface: Run Controls

Selecting Startup Run Options You may specify startup run options in the Preferences dialog box. The run options specified are set for all subsequent jobs. 1

Select File ⇒ Preferences from the main menu.

2

EM

The Preferences dialog box appears on your display as shown on page 284. Enter the command line run options you wish to have as default in the Startup Options text entry box. Any options you enter will be set when a new job window is invoked. For example, entering “-vd” sets the Verbose and De-embed options. For details about command line options, see “The em Command Line,” page 287. Clicking on the Set to Top Window button enters any options presently set in the job window into the Startup Options text entry box. For example, if the Verbose, De-embed and Make emvu file options are set in the main job window and the Quad Precision option is selected in the Additional Options dialog box when you click on the Set to Top Window button, then "-djqv" appears in the Startup Options text entry box. 3

Click on the OK command button. This closes the dialog box and applies your entries. These preferences are used in all subsequent em jobs, including additions to the run list, but not for the present job.

Setting Up a Default Simple Sweep for Analyses You may specify a default simple sweep in the Preferences dialog box. The simple sweep specified will be used in any subsequent em jobs. 4

Enter the Start, Stop and Step values in the appropriate text entry boxes in the Startup Simple Sweep section of the Preferences dialog box. If you wish to analyze at only one default frequency, enter that frequency in the Start text entry box. Otherwise, Start provides the beginning frequency, Stop the ending frequency and Step the interval between analysis frequencies. For instance

285

Em User’s Manual if the Frequency Units were set to GHz, then the values of 2, 10, 2 would start an analysis at 2 GHz and end at 10 GHz with steps of 2 GHz (e.g. 2, 4, 6, 8 and 10 GHz) 5

Select the desired frequency units from the drop list. You may select from Hz, KHz, MHz, GHz, THz, and PHz. This sets the units used for the specified frequencies.

6

Click on the OK command button. This closes the dialog box and applies your entries. These preferences are used in all subsequent em jobs, including additions to the run list, but not for the present job.

286

Appendix I The em Command Line

EM

Appendix I

The em Command Line

This appendix details the em command line. The command line appears in the Run List in the main em window. This chapter also serves as syntax guide for a batch file or for the OPT field in a GEO command in a network file. If the option is available in the interactive interface, the location is identified. Options discussed here, that are not available through em Control, may be entered in the Advanced Options text entry box in the Additional Options dialog box. These will be identified as advanced options. NAME: em - Electromagnetic analysis of planar circuits. SYNTAX: em [-options] [files]

287

Em User’s Manual where [options] is a list of command line options and [files] is a list of input and output files. The [options] list is made up of one or more groups of options, with each group preceded by a minus sign (-). For example, the following are equivalent ways to express the same list: -vdmjbx5 -vd -mjb -x5 -v -d -m -j -b -x5

Options -h

List frequently used options to your screen. This is only available as an advanced option.

-m

Memory saver option. The system matrix is filled in single precision which reduces memory requirements for storing the matrix to one half of that of double precision. On some computers, matrix solution time is also faster. If this option is not used, the matrix is stored in double precision. This option affects only the matrix storage and the matrix solution. The precision of the matrix fill calculations is not affected (see “-q”). This is available as the Memory Save option in the Options section of the job window. Available only for geometry file analyses.

-q

Quadruple precision option. If cell size is less than about 10-5 wavelengths and vias are used in the circuit, numerical precision can cause numerical error. This option switches the matrix fill calculations to quadruple precision so that very low frequency analysis is accurate. For example, for a cell size of 1 micron and a frequency of 1 GHz, errors are often seen, and the -q option should be used. Be aware, however, that Quad Precision can slow the analysis down substantially; use the option only when necessary. Analyzing the circuit at a higher frequency is often a better solution.

288

Appendix I The em Command Line This option affects only the matrix fill calculations. The matrix is still stored and solved in double (or single, if “-m”) precision. This option and “-m” are completely independent. This is available as the Quad Precision checkbox in the Additional Options dialog box. Available only for geometry file analyses. N-port circuit parameters are stored in Y or Z parameter form. The Y and Z parameters are normalized to 1 ohm independent of port terminations. If no option is specified, S-parameters are stored. These options are available under the Parameters section of the Additional Options dialog box. Available only for geometry file analyses. -C

Store the circuit response information in Super-Compact format. If no option specified, Touchstone format is used. This option is available under the File Format section of the Additional Options dialog box. Available only for geometry file analyses.

-n

Specify a name for the output SPICE sub-circuit (see -x) or for the S-parameter output data. For example, “em -x -nModel_name”. This is an advanced option and is available only by editing the Advanced Options text entry box in the Additional options dialog box.

-r

Specify the file for circuit response (usually S-parameter) data. Specifying a command line file name which ends with a period followed by “nd” or “d” is equivalent to specifying the option. This option is usually used when the desired response file does not end with “.nd” or “.d”, for example, “-rAnswer.s2p”. This is an advanced option and is available only by editing the Advanced Options text entry box in the Additional options dialog box. Note that this option will only function if all output file types are turned off in the Output Files dialog box.

-g

Specify the file for circuit geometry. Default file is “cir.geo”. Specifying a command line file name which ends with a period followed by “geo” is equivalent to specifying the option. For example, “-gFile.geo”.

289

EM

-y -z

Em User’s Manual This option is available in the Geometry File text entry box in the job window. Available only for a geometry file analysis. -a

Specify the analysis control file. Default file is “ctl.an.” Specifying a command line file name which ends in a “.an” is equivalent to specifying the option. If the option “-astdin” is specified or if the analysis control file does not exist, the user is prompted for a start, stop and step frequency in MHz. These options are available in the Frequency Control section of the job window.

-v

Verbose mode. Display messages during program execution describing the current state of the analysis. This is available in the Options sections of the job window. Available for both geometry file and network file analyses.

-j

Outputs current density information. Specifying a command line file name which ends with “.jxy” is equivalent to specifying the option. If the “.jxy” file already exists, it is renamed “.jxb” and output is sent to the new, empty, “.jxy” file. In this case, any previous “.jxb” file is lost. This option is available by selecting the Make emvu file option in the job window or by specifying a “.jxy” file in the Output Files dialog box. Available only for geometry file analyses.

-J

Outputs current density file containing subsectioning information only. Emvu will show circuit as completely red, but can be useful to view subsectioning. If the “.jxy” file already exists, it is renamed “.jxb” and output is sent to the new, empty, “.jxy” file. In this case, any previous “.jxb” file is lost. This option is available by selecting Generate Subsections Only in the Additional Options dialog box. Available only for geometry file analyses.

-x

290

SPICE lumped model synthesis. Requires at least two frequencies of analysis. Output is a lumped model suitable for incorporating as a “.subckt” directly in a SPICE deck. If followed by a single number, 0-7, that number is taken as the number of digits to the right of the decimal place to be used for formatting lumped element values. Default is 2. Output file name is “spice.lc”, unless there is a file ending in “.lc” on the command line.

Appendix I The em Command Line This option is available by selecting a “.lc” file in the Output Files dialog box and entering the values in the SPICE dialog box accessible in the Analysis Control dialog box. However, if you wish to use the number of digits you must enter -xn in the Advanced text entry box in the Additional Options dialog box. Available only for geometry file analyses. -E

This option is available as the Edge Mesh option in the Additional Options dialog box. Available only for geometry file analyses. -F

Force multi-frequency caching (MFC) to run above the cutoff frequency for box resonances. By default, the MFC algorithm computes a cutoff frequency above which box resonances may occur. MFC is then enabled for all frequencies up to the cutoff frequency. This option forces MFC to be enabled at all frequencies. You may use this option when there are no box resonances present in the frequency band over which you are analyzing. Note, however, if you use this option and there are box resonances present in the analysis band, the s-parameter results over the entire band may be corrupted.

-X

SPICE distributed N-coupled line RLCG matrix synthesis. The geometry must be an N-coupled line. Only a single frequency need be specified. Output are the L and C N x N matrices for the N-coupled line. If there is metal loss, R is also generated. If there is dielectric loss, G is also generated. Output file name is “spice.lct”, unless there is a file ending in “.lct” on the command line. This option is available by selecting a “.lct” file in the Output Files dialog box. Available only for geometry file analyses.

-N

Calculates the number of subsections followed by an estimate of the number of Mbytes of memory required. The analysis is not performed. This option is available by selecting Calculate Memory Usage in the Additional Options dialog box. Available only for geometry file analyses.

291

EM

All Manhattan polygons are treated as if they were non-Manhattan polygons. In other words, the edge subsections are always one cell wide regardless of X Min or Y Min. When used in conjunction with large X Min or Y Min values, this option can be very useful in reducing the number of subsections but still maintaining the edge singularity.

Em User’s Manual -d

The circuit is automatically de-embedded to the specified reference planes (see the geometry file description). This is the only case in which the reference plane information is used. Specifying a command line file name which ends with a period followed by “d” is equivalent to specifying the option. This option is available by selecting the De-Embed option in the job window or by specifying a “.d” file in the Output Files dialog box. Available only for geometry file analyses.

-b

Detects potential box resonances and prints out a warning message just before the frequency data in the output file. This option is available as the Detect Box Resonances option in the Additional Options dialog box. Available only for geometry file analyses.

-e

This option disables the detection of polygon edges on other than the present level for subsectioning purposes. If thin dielectric layers are used (for example, capacitor dielectrics), this option is not recommended. May result in a less accurate, but faster analysis. This option is only available by entering it in the Advanced text entry box in the Additional Options dialog box. Available for both geometry and network file analyses.

-P

Outputs higher precision numbers (more significant figures) to the response file. This option is available as the High Precision Output option in the Additional Options dialog box. Available only for geometry file analyses.

-p

Outputs a special high precision real-imaginary S-parameter file (“.pd” or “.pnd” file) to be used by the networking capability. This option is available by specifying a “.pd” or “.pnd” file in the Output Files dialog box. Available only for geometry file analyses.

-R

Outputs real/imaginary data. This option is available as the Real/Imag option in the Additional Options dialog box. Available only for geometry file analyses.

-DB 292

Outputs magnitude/dB data.

Appendix I The em Command Line This option is available as the Mag/dB option in the Additional Options dialog box. Available only for geometry file analyses. -f

This is only available as an advanced option. -ver

Prints out the present version of em and license id and exits. Must be the first option on the command line. All other options are ignored. This option is available only as an advanced option. All other options must be off, and this option entered into the Advanced text entry box in the Additional options dialog box for it to function properly.

-Oforcerun

Ignore any existing em analysis data. All em analyses are run at all frequencies. This option is available as the Force Running option in the Additional Options dialog box. Available only for network file analyses.

-Ofs

Specify the field size (number of characters) for network output results. The value must be > 2. For example, if -Ofs7 is specified: • • •

2 is stored as 2.00000 -3 is stored as -3.0000 4e-9 is stored as 4.00e-9

The default field size is 8 for S-parameter magnitudes and 6 for S-parameter phases. This option is only available by entering it in the Advanced text entry box in the Additional Options dialog box. Available only for network file analyses. -Olast

Exclude all data sets present in an input file except the final set with a tag which matches the tag specified in the SNP statement. See “The CKT Data Block,” page 142 for details.

293

EM

Outputs comma separated value data for use in common spread sheet programs such as Excel by Microsoft Corporation. The output file will be named “basename.csv.” For example, if you are analyzing “steps.geo” with the -f option, the output file would be named “steps.csv.” This filename extension is reserved for Excel.

Em User’s Manual This option is available as the Use last data sets only option in the Additional Options dialog box. Available only for network file analyses. -Onocheck

Do not check for geometry file consistency. All data sets in an existing em response file are read regardless of consistency. See “Geometry File Consistency,” page 148 for details. This option is available as the Do not check for consistency option in the Additional Options dialog box. Available only for network file analyses.

-SNP

This option outputs a Touchstone format frequency sorted response file. The Touchstone format frequency sorted response files provide Touchstone format S-Parameter data with the following characteristics: •

File contains only S-parameter data in Touchstone format.

•

Data is sorted by frequency.

•

File contains only data which is consistent with the present analysis.

•

File is updated on a frequency-by-frequency basis.

•

File contains de-embedded results if de-embedding was enabled, otherwise it contains non-de-embedded results.

•

File contains -ohm s-parameters, provided that all ports in the circuit are terminated with ohms. If not, the file contains 50ohm s-parameters.

Input/Output Files The input and output files specified on the em command line vary depending upon the type of analysis being performed. For a network file analysis, you must specify a network (“.net”) input file on the command line. Generally, this is the only file specified on the command line when analyzing a network file. All other input and output files are usually specified within the network file. The only exception is an analysis control file (“.an”). If

294

Appendix I The em Command Line you specify an analysis control file along with a network file on the em command line, em will ignore all information contained in the FREQ block of the network file and will instead use the information contained in the analysis control file. The following files may be specified on the batch command line: Input file. The file name must end with “.an”. This file can set the frequencies for analysis, among other things. For executing from a batch file, if the file does not exist, you will receive an error message. When performing a network file analysis, if you specify an analysis control file in the command line, em will ignore all information contained in the FREQ block of the network file and will instead use the information contained in the analysis control file. This is the equivalent of selecting Analysis File under Frequency Control in the job window. Circuit Geometry File

Input file. The file name is required to end with “.geo.” The circuit geometry file can be created with a geometry capture program such as xgeom, see the Xgeom User’s Manual. If the circuit geometry file does not exist, em terminates.

Network File

Input file. The file name is required to end with “.net.” The network file can be created with an ASCII text editor, such as Notepad or Vi. For details on the network file, see Chapter 11, “The em Network File.”

Circuit Response File

Output file. The file name should end with “.nd” or “.d”. If the file ends with “.d”, automatic de-embedding is enabled. Otherwise, the file name must be specified with the -r option. This file contains the N-port circuit parameters (e.g., Sparameters) of the circuit being analyzed. This file can be used for input directly to any of a number of high frequency circuit analysis programs. If the analysis is run with High Precision invoked (-p), the resulting circuit response file will end with “.pnd” or “.pd” extension. If the -SNP option is used, then a Touchstone format frequency sorted response file is output with the extension ".sp" or ".s" where is the number of ports when the number of ports is less than 10 and is the number of ports when the number of ports is greater than 10.

295

EM

Analysis Control File

Em User’s Manual Current Density File

Output file. The file name usually ends with “.jxy”. This file stores current density information on the file for later viewing by emvu. Analysis time is increased by this option and the file can take up a large amount of disc space if the circuit is large with many ports.

SPICE File

Output file. The file name should end with a “.lc” or a “.lct.” The “.lc” file contains a lumped model of inductors, capacitors, resistors and mutual inductors. The “.lct” file contains LCRG matrices for N-coupled transmission lines. These matrices represent the distributed parameters of the transmission lines. See Chapter 16, “SPICE Lumped Model Synthesis,” for a detailed discussion of the SPICE options.

296

Appendix II The Analysis Control File Format

EM

Appendix II

The Analysis Control File Format

The analysis control file controls the frequencies used for analysis as well as other analysis parameters. The file is required to have a name ending with “.an”. If no file name is specified in the command line, em looks for “ctl.an” in the current directory. If that file is not found, em returns an error message. Any line with a first non-space character of “!” is ignored. Any blank line is also ignored. Comments following any complete line of data are allowed. In the keywords that follow, only the specified number of letters (3 or 4) are significant. Upper and lower case letters are allowed. Additional letters may be used but they do not alter the program’s execution. For example, “VER”, “VERSION” and “VERTRFGH” all have the same effect. There may be no more than 255 characters per line.

297

Em User’s Manual VER

Any line with the first three characters “VER” is taken to specify the em version number. Up to five characters following the end of the first space characters following “VER” are read into em. This information is required for compatibility with future versions. The VER line should be the first non-comment line in the file.

ANN

When the response file is created, the analysis parameter file is searched for any lines beginning with ANN. The remainder of any lines found are listed in the heading of the circuit response file. This is useful for the automatic documentation of the em analysis output.

HZ KHZ

Frequency units can be specified by a line with “HZ”, “KHZ”, “MHZ”, “GHZ”, “THZ”, or “PHZ”. The frequency units can be changed as often as desired within the same analysis file. All frequencies specified on a FRE line must be in the units most recently specified. If no units have been specified, MHZ is assumed. If no frequency units were specified in the command line, the frequency units in effect at the end of the analysis file is used for the circuit response file. Frequency units specified in the command line have no effect on how the analysis file is read.

MHZ HZ

SWEEP

A line starting with SWEEP specifies a list of analysis frequencies. SWEEP may be followed by one, two or three numbers. One or two numbers specify one or two frequencies. Three numbers specify a start, stop and step. The start, stop and step are all checked for error conditions. As many SWEEP lines may be used as is needed. When the file is complete, the frequency list is sorted. Duplicate frequencies are not removed. There is no limit on the number of frequencies.

FRE

Same as the SWEEP keyword.

LSWEEP

Syntax: LSWEEP f1 f2 Nfreq Linear frequency sweep from f1 to f2. Step size is equal to(f2-f1)/(Nfreq-1).

ESWEEP

Syntax: ESWEEP f1 f2 Nfreq. Exponential frequency sweep from f1 to f2 with a common ratio between the Nfreq frequency points.

298

Appendix II The Analysis Control File Format Followed by a floating point number, specifies the effective dielectric constant (Eeff) used to calculate the wavelength for satisfying the subsections/wavelength parameter. If not specified, or if it is less than 1.0, the parameter is ignored and a simple estimate of Eeff is used.

END

Sort and analyze all frequencies (not yet analyzed) which precede the END keyword. Used to force a particular order of frequencies.

STEP

Followed by as many discrete frequency points as desired.

FMAX

The subsectioning parameter “subsections/wavelength” normally uses the highest analysis frequency to determine the wavelength. However, this may be changed by using the keyword “FMAX” followed by a frequency in the ctl.an file. That frequency (in the units most recently specified) is now used for the wavelength determination instead of the highest frequency of analysis. Thus, the same subsectioning can be used for several analyses which differ in the highest frequency being analyzed.

CMIN

Followed by a number, specifies the smallest capacitor allowed for inclusion in a SPICE lumped model, in pF.

LMAX

Followed by a number, specifies the largest inductor allowed for inclusion in a SPICE lumped model, in nH.

RMAX

Followed by a number, specifies the largest resistor allowed for inclusion in a SPICE lumped model, in ohms.

KMIN

Followed by a number, specifies the smallest mutual inductance allowed for inclusion in a SPICE lumped model, dimensionless ratio

RZERO

Followed by a number, specifies the resistor to go in series with all lossless inductors, in ohms. Needed for some versions of SPICE.

AUTO:

AUTO NET=GEO N=Nfreq f1 f2 prec

EM

EEFF

Automatic frequency selection using the geometry file as the basis. Em begins by analyzing at f1 and f2. It then analyzes at Nfreq frequencies between f1 and f2. The prec field specifies the frequency grid upon which frequencies are selected. For

299

Em User’s Manual example, if prec = 0.10, f1 = 1.0 and f2 = 2.0, the algorithm is constrained to the following frequencies: 1.00, 1.10, 1.20, 1.30, 1.40, 1.50, 1.60, 1.70, 1.80, 1.90 and 2.00. FINDMIN:

FINDMIN NET=GEO param [MAX=Nfreq] f1 f2 prec FINDMIN finds the frequency at which the minimum frequency response of the geometry file occurs. The param field specifies a basis S-, Y- or Z-parameter using one of the following formats: pxy or px_y, where p is S-, Y- or Z, and x,y are a pair of port indices. The px_y format must be used when a port index with two or more digits is referenced. For example, S[port 1 - port 2] may be specified as S12 or S1_2, but S[port 15 - port 1] may only be specified as S15_1. The search for the minimum is constrained to frequencies which fall on a grid controlled by prec, f1 and f2 (see description of AUTO). If Nfreq is specified, the total number of frequency points analyzed is limited to the endpoints f1 and f2, plus Nfreq points between f1 and f2. FINDMAX: FINDMAX NET=GEO param [MAX=Nfreq] f1 f2 prec FINDMAX is identical to FINDMIN except that it finds the frequency at which the maximum frequency response of the geometry file occurs.

300

Appendix III LEVEL1 and LEVEL1plus

EM

Appendix III

LEVEL1 and LEVEL1plus

This appendix describes the restrictions on the software for the LEVEL1 and LEVEL1plus suites.

LEVEL1 Suite The LEVEL1 suite includes the following Sonnet products, with the limitations cited below: xgeom, em, emgraph, emvu, dxfgeo, and patvu. The circuit network capability is available as an add-on purchase, but is not available in the LEVEL1 demo. •

One metalization layer available. The full Sonnet suite allows an unlimited number of metalization layers; LEVEL1 is limited to one metalization layer. The option to add another dielectric layer, and hence another metalization layer is not available. The Add Below and Add Above buttons in the Dielectric Layers dialog box in xgeom are disabled.

301

Em User’s Manual •

Maximum of two dielectric layers available. The full Sonnet suite allows an unlimited number of dielectric layers; LEVEL1 is limited to two dielectric layers. The Add Below and Add Above buttons in the Dielectric Layers dialog box in xgeom are disabled.

•

64 Megabyte memory limit. The full Sonnet suite allows use of an unlimited memory space, although most users limit the memory to the size of their physical memory. LEVEL1 limits you to the use of 64 Megabytes of memory regardless of the memory available.

•

Dielectric bricks are not available. The full Sonnet suite allows for the use of dielectric bricks throughout a circuit. Dielectric bricks are not available for the LEVEL1 suite. The Brick Mode button in the xgeom tool box and the Modify ⇒ Convert to Bricks menu items are disabled in xgeom.

•

Auto-grounded ports are not available. The full Sonnet suite allows for the use of an unlimited number of auto-grounded ports. Auto-grounded ports are not available in the LEVEL1 suite. The Type drop list in the Port Attributes dialog box in xgeom is disabled.

•

Parallel subsections are not available. The full Sonnet suite allows you to remove parallel subsections where there is very little transverse current to reduce the number of subsections and improve processing time. Parallel subsections are not available in the LEVEL1 suite. The Parameters ⇒ Parallel Subsections menu item in xgeom is disabled.

•

Vias not available. The full Sonnet suite allows an unlimited number of vias which allow current to flow in the Z-direction between metallization layers. LEVEL1 suite does not allow the use of vias. In xgeom, the Add Vias button on the tool box, the Tools ⇒ Add Via menu item, and the Modify ⇒ Add Vias to All menu item are disabled.

302

Appendix III LEVEL1 and LEVEL1plus •

Kinetic inductance is not available. The full Sonnet suite allows the user to specify a kinetic inductance for a metal type for superconductor applications. This parameter is not available for the LEVEL1 suite. The Ls parameter in the Metal Types dialog box in xgeom is disabled. Maximum of 4 ports available. The full Sonnet suite allows an unlimited number of ports in a circuit. A maximum of 4 ports are allowed in the LEVEL1 suite. After 4 ports have been added to a circuit in xgeom, the Add Port button on the tool box and the Tools ⇒ Add Port menu item are disabled.

•

The variables XMIN and YMIN are not available. The full Sonnet suite allows you to control how the circuit is subsectioned by allowing you to set a minimum size for subsectioning in the x and y directions for any given polygon. In the LEVEL1 suite these values are both set to the default value of 1. The X Min and Y Min text entry boxes in the Metalization Attributes dialog box in xgeom are disabled.

•

The variables XMAX and YMAX are not available. The full Sonnet suite allows you to control how the circuit is subsectioned by allowing you to set a maximum size for subsectioning in the x and y directions for any given polygon. In the LEVEL1 suite these values are both set to the default value of 100. The X Max and Y Max text entry boxes in the Metalization Attributes dialog box in xgeom are disabled.

•

The variable Max. Subsection Size is not available. The full Sonnet suite allows you to control how the circuit is subsectioned by allowing you to set a maximum size for subsectioning in terms of subsections/ lambda. In the LEVEL1 suite this value is set to the default value of 20. The Max. Subsection Size text entry box in the Box Parameters dialog box in xgeom is disabled.

303

EM

•

Em User’s Manual •

Multi-Frequency Caching run option is not available in em Control. The full Sonnet suite provides the Multi-Frequency run option for em, which precomputes frequency independent data to save on processing time. The checkbox in the em Control job window is disabled.

LEVEL1plus Suite The limitations on the LEVEL1plus suite are the same as the LEVEL1 suite with the following exceptions: •

128 Megabyte memory limit. The full Sonnet suite allows use of an unlimited memory space, although most users limit the memory to the size of their physical memory. LEVEL1plus limits you to the use of 128 Megabytes of memory (twice that of LEVEL1) regardless of the memory available.

•

Vias are available. The full Sonnet suite allows an unlimited number of vias which allow current to flow in the Z-direction between metallization layers. LEVEL1plus suite also allows the use of vias (up or down).

•

Maximum of 6 ports available. The full Sonnet suite allows an unlimited number of ports in a circuit. A maximum of 6 ports (2 more than LEVEL1 suite) is allowed in the LEVEL1plus suite. After 6 ports have been added to a circuit in xgeom, the Add Port button on the tool box and the Tools ⇒ Add Port menu item are disabled.

•

Internal Ports available. The full Sonnet suite allows an unlimited number of internal ports. LEVEL1plus suite allows internal ports. Note that these ports are counted as part of the 6 port limit.

304

Appendix IV Warning and Error Messages

EM

Appendix IV

Warning and Error Messages

The following is a list of error and warning messages that may be generated by em. When warning messages occur, em continues to run. A warning message should be considered important to note, but does not necessarily mean that you have done anything wrong. When error messages occur, em does not continue to run.

305

Em User’s Manual

Warning Messages Below use of Memory Saver (-m) may result in error. At extremely low frequencies, when the subsections size is on the order of 0.0001 wavelength or smaller, single precision may not be enough to allow the matrix to be properly inverted. In this case, very strange results are generated. For example, S21 may be different from S12. In such a case, eliminate the -m option and reanalyze.

Circuit has metal with no subsections. This means that a polygon has been found that does not contain any subsections. Look at the xgeom file at these <x,y> locations. One reason a polygon may not have metal is because it is outside of the xgeom box. Another reason might be because the polygon is too small. Any polygon that is smaller than 1 cell by 2 cells may have missing subsections. No current is allowed to flow in the X or Y direction in these locations.

Circuit outside of box at (<x>, ), level . All circuit outside of box is ignored. This means that em found part of the circuit to be outside the xgeom box. To find the problem area, bring up xgeom and go to level . Make sure your origin is set to “Top Left” (using View ⇒ Origin). Then use the ruler to determine the location of the coordinates given by <x> and in the warning message. If the circuit is only slightly outside the box (less than a half a cell), this message may be ignored. To correct the problem, you may snap the offending polygon(s) to the grid (using Edit ⇒ Snap).

No subsections in rectangular area <x,y>. See “Circuit has metal with no subsections.”

306

Appendix IV Warning and Error Messages <Side> box wall de-embedding merit down to <x>%.

Subsections/wavelength value of specified in file . Using required minimum of 6 subsections/wavelength for analysis. The subsections/wavelength parameter was less than 6 in the geometry file being analyzed. The em analysis requires that the parameter have a minimum value of 6 subsections/wavelength. This minimum value is used in place of the value specified in the geo file.

The ‘-q’ option has no effect on this computer. Some computers do not support quad precision. This message occurs when quad precision is attempted on a computer that does not support quad precision.

The thickness of layer is less than 0.05 uM. Ultra-thin layers may result in numerical precision problems. Precision problems have been found with ultra-thin layers. When this occurs, the data may be incorrect. Try setting the thickness a little larger than 0.05 microns.

307

EM

This message means that something may be wrong with de-embedding. Some possible causes for this problem are box resonances, higher order propagating modes, and short reference planes or calibration standards. Try taking out the reference planes or making them longer. If you do not have reference planes, then try specifying a calibration standard for the side specified in the error message. Usually a reference plane or a calibration standard length of 2-3 substrate thicknesses is sufficient.

Em User’s Manual Transmission Line SPICE model requires even number of ports. Found ports. Your “.geo” file must be an N-coupled line with ports 1 through N as input and ports N+1 through 2N as output. The input of line M should be port M and its output should be port M+N. The software does not check for this condition, but issues a warning message if the number of ports is not an even number. This restriction does not apply to generating “.lc” files, only generating “lct” files. There is no limit on N.

<X|Y> cell size is greater than wavelengths (<m> ) at highest frequency. If the subsections/wavelength parameter is K, then the maximum allowed subsection size is lambda/K. Since the smallest possible subsection size is equal to the cell size, the maximum allowed cell size is also lambda/K. This warning message is output when the cell size is greater than lambda/K. It indicates that your cell size in the <X|Y> direction may be too large and may result in analysis error.

Error Messages Bottom ref. plane must be more than 3 cells long. The reference plane associated with the bottom side of the box is too short for accurate de-embedding. Usually the easiest way to fix this problem is to remove the reference plane. De-embedding is still valid for zero-length reference planes.

Dielectric constant of layer <= 0.0. The dielectric constant must be non-zero and positive.

308

Appendix IV Warning and Error Messages Hardware key not found. This message applies only to PCs. Make sure your key is properly connected to your parallel port. If you have multiple keys connected, try switching them around. EM

Left ref. plane must be more than 3 cells long. The reference plane associated with the left side of the box is too short for accurate de-embedding. Usually the easiest way to fix this problem is to remove the reference plane. De-embedding is still valid for zero-length reference planes.

Lumped spice model must have even number of analysis frequencies. The “-x” option requires an even number of frequencies for analysis. Change your analysis control file so there are an even number of frequencies.

Port is below line of symmetry. The analysis has determined that the circuit has a port below the line of symmetry and symmetry is enabled. This is not allowed. When symmetry is enabled, all ports must be located on or above the line of symmetry. To perform the analysis, either remove the port below the symmetry line, or disable symmetry.

Port is on ground plane. The analysis has determined that the circuit has an invalid port located on the ground plane. Box-wall, auto-grounded, and ungrounded-internal ports cannot be located on the ground plane. Via-ports can be located on the ground plane, but the port MUST be attached to an edge via (the edge-via and the port must be on the same polygon edge).

309

Em User’s Manual Port is not connected between two polygons. The analysis has determined that the circuit has a standard port connected to the edge of a single polygon, located in the interior of the metal box. This is not allowed. Box-wall ports must be located on a box-wall. Ungrounded-internal ports must be connected between two abutting polygons.

Relative permeability of layer <= 0.0. The relative permeability of any layer must be non-zero and positive.

Right ref. plane must be more than 3 cells long. The reference plane associated with the right side of the box is too short for accurate de-embedding. Usually the easiest way to fix this problem is to remove the reference plane. De-embedding is still valid for zero-length reference planes.

Subsection <X | Y> dimension is too large at highest frequency. This message is printed when the cell size of your circuit is larger than 1/3 wavelengths. Results from such an analysis would be incorrect. Check the units used in xgeom and your frequency units in the ctl.an file. If these are correct, you will need to use a smaller cell size or a lower frequency.

Top ref. plane must be more than 3 cells long. The reference plane associated with the top side of the box is too short for accurate de-embedding. Usually the easiest way to fix this problem is to remove the reference plane. De-embedding is still valid for zero-length reference planes.

De-embedding Error Codes There are certain situations, discussed in detail in Chapter 7, “De-embedding Guidelines,” for which em is unable to obtain accurate de-embedded results. Em will usually, but not always, detect these situations and replace any suspect results 310

Appendix IV Warning and Error Messages with an error message. The format of the error message is “undefined: ”, where is a code which indicates the reason that em is unable to determine the de-embedded results. Table 9 describes the various error codes which may be displayed by em.

Code

De-embedded S-Parameters

Description

nd

N/A

Port was not de-embedded. No data is available.

mp

Valid

Multiple ports on same box wall.

sl

Caution

Length of first de-embedding standard is too short.

nl

Valid

Length of first standard is multiple of half wavelength.

mv

Valid

Multiple values of Eeff or Z0 for a single port number.

bd

Caution

Bad Eeff or Z0 data due to unknown reason.

The second column of Table 10, labeled “De-embedded S-Parameters”, gives the status of the de-embedded S-parameters corresponding to each error code. Error code “nd” indicates that the port was not de-embedded, therefore the status is not applicable. Error codes “mp”, “nl” and “mv” have a status of “Valid”. This indicates that while em was not able to determine Eeff or Z0, the de-embedded Sparameter results are completely valid. Error codes “sl” and “bd” have a status of “Caution”. This indicates that you should be cautious about using the deembedded S-parameter results as they may be corrupt. The “nd” error code indicates that the port cannot be de-embedded. Via ports are the only port type available in em that cannot be de-embedded. Thus, you will get this error code only when de-embedding circuits which contain via ports.

311

EM

Table 10 Codes displayed for indeterminate de-embedded results

Em User’s Manual The “mp” error code indicates that em is unable to determine Eeff and Z0 because the circuit has multiple ports on the same side of the box. The reason for this is that more than one value is required to describe the multiple modes associated with coupled transmission lines. The “sl” code indicates that the length of the first de-embedding standard is too short. We recommend that the length be at least one substrate thickness. See the section “Reference Plane Length Minimums,” page 82 for details. The “nl” code indicates that the length of the first de-embedding standard is a multiple of a half wavelength. In this case, em is unable to determine Eeff and Z0, but the de-embedded S-parameter results are completely valid. See the section “Reference Plane Lengths at Multiples of a Half-Wavelength,” page 84 for details. The “mv” code indicates that a single port number is used for multiple ports in the circuit, and that the Eeff and Z0 values vary for the different ports. Finally, the “bd” error code indicates that em is unable to determine Eeff and/or Z0 for an unknown reason. Low precision and box resonances in the calibration standards are sources of error that occasionally lead to the “bd” code.

312

Appendix V Sonnet Bibliography

EM

Appendix V

Sonnet Bibliography

Sonnet Theory [1]

Aki Kogure, "Automatic SPICE Models and S-Parameters Analysis," Design Wave Magazine, No. 20, March 1999, pp. 145 - 151. (Japanese Article)

[2]

Shigeki Nakamura, "Top Interview: Electromagnetic Analysis is not Difficult Big Rush to Install PC Version," Electronic Products Digest, Vol. 16, No. 1, January 1999, page 48. (Japanese Article)

[3]

James C. Rautio, “Comments on ‘Revisiting Characteristic Impedance and Its Definition of Microstrip Line with a Self-Calibrated 3-D MoM Scheme,’ ” IEEE Transactions on Microwave Theory and Techniques, Vol. 47, No. 1, January 1999, pp. 115 - 117.

[4]

Aki Kogure, “Why Electromagnetic Analysis is Necessary,” Design Wave Magazine, Vol. 1, No. 19, January 1999, pp. 27 - 38. (Japanese Article)

313

Em User’s Manual

314

[5]

James C. Rautio and George Matthaei, “Tracking Error Sources in HTS Filter Simulations,” Microwaves and RF, Vol. 37, No. 13, December 1998, pp. 119 130.

[6]

J.C. Rautio, “-Electromagnetic Analysis for Microwave Applications,” Computational Electromagnetics and Its Applications, Vol. 5, Boston: Kluwer Academic Publishers, 1997, pp. 80-96.

[7]

Yasumasa Noguchi, Shin-ichi Nakao, Hideaki Fujimoto and Nobuo Okamoto, “Characteristics of Shielded Coplanar Waveguides on Multilayer Substrates,” Electronic Information and Communications Univerisity Meeting, Electronics Society Conference, June 29, 1998. (Japanese Article)

[8]

Aki Kogura, “Sonnet, KCC Electromagnetic Analysis Software for Antenna Analysis,” Electronics Update, Vol. 13, No. 146, 1998, pp. 58-59. (Japanese article.)

[9]

Erik H. Lenzing and James C. Rautio, “A Model for Discretization Error in Electromagnetic Analysis of Capacitors,” IEEE Transactions on Microwave Theory and Techniques, Vol. 46, No. 2, February 1998, pp. 162-166.

[10]

J. C. Rautio, “Electromagnetic Analysis for Microwave Applications,” Computational Electromagnetics and Its Applications, Kluwar Academic Publishers, pp. 80-96.

[11]

J. C. Rautio, “Retracing Key Moments In the Life of Maxwell,” Microwaves & RF,” Vol. 36, No. 11, November 1997, pp. 35-51.

[12]

Keisuke Ogawa, Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Influence of Microstrip Conductor Offset in Microstrip Transmission Line,” Faculty of Science and Technology, Science University of Tokyo, Japan Institute for Interconnecting and Packaging Electronic Circuits, 11th JIPC Annual Meeting, March 1997, pp. 83-84. (Article in Japanese.)

[13]

Yusuke Hamada, Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Input Impedance of Equipment Housing with an Aperture for EMI Estimation Inside the Housing,” Faculty of Science and Technology, Science University of Tokyo, Technical Report of the Institute of Electronics, Information, and Communication Engineers of Japan, EMCJ 97-29, July 1997, pp. 45-50. (Article in Japanese.)

Appendix V Sonnet Bibliography Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Susceptibility inside Equipment Housing with an Aperture,” Faculty of Science and Technology, Science University of Tokyo, The Journal of Japan Institute for Interconnecting and Packaging Electronic Circuits, Vol. 12, No. 5, August 1997, pp. 369-373. (Article in Japanese.)

[15]

Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Analysis of Electromagnetic Field Coupled through an Aperture of Equipment Housing,” Faculty of Science and Technology, Science University of Tokyo, The Transaction of the Institute of Electronics, Information, and Communication Engineers of Japan, Vol. J80-B-11, No. 9, September 1997, pp. 809-811. (Article in Japanese.)

[16]

Keisuke Ogawa, Yasuhiro Kido, Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Radiated Emission from PCB with Microstrip Conductor Offset,” Faculty of Science and Technology, Science University of Tokyo, Technical Report of the Institute of Electronics, Information, and Communication Engineers of Japan, Vol. EMCJ 97-77, November 1997, pp. 2329. (Article in Japanese.)

[17]

Yusuke Hamada, Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Influence of a PCB Inside Equipment Housing with an Aperture on Resonant Modes,” Faculty of Science and Technology, Science University of Tokyo, Technical Report of the Institute of Electronics, Information, and Communication Engineers of Japan, EMCJ 97-78, November 1997, pp. 31-37. (Article in Japanese.)

[18]

Hiroaki Kogure, Yusuke Hamada, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Resonant Modes inside Equipment Housing and Susceptibility of Printed Circuit,” Faculty of Science and Technology, Science University of Tokyo, Japan Institute for Interconnecting and Packaging Electronic Circuits, Papers of Electromagnetic Behavior Society, Vol. 6, No. 3, November 1997, pp. 1-5. (Article in Japanese.)

[19]

Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Analysis of Electromagnetic Field inside Equipment Housing with an Aperture,” Faculty of Science and Technology, Science University of Tokyo, The Institute of Electronics, Information, and Communication Engineers of Japan, Transaction on Communications, Vol. E80-B, No. 11, November 1997, pp. 1620-1624. (Article in Japanese.)

315

EM

[14]

Em User’s Manual

316

[20]

J. C. Rautio, “Electromagnetic Analysis for Microwave Applications,” NASA CEM (Computational Electromagnetics) Workshop, Newport News, VA, May 1996.

[21]

J. C. Rautio, “Seven Years Later,” Applied Microwave and Wireless, November/ December 1996, pp. 99-100.

[22]

J. C. Rautio, “Questionable Reviews,” The Institute (IEEE newspaper), Jan. 1996, pg. 11.

[23]

J. C. Rautio, “An Investigation of an Error Cancellation Mechanism with Respect to Subsectional Electromagnetic Analysis Validation,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 6, No. 6, November 1996, pp. 430-435.

[24]

J. C. Rautio, “The Microwave Point of View on Software Validation,” IEEE Antennas and Propagation Magazine, Vol. 38, No. 2, April 1996, pp. 68-71.

[25]

J. C. Rautio and Hiroaki Kogure, “EMI Applications Of The Electromagnetic Analysis By The Method Of Moments-Electromagnetic Analysis Applied To Analog And Digital PCB Design,” JPCA Show 96 Text: Today and Tomorrow of EMI Design, pp. 11-19.

[26]

Hiroaki Kogure, “Susceptibility inside Equipment Housing with a Slot,” Faculty of Science and Technology, Science University of Tokyo, 7th Workshop, WG-1, pp. 4-5, Jan 1996. (Article in Japanese.)

[27]

Hiroaki Kogure, Keisuke Ogawa, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Analysis of Radiation from Printed Circuit Board with a Slot,” Papers of Electromagnetic Behavior Society, Vol. 5, No. 3, pp. 7-12, February 1996. (Article in Japanese.)

[28]

Hiroaki Kogure, Kohji Koshiji and Eimei Shu, “Electromagnetic Simulation by MoM and TLM Method,” 10th JIPC Annual Meeting, Proceedings, pp. 169-170, March 1996, Tokyo.

[29]

Hiroaki Kogure, Keisuke Ogawa, Kohji Koshiji and Eimei Shu, “Susceptibility inside Equipment Housing with a Slot,” 10th JIPC Annual Meeting, Proceedings, pp. 185-186, March 1996, Tokyo.

[30]

Hiroaki Kogure, Keisuke Ogawa, Kohji Koshiji and Eimei Shu, “Susceptibility inside Equipment Housing with a Slot,” Proceedings of The 1996 IEICE General Conference, pg. 318, March 1996.

Appendix V Sonnet Bibliography Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Susceptibility of the Multilayer Printed Circuit inside the Equipment Housing,” Communication Engineers of Japan Technical Report of the IEICE, EMCJ96-19 (1996-07), pp. 13-18, July 1996.

[32]

Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Susceptibility of Multilayer Printed Circuit inside Equipment Housing,” Asia-Pacific Conference on Environmental Electromagnetics (CEEM 96) Xi'an China, November 1996, pp. 263-266.

[33]

Hiroaki Kogure, Keisuke Ogawa, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Electromagnetic Distribution of Multilayered Printed Circuit - Analysis of coupling to other layers,” Japan Institute for Interconnecting and Packaging Electronic Circuits, Papers of Electromagnetic Behavior Society, Vol. 5, No. 2, pp. 1-10, Oct. 1995. (Article in Japanese.)

[34]

Hiroaki Kogure, Keisuke Ogawa, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Analysis of Current Distribution of Multilayered Printed Circuit,” Communications Engineers of Japan Technical Report of IEICE, EMCJ95-19 (1995-07), pp. 1-8, July 1995. (Article in Japanese.)

[35]

J. C. Rautio, “EM-Analysis Error Impacts Microwave Designs,” Microwaves and RF, September 1996, pp. 134-144.

[36]

James R. Willhite, “Turning Clean Theory into Reality,” Wireless Design and Development, March 1996, Vol. 4, No. 3, pp. 19-20.

[37]

J. C. Rautio, “Response #2. Comments on Zeland's Standard Stripline Benchmark Results - MIC Simulation Column,” International Journal of Microwave and Millimeter- Wave Computer-Aided Engineering, Vol. 5, No. 6, November 1995, pp. 415-417.

[38]

J. C. Rautio, “EMI Analysis from a Wireless Telecommunication and RF Perspective,” Proceedings of the 1995 Nepcon West Conference, Anaheim, CA, USA, pp. 749-755.

[39]

J. C. Rautio and Hiroaki Kogure, “An Overview of the Sonnet Electromagnetic Analysis,” Proceedings of the 1994 IEICE Fall Conference, Tokyo, pp. 325-326.

[40]

J. C. Rautio, “An Ultra-High Precision Benchmark For Validation Of Planar Electromagnetic Analyses,” IEEE Tran. Microwave Theory Tech., Vol. 42, No. 11, Nov. 1994, pp. 2046-2050.

317

EM

[31]

Em User’s Manual

318

[41]

J. C. Rautio, “A Precise Benchmark for Numerical Validation,” IEEE International Microwave Symposium, Workshop WSMK Digest, Atlanta, June 1993.

[42]

“Comparison of Strategies for Analysis of Diagonal Structures,” Sonnet Application Note 51-02.

[43]

J. C. Rautio, “MIC Simulation Column - A Standard Stripline Benchmark,” International Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, Vol. 4, No. 2, April 1994, pp. 209-212.

[44]

J. C. Rautio, “Response #3. Standard Stripline Benchmark - MIC Simulation Column,” International Journal of Microwave and Millimeter-Wave ComputerAided Engineering, Vol. 5, No. 5, September 1995, pp. 365-367.

[45]

J. C. Rautio, “Some Comments on Approximating Radiation,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 4, No. 2, 1994, pp. 454-457.

[46]

J. C. Rautio, “Synthesis of Lumped Models from N-Port Scattering Parameter Data,” IEEE Tran. Microwave Theory Tech., Vol. 42, No. 3, March 1994, pp. 535-537.

[47]

J. C. Rautio, “Educational Use of a Microwave Electromagnetic Analysis of 3-D Planar Structures,” Computer Applications in Engineering Education, Vol. 1, No. 3, 1993, pp. 243-254.

[48]

J. C. Rautio, “Characterization of Electromagnetic Software,” 42nd ARFTG Conference Digest, San Jose, CA, Dec. 1993, pp. 81-86.

[49]

J. C. Rautio, “Some Comments on Electromagnetic De-Embedding and Microstrip Characteristic Impedance” International Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, Vol. 3, No. 2, April 1993, pp. 151-153.

[50]

J. C. Rautio, “Some Comments on Electromagnetic Dimensionality,” IEEE MTT-S Newsletter, Winter 1992, pg. 23.

[51]

J. C. Rautio, “Sonnet Software Reveals Tangential Fields,” EEsof Wavelengths, Vol. 9, No. 1, March 1993, pg. 12.

[52]

J. C. Rautio, “Sonnet Introduces Antenna Pattern Visualization in New Release,” EEsof Wavelengths, Vol. 9, No. 2, June 1993, pg. 21.

Appendix V Sonnet Bibliography J. C. Rautio, “EEsof Joins Forces With Sonnet Software,” EEsof Wavelengths, Vol. 8, No. 3, Sept. 1992, pg. 14.

[54]

J. C. Rautio, “Electromagnetic Design of Passive Structures - Emerging Technology in Microwave CAD,” IEEE MTT-S Newsletter, Fall 1990, pp. 2122.

[55]

J. C. Rautio, “Electromagnetic Microwave Design,” RF/Microwave Applications Conference, Santa Clara, CA, March 1992, pp. 105-109.

[56]

J. C. Rautio, “Experimental Validation of Microwave Software,” IEEE International Microwave Symposium, Panel Session PSB Digest, Albuquerque, June 1992.

[57]

J. C. Rautio, “Current Developments in 3-D Planar Microwave Electromagnetics,” Microwave Hybrid Circuits Conference, Oct. 1991, Arizona.

[58]

J. C. Rautio, “Current Developments in 3-D Planar Microwave Electromagnetics,” Microwave Hybrid Circuits Conference, Oct. 1992, Arizona.

[59]

J. C. Rautio, “Current Developments in 3-D Planar Microwave Electromagnetics,” Microwave Hybrid Circuits Conference, Oct. 1993, Arizona.

[60]

J. C. Rautio, “Current Developments in 3-D Planar Microwave Electromagnetics,” Microwave Hybrid Circuits Conference, Oct. 1994, Arizona.

[61]

J. C. Rautio, “Experimental Validation of Electromagnetic Software,” International Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, Vol. 1, No. 4, Oct. 1991, pp. 379-385.

[62]

J. C. Rautio, “Electromagnetic Microwave Analysis,” IEEE International Microwave Symposium, Workshop WSA Digest, Albuquerque, June 1992.

[63]

J. C. Rautio, “EM Visualization Assists Designers,” Microwaves and RF, Nov. 1991, pp. 102-106.

[64]

J. C. Rautio, “Reviewing Available EM Simulation Tools,” Microwaves & RF, June 1991, pp. 16A-20A.

[65]

“Generating Spice Files Using the em Electromagnetic Analysis,” Sonnet Application Note 104a, Dec. 1998.

[66]

J. C. Rautio, “A New Definition of Characteristic Impedance,” MTT International Symposium Digest, June 1991, Boston, pp. 761-764.

319

EM

[53]

Em User’s Manual

320

[67]

J. C. Rautio, “A De-Embedding Algorithm for Electromagnetics,” International Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, Vol.1, No. 3, July 1991, pp. 282-287.

[68]

J. C. Rautio, “Triangle Cells in an Electromagnetic Analysis of Arbitrary Microstrip Circuits,” MTT International Microwave Symposium Digest, Dallas, June 1990, pp. 701-704.

[69]

J. C. Rautio, “Experimental Validation of Microwave Software,” 35th ARFTG Conference Digest, Dallas, May 1990, pp. 58-68. (Voted best paper at the conference.)

[70]

J. C. Rautio, “Preliminary Results of a Time-Harmonic Electromagnetic Analysis of Shielded Microstrip Circuits,” 27th ARFTG Conference Digest, Dallas, Dec. 1986. (Voted best paper at the conference.)

[71]

J. C. Rautio, “An Experimental Investigation of the Microstrip Step Discontinuity,” IEEE Tran. Microwave Theory Tech., Vol. MTT-37, Nov. 1989, pp. 1816-1818.

[72]

J. C. Rautio, “A Possible Source of Error in On-Wafer Calibration,” 34th ARFTG Conference, Ft. Lauderdale, FL, Dec. 1989, pp. 118-126.

[73]

J. C. Rautio, “Microstrip Program Improves Accuracy of Circuit Models,” Microwaves & RF, Vol. 27, No. 12, pp. 89-96, Nov. 1988.

[74]

J. C. Rautio, “Reflection Coefficient Analysis of the Effect of Ground on Antenna Patterns,” IEEE Antennas and Propagation Society Newsletter, Feb. 87, pp. 5-11.

[75]

J. C. Rautio and R. F. Harrington, “An Electromagnetic Time-Harmonic Analysis of Shielded Microstrip Circuits,” IEEE Trans. Microwave Theory Tech., Vol. MTT-35, pp. 726-730, Aug. 1987.

[76]

J. C. Rautio and R. F. Harrington, “An Efficient Electromagnetic Analysis of Arbitrary Microstrip Circuits,” MTT International Microwave Symposium Digest, Las Vegas, June 1987, pp. 295-298.

[77]

J. C. Rautio and R. F. Harrington, “Results and Experimental Verification of an Electromagnetic Analysis of Microstrip Circuits,” Trans. of The Society for Computer Simulation, Vol. 4, No. 2, pp. 125-156, Apr. 1987.

Appendix V Sonnet Bibliography J. C. Rautio, “A Time-Harmonic Electromagnetic Analysis of Shielded Microstrip Circuits,” Ph. D. Dissertation, Syracuse University, Syracuse, NY, 1986.

[79]

J. C. Rautio, “Preliminary Results of a Time-Harmonic Electromagnetic Analysis of Shielded Microstrip Circuits,” ARFTG Conference Digest, Baltimore, pp. 121-134, June 1986. (Voted best paper at the conference.)

[80]

J. C. Rautio, “Techniques for Correcting Scattering Parameter Data of an Imperfectly Terminated Multiport When Measured with a Two-Port Network Analyzer,” IEEE Trans. Microwave Theory Tech., Vol. MTT-31, May 1983, pp. 407-412.

[81]

R. F. Harrington, Time-Harmonic Electromagnetic Fields, New York: McGraw-Hill, 1961, section 8-11, pg. 8.

Sonnet Applications [82]

John F. Sevic, "A Sub 1 Ω Load-Pull Quarter-Wave Prematching Network Based on a Two-Tier TRL Calibration," Microwave Journal, Vol. 42, No. 3, March 1999, pp. 122-132.

[83]

John W. Bandler, Radoslaw M. Biernacki, Shao Hua Chen, "Parameterization of Arbitrary Geometrical Structures for Automated Electromagnetic Optimization," International Journal of RF and Microwave Computer-Aided Engineering, Vol. 9, No. 2, March 1999, pp. 73 - 85.

[84]

Jack Browne, "Technology Fuels Firm’s Entry Into Filter Market," Microwaves & RF, Vol. 38, No. 1, January 1999, pp. 113-118.

[85]

Mansoor K. Siddiqui, Arvind K. Sharma, Lenonardo G. Callejo, and Richard Lai, "A High-Power and High-Efficiency Monolithic Power Amplifier at 28 GHz for LMDS Applications," IEEE Transactions on Microwave Theory and Techniques, Vol. 46, No. 12, December 1998, pp. 2226 - 2232.

[86]

Søren F. Peik, Raafat R. Mansour, and Y. Leonard Chow, "Multidimensional Cauchy Method and Adaptive Sampling for an Accurate Microwave Circuit Modeling," IEEE Transactions on Microwave Theory and Techniques, Vol. 46, No. 12, December 1998, pp. 2364 - 2371.

321

EM

[78]

Em User’s Manual

322

[87]

Mohamed H. Bakr, John W. Bandler, Radoslaw M. Biernacki, Shao Hua (Steve) Chen, and Kaj Madsen, "A Trust Region Aggressive Space Mapping Algorithm for EM Optimization," IEEE Transactions on Microwave Theory and Techniques, Vol. 46, No. 12, December 1998, pp. 2412 - 2425.

[88]

Y. C. Chen, L. Ingram, R. Lai, M. Barsky, R. Grunbacher, T. Block, H. C. Yen, and D. C. Streit, “A 95-GHz InP HEMT MMIC Amplifier with 427-mW Power Output,” IEEE Microwave and Guided Wave Letters, Vol. 8, No. 11, November 1998, pp. 399 - 401.

[89]

R. Lai, M. Barsky, T. Huang, M. Sholley, H. Wang, Y. L. Kok, D. C. Streit, T. Block, P. H. Liu, T. Gaier, and L. Samoska, “An InP HEMT MMIC LNA with 7.2-dB Gain at 190 GHz,” IEEE Microwave and Guided Wave Letters, Vol. 8, No. 11, November 1998, pp. 393-395.

[90]

Huei Wang, Richard Lai, Yon-Lin Kok, Tian-Wei Huang, Michael V Aust, Yaochung C. Chen, Peter H. Siegel, Todd Gaier, Robert J. Dengler, and Barry R. Allen, “A 155-GHz Monolithic Low-Noise Amplifier,” IEEE Transactions on Microwave Theory and Techniques, Vol. 46, No. 11, November 1998, pp. 16601666.

[91]

Charles Trantanella, Mitch Shifrin, and Brian Bedard, “Low Cost, Plastic Encapsulated Mixers for C/X-Band Applications,” IEEE GaAs IC Symposium Technical Digest 1998, November 1998, pp. 131-134.

[92]

Jakub J. Kucera and Urs Lott, “A 1.8 dB Noise Figure Low DC Power MMIC LNA for C-Band,” IEEE GaAs IC Symposium Technical Digest 1998, November 1998, pp. 225-228.

[93]

David E. Meharry, “Multi-Octave Transformer Coupled Differential Amplifier for High Dynamic Range,” IEEE GaAs IC Symposium Technical Digest 1998, November 1998, pp. 221-224.

[94]

J.S. Hong, M.J. Lancaster, R.B. Greed, D. Voyce, D. Jedamzik, J.A. Holland,H.J. Chaloupka, Jean-Claude Mage "Thin Film HTS Passive Microwave Components for Advanced Communication Systems", Accepted for IEEE Trans. on Applied Superconductivity.

[95]

Nanju Na, Kwang Lim Choi and Madhavan Swaminatham, "Characterization of embedded resistors for high frequency wireless applications," 1998 IEEE Radio and Wireless Conference Proceedings, August 1998, pp. 117-120.

Appendix V Sonnet Bibliography Tony Yeung, Jack Lau, H.C. Ho, and M.C. Poon, "Design Condsiderations for Extremely High-Q Integrated Inductors and Their Application in CMOS RF Power Amplifiers," 1998 IEEE Radio and Wireless Conference Proceedings, August 1998, pp. 265-268.

[97]

Brian K. Kormanyos, Ronald W. Kruse, and Debra R. Follensbee, "A High Efficiency MMIC Power Amplifier for Phased Array Antenna Applications," 1998 IEEE Radio and Wireless Conference Proceedings, August 1998, pp. 333334.

[98]

Robert W. Jackson and Zhaoyang Wang, “Circuit Model for Coupling Between MMIC’s in Multichip Modules Including Resonance Effects,” IEEE Transactions on Microwave Theory and Techniques, Vol. 46, No. 7, July 1998, pp. 959-965.

[99]

John H. Mayer, "Simulation advances accelerate RF designs," Military & Aerospace Electronics, June 1998, pp. 19 -22.

[100]

Alan L. L. Pun, Tony Yeung, Jack Lau, François J. R. Clément, and David K. Su, "Substrate Noise Coupling Through Planar Spiral Inductor," IEEE Journal of Solid-State Circuts, Vol. 33, No. 6, June 1998, pp. 877-884.

[101]

Kyu Yong Kim, Yong Chung and Yong Su Choe, “Low Side Lobe Series-fed Planar Array at 20 GHz,” IEEE 1998 AP-S International Symposium, Altlanta, Georgia, June 21 - 26, 1998, pp. 1196 - 1199.

[102]

A. Torabian and Y. L. Chow, “Rapid Analysis of High Q and High Order Patch Filters,” IEEE 1998 AP-S International Symposium, Altlanta, Georgia, June 21 26, 1998, pp. 1906 - 1909.

[103]

Z. Wang and R. W. Jackson, “A CAD Algorithm for Coupling Between Dielectric Covered MMICs in Multi-Chip Assemblies,” 1998 IEEE MTT-S International Symposium Digest, Vol. 1, June 1998, pp. 33-36.

[104]

D. Prieto, J.C. Cayrou, J.L. Cazaux, T. Parra, and J. Graffeuil, “CPS Structure Potentialities for MMICs: A CPS/CPW Transition and a Bias Network,” 1998 IEEE MTT-S International Symposium Digest, Vol. 1, June 1998, pp. 111-114.

[105]

J.S. Hong, M.J. Lancaster, D. Jedamzik and R.B. Greed, “8-Pole Superconducting Quasi-Elliptic Function Filter for Mobile Communications Application,” 1998 IEEE MTT-S International Symposium Digest, Vol. 1, June 1998, pp. 367-370.

323

EM

[96]

Em User’s Manual

324

[106]

T. Gokdemir, S. Nam, A. E. Ashtiani, I. D. Robertson and Ulun Karacaoglu, “Millimeter-Wave Monolithic Balanced BPSK Modulator Using a Miniaturized Backward-Wave Coupler” 1998 IEEE MTT-S International Symposium Digest, Vol. 2, June 1998, pp. 877-880.

[107]

G Subramanyam, F. Van Keuls and F. A. Miranda, “A Novel K-Band Tunable Microstrip Bandpass Filter Using a Thin Film HTS/Ferroelectric/Dielectric Multilayer Configuration” 1998 IEEE MTT-S International Symposium Digest, Vol. 2, June 1998, pp. 1011-1014.

[108]

F. Rouchaus, V. Madrangeas, M. Aubourg, P. Guillon, B. Theron, M. Maigan, “New Classes of Microstrip Resonators for HTS Microwave Filters Applications” 1998 IEEE MTT-S International Symposium Digest, Vol. 2, June 1998, pp. 1023-1026.

[109]

A. Fathy, V. Pendrick, G. Ayers, B. Geller, Y. Narayan, B. Thaler, H. D. Chen, M. J. Liberatore, J. Prokop, K. L. Choi, M. Swaminathan, “Design of Embedded Passive Components in Low-Temperature Cofired Ceramic on Metal (LTCC-M) Technology,” 1998 IEEE MTT-S International Symposium Digest, Vol. 3, June 1998, pp. 1281-1284.

[110]

Brad Heimer and Thomas Budka, “Methodology for Creating Embedded Transmission Line 90° Bend and Shunt Capacitor Models,” 1998 IEEE MTT-S International Symposium Digest, Vol. 3, June 1998, pp. 1297-1300.

[111]

Yon-Lin Kok, Pin-Pin Huang, Huei Wang, Barry R. Allen, Richard Lai, Mike Sholley, Todd Gaier and I. Mehdi, “120 and 160 GHz Monlithic InP-based HEMT Diode Sub-harmonic Mixer,” 1998 IEEE MTT-S International Symposium Digest, Vol. 3, June 1998, pp. 1723-1726.

[112]

M. H. Bakr, J. W. Bandler, R. M. Biernacki, S. H. Chen and K. Madsen, “A Trust Region Aggressive Space Mapping Algorithm for EM Optimization,” 1998 IEEE MTT-S International Symposium Digest, Vol. 3, June 1998, pp. 1759-1762.

[113]

K.-F. Lau, L. Liu, and S. Dow, “Recent MMW Technology Development its Military and Commercial Applications,” 1998 IEEE Radio Frequency Integrated Circuits Symposium Digest of Papers, June 1998, pp. 87-90.

[114]

Zhaofeng Zhang, Alan Pun, Jack Lau, “Interference Issues in Silicon RFIC Design,” 1998 IEEE Radio Frequency Integrated Circuits Symposium Digest of Papers, June 1998, pp. 119-122

Appendix V Sonnet Bibliography D. Staiculescu, A. Pham, J. Laskar, S. Consolazio and S. Moghe, “Analysis and Performance of BGA Interconnects for RF Packaging,” 1998 IEEE Radio Frequency Integrated Circuits Symposium Digest of Papers, June 1998, pp. 131134.

[116]

F.A. Miranda, F.W. Van Keuls, R.R. Romanofsky, and G. Subramanyam, “Tunable Microwave Components for Ku and K band Satellite Communications,” (accepted by Integrated Ferroelectrics).

[117]

F.W. Van Keuls, R.R. Romanofsky, and F.A. Miranda, “Several Microstrip-Based Conductor/ Thin Film Ferroelectric Phase Shifter Designs Using (YBa2Cu3O7d,Au)/SrTiO3/LaAlO3 Tunable Ring Resonators,” (accepted by Integrated Ferroelectrics).

[118]

F.W. Van Keuls, R.R. Romanofsky, N.D. Varaljay, F.A. Miranda, C.L. Canedy, S. Aggarwal, T. Venkatesan, and R. Ramesh, “A Ku-Band Gold/BaxSr1-xTiO3/ LaAlO3 Thin Film Conductor/Ferroelectric Microstripline Phase Shifter for Room Temperature Phased Array Applications,” (submitted to Microwave and Optical Technology Letters)

[119]

Zhaoyang Wang, and Robert W. Jackson, "A CAD Algorithm for Coupling Between Dielectric Covered MMICs in Multi-Chip Assemblies", to appear in IEEE Microwave Theory and Techniques Symposium Digest, June 1998.

[120]

G. Subramanyam, F.W. Van Keuls and F.A. Miranda, “A Novel Tunable Microstrip Bandpass Filter Using a Thin Film HTS/Ferroelectric/Dielectric Multilayer Configuration,” (accepted by IEEE Transactions on Microwave Theory and Techniques).

[121]

Guru Subramanyam, Fred Van Keuls, and Félix A. Miranda, “A K-Band Tunable Microstrip Bandpass Filter Using a Thin-Film Conductor/Ferroelectric/Dielectric Multilayer Configuration,” IEEE Microwave and Guided Wave Letters, Vol. 8, No. 2, February 1998, pp. 78 - 80.

[122]

Lei Zhu and Ke Wu, “Revisiting Characteristic Impedance and Its Definition of Microstrip Line with a Self-Calibrated 3-D MoM Scheme,” IEEE Microwave and Guided Wave Letters, Vol. 8, No. 2, February 1998, pp. 87 - 89.

[123]

Jia-Sheng Hong and Michael J. Lancaster, “Cross-Coupled Microstrip HairpinResonator Filters,” IEEE Transactions on Microwave Theory and Techniques, Vol. 46, No. 1, January 1998, pp. 118-122.

325

EM

[115]

Em User’s Manual

326

[124]

F.W. Van Keuls, F.A. Miranda, R.R. Romanofsky , C. H. Mueller, R. E. Treece and T.V. Rivkin, “(YBa2Cu3O7-d, Au)/SrTiO3/LaAlO3 thin film conductor/ ferroelectric phase shifters and their potential for phased array applications,” Appl. Phys. Lett. 71, 3075 (1997).

[125]

Brad Ryan Heimer, Lu Fan, and Kai Chang, “Uniplanar Hybrid Couplers Using Asymmetrical Coplanar Striplines,” IEEE Transactions on Microwave Theory and Techniques, Vol. 45, No. 12, December 1997, pp. 2234-2240.

[126]

Jia-Sheng Hong, Michael J. Lancaster, “Theory and Experiment of Novel Microstrip Slow-Wave Open-Loop Resonator Filters,” IEEE Transactions on Microwave Theory and Techniques, Vol. 45, No. 12, December 1997, pp. 23582365.

[127]

Pin-Pin Huang, Tian-Wei Huang, Huei Wang, Eric W. Lin, Yonghui Shu, Gee. S. Dow, Richard Lai, Michael Biedenbender, and Jeffrey H. Elliot, “A 94-GHz 0.35-W Power Amplifier Module,” IEEE Transactions on Microwave Theory and Techniques, Vol. 45, No. 12, December 1997, pp. 2418-2423.

[128]

Daisy L. Ingram, D. Ian Stones, Jeffrey H. Elliot, Huei Wang, Richard Lai, and Michael Biedenbender, “A 6-W Ka-Band Power Module Using MMIC Power Amplifiers,” IEEE Transactions on Microwave Theory and Techniques, Vol. 45, No. 12, December 1997, pp. 2424-2430.

[129]

Mark S. Mirotznik and Dennis Prather, “How to choose EM software,” IEEE Spectrum, Vol. 34, No. 12, December 1997, pp. 53-58.

[130]

“CAD Roundtable: benchmarking the future of design,” Microwave Engineering Europe, November 1997, pp. 31-42.

[131]

Alan Conrad and Jack Browne, “EM Tools Enhance Simulation Accuracy,” Microwaves & RF, Vol. 36, No. 11, November 1997, pp. 133-136.

[132]

Robert W. Jackson and Ryosuke Ito, “Modeling Millimeter-Wave IC Behavior for Flipped-Chip Mounting Schemes,” IEEE Transactions on Microwave Theory and Techniques, Vol. 45, No. 10, October 1997, pp. 1919-1925.

[133]

Robert W. Jackson and Sambarta Rakshit, “Microwave-Circuit Modeling of High Lead-Count Plastic Packages,” IEEE Transactions on Microwave Theory and Techniques, Vol. 45, No. 10, October 1997, pp. 1926-1933.

[134]

Robert W. Jackson and Ryosuke Ito, "Microwave Modeling of Flipped Chip Packaging Schemes," IEEE Trans. Microwave Theory and Techniques, October 1997.

Appendix V Sonnet Bibliography Ryosuke Ito and Robert W. Jackson, "Circuit Modeling of Isolation in Flip-Chip Microwave Integrated Circuits", 1997 Conference on the Electrical Performance of Electronic Packaging Proceedings, San Jose., pp.217-220, October 1997.

[136]

I. Toyoda, T. Tokumitsu, and M. Aikawa, “A Basic Concept of Microwave Design Automation Based on Three-dimensional Masterslice MMIC technology,” in 27th European Microwave Conf. Proc., Sept. 1997. (To be published)

[137]

K. Nishikawa, I. Toyoda, and T. Tokumitsu, “Miniaturized three-dimensional MMIC K-band upconverter,” IEEE Microwave and Guided Wave Letter, 1997.

[138]

George L. Matthaei, Neal O. Fenzi, Roger J. Forse, and Stephan M. Rolhing, “Hairpin-Comb Filters for HTS and Other Narrow-Band Applications,” IEEE Transactions on Microwave Theory and Techniques, Vol. 45, No. 8, August 1997, pp. 1226-1231.

[139]

T. Tokumitsu, K. Nishikawa, K. Kamogawa, I. Toyoda, and K. Nishimura, “Three-dimensional MMIC technology and application to millimeter-wave MMIC's,” 1997 Topical Symposium on Millimeter Waves Digest, July 1997.

[140]

“Sorting Through the Myriad of Software Options,” Wireless Systems Design, Master Reference, Vol. 2, No. 8, July 1997, pp. 78-81.

[141]

Marinus (Ron) Korber, Jr., “New Microstrip Bandpass Filter Topologies,” Microwave Journal, Vol. 40, No. 7, July 1997, pp. 138-144.

[142]

K. Kamogawa, K. Nishikawa, C. Yamaguchi, M. Hirano, I. Toyoda, and T. Tokumitsu, “A Very Wide-tuning Range 5-GHz-band Si Bipolar VCO Using Three-dimensional MMIC technology,” in IEEE International Microwave Symposium Digest, June 1997, pp. 1221-1224.

[143]

I. Toyoda, K. Nishikawa, T. Tokumitsu, C. Yamaguchi, M. Hirano, and M. Aikawa, “Three-dimensional Masterslice MMIC on Si Substrate,” in 1997 IEEE Radio Frequency Integrated Circuits Symposium Digest, June 1997, pp. 113-116.

[144]

Gregory L. Hey-Shipton, Neal O. Fenzi, and Kurt F. Raihn, “HTS Diplexer & Low Noise Amplifier RF Module,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 1, pp. 295-298.

[145]

Shen Ye and Raafat R. Mansour, “A Novel Split-Resonator High Power HTS Planar Filter,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 1, pp. 299-301.

327

EM

[135]

Em User’s Manual

328

[146]

Michael J. Lee and Joseph A. Faulkner Jr., “Power Combining Port Impedance Model,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 2, pp. 543-546.

[147]

J. S. Hong and M. J. Lancaster, “Microstrip Slow-Wave Open-Loop Resonator Filters,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 2, pp. 713-716.

[148]

T. Gokdemir, U. Karacaoglu, D. Budimir, S. B. Economides, A. Khalid, A. A. Rezazadeh and I. D. Robertson, “Multilayer Passive Components for Uniplanar Si/SiGe MMICs,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 2, pp. 761-764.

[149]

M. N. Tutt, H. Q. Tserng and A. Ketterson, “A Low Loss, 5.5 GHz - 20 GHz Monolithic Balun,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 2, pp. 933-936.

[150]

Kunihiko Sasaki, Junshi Utsu, Kazuoki Matsugatani, Kouichi Hoshino, Takashi Taguchi, and Yoshiki Ueno, “InP MMICs for V-Band FMCW Radar,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 2, pp. 937-940.

[151]

Michael Case, Mehran Matloubian, Hsiang-Chih Sun, Debabani Choudhury, and Catherine Ngo, “High-Performance W-Band GaAs PIN Diode Single-Pole Triple-Throw Switch CPW MMIC,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 2, pp. 1047-1050.

[152]

Y. Hwang, J. Lester, G. Schreyer, G. Zell, S. Schrier, D. Yamauchi, G. Onak, B. Kasody, R. Kono, Y. C. Chen, and R. Lai, “60 GHz High-Efficiency HEMT MMIC Chip Set Development for High-Power Solid State Power Amplifier,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 3, pp. 1179-1182.

[153]

D. L. Ingram, D. I. Stones, T.W. Huang, M. Nishimoto, H. Wang, M. Siddiqui, D. Tamura, J. Elliot, R. Lai, M Biedenbender, H. C. Yen, and B. Allen, “A 6 Watt Ka-Band MMIC Power Module Using MMIC Power Amplifiers,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 3, pp. 1183-1186.

[154]

H. Wang, R. Lai, Y. C. Chen, Y. L. Kok, T. W. Huang, T. Block, D. Streit, P. H. Liu, P. Siegel, and B. Allen, “A 155-GHz Monolithic InP-Based HEMT Amplifier,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 3, pp. 1275-1278.

Appendix V Sonnet Bibliography Michael K. Waldo, Irving Kaufman, and Samir El-Ghazaly, “Coplanar Waveguide Technique for Measurement of Dielectric Constant or Thickness of Dielectric Films,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 3, pp. 1339-1342.

[156]

Robert W. Jackson and Zhaoyang Wang, “Circuit Based Model for Coupling Between MMICs in Multi-Chip Assemblies,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 3, pp. 1377-1380.

[157]

J. J. Komiak, S. C. Wang, and T. J. Rogers, “High Efficiency 11 Watt Octave S/ C-Band PHEMT MMIC Power Amplifier,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 3, pp. 1421-1424.

[158]

A. R. Barnes et. al., “A 6-18 GHz Broadband High Power MMIC for EW Applications,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 3, pp. 1429-1432.

[159]

Yon-Lin Kok, Mark DuFault, Tian-Wei Huang, and Heui Wang, “A Calibration Procedure for W-band On-Wafer Testing,” IEEE MTT-S 1997 International Microwave Symposium Digest, Vol. 3, pp. 1663-1666.

[160]

Daniel G. Swanson, Jr., “Optimizing Combline Filter Designs Using 3D FieldSolvers,” IEEE MTT-S International Microwave Symposium Workshops, WMA: State-of-the-art Filter Design Using EM and Circuit Simulation Techniques, June 1997.

[161]

J. W. Bandler, “EM Optimization Using Space Mapping,” IEEE MTT-S International Microwave Symposium Workshops, WMA: State-of-the-art Filter Design Using EM and Circuit Simulation Techniques, June 1997.

[162]

George L. Matthaei, “Some CAD Techniques for Planar Microwave Filter Design and Some Observations Regarding Dispersion,” IEEE MTT-S International Microwave Symposium Workshops, WMA: State-of-the-art Filter Design Using EM and Circuit Simulation Techniques, June 1997.

[163]

W. R. Gaiewski, L.P. Dunleavy, and A. Castro, Jr., “Analysis and Measurement of Mode Polarizers in Square Waveguide,” IEEE Trans. on Microwave Theory and Techniques, Vol. 45, No. 6, June 1997, pp. 997-1000.

[164]

G. Avitabile, A. Cidronali, and C. Salvador, “Equivalent Circuit Model of GaAs MMIC-Coupled Planar Spiral Inductors,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 7, No. 4, July 1997, pp. 318-326.

329

EM

[155]

Em User’s Manual

330

[165]

Dr. James Willhite, “Three-Dimensional EM Software for PCs,” Microwave Journal, Vol. 40, No. 5, May 1997, pp. 354-357.

[166]

Nitin Jain and Peter Onno, “Methods of Using Commercial Electromagnetic Simulators for Microwave and Millimeter-Wave Circuit Design and Optimization,” IEEE Trans. on Microwave Theory and Techniques, Vol. 45, No. 5, May 1997, pp. 724-746.

[167]

John W. Bandler, Radoslaw M. Biernacki, Shao Hua Chen and Ya Fei Huang, “Design Optimization of Interdigital Filters Using Aggressive Space Mapping and Decomposition,” IEEE Trans. on Microwave Theory and Techniques, Vol. 45, No. 5, May 1997, pp. 761-769.

[168]

Shin Ye and Raafat R. Mansour, “An Innovative CAD Technique for Microstrip Filter Design,” IEEE Trans. on Microwave Theory and Techniques, Vol. 45, No. 5, May 1997, pp. 780-786.

[169]

Gregory L. Creech, Bradley J. Paul, Christopher D. Lesniak, Thomas J. Jenkins and Mark C. Calcatera, “Artificial Neural Networks for Fast and Accurate EMCAD of Microwave Circuits,” IEEE Trans. on Microwave Theory and Techniques, Vol. 45, No. 5, May 1997, pp. 794-802.

[170]

Jia-Sheng Hong and Michael J. Lancaster, “Investigation of Microstrip PseudoInterdigital Bandpass Filters Using a Full-Wave Electromagnetic Simulator,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 7, No. 3, May 1997, pp. 231-340.

[171]

Noyan Kinayman and M. I. Aksun, “Efficient Use of Closed-Form Green's Functions for the Analysis of Planar Geometries with Vertical Interconnections,” IEEE Trans. on Microwave Theory and Techniques, Vol. 45, No. 5, May 1997, pp. 593-603.

[172]

Noyan Kinayman and M. I. Aksun, “On the Fast Track - Supercomputing at SU's Info Mall Helps Launch Businesses,” Syracuse Herald American, Syracuse OnLine supplement, March 19, 1995, pg. 16.

[173]

Victor Perrote, “Wireless Applications Spur EM Applications,” Microwaves & RF, April 1997, pg. 17.

[174]

George Jankovic, “Wireless on the Web,” Applied Microwave & Wireless, Vol. 9, No. 2, March/April 1997.

Appendix V Sonnet Bibliography Andreas Vogt and Wilhelm Jutzi, “An HTS Narrow Bandwidth Coplanar Shunt Inductively Coupled Microwave Bandpass Filter on LaAlO3,” IEEE Trans. on Microwave Theory and Techniques, Vol. 45, No. 4, April 1997, pp. 492-497.

[176]

Keisuke Ogawa, Hiroaki Kogure, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Influences of Finite Ground Conductor Width and Microstrip Conductor Offset on Characteristic Impedance of Microstrip Line,” Papers of Electromagnetic Behavior Society, Vol. 6. No. 2, pp. 14-20, November 1996. (Article in Japanese.)

[177]

Kenich Kamitani, Naoko Yoshita, Hideki Nakano, Kohji Koshiji and Eimei Shu, “Multilayered printed antenna with dotmatrix-like director - Relation between the characteristics and the dot density,” Papers of Electromagnetic Behavior Society, Vol. 6., No. 2, pp. 27-33, November 1996.

[178]

Lu Fan, Kai Chang, “Uniplanar Power Dividers Using Coupled CPW and Asymmetrical CPS for MIC's and MMIC's,” IEEE Transactions on Microwave Theory and Techniques, December 1996, pp. 2411-2420.

[179]

John W. Bandler, Radoslaw M. Biernacki, Shao Hua Chen, Piotr A. Grobelny, “Optimization Technology for Nonlinear Microwave Circuits Integrating Electromagnetic Simulations,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, January 1997, pp. 6-28.

[180]

Jack Browne, “Evaluate RF Designs With An EM Simulator,” Microwaves and RF, January 1997, pp. 123-124.

[181]

Jack Browne, “EM Simulators Run Under PC Windows,” Microwaves and RF, December 1996, pp. 163-164.

[182]

Nitin Jain and Peter Onno, “Use EM Software For Component Optimization,” Microwaves and RF, January 1997, pp. 65-74.

[183]

I. Toyoda, T. Tokumitsu, and M. Aikawa, “Highly Integrated Three-dimensional MMIC Single-chip Receiver and Transmitter,” IEEE Trans. Microwave Theory Tech., Vol. 44, No. 12, pp. 2340-2346, December 1996.

[184]

Nitin Jain and Peter Onno, “EM Software Aids Microwave Characterization,” Microwaves and RF, December 1996, pp. 98-108.

[185]

Jack Browne, “Top Products of 1996,” Microwaves and RF, December 1996, pp. 189-199.

331

EM

[175]

Em User’s Manual

332

[186]

Bill Oldfield, “The Stripline Forward Coupler,” Microwave Engineering Europe, February/March 1996, pp. 39-40.

[187]

Jack Browne, “The Changing Colors of Microwave CAE,” Microwaves and RF, November 1996, pg. 17.

[188]

Janine Sullivan, Alan Conrad and Jack Browne, “Software Tools Grow With The Power Of The PC,” Microwaves and RF, November 1996, pp. 31-37.

[189]

C.W. Turner, “V-Shaped transmission-lines for superconducting circuits and MMICs,” Microwave Engineering Europe, October 1996, pp. 43-48.

[190]

Charles J. Trantanella, “Modeling and Simulation of MMICs and Interconnects in Microwave Packages,” DTIC (Defense Technical Information Center) or NTIS (National Technical Information Service) Report # ADA321689, November 1996.

[191]

David Sanchez-Hernandez and Ian D. Robertson, “Some Experimental Results of Printed Antennas for the Benchmarking of Sonnet em Electromagnetic Simulator,” International Journal of Microwave and Millimeter-Wave ComputerAided Engineering, Vol. 6, No. 6, November 1996, pp. 419-429.

[192]

Robert W. Jackson and S. Rakshit, “Microwave Modeling of an Elevated Paddle Surface Mount Package," 1996 Conference on the Electrical Performance of Electronic Packaging Proceedings, Napa Valley, pp. 57 - 62, October 1996.

[193]

David Sanchez-Hernandez, Q.H. Wang, Ali A. Rezazadeh and Ian D. Robertson, “Millimeter-Wave Dual-Band Microstrip Patch Antennas Using Multilayer GaAs Technology,” IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 9, September 1996, pp. 1590-1593.

[194]

Kenji Kamogawa and Tsuneo Tokumitsu, “A Novel Antenna Using Ceramic/ Polyimide Multilayer Dielectric Substrate,” Technical Report of IEICE MW9547 (1995-07).

[195]

Kenji Kamogawa, Tsuneo Tokumitsu, and Masayoshi Aikawa, “A Novel Microstrip Antenna Using Alumina-ceramic/Polyimide Multilayer Dielectric Substrate,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 1, pp. 71-74.

[196]

Shinji Mino, Yasufumi Yamada, Yuji Akahori, Mitsuho Yasu and Kazuyuki Moriwaki, “Loss Reduction in a Coplanar Waveguide on a Planar Lightwave Circuit (PLC) Platform by Quenching,” Journal of Lightwave Technology, Vol. 14, No. 8, August 1996.

Appendix V Sonnet Bibliography Robert Howald and Chris McDonnell, “Design and Simulation of an Inhomogeneous Coupled-Line Bandpass Filter,” Microwave Journal, July 1996, pp. 64-74.

[198]

Aditya Gupta, Mike Salib and Andy Ezis, “A High Efficiency 1.8 W, 6 to 18 GHz HBT MMIC Power Amplifier,” Microwave Journal, August 1996, pp. 2026.

[199]

Raafat R. Mansour, Shen Ye, Van Dokas, Bill Jolley, Glenn Thomson, WaiCheung Tang and Chandra M. Kudsia, “Design Considerations of Superconductive Input Multiplexers for Satellite Applications,” IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 7, July 1996, pp. 1213-1228.

[200]

T. Tokumitsu, M. Hirano, K. Yamasaki, C. Yamaguchi, and M. Aikawa, “Highly Integrated 3-D MMIC Technology Being Applied to Novel Masterslice GaAsand Si- MMIC's (Invited Paper),” in IEEE GaAs IC Symposium Digest, November 1996, pp. 151-154.

[201]

Kenjiro Nishikawa, Tsuneo Kokumitsu, and Ichihiko Toyoda, “Miniaturized Wilkinson Power Divider Using Three-Dimensional MMIC Technology,” IEEE Microwave and Guided Wave Letters, Vol. 6, No. 10, Oct. 96, pp. 372-374.

[202]

M. Aikawa, T. Tokumitsu, and K. Nishikawa, “Advanced MMIC Technology for the Next Generation 3D MMICs and Master-slice Technology (Invited Paper),” in 26th European Microwave Conf. Proc., September 1996, pp. 748-753.

[203]

K. Nishikawa, K. Kamogawa, T. Tokumitsu, M. Aikawa, M. Hirano, and S. Sugitani, “Highly Integrated Three-dimensional MMIC 20-GHz Single-chip Receiver,” in 26th European Microwave Conf. Proc., September 1996, pp. 199203.

[204]

R.R. Mansour, “Design of superconductive multiplexers using single-mode and dual- mode filters,” IEEE Trans. Microwave Theory Tech., vol. 42, pp. 14111418, July 1994.

[205]

S.H. Talisa et al., “High-temperature superconducting four-channel filter bank,” IEEE Trans. Appl. Superconduct., vol. 5, no. 2, pp. 2079-2082, June 1995.

[206]

Salvador H. Talisa, Michael A. Janocko, D.L. Meier, John Talvacchio, C. Moskowitz, D.C. Buck, R.S. Nye, S.J. Pieseski and George R. Wagner, “High Temperature Superconducting Space-Qualified Multiplexers and Delay Lines,” IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 7, July 1996, pp. 1229-1239.

333

EM

[197]

Em User’s Manual

334

[207]

George L. Matthaei, Stephan M. Rohlfing and Roger J. Forse, “Design of HTS, Lumped-Element, Manifold-Type Microwave Multiplexers,” IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 7, July 1996, pp. 1313-1321.

[208]

Kurt F. Raihn, Neal O. Fenzi, Gregory L. Hey-Shipton, Elna R. Saito, P. Vince Loung and David L. Aidnik, “Adaptive High Temperature Superconducting Filters for Interference Rejection,” IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 7, July 1996, pp. 1374-1381.

[209]

Robert W. Jackson, “A Circuit Topology for Microwave Modeling of Plastic Surface Mount Packages,” IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 7, July 1996, pp. 1140-1146.

[210]

Dan Swanson, “Multilayer Transitions in FR4,” 1996 Wireless Workshop, Sedona, AZ, October 1996 (also available in the “Library” at http:// www.rfglobalnet.com, 1997).

[211]

Robert Jackson, “Modeling & Application of Plastic Surface Mount Packages on Typical PCBs,” 1996 Wireless Workshop, Sedona, AZ, October 1996.

[212]

Robert Jackson, "Modeling Millimeterwave IC Behavior for Flipped Chip Mounting Schemes," invited paper for the 1996 WRI International Symposium on "Directions for the Next Generation of MMIC Devices and Systems", N.Y., N.Y., September 1996.

[213]

Jan Snel, “Ceramic Multilayer Microwave Components Work at the Philips Ceramic Innovation Centre,” IEEE MTT-S 1996 Multilayer Microwave Circuits Workshop, pp. 217-227.

[214]

Anthony M. Pavio, “Multilayer Couplers, Hybrids and Baluns,” IEEE MTT-S 1996 Multilayer Microwave Circuits Workshop, pp. 183-203.

[215]

D. Mirshekar-Syahkal, "Computation of Equivalent Circuits of CPW Discontinuities Using Quasi-Static Spectral Domain Method," IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 6, June 1996, pp. 979-984.

[216]

Zhi-Yuan Shen, Charles Wilker, Philip Pang and Charles Carter, III, “HighPower HTS Planar Filters with Novel Back-Side Coupling,” IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 6, June 1996, pp. 984-986.

[217]

T. Tokumitsu, K. Nishikawa, K. Kamogawa, I. Toyoda, and M. Aikawa, “Threedimensional MMIC Technology for Multifunction Integration and Its Possible Application to Masterslice MMIC,” in IEEE 1996 Microwave and MillimeterWave Monolithic Circuits Symposium Digest, June 1996, pp. 85-88.

Appendix V Sonnet Bibliography Long Tran, Michael Delaney, Russ Isobe, Derek Jang and Julia Brown, “Frequency Translation MMICs Using InP HEMT Technology,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 1, pp. 261-264.

[219]

Paul D. Cooper, Patricia A. Piacente and Robert J. Street, “Multichip-on-Flex Plastic Encapsulated MHDI-Low Cost Substrateless Manufacturing for Microwave and Millimeterwave Modules,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 1, pp. 219-222.

[220]

J.A. Lester, Y. Hwang, J. Chi, R. Lai, M. Biedenbender and P.D. Chow, “Highly Efficient Compact Q-Band MMIC Power Amplifier Using 2-Mil Substrate and Partially- Matched Output,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 1, pp. 153-155.

[221]

Y. Hwang, P.D. Chow, J. Lester, J. Chi, D. Garske, M. Biedenbender and R. Lai, “Fully-Matched, High-Efficiency Q-Band 1 Watt MMIC Solid State Power Amplifier,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 1, pp. 149-152.

[222]

Francois Colomb, Kevin Eastman and John Roman, “Characterization of Metal on Elastomer Vertical Interconnections,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 1, pp. 75-77.

[223]

Long Tran, Russ Isobe, Michael Delaney, Rick Rhodes, Derek Jang, Julia Brown, Loi Nguyen, Minh Le, Mark Thompson and Takyiu Liu, “High Performance, High Yield Millimeter-Wave MMIC LNAs Using InP HEMTs,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 1, pp. 9-12.

[224]

J. W. Bandler, R. M. Biernacki and S. H. Chen, “Parameterization of Arbitrary Geometrical Structures for Automated Electromagnetic Optimization,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 2, pp. 10591062.

[225]

D. Sturzebecher, J. Leen, R. Cadotte, J. DeMarco, T. D. Ni, T. Higgins, M. Popick, M. Cummings, B. VanMeerbeke, T. Provencher, B. Kimble, K. Shalkhauser and R. Simons, “20 GHz LTCC Phased Array Module,” IEEE MTTS 1996 International Microwave Symposium Digest, Vol. 2, pp. 991-994.

[226]

I. Toyoda, T. Tokumitsu, and M. Aikawa, “Highly integrated three-dimensional MMIC single-chip receiver and transmitter,” 1996 IEEE MTT-S International Microwave Symposium Digest, June 1996, pp. 1209-1212.

335

EM

[218]

Em User’s Manual

336

[227]

Nitin Jain and Peter Onno, “High Power 6-18 GHz H/V Switch Designed in Channelized Wafer Scale Fabrication Process,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 2, pp. 955-958.

[228]

Hiroaki Tanaka, Yutaka Sasaki, Takuya Hashimoto, Yoshikazu Yagi and Youhei Ishikawa, “Miniaturized 90 Degree Hybrid Coupler Using High Dielectric Substrate for QPSK Modulator,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 2, pp. 793-796.

[229]

Der-Woei Wu, “A High-Efficiency HBT Cellular Power Amplifier with Integrated Matching Networks,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 2, pp. 767-770.

[230]

J.W. Bandler, R.M. Biernacki and S.H. Chen, “Fully Automated Space Mapping Optimization of 3D Structures,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 2, pp. 753-756.

[231]

G.L. Creech, B. Paul, C. Lesniak, T. Jenkins, R. Lee and M. Calcatera, “Artificial Neural Networks for Accurate Microwave CAD Applications,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 2, pp. 733-736.

[232]

G.L. Matthaei, N.O. Fenzi, R. Forse and S. Rohlfing, “Narrow-Band HairpinComb Filters for HTS and Other Applications,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 2, pp. 457-460.

[233]

S. Chaki, T. Takagi, Y. Tsukahara, H. Matsubayashi, N. Andoh, Y. Sasaki and M. Otsubo, “A Miniaturized X-band 4-Stage LNA Designed Using a Novel Layout Optimization Technique,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 3, pp. 1213-1216.

[234]

Mark D. DuFault and Arvind K. Sharma, “Millimeter-Wave Hemt Noise Models Verified Thru V-Band,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 3, pp. 1321-1324.

[235]

Mark D. DuFault and Arvind K. Sharma, “A Novel Calibration Verification Procedure for Millimeter-Wave Measurements,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 3, pp. 1391-1394.

[236]

Tsang-Der Ni, James DeMarco, Dana Sturzebecher and Mike Cummings, “High Frequency Hermetic Packages Using LTCC,” IEEE MTT-S 1996 International Microwave Symposium Digest, Vol. 3, pp. 1627-1630.

Appendix V Sonnet Bibliography F. Schnieder, R. Doerner and W. Heinrich, “High-Impedance Coplanar Waveguides with Low Attenuation,” IEEE Microwave and Guided Wave Letters, Vol. 6, No. 3, March 1996, pp. 117-119.

[238]

Chappell Brown, Silicon inductors boost RF design, EE Times, May 20, 1996.

[239]

T. Tokumitsu, M. Aikawa, and K. Kohiyama, “Three-dimensional MMIC Technology: A possible solution to masterslice MMIC's on GaAs and Si,” IEEE Microwave Guide Wave Letter, Vol. 5, No. 11, pp. 411-413, November 1995.

[240]

J.S. Hong and M.J. Lancaster, “Microstrip Bandpass Filter Using Degenerate Modes of a Novel Meander Loop Resonator,” IEEE Microwave and Guided Wave Letters, Vol. 5, No. 11, November 1995, pp. 371-372.

[241]

Daniel G. Swanson. Jr., “Guest Editorial,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 5, No. 5, September 1995, pg. 301.

[242]

Nitin Jain and Peter Onno, “Efficient Use of Commercial Electromagnetic Simulators for Microwave and Millimeter-Wave Circuits,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 5, No. 5, September 1995, pp. 302-323.

[243]

John W. Bandler, Radoslaw M. Biernacki, Shao Hua Chen, William J. Getsinger, Piotr A. Grobelny, Charles Moskowitz, and Salvador H. Talisa, “Electromagnetic Design of High-Temperature Superconducting Microwave Filters,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 5, No. 5, September 1995, pp. 331-343.

[244]

Daniel G. Swanson. Jr., “Optimizing a Microstrip Bandpass Filter Using Electromagnetics,” International Journal of Microwave and Millimeter-Wave Computer- Aided Engineering, Vol. 5, No. 5, September 1995, pp. 344-351.

[245]

George L. Matthaei and Roger J. Forse, “A Note Concerning the Use of Field Solvers for the Design of Microstrip Shunt Capacitances in Lowpass Structures,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 5, No. 5, September 1995, pp. 352-358.

[246]

Inder J. Bahl (coordinated by), “MIC Simulation Column," International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 5, No. 5, September 1995, pp. 359-367.

337

EM

[237]

Em User’s Manual

338

[247]

Rolf H. Jansen, “Computer - aided design of microwave and millimeterwave integrated circuits - progress during the last decade and future perspectives,” 25th European Microwave Conference 1995, Conference Proceedings, pp. 93-100.

[248]

Daniel G. Swanson, Jr., “First Pass CAD of Microstrip Filters Cuts Development Time,” Microwave Journal, August 1995.

[249]

Martin I. Herman, Karen A. Lee, Elzbieta A. Kolawa, Lynn E. Lowry and Ann N. Tulintseff, “Novel Techniques for Millimeter-Wave Packages,” IEEE Trans. on Microwave Theory and Techniques, Vol. 43, No. 7, July 1995, pp. 1516-1523.

[250]

John N. Poelker and Ralston S. Roberson, “A Comparison of Planar Doped Barrier Diode Performance Versus Schottky Diode Performance in a Single Balanced, MIC Mixer with Low LO Drive,” IEEE Trans. on Microwave Theory and Techniques, Vol. 43, No. 6, June 1995, pp. 1241-1246.

[251]

J.W. Bandler, R.M. Biernacki, Q. Cai and S.H. Chen, “Cost-Driven PhysicsBased Large-Signal Simultaneous Device and Circuit Design,” 1995 IEEE International Microwave Symposium Digest, Orlando, FL, May 1995, pp. 14431446.

[252]

Makoto Hirano, Kenjiro Nishikawa, Ichihiko Toyoda, Shinji Aoyama, Suehiro Sugitani and Kimiyoshi Yamasaki, “Three-Dimensional Passive Circuit Technology For Ultra-Compact MMICs,” 1995 IEEE International Microwave Symposium Digest, Orlando, FL, May 1995, pp. 1447-1450.

[253]

Makoto Hirano, Kenjiro Nishikawa, Ichihiko Toyoda, Shinji Aoyama, Suehiro Sugitani and Kimiyoshi Yamasaki, “Three-Dimensional Passive Circuit Technology For Ultra-Compact MMICs,” IEEE Trans. on Microwave Theory and Techniques, Vol. 43, No. 12, Dec. 1995, pp. 2845-2850.

[254]

Ichihiko Toyoda, Makoto Hirano, and Tsuneo Tokumitsu, “An Ultra-Wideband Miniature Balun for 3-Dimensional MMICs,” 1994 Asia-Pac. Microwave Conf. proc., Dec. 1994, pp. 511-514.

[255]

I. Toyoda, T. Hirota, T. Hiraoka, and T. Tokumitsu, “Multilayer MMIC BranchLine Coupler and Broad-Side Coupler.” IEEE MMWMC Dig., S-5, June 1992, pp. 79-82.

[256]

Makoto Hirano, Ichihiko Toyoda, Masami Tokumitsu and Kazuyoshi Asai, “Folded U-Shaped Micro-Wire Technology for GaAs IC Interconnections,” 1996 IEEE International Microwave Symposium Digest, San Francisco, June 1996, pp. 1153-1156.

Appendix V Sonnet Bibliography Makoto Hirano, Ichihiko Toyoda, Masami Tokumitsu and Kazuyoshi Asai, “Folded U-Shaped Micro-Wire Technology for GaAs IC Interconnections,” IEEE Trans. on Microwave Theory and Techniques, Vol. 44, No. 12, Dec. 1996, pp. 2347-2353.

[258]

Satoshi Yamaguchi, Yuhki Imai, Tsugumichi Shibata, Taiichi Otsuji, Makoto Hirano and Eiichi Sano, “An Inverted Microstrip Line IC Structure for Ultrahigh-speed Applications,” 1995 IEEE International Microwave Symposium Digest, Orlando, FL, May 1995, pp. 1643-1646.

[259]

R. H. Blick, R. J. Haug, D.W. van der Weide, K. von Klitzing, and K. Eberl, “Photon- assisted tunneling through a quantum dot at high microwave frequencies,” Applied Physics Letters, Dec. 1995.

[260]

D.W. van der Weide, “Delta-doped Schottky diode nonlinear transmission lines for 480-fs, 3.5-V transients,” Appl. Phys. Lett. 65 (7), August 1994.

[261]

K. F. Raihn, N. O. Fenzi, E. R. Soares, and G. L. Matthaei, “An Optical Switch for High Temperature Superconducting Microwave Band Reject Resonators,” IEEE MTT-S 1995 International Microwave Symposium, Orlando, Florida, pp. 187-190.

[262]

J. A. Costello, M. Kline, F. Kuss, W. Marsh, R. Kam, B. Rasano, M. Berry, and N. Koopman, “The Westinghouse High Density Microwave Packaging Program,” IEEE MTT-S 1995 International Microwave Symposium, Orlando, Florida, pp. 177-180

[263]

M. A. Schamberger, and A. K. Sharma, "A Generalized Electromagnetic Optimization Procedure for the Design of Complex Interacting Structures in Hybrid and Monolithic Microwave Integrated Circuits,” IEEE MTT-S 1995 International Microwave Symposium, Orlando, Florida, pp. 1191-1194.

[264]

C. M. Jackson, T. Pham, Z. Zhang, A. Lee, and C. Pettiete-Hall, “Model for a Novel CPW Phase Shifter,” IEEE MTT-S 1995 International Microwave Symposium, Orlando, Florida, pp. 1439-1442.

[265]

J. W. Bandler, R. M. Biernaki, S. H. Chen, R. H. Hemmers, and K. Madsen, “Aggressive Space Mapping For Electromagnetic Design,” IEEE MTT-S 1995 International Microwave Symposium, Orlando, Florida, pp. 1455-1458.

[266]

I. Toyoda, T. Hirota, T. Hiraoka, and T. Tokumitsu, “Multilayer MMIC BranchLine Coupler and Broad-Side Coupler.” IEEE MMWMC Dig., S-5, June 1992, pp. 79-82.

EM

[257]

339

Em User’s Manual

340

[267]

I. Toyoda, M. Hirano, and T. Tokumitsu, “Three-dimensional MMIC and Its Application: An Ultra-wideband Miniature Balun,” IEICE Trans. Elec.,Vol. E78C, no. 8, pp. 919-924, August 1995.

[268]

J. W. Bandler, R. M. Biernaki, Q. Cai, S. H. Chen, P. A. Grobelny, and D. G. Swanson Jr., “Heterogeneous Parallel Yield-Driven Electromagnetic CAD,” IEEE MTT-S 1995 International Microwave Symposium, Orlando, Florida, pp. 1085-1088.

[269]

Y. Tsukahara, S. Chaki, Y. Sasaki, K. Nakahara, N. Andoh, H. Matsubayasi, N. Tanino, and O. Ishihara, “A C-Band 4-Stage Low Noise Miniaturized Amplifier Using Lumped Elements,” IEEE MTT-S 1995 International Microwave Symposium, Orlando, Florida, pp. 1125-1128.

[270]

J. W. Bandler, R. M. Biernaki, Q. Cai, S. H. Chen, and P. A. Grobelny, “Integrated Harmonic Balance and Electromagnetic Optimization with Geometry Capture,” IEEE MTT-S 1995 International Microwave Symposium, Orlando, Florida, pp. 793-796.

[271]

Dan Swanson, “Optimizing Microstrip Filters Using Electromagnetics,” IEEE MTT Symposium Workshop WMFE Digest, May 1995, Orlando, Florida.

[272]

S. H. Chen, “Automated EM Optimization of Linear and Nonlinear Circuits with Geometry Capture for Arbitrary Planar Structures,” IEEE MTT Symposium Workshop WFFE Digest, May 1995, Orlando, Florida.

[273]

Anthony M. Pavio, “The Electromagnetic Analysis and Optimization of a Broad Class of Problems Using Companion Models.” IEEE MTT Symposium Workshop WFFE Digest, May 1995, Orlando, Florida.

[274]

Nitin Jain, “Automated Circuit Design Using Commercial EM Simulators,” IEEE MTT Symposium Workshop WFFE Digest, May 1995, Orlando, Florida.

[275]

Marc Goldfarb, “CAD Methodology for Commercial Applications,” IEEE MTT Symposium Workshop WFFE Digest, May 1995, Orlando, Florida.

[276]

Alan Conrad, Jack Browne, “EM Simulator Enhances Performance of Microwave Circuits,” Microwaves & RF, April 1995, pp. 200-207.

[277]

John W. Bandler, Radoslaw M. Biernacki, Shao Hua Chen, Piotr A. Grobelny and Ronald H. Hemmers, “Space Mapping Technique for Electromagnetic Optimization,” IEEE Trans. on Microwave Theory and Techniques, Vol. 42, No. 12, December 1994.

Appendix V Sonnet Bibliography D.W. van der Weide, “Planar antennas for all-electronic terahertz systems,” Optical Society of America, Vol. 11, No. 12, December 1994.

[279]

D.W. van der Weide, R.H. Blick, F. Keilmann, and R.J. Haug, “Electronic Picosecond-pulse Interferometer Probing the Millimeter-wave Response of a Quantum- Dot System,” Summary for OSA Topical Meeting on Ultrafast Electronics and Optoelectronics/Quantum Optoelectronics, March 13-17, 1995, Dana Point, CA, USA.

[280]

-, “Focus on CAD/CAE: A Benchmark guide Through The EM Simulation Maze,” Microwave Engineering Europe, May 1995, pp. 23-26.

[281]

-, “CAD Benchmark: Electromagnetic Simulators,” Microwave Engineering Europe, Nov. 1994, pp. 11-20.

[282]

-, “EM CAD Benchmark: The Vendors Respond,” Microwave Engineering Europe, December/January 1995, pg. 12.

[283]

-, “Entry Level CAD/CAE: An Independent Review,” Microwave Engineering Europe, Nov. 1992, pp. 11-19.

[284]

-, “CAD Review: The 7 GHz Doubler Circuit,” Microwave Engineering Europe, May. 1994, pp. 43-53.

[285]

G. L. Matthaei and G. Hey-Shipton, “Concerning the Use of High-Temperature Superconductivity in Planar Microwave Filters,” IEEE Trans. Microwave Theory Tech., Vol. MTT-42, No. 7, July 1994, pp. 1287-1294.

[286]

J. Bandler, et al., “Microstrip Filter Design Using Direct EM Field Simulation” IEEE Trans. Microwave Theory Tech., Vol. MTT-42, No. 7, July 1994, pp. 13531359.

[287]

D. G. Swanson, Jr., “Using A Microstrip Bandpass Filter To Compare Different Circuit Analysis Techniques,” International Journal of Microwave & MillimeterWave Computer-Aided Engineering, Vol. 5, No. 1, Jan. 1995, pp. 4-12.

[288]

U. L. Rohde, “Feedback,” Microwaves and RF, Aug. 1994, pg. 13.

[289]

Jack Browne, “Simulation for Wireless Markets,” Microwaves and RF, Aug. 1994, pg. 17.

[290]

Jack Browne, “Crosstalk,” Interview with Jim Rautio, Microwaves and RF, Aug. 1994, pp. 47-48.

341

EM

[278]

Em User’s Manual

342

[291]

T. Winslow, “Response #2: MMIC Miniature Filter,” International Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, Vol. 5, No. 1, Jan. 1995, pp. 45-49.

[292]

J. M. Carroll and K. Chang, “Full Wave Convergence Analysis of Microstrip Transmission Parameters,” International Journal of Microwave & MillimeterWave Computer-Aided Engineering, Vol. 4, No. 2, pp. 140-147, April 1994.

[293]

V. K. Sadhir, I. J. Bahl, and D. A. Willems, “CAD Compatible Accurate Models of Microwave Passive Lumped Elements for MMIC Applications,” International Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, Vol. 4, No. 2, pp. 148-162, April 1994.

[294]

J. M. Carroll and K. Chang, “Full-Wave Convergence Analysis of Microstrip Transmission Parameters,” International Journal of Microwave and MillimeterWave Computer-Aided Engineering, Vol. 4, No. 2, 1994, pp. 140-147.

[295]

M. Goldfarb and A. Platzker, “The Effects of Electromagnetic Coupling on MMIC Design,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol. 1, No. 1, Jan. 1991, pp. 38-47.

[296]

D. Swanson, D. Baker, and M. O'Mahoney, “Connecting MMIC Chips to Ground in a Microstrip Environment,” Microwave Journal, December 1993, pp. 58-64.

[297]

D. Swanson, “Simulating EM Fields,” IEEE Spectrum Magazine, November 1991, pp. 34-37.

[298]

R.J. Furlow, “MIC Simulation Column,” International Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, Vol. 1, No. 4, Oct. 1991, pp. 412-413. (Diagonal filter, meander line from R. Furlow).

[299]

R. J. Furlow, “MIC Simulation Column,” International Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, Vol. 2, No. 2, Apr. 1992, pp. 121-122. (Diagonal filter measured data from R. Furlow).

[300]

D. G. Swanson. and R. J. Forse, “An HTS End-Coupled CPW Filter at 35 GHz,” IEEE International Microwave Symposium, May 1994, San Diego, pp. 199-202.

[301]

J. W. Bandler, et. al., “Exploitation of Coarse Grid for Electromagnetic Optimization,” IEEE International Microwave Symposium, May 1994, San Diego, pp. 381-384.

Appendix V Sonnet Bibliography J. W. Bandler, et. al., “Electromagnetic Design of High-Temperature Superconducting Microwave Filters,” IEEE International Microwave Symposium, May 1994, San Diego, pp. 993-996.

[303]

C. Sinclair, “A Coplanar Waveguide 6-18 GHz Instantaneous Frequency Measurement Unit for Electronic Warfare Systems,” IEEE International Microwave Symposium, May 1994, San Diego, pp. 1767-1770.

[304]

M. Gillick and I. D. Robertson, “Ultra Low Impedance CPW Transmission Lines for Multilayer MMIC's,” IEEE International Microwave Symposium, June 1993, Atlanta, pp. 145-148.

[305]

D. Willems and I. Bahl, “A MMIC Compatible Coupled Line Structure that uses Embedded Microstrip to Achieve Extremely Tight Couplings,” IEEE International Microwave Symposium, June 1993, Atlanta, pp. 581-584.

[306]

H. Wang, et. al., “A High Gain Low Noise 110 GHz Monolithic Two-stage Amplifier,” IEEE International Microwave Symposium, June 1993, Atlanta, pp. 783-785.

[307]

J. W. Bandler, et. al., “Minimax Microstrip Filter Design using Direct EM Field Simulation,” IEEE International Microwave Symposium, June 1993, Atlanta, pp. 889-892.

[308]

J. W. Bandler, et. al., “Multilevel Multidimensional Quadratic Modeling for Yield- Driven Electromagnetic Optimization,” IEEE International Microwave Symposium, June 1993, Atlanta, pp. 1017-1020.

[309]

R. R. Mansour, F. Rammo, and V. Dokas, “Design of Hybrid-Coupled Multiplexers and Diplexers using Asymmetrical Superconducting Filters,” IEEE International Microwave Symposium, June 1993, Atlanta, pp. 1281-1284.

[310]

A. K. Rayit and N. J. McEwan, “Coplanar Waveguide Filters,” IEEE International Microwave Symposium, June 1993, Atlanta, pp. 1317-1320.

[311]

C. Sinclair and S. J. Nightingale, “An Equivalent Circuit Model for the Coplanar Waveguide Step Discontinuity,” IEEE International Microwave Symposium, June 1992, Albuquerque, pp. 1461-1464.

[312]

D. Willems and I. Bahl, “An MMIC-Compatible Tightly Coupled Line Structure Using Embedded Microstrip,” IEEE Trans. Microwave Theory Tech., Vol. MTT41, No. 12, pp. 2303-2310, Dec. 1993.

343

EM

[302]

Em User’s Manual

344

[313]

D. Swanson, “Experimental Validation: Measuring a Simple Circuit,” IEEE International Microwave Symposium, Workshop WSMK Digest, Atlanta, June 1993.

[314]

J. Bandler, “Analog Diagnosis and Optimization Technology for Experimental Validation,” IEEE International Microwave Symposium, Workshop WSMK Digest, Atlanta, June 1993.

[315]

R. Y. Shimoda, “Critical Issues in Experimental Validation,” IEEE International Microwave Symposium, Panel Session PSB Digest, Albuquerque, June 1992.

[316]

I. Bahl and D. WIllems, “Critical Issues in Experimental Validation,” IEEE International Microwave Symposium, Panel Session PSB Digest, Albuquerque, June 1992.

[317]

M. Goldfarb, “Verification Structures for Passive Element Model Development,” IEEE International Microwave Symposium, Panel Session PSB Digest, Albuquerque, June 1992.

[318]

D. Swanson, “Designing Microwave Components Using Electromagnetic Field Solvers,” IEEE International Microwave Symposium, Workshop WSA Digest, Albuquerque, June 1992.

[319]

D. Swanson, “Electromagnetic Software Simulation,” Emerging Microwave Technologies and Applications Conference, Stanford, CA, March 1991.

[320]

L. P. Dunleavy and R. Wenzel, “Use of EM Analysis to Study Shielding Effects in Microstrip Circuits,” South Con Conference Digest, Orlando, FL, March 1994.

[321]

W. Gaiewski and L. Dunleavy, “Design and Analysis of Spiral Inductors on Silicon,” Interim Technical Report, Univ. S. Florida, May 1994.

[322]

P. A. MacDonald, “Characterization of Microstrip Discontinuities on LaAlO3,” IEEE International Microwave Symposium, June 1991, Boston, pp. 1341-1344.

[323]

D. G. Swanson, Jr., “Grounding Microstrip Lines with Via Holes,” IEEE Trans. Microwave Theory Tech., Vol. MTT-40, No. 8, pp. 1719-1721, Aug., 1992.

[324]

D. G. Swanson, “Electromagnetic Simulation of Microwave Components,” 37th ARFTG Conference, Boston, June 1991, pp. 3-9.

[325]

W. Oldfield, et. al., “Simple Microstrip Structures Calculated Vs. Measured,” 37th ARFTG Conference, Boston, June 1991, pp. 10-20.

Appendix V Sonnet Bibliography [326]

M. E. Goldfarb and R. A. Pucel, “Modeling Via Hole Grounds in Microstrip,” IEEE Microwave and Guided Wave Letters, Vol. 1, No. 6, June 1991, pp. 135137.

[327]

D.N. Meeks, "Re-Normalizing the Scattering Parameters," RF Design, October 1985, pp. 41-42. EM 345

Em User’s Manual

346

Index

Index

EM

A accuracy cell size 233 Add command button 265 Add IFS command button 267 Additional Options dialog box geometry file analysis 39 network file analysis 261 advanced options 262 Advanced Options dialog box 198 air bridges 165, 203 amp.geo 127 amp.rsp 129 analysis continuing 277 geometry file 245–255 network file 97, 257–262 output window 277 pausing 276 running 272–277 starting 276 stopping 277 Analysis Control dialog box 250 analysis control file 6, 90, 148, 156, 157, 249, 260 comments 6, 269 editing 251, 264 format 297–299 geometry file analysis 245, 250

intelligent frequency selection 129 internal 100 network file 144, 250, 257 SPICE file 189, 268 analysis control keywords 154 ANN 154 AUTO 154 END 154 ESWEEP 154 FINDMAX 155, 300 FINDMIN 155, 300 LSWEEP 154 STEP 154 SWEEP 154 analysis controls, See frequency control analysis frequencies default 285 units 249, 286 ANG 138, 159 anisotropic dielectric bricks 172 ANN 154, 298 antennas 177–185 anti-comment symbol 135 AUTO 128, 154, 157, 267 auto-grounded ports 53, 54, 103, 302 de-embedding 70, 76

B $BASE 146, 150 $BASE_new.rsp 146 347

Em User’s Manual balanced ports 59, 203, 204 bandpass filter 111 batch 272 benchmark 227–232 accuracy 227 residual error 229 box resonance 216, 218, 292 box-wall ports 50 br32.geo 197 bricks see dielectric bricks byte-reversal network 196

C cache directory specifying 283 cache memory limiting 283 caching 17, 20–21 directory 284 memory 284 parameters 283 CAE software 8 calculate memory usage option 253 CAP 138, 143 capacitors 89, 143, 173, 187–201, 257, 296 cascade.net 94 cascading data files 89, 90 cell size 24, 25, 28, 162, 175, 234, 235, 243 accuracy 233 determining error 231 error message 310 quad precision 254, 288 selecting 28–31 subsectioning 27, 35 vias 166 circuit analyses 6, 89, 90, 94, 97, 101, 133 circuit geometry file 6, 90, 148, 295 circuit response file 6, 295 348

circuit theory simulators 8, 240 CKT 95, 100, 136, 142, 145 $BASE 146, 150 defnp 145 elements 144 filename 145 netname 145 nodes 145 CMIN 269, 299 cocross.geo 207 COM 160 combine.net 101 combine.rsp 102 combined circuit analyses 90 combining data files 89 comma separated values 150 command buttons Add 265 Add IFS 267 Comments 269 Output Files 271 Set to Top Window 285 SPICE 269 command line options 288–293 -N 238 -q 189, 235, 243 comment line 94, 135 Comments command button 269 Compact 289 complex sweep 249, 264 coplanar 59 cross junction 203, 206 short 203 waveguide 203 cosht.geo 205 cosht_sy.geo 206 coupling mechanism 114

Index cross junction coplanar 203 cross-talk 187 CSV 150 CTL 100 current density file 5, 252, 253, 271, 290, 296 cvia.geo 168

.d 102 data blocks 94 CKT 95, 100, 136, 142, 143, 145 DIM 95, 136, 138, 139 FILEOUT 95, 136, 149, 150 FREQ 95, 100, 128, 136 overriding 157 syntax 153 OUT 136 VAR 136, 140 data files 89 cascading 90 data tags 146 DATA_TAG 146 DB 159 de-embed option 251, 271 de-embedded data 271 de-embedding 17, 18, 61–88, 292 auto-grounded ports 70, 76 box-wall ports 67, 73 coupled transmission lines 74 enabling 62 error codes 78 example 62 guidelines 81 higher order modes 88 output format 77 port discontinuities 66 reference planes 72, 81, 84

EM

D

ungrounded-internal ports 69 Defnp 145 Delim 160 detect box resonance option 254 diagonal fill 163, 175 dialog boxes Additional Options geometry file analysis 39 network file analysis 261 Advanced Options 198 Analysis Control 250 Frequency Control 265 Intelligent Frequency Control 267 main window with run list 273 Open File 281, 282, 283 Preferences 283–286 Save As 283 Select Output Files 271 SPICE 269 dielectric bricks 171–175, 302 air 175 applications 173 de-embedding 174 ebridge 175 limitations 175 parameters 173 patgen 175 subsectioning 173 vias 174 dielectric constant 235 dielectric layer thickness 235 digital interconnect 187 digital, high speed 196 DIM 95, 136, 137, 138 discontinuity 17 coplanar cross junction 203, 206 coplanar short 203

349

Em User’s Manual lumped model example 192 lumped models 187 port 88 disk swap 24 distributed parameters 188, 296 DMAC 227 do not check for consistency option 261 DUT 61

E -E 291 ebridge dielectric bricks 175 edge mesh option 37, 254, 291 edge-coupled bandpass filter 111 edge-vias 166 editing analysis control file 251 analysis controls 264 frequency control 268 geometry file 248 network file 259 electric fields 223 electrically thick conductors 42 em ANN 154 approximations 233 AUTO 154, 157 description 1 END 154 ESWEEP 154 FINDMAX 155, 157, 300 FINDMIN 155, 157, 300 frequency selection feature 125 input files 4 interface 245–283 invoking 246 LSWEEP 154 350

network file 133 output frequency interpolation 156 output files 4 reduce circuit size 24 speed 24 STEP 154 SWEEP 154 theory 2–4, 8 timing 237 emgraph invoking 278 plot 130 emvu 31, 205, 225 current density file 252, 271, 290, 296 END 154, 299 error messages 305, 308–310 error, residual 229 ESWEEP 154, 266, 298 example files amp.geo 127 amp.net 128 amp.rsp 129 benchmark s100.geo 229 s25.geo 229 s50.geo 229 bpfilter 111–123 br32.geo 197 cascade.net 94 cocross.geo 207 combine.net 101 cosht.geo 205 cosht_sy.geo 206 cvia.geo 168 filter.geo 64 findmax.net 131 gap20.geo 225

Index

F ! 134, 135 fields, viewing 223 filename 145, 159 FILEOUT 95, 136, 149, 150, 151 files input 297 output 270, 297 fill diagonal 163 filter example 111–123 filter.geo 64 filter.net 146 filter_new.rsp 146 FINDMAX 130, 155, 157, 268, 300 findmax.net 131 findmax.rsp 132

FINDMIN 130, 155, 157, 267, 300 FMAX 299 force running option 261 FRE 298 Free Space metal type 220, 222 FREQ 95, 100, 128, 136, 138, 153 overriding 157 frequency interpolation 156 points 89 resolution 125 response 125 selection 125 sorted sweeps 155 subsectioning 270 frequency control adding 265 analysis control file 250 complex sweep 249, 264 editing 264, 268 exponential sweep 266 information 263 intelligent frequency selection 267 internal sweep 260 linear sweep 266 saving 270 separator 268 simple sweep 248 single 265 specifying 248, 260 SPICE 268 sweep 265 Frequency Control dialog box 265 fringing fields 88 FTYP NET 94 full analysis mode 253

EM

lumped.geo 106 lumped2.geo 109 lumped2.net 109 open_120.geo 166, 217 openloss.geo 220 openmite.geo 163 package.geo 213 patch.geo 52, 182 raystub.geo 169 res16.d 92 res67.geo 101 steps.geo 192 steps_sy.geo 22 tane.geo 225 thkstep.geo 211 thkthru.geo 210 tripat.geo 183 via.geo 166 viaports.geo 52 exponential sweep 154, 266

351

Em User’s Manual G .geo, see geometry file GABMAC 227 gap20.geo 225 generate subsections only mode 253 GEO 144, 148 geometry file 6, 90, 148 editing 248 selecting 247 geometry file analysis 245–255 run options 251–255 GHZ 298 ground via 165, 168

H header line 94, 134, 135 high precision data file 271, 272 high speed digital example 196 HZ 298

I IM 159 Impedance 151 IND 138, 143 inductors 89, 143 infinite array 178 input files 297 Intelligent Frequency Control dialog box 267 intelligent frequency controls 89 AUTO 128 automatic 267 find maximum 268 find minimum 267 FINDMAX 130 FINDMIN 130 specifying 267 interface basics 15

352

internal sweep 260 interpolating frequencies 156 interpolation 112, 117 invoking em 246

J .jxy, see current density file job file creating a new 280 loading 282 opening 280 renaming 283 saving 283

K KHZ 298 kinetic inductance 44, 303 KMIN 269, 299

L LEVEL1 301–303 auto-grounded ports 302 dielectric bricks 302 dielectric layers 301, 302 kinetic inductance 303 maximum number of ports 303 maximum subsection size 303 memory limit 302 metalization layer 301 parallel subsections 302 vias 302 XMAX 303 XMIN 303 YMAX 303 YMIN 303 LEVEL1plus 301, 304 maximum number of ports 304 memory limit 304

Index

M MAG 159 magnetic wall 22 main menu accesses 11 main window with run list 273 make emvu file option 252, 271 Manhattan Polygon 34 maximum frequency 234 maximum subsection size 31, 37, 303 Meas 158 memory available 239 memory save option 252 menu bar accesses 11 messages 305–310 metal, thick 209 metalization layer 301

metallization loss 41 metallization thickness 42, 234 MFC 17, 283 MHZ 298 modes higher order 88 slot line 203, 206 TEM 88 multi-frequency caching 17, 20–21 directory 284 memory 284 parameters 283

EM

vias 304 linear sweep 266 LMAX 269, 299 LNG 138 loss 41 low frequency 42 related to frequency 42 LSWEEP 154, 266, 298 lumped elements capacitors 89, 138, 143 inductors 89, 138, 143 inserting 89, 102 resistors 89, 138, 143 transmission lines 89, 143 lumped model 187, 296 lumped.geo 106 lumped2.geo 109 lumped2.net 109

N N-coupled line 188, 191, 296 Netname 145, 158 network file 112, 118, 133 analysis 90, 97, 120, 257–262 comment line 94, 135 comment lines 94 data blocks 94, 136 detailed description 94 editing 259 frequency control 260 header line 94, 134, 135 internal sweep 260 interpolation 112 run options 260–262 selecting 258 nodes CKT 145 normalizing impedances 55 number of points automatic 267 exponential sweep 266 find maximum 267 find minimum 267 linear sweep 266 353

Em User’s Manual numerical precision 235

O Open File dialog box 281, 282, 283 open_120.geo 166, 217 openloss.geo 220 openmite.geo 163 options startup 285 OUT 136, 158 output files 297 .d extension 102, 271 .jxy extension 271 .lc extension 272 .lct extension 272 .nd extension 271 .pd extension 102, 272 .pnd extension 271 combine.rsp 102 default 271 lumped.rsp 106 lumped2.rsp 110 specifying 270 Output Files command button 271 output window 277 closing 277 opening 277 saving the contents 278

P .pd 102 package resonances 213–216 package.geo 213 parallel subsections 17, 26, 302 parameter 138 CKT 145 DIM 138 FILEOUT 150 354

FREQ 153 parameter type 254 patch antenna 177 patch.geo 52, 182 patgen dielectric bricks 175 perturbational approach 43 phased arrays 177 PHZ 298 port discontinuity 17 ports 49–59, 303, 304 auto-grounded 53, 54, 103, 302 balanced 59, 203, 204, 209 box-wall 50, 67, 73 discontinuities 66, 88 normalizing impedances 55 push-pull 59, 204, 209 push-push 209 renumbering 58 unbalanced 209 ungrounded-internal 51, 69, 103, 107 via 52 precision numerical 189, 235 quadruple 189, 235, 243 single 23 SPICE 189, 194 Preferences dialog box 283–286 probes 165 PSPICE 150, 191 push-pull ports 59, 204, 209 push-push ports 209

Q -q 189, 235, 243 quad precision 189, 235, 243 cell size 254, 288 option 254

Index R

EM

radiation 177, 177–185 raystub.geo 169 RE 159 reactance 43 reactive surface impedance 43 reference planes 72, 84 de-embedding without 82 short length 82 remove top cover 220, 222 RES 138, 143 res16.d 92 res67.geo 101 residual error 229 resistance 42 resistors 89, 143 thin film 41, 42, 90 resonance box 216, 218, 292 response data plotting 278, 279 response file 6 RMAX 269, 299 run list viewing 272 run options advanced 262 calculate memory usage mode 253 de-embed 251, 271 detect box resonance 254 do not check for consistency 261 edge mesh 37, 254 force running 261 full analysis mode 253 generate subsections only 253 geometry file analysis 251–255 high precision 254

make emvu file 252 memory save 252 network file 260–262 parameter type 254 quad precision 254 startup 285 use last data sets only 261 verbose 251, 261 RZERO 269, 299

S s100.geo 229 s25.geo 229 s50.geo 229 Save As dialog box 283 SC 150 Select Output Files dialog box 271 sense layer 223 separator 268 Set to Top Window command button 285 simple sweep 248 default 285 single 265 single precision 23 slot line mode 203, 206 SNP 144, 293 S-Parameters 7 generalized 55 SPARCstation 24 speed 24 SPICE 150, 187 command button 269 dialog box 269 option 290 output file 272 parameters 268 CMIN 269 KMIN 269 355

Em User’s Manual LMAX 269 RMAX 269 RZERO 269 spiral inductors 165 STEP 154, 265, 299 steps.geo 192 steps_sy.geo 22 stripline benchmark 227 subdivision 112 subs/lambda 31, 303 subsectional vias 166 subsectioning 24, 233 cell size 27, 35 frequency 270 subsections 27, 31, 291 Manhattan polygons 34 maximum 25 minimum 25 Non-Manhattan polygons 36 of polygon 32 reduce number 24 XMIN 36 YMIN 34 Super-Compact 289 superconductor 44, 303 surface reactance 43, 44 surface resistance 42 swapping 24 SWEEP 154, 265, 298 symmetry 17, 21, 206, 243

T T attenuator 97, 102 with ungrounded-internal ports 107 tane.geo 225 tangential electric field 223 TEM modes 88 theory 2–4, 8 356

thick metal 58, 209 thickness metal 42, 234 thin film resistor 41, 90 two connected 91 thkstep.geo 211 thkthru.geo 210 3-D 165 THZ 298 timing 237 TLIN 143 TLINP 144 TLM 173, 177 top cover 220, 222 resistive 227 Touchstone 150 Transmission Line Matrix technique 173, 177 transmission lines 89, 143 N-coupled 188, 296 structures 102 triangle subsection 163–164 tripat.geo 183 two-port circuit 91 two-port T attenuator 97

U ungrounded-internal ports 103, 107 units frequency 249, 286 use last data sets only option 261 user guided subdivision 112 user interface 15

V VAR 136, 140 verbose option 251, 261, 290 via 165, 165, 302, 304 cell size 166

Index

EM

conical 168 edge 166 example 166 ground 168 inside dielectric bricks 174 level-to-level 165, 203 loss 166 posts 166 precision problem 235 restrictions 166 subsectional 166 via ports 52 via.geo 166 viaports.geo 52

W The "Wall" 240 warning messages 305–307 warnings 108, 128, 248, 259 waveguide simulator 178 wire bonds 165

X -x 290 xgeom 22 Free Space 220 geometry file 6 menu bar 11 XMAX 37, 303 XMIN 34, 36, 37, 303

Y YMAX 37, 303 YMIN 34, 36, 37, 303

Z Z current 165 Z-Partitions 174 357

Em User’s Manual

358

Sonnet Application Notes

Sonnet Application Notes

Table of Contents

Table of Contents SAN-101A:

Precise Electromagnetic Analysis of Lange Couplers Using em . . . . . . . 1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Previous Modeling Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Thick Metal Approximation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Simplified Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

SAN-102A:

Comparison of Analysis Strategies for Diagonal Structures in em . . . . 11 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Initial Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Rotating the Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Analyzing Half of the Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

SAN-104B:

Generating PSpice Files Using Electromagnetic Analysis . . . . . . . . . . 19 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Class of Problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Using The SPICE Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 An Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 dxlv

APP NOTES

Implementation and Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Sonnet Application Notes Other Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

dxlvi

SAN-101A: Precise Electromagnetic Analysis of Lange Couplers Using em

Precise Electromagnetic Analysis of Lange Couplers Using em

APP NOTES

SAN-101A:

Summary This application note describes a technique to accurately characterize the coupling characteristics of the Lange coupler in em. The technique may be extended to any problem set including coupled lines separated by distances on the order of the metal thickness.

Introduction One of the best known and popular planar power divider networks is the Lange coupler. This structure is used on many microwave and millimeterwave MMICs and hybrids.

547

::

Sonnet Application Notes Circuit theory predictions tend to characterize these structures as overcoupled. Planar electromagnetic analysis using zero metal thickness tends to underpredict the degree of coupling.

Thru S21 measured

em (thick metal)

measured

em (thick metal)

Coupled S21

Lange coupler data courtesy of the Lockheed Martin Electronics Laboratory, Syracuse, NY.

Lange couplers are used in a wide range of both hybrid and MMIC microwave and millimeterwave components for signal splitting and combining. Typically, Lange couplers are utilized to provide equal power with quadrature phase splitting. In power amplifier development, where multiple components are combined using an array of input and output couplers to achieve high power levels, the accuracy of the coupler performance is critical to maximizing overall output power and efficiency. Any unbalance in the performance of the Lange coupler will result in power dumping to the Lange terminating resistor, thereby resulting in lower than expected output power and efficiency.

548

SAN-101A: Precise Electromagnetic Analysis of Lange Couplers Using em This application note will show that em accurately predicts coupler performance through the use of a thick metal model for the coupled line section of the Lange coupler. This results in significantly improved probability of first pass success. The technique to be outlined below can be applied to any structure where metal thickness may impact circuit performance.

Previous Modeling Approach

APP NOTES

The figure below shows the frequency response for the direct and coupled port transmission characteristics of a 60 GHz Lange coupler fabricated on 2 mil GaAs. Neither the circuit theory nor the electromagnetic simulation accurately predict the Lange coupler performance.

Thru S21

Measured Data

Circuit Theory

Coupled S21 em (zero thickness)

S-parameter data of 60 GHz Lange coupler thru and coupled port response. Circuit theory, em analysis, and measured performance are shown.

549

::

Sonnet Application Notes

Thick Metal Approximation Tightly coupled transmission line structures typically have metal thicknesses on the order of the line spacings. This results in parasitic capacitances between parallel sides of the coupled lines which can substantially impact coupling. This is evident in the electromagnetic analysis response of the 60 GHz Lange coupler with zero thickness metallization, which predicts significantly under coupled performance for the coupled port. In order to more accurately predict the performance of the Lange coupler in the electromagnetic simulation, a thick metal model for the coupled line section has been developed. This is effected by creating a second metallization pattern identical to the first, spacing the two structures to accurately reflect the desired metal thickness, and connecting the two structures with vias. For metallization where significant etchback occurs, slanted walls can be implemented to more accurately predict the results of the impact of metal thickness on coupling. The figure below shows a simplified cross-section model of how to approximate nonvertical wall metallizations.

<=>

<=> Simplified representation of the implementation of non-vertical metallization faces in xgeom. The left side figures show the metallization geometries involved in the approximation of the metallization cross-sections shown in the right side figures.

As stated above, the metal thickness is modeled by spacing the two metallization patterns appropriately. This is accomplished by incorporating an additional dielectric layer between the substrate material and the open region (air) above it. This additional dielectric layer is modeled simply as an air layer with the thickness

550

SAN-101A: Precise Electromagnetic Analysis of Lange Couplers Using em of the metal being modeled. The figure below shows the xgeom dielectrics window from the 60 GHz Lange coupler geometry file showing the additional dielectric layer used to simulate the metal thickness.

Additional 3.4 µm thick dielectric layer to allow simulation of thick metallization structures.

APP NOTES

Xgeom dielectrics window from the .geo file of the 60 GHz Lange coupler.

551

::

Sonnet Application Notes

Implementation and Results The figure below shows a close-up view of the coupled line section of a Lange coupler with the thick metal model implemented. In this figure, plated metal is modeled on both layers, and airbridge metal is used on the top layer only.

airbridge metal

plated metal

Detail of geometry file of Lange coupler thick metal implementation. Vias connect between layers separated by thickness to be modeled.

The edge vias connecting the two metallization patterns are clearly evident. In this case, vias at the ends of the thick metal transmission line structures are sufficient for accurate modeling. (See Chapter 18, “Thick Metal with Arbitrary CrossSection” in the Em User’s Manual for details on the correct use of vias in modeling thick metal.) The feed structure on the four ports of the Lange coupler are not implemented using the thick metal model. This simplification does not add appreciable error to the modeled results, and reduces the analysis time.

552

SAN-101A: Precise Electromagnetic Analysis of Lange Couplers Using em The plot below shows a performance comparison between the measured thru and coupled port transmission characteristics and the predicted result from the electromagnetic analysis run with the thick metal model. Improved correlation between predicted and measured results are shown.

Thru S21 em (thick metal) measured

APP NOTES

measured em (thick metal)

Coupled S21 Measured and em data for 60 GHz Lange coupler thru and coupled port data.

Simplified Analysis The previous work results in an excellent level of accuracy in the analysis of Lange couplers as is required for final verification prior to a design release. However, the time required to perform the analysis is commensurate with the level of accuracy required. This section details a procedure for simplifying the Lange coupler analysis, providing a high level of accuracy in a correspondingly reduced time span. This approach is amenable to the preliminary design phase, where optimization of design parameters (i.e. - line width and spacing as well as coupled line section length) can occur quickly. Final analysis in the mode discussed previously would be performed following the preliminary design phase.

553

::

Sonnet Application Notes The following elements were used in the simplification of the analyses: 1) untwisting the coupled lines, 2) via removal, 3) port renumbering, 4) box size and metal cell count minimization, 5) cal standard modification, 6) parallel subsection removal, and 7) invocation of the -E edge-mesh option. The figure below presents the xgeom .geo file of a coupled line section similar to that used in the analysis of the 60 GHz Lange coupler.

Xgeom .geo file of a four coupled line structure used in the simplified em analysis of a Lange coupler with metal thickness. Note the port renumbering and lack of vias to speed the analysis.

Element one above utilizes the inverse of the approach J. Lange [1] used in the development of the Lange coupler. See the Sonnet User’s Manual or contact Sonnet Software, Inc. for further information concerning elements two through seven listed above. Table 1 presents the relative performance of each of the approaches: 1) full Lange coupler analysis using a zero thickness metal model, 2) full Lange coupler analysis utilizing the thick metal model, and 3) the simplified coupled line structure. Thru and coupled port magnitude and phase at a single frequency (60 GHz) and analysis time per frequency data is presented to provide a trade-off matrix between speed and accuracy. This analysis work was performed using em Version 4.0 and timing performed on a Sun Microsystems SPARCstation 20. Relative analysis times running under Version 5.1 or above on Windows should be similar or faster.

554

SAN-101A: Precise Electromagnetic Analysis of Lange Couplers Using em

Summary A highly accurate method to model the performance of a Lange coupler through the use of a thick metal model has been presented. Methods to reduce the analysis time with the goal of allowing efficient optimization of critical Lange coupler parameters have also been outlined. This approach to modeling thick transmission lines, where parasitic coupling along the vertical face of closely spaced coupled line sections may impact performance, is directly applicable to other structures such as edge-coupled bandpass filters, interdigital capacitors and edge-coupled DC blocks. APP NOTES

References [1]

1. J. Lange, “Integrated Stripline Quadrature Hybrids,” IEEE Transactions on Microwave Theory and Techniques, December 1969.

Table 1: Comparison of em Modeling Approaches for 60 GHz Lange Couplers

Coupler Type

Relative Phase (o)

Thru S21 (dB) @60 GHz

Coupled S21 (dB) @60 GHz

Analysis Time per Frequency (min:sec)

1) Full Lange coupler, zero thickness metal model

92.1

-3.13

-4.24

21:16

2) Full Lange coupler, thick metal model

89.9

-3.25

-3.61

31:29

3) Coupled line section only

91.0

-3.12

-3.66

0:51

4) Measured data

90.1

-3.21

-3.44

-

555

::

Sonnet Application Notes

556

SAN-102A: Comparison of Analysis Strategies for Diagonal Structures in em

Comparison of Analysis Strategies for Diagonal Structures in em

APP NOTES

SAN-102A:

Summary The em electromagnetic analysis works best with rectangular (Manhattan) structures since such structures are easily subsectioned. Diagonal (nonManhattan) structures are more difficult to analyze, requiring additional subsections or the inclusion of diagonal subsections. Both increase analysis time and memory requirements. In this application note, we explore analysis techniques for a diagonally oriented 35 GHz band pass filter. Analysis is performed on the filter as originally drawn and results are compared to measured data. Then, several different approaches to analyzing the filter seeking to reduce analysis time are evaluated. Timing and accuracy comparisons between these, the baseline and measured results are made.

557

::

Sonnet Application Notes It will be shown that modifying the geometry to make use of the inherent Manhattan nature of the structure results in significantly reduced analysis times and memory (RAM) requirements with minimal degradation in the resulting accuracy.

Introduction The em electromagnetic analysis subsections structures created on an underlying, user-defined rectangular grid. Structures which have rectangular edges, otherwise known as Manhattan, are most easily, quickly and accurately analyzed. Diagonal, or non-Manhattan, structures, however, are not as easily analyzed. Either diagonal subsections must be added to allow current to flow diagonally along a polygon edge or a smaller grid/cell size used to improve the edge definition. Either results in increased analysis time and hardware requirements (RAM use). To explore the impact of various analysis approaches on diagonal structure analysis time and accuracy, we will analyze a 35 GHz filter fabricated on 100 micron GaAs. Complete dimensional data on the filter is provided in [1]. Measured data and results from other software packages have been published [2], making the filter a good candidate for validation and comparison. The work in [1] and [2] was reported by R. Furlow of Boeing.

558

SAN-102A: Comparison of Analysis Strategies for Diagonal Structures in em Different approaches used to analyze this structure with em range in analysis time from 8 seconds to 6 minutes per frequency. All analyses were performed on a 400 MHz Pentium II processor running Sonnet Release 6.0. Loss and de-embedding are included in the analysis.

Initial Analysis

APP NOTES

The xgeom layout of the baseline filter is shown in the figure below. Cell size was chosen to allow diagonal subsections to exactly follow the long diagonal edges of the filter. However, the ends of the filter are now only approximated, as can be seen with close inspection of the circuit as shown in the magnified section. This shortens the subsectioned metal at each end of the resonator by about 1/2 cell, so the drawing of the resonator was lengthened by 1/2 cell at each end as compensation.

The 35 GHz bandpass filter [1] layout as shown in xgeom.

The plot below shows the result of the em analysis compared with measurement and with circuit theory analysis. Analysis time was 6 minutes per frequency and required 66 MB of RAM to analyze. 559

::

Sonnet Application Notes The error in the em analysis is primarily in the center frequency while the circuit theory analysis is in error in both frequency and response shape. The circuit theory also predicts incorrect loss. The error in the center frequency in the em data is approximately 1%.Unfortunately, the cost of this 1% accuracy is 6 minutes of analysis time per frequency point and the need for 66 MB of RAM. Can we do any better?

0.3

Measured Em Analysis (1%)

0.2 S21 0.1

Circuit Theory (5%)

0.0 0

10

20

30

40

50

F(GHz) Em, circuit theory and measured results [2] for the 35 GHz bandpass filter.

Rotating the Filter Rotating the filter so that the resonator sections lie horizontal to the y-axis offers several advantages in an em analysis. First, modeling of the resonator becomes more accurate as they can be located exactly on the underlying grid. Also, since only the small elements associated with the input and output fees require diagonal subsections, RAM and analysis time requirements ought to decrease.

560

SAN-102A: Comparison of Analysis Strategies for Diagonal Structures in em The figure below shows the xgeom layout of the filter rotated by 30 degrees so that the filter section is horizontal. The filter resonator sections are now orthogonal whereas the feed lines are on a diagonal.

APP NOTES

Xgeom layout of the 35 GHz bandpass filter rotated 30o so that the filter resonators are horizontal.

561

::

Sonnet Application Notes The plot below shows the analysis results for this structure. We see improved agreement between the measured and modeled data. Analysis time reduced to 1 minute 24 seconds per frequency and the analysis required 31 MB of RAM to solve. Simply by changing the way we looked at the circuit, we decreased analysis time by 4X and RAM requirements by 2X.

1.0 0.8 Measured 0.6 |S21|

em (1%)

0.4 Circuit Theory (5 %)

0.2 0.0

20

25

30 Frequency (GHz)

35

The results of the em and circuit theory analysis compared to measured data. The rotated filter analysis shows very good correlation to the measured result.

562

40

SAN-102A: Comparison of Analysis Strategies for Diagonal Structures in em

Analyzing Half of the Filter A review of the original and the Manhattan filter shows that they are symmetric about a vertical center line. This means we can divide the filter in half, as shown below and analyze it, then cascade the results together using a Network file analysis to produce the full filter response.

APP NOTES

The geometry file for the half filter.

Cutting the number of subsections by half results in an analysis time of 39 seconds per frequency and requires only 22 MB of RAM. This is a 2X reduction in analysis time per frequency and a 1.4X reduction in required RAM while producing almost identical S-parameter data. An added benefit of analyzing the filter in pieces is that since we now have a broadband (non-resonant) structure, we can analyze at fewer frequencies....then interpolate to obtain the greater number of frequencies required to generate the complete filter response. This effectively increased the savings in analysis time, also. In this case, we analyzed every 5 GHz, 1/5 the required number of frequencies for the full filter. Adding in the interpolation effect, analyzing half the filter decreased the analysis time by 10X. This works out to be an equivalent time of about 8 seconds per frequency.

Conclusion We have demonstrated several strategies for analysis of a diagonal structure using Sonnet. Analysis times ranged from 8 seconds to 6 minutes, depending on the strategy used and the accuracy required. 563

::

Sonnet Application Notes

References [1] I. Bahl, “MIC Simulation Column,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol.1, No. 4, pp. 412-419, October 1991. [2] I. Bahl, “MIC Simulation Column,” International Journal of Microwave and Millimeter-Wave Computer-Aided Engineering, Vol.2, No. 2, pp. 116-130, April 1992.

564

SAN-104B: Generating PSpice Files Using Electromagnetic Analysis

Generating PSpice Files Using Electromagnetic Analysis

APP NOTES

SAN-104B:

Summary This application note describes how to use em to automatically derive SPICE models from a full wave electromagnetic analysis. The model generated is compatible with OrCAD Pspice1 and OrCAD PSpice A/D as well as other popular time domain circuit simulation packages. This capability is useful for circuits which are small with respect to the wavelength of the highest frequency of interest. This includes structures such as discontinuities like step, tee, and cross junctions. The primary application is expected to be the generation of SPICE models to predict cross-talk and propagation delay in high-speed digital interconnects. Use of this command option causes em to automatically take the results of the electromagnetic analysis of a circuit and synthesize a lumped element equivalent model of inductors, capacitors, resistors, and mutual inductors. This information is then formatted into an ASCII SPICE ".subckt" definition ready for inclusion in a SPICE input file. This application note provides the basic elements of an em spice analysis through some simple directions and an example. 565

::

Sonnet Application Notes

Introduction Em includes an option which synthesizes SPICE models2,3. This allows a given circuit layout to be analyzed electromagnetically with the result used to synthesize a SPICE lumped element model. The SPICE model consists of resistors, inductors, capacitors, and mutual inductors and is written to disc in ASCII format. The sub-circuit can then be incorporated in a complete circuit and analyzed using, for example, PSpice or PSpice A/D, to obtain time domain responses. The primary use of this capability is intended to be cross-talk analysis of high speed digital circuits. The examples in this note are compatible with Pspice but are general enough to be used in other SPICE type simulators after some modifications.

Class of Problems The SPICE generation capability is intended for any circuit which is small with respect to the wavelength of the highest frequency of excitation. Typically, 1/10th wavelength is an appropriate limit. If a circuit is too large, split it into two or more circuits and analyze each circuit separately. The model generated by the analysis includes any lumped elements (including mutual inductors) between any ports of the circuit layout. Lumped elements from any port to ground are also included. The synthesis capability does not allow internal nodes (i.e., nodes in a circuit which are not connected to a port in the layout; series RL, LC, and RC are considered single lumped elements). Any circuit which requires internal nodes for an accurate model should have the appropriate point specified as a port. All ports must have a ground reference. The SPICE file generation capability is usually not appropriate for microwave circuits as such circuits are usually larger than a small fraction of a wavelength. However, there are exceptions. For example, a de-embedded step discontinuity has zero physical size and can be used with this capability. The SPICE model synthesis capability is fast enough that it can be used on circuits of up to several hundred ports.

566

SAN-104B: Generating PSpice Files Using Electromagnetic Analysis

Using The SPICE Option The spice synthesis needs electromagnetic results at two frequencies to create a lumped equivalent model. The user must select two frequencies for analysis and specify them either through a simple sweep or referring to an analysis control file in the Frequency Control section of the em Control window. (See the Sonnet Em User’s Manual for details if needed). Selection of the proper frequencies to use can be somewhat difficult for certain structures so here are a few guidelines to help with the selection: Select the two frequencies so that they are separated by at least 10%. Also, make sure they do not exceed (or even come close to) the highest useful frequency, where the circuit has become a sizable fraction of a wavelength.

•

Make sure the frequency is not too low. When the cell size is less than 0.00001 wavelength, numerical precision can be a problem. For example, if the cell size (or vertical via length) is 1 mm, it would be unwise to analyze below 1 MHz. If you are approaching the lower frequency limit, or want to test for a numerical problem, you can use the Memory Saver option in em Control to turn on single precision (double precision is default) and see if the result changes.

•

After completing the analysis, always do a “reality check” for reasonable values. If you have bad data, one of the above problems is likely to be at fault. To be absolutely sure your results are good, select a second pair of frequencies, different from the first pair by, say, a factor of two, and re-analyze the circuit. You should obtain almost the same answer.

APP NOTES

•

To create a SPICE model from an em analysis, enter “-pspice” in the Advanced Text entry box in the Additional Options dialog box of the em Control interface. Here “Number” is the number of significant digits printed for the values of the components in the final lumped model. For example specifying “-pspice4” will produce components in the output file such as, C_C1

1

2

2.534pf

567

::

Sonnet Application Notes Capacitors and inductors that would be printed out as zero at the specified precision are excluded from the listing. The option “-pspice” is equivalent to “pspice6”. Output is sent to a “.psp” file with the same prefix. For example, the input file is “pspice.geo” so the spice output file is “pspice.psp.”

An Example The example circuit is shown in the figure below. The circuit layout can be imported from GDSII, DXF, Series 4, or entered directly using xgeom. The ports are then specified and will translate to nodes in the lumped element netlist. This circuit is a simple example of three coupled printed circuit traces connected to three terminals of some pc-board mounted device. It is expected that there will be significant crosstalk between the lines and added propagation delay due to the discontinuities.

Geometry of example circuit under investigation shown using Sonnet xgeom . This circuit is on a 25 mil thick substrate of relative dielectric constant 9.8. em simulation of this circuit will produce scattering parameters as well as a complete spice compatible subcircuit.

568

SAN-104B: Generating PSpice Files Using Electromagnetic Analysis This example geometry is in the Sonnet example directory (pspice.geo). For this example, the two frequencies chosen were 100 and 125 MHz. You may specify these frequencies either by using the analysis control file, “pspice.an” or by entering a two point sweep in the Frequency Control section of the em Control window. Use of the analysis control file is shown below.

APP NOTES

The em Control window showing the use of the analysis control file.

Remember that you must click on Additional Options command button and enter “-pspice” in the Advanced Options text entry box in the dialog box which appears on your display.

569

::

Sonnet Application Notes The following is the resulting SPICE model in ".subckt" form (... indicates information left out): * Limits: C>0.1 pF, L<100 nH, R<1000 Ohms, K>0.01. * Analysis frequencies: 100.000000, 125.000000 MHz .subckt pspice 1 2 3 4 5 6 GND C_C1 1 4 2.73473pf C_C2 1 6 1.39261pf C_C3 2 GND 0.239061pf C_C4 2 3 2.72413pf C_C5 2 5 1.38228pf C_C6 3 GND 0.197708pf C_C7 3 5 0.552376pf C_C8 4 GND 0.285267pf C_C9 4 6 0.509081pf C_C10 5 GND 0.196357pf C_C11 6 GND 0.287491pf L_L1 1 2 4.93059nh L_L2 3 4 5.57302nh L_L3 5 6 5.55392nh Kn_K1 L_L1 L_L2 0.354279 Kn_K2 L_L1 L_L3 0.353451 Kn_K3 L_L2 L_L3 0.212225 .ends pspice

The file containing this model can be simply included in the spice circuit file with a ".INC" statement and then connected as needed into the spice netlist. Alternatively, a symbol can be created for this model and used in a schematic capture program such as OrCAD Schematics1. Schematics contains a Symbol Creation Wizard feature which allows symbols to be created quickly from existing models. For smaller models, there is an option to enter the components individually using the schematic entry method. This was done for this example and is shown in the figure below.

570

SAN-104B: Generating PSpice Files Using Electromagnetic Analysis

APP NOTES

Inductor Coupling Block

Netlist generated for half of the circuit in the first figure. Only the circuit for ports 1,2,3,4 is shown here. Note the inductor coupling block highlighted in the diagram. This coupling along with all other parasitics are computed automatically by em. The circuit is displayed using OrCAD Schematics.

571

::

Sonnet Application Notes The circuit shown above is a complete circuit ready for analysis by PSpice or PSpice A/D. Only the generator, which is a digital clock for this example, and the port terminations needed to be added to the circuit generated by em. The generator was connected to port 3 to check for crosstalk between port 3 and port 1. An example Pspice analysis is shown below.

Time domain response of the circuit showing the crosstalk between the lines 1-2 and 3-4. Nearly 2 volts is generated at port 1 from a 5 volt clock signal at port 3. The data is displayed using OrCAD Probe.

Other Techniques Classical techniques use, for example, just an electrostatic or just a magnetostatic analysis to derive a model. This is adequate for uniform transmission lines embedded in homogenous dielectric (no different layers). In an arbitrary predominantly planar circuit, as we have here, a single static analysis provides only half a circuit model, just the capacitors or just the inductors. Since em is a full dynamic analysis, both the inductive and capacitive portion of the model are obtained with one analysis (at two frequencies).

572

SAN-104B: Generating PSpice Files Using Electromagnetic Analysis In addition, the techniques usually used for the static analyses are of a volume griding variety (e.g., finite elements, finite difference). Even under the simplifications allowed by static analysis, circuits more complicated than shown in Figures 1 and 2 quickly go beyond the capability of such software tools. However, because em is a surface meshing analysis, it can do the circuit of Figure 1 in 1 minute on a 200 MHz Pentium using about 1 Mbyte of memory. The SPICE file is generated after analyzing two frequencies, or in about 2 minutes.

APP NOTES

Note that the experimental approach to modeling this circuit would involve building the circuit, measuring a 6 port structure, developing and entering an appropriate model in a circuit simulator, and optimizing each of the variables for a best fit. Such a task is time consuming and error prone. In contrast, by using the em analysis, the total, end-to-end time was about one hour. This includes the time required for manual circuit layout capture and inspection of the final results.

Conclusion We have shown how the SPICE model synthesis capability of em can be used to quickly derive lumped models of complex circuits, provided the circuits are small with respect to wavelength. The only input information required is the circuit layout. The lumped model is synthesized based on results of a complete electromagnetic analysis of the layout. The capability can be used on typical workstations to model circuits of up to several hundred ports, allowing timely generation of models of complex circuits which can not be accurately evaluated in any other way. [1] PSpice, Schematics, and Probe are products of OrCAD Corp. 9300 S.W. Nimbus Avenue, Beaverton, OR 97008-9625 [2] Sonnet User’s Manual, Sonnet Software, 1020 Seventh North Street, Suite 210, Liverpool, NY 13088 [3] J.C. Rautio, “Synthesis of Lumped Models from N-Port Scattering Parameter Data,” IEEE Tran. Microwave Theory Tech., Vol. 42, No. 3, March 1994, pp. 535-537.

573

::

Sonnet Application Notes

574