Theses 8 : Alejandro Nieto Gutiérrez
This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA
Overview
Download & View Theses 8 : Alejandro Nieto Gutiérrez as PDF for free.
More details
- Words: 31,980
- Pages: 186
Alejandro Nieto Gutiérrez
Hardware design of demand as frequency controlled reserve Master’s Thesis, January 2008
Alejandro Nieto Gutiérrez
Hardware design of demand as frequency controlled reserve Master’s Thesis, January 2008
Hardware design of demand as frequency controlled reserve
This report was drawn up by: Alejandro Nieto Gutiérrez Supervisor(s): Zhao Xu Tonny W. Rasmussen
Ørsted•DTU Centre for Electric Technology (CET) Technical University of Denmark Elektrovej Building 325 2800 Kgs. Lyngby Denmark www.oersted.dtu.dk/cet Tel: (+45) 45 25 35 00 Fax: (+45) 45 88 61 11 E-mail: [email protected]
Release date:
05-01-2008
Category:
1 (offentlig)
Edition:
1st edition
Comments:
This report is part of the requirements to achieve the Erasmus Final Thesis Project at the Technical University of Denmark. This report represents 35 ECTS points or at least 6 full-time months of dedication.
Rights:
© Alejandro Nieto Gutiérrez, 2008
HOME UNIVERSITY INFORMATION
Titulo: Hardware design of demand as frequency controlled reserve Prototipo de controlador de consumo de cargas según la frecuencia de la red Autor: Alejandro Nieto Gutiérrez Supervisores: Zhao Xu Tonny W. Rasmussen Censor:
Nota:
Universidad: DTU, Lyngby.
Departamento: Ørsted•DTU
Duración, carga de trabajo: 35 ETCS. Six months full-time project
Tipo: Master Thesis project, Proyecto Final de Carrera.
Fecha de inicio: 1 de Febrero de 2007 Fecha de finalizacion: 5 de Enero de 2008
i
Home University Information Resumen en Español: El proyecto consiste en el desarrollo de un medidor de frecuencia de la red eléctrica desde cero. La red eléctrica ante desequilibrios de la producción y consumo de energía puede reaccionar de maneras indeseadas. El controlador desarrollado en este proyecto es capaz de medir los cambios en la frecuencia de la red eléctrica y a partir de los datos obtenidos reaccionar desconectado la carga y así disminuyendo el consumo eléctrico. Las cargas propicias para estos propósitos son las cargas, y más específicamente, electrodomésticos, que pueden ser desconectados de la red eléctrica sin que sea apreciado por parte de los usuarios. Estos electrodomésticos pueden ser por ejemplo calentadores de agua eléctricos, frigoríficos o también bombas de calor o climatizadores. Según la carga a controlar cortes de unos segundos serán inapreciables por parte del usuario pero si por parte de la red eléctrica. El uso generalizado de esta tecnología podrá reducir costes en el mantenimiento de centrales eléctricas de reserva, que en estos momentos, permanecen a la espera de que se produzca un exceso de consumo para así empezar su producción. El mantenimiento de estas centrales es caro y podrían ser aprovechadas para la producción regular de energía. Otra ventaja de este controlador puede ser su capacidad de reacción ante producciones de energía irregulares como pueden ser las fuentes de energía renovables. Estas fuentes de energía pueden mermar su producción sin dejar suficiente tiempo a la red para recuperarse, mediante el uso de esta tecnología, se podrá asumir gran parte de estas inestabilidades dejando un mayor margen de tiempo para la recuperación de la red. En este documento se explica el proceso completo del desarrollo del controlador de cargas según la frecuencia. Está contemplado desde los inicios del desarrollo, con la selección de componentes, hasta la programación del código fuente del micro controlador, pasando por el desarrollo y diseño del hardware adicional utilizado para completar los propósitos.
Palabras clave: Microcontroladores, medidor de frecuencia, programación C, DSP, DSC, Code Composer Studio, filtro paso banda, fuente de alimentación, ADC, hardware, software, medidor de potencia RMS, Event Manager, Reserva eléctrica, equilibrio consumoproducción, Energías renovables, fuente de alimentación.
ii
ABSTRACT This document is the final report for “Hardware design of demand as frequency controlled reserve” project, carried out by Alejandro Nieto Gutiérrez, Erasmus student from Spain. The aim of this project is to develop a load controller, which is able to monitor the frequency and power of the power grid, from the wall outlet, and react with frequency changes, as well as prepare for recording the acquired data for subsequent studies. The energy Balance between consuming and producing units need to be maintained continuously, or otherwise the system will became unstable ultimately resulting in a total system collapse. This project will react with the power grid imbalances according to the frequency measured at the wall outlet. If necessary it will disconnect the load reducing the consumption, which will help to avoid a power grid imbalance as well as recover the power grid balance within instability. The introduction and the theoretical overview about concepts present in the project process are included into the first chapter. The hardware part of the project, design and development is described in the second chapter. Software development process, in instance algorithms used and programming of the application and other related matters, is explained in the third chapter. The last chapters are the future work chapter, which describes a recommended way to follow the development of the load controller project, the conclusions chapter, which gathers all the conclusions obtained once this load controller prototype is finish, the bibliography and finally the appendix, which contains all the additional information that is not included in neither of previous sections, like the tables of acquired data, as well as the full source code of application and the hardware schemes.
iii
ACKNOWLEDGES
I specially want to acknowledge the invaluable help and understanding given by my tutor and co-tutor, Zhao Xu and Tonny W. Rasmussen, without their help the accomplishment of this project would not be possible. I want to thank Knud Ole Helgesen Pedersen for the help given and also for teaching me the working of microcontrollers in the January 2007 course. I also want to thank everybody in DTU University and in Denmark in general, who made me feel like at home, in spite of being in other country. Finally I want to thank to my family for their support and my home university for given me this opportunity.
v
CONTENTS
1 Introduction..................................................................................................... 1 1.1. The Power grid.......................................................................................... 1 1.2. How the power grid can be helped ........................................................... 2 1.3. The aim and objectives of the project ....................................................... 3 1.4. Block Diagram of Load Controller ........................................................... 4 1.5. Frequency measurement overview ........................................................... 5 1.6. Frequency decay measurement overview ................................................. 5 1.7. RMS Current, voltage and power measurement overview ....................... 6 2 Algorithms, procedures and results. ............................................................. 9 2.1. Frequency measurement algorithm ........................................................... 9 2.1.1. Averaging measures ........................................................................ 16 2.1.2.
Maximum theoretical Accuracy...................................................... 18
2.1.3.
Soft frequency filter ........................................................................ 21
2.1.4.
Frequency sensor range .................................................................. 22
2.1.5.
Accuracy ......................................................................................... 23
2.2. Frequency decay measurement ............................................................... 24 2.3. Control logic Type I: Frequency Thresholds .......................................... 26 2.4. Control logic Type II: Temperature Set points change........................... 30 2.5. Voltage, current and Power measurement .............................................. 31 2.5.1. Sample acquisition .......................................................................... 31 2.5.2.
Acquired measures .......................................................................... 32
2.5.3.
RMS Current, voltage and power calculation ................................. 36
2.6.
Real-time stamp ...................................................................................... 42
3 Hardware ....................................................................................................... 45 3.1. DSC......................................................................................................... 45 3.1.1. DSC prerequisites ........................................................................... 46 3.1.2.
DSC Technical features .................................................................. 48
3.1.3.
DSC Hardware Description ............................................................ 50
vii
Contents 3.1.3.1. C28x CPU .................................................................................. 50 3.1.3.2. Memory Bus (Harvard Bus Architecture) .................................. 50 3.1.3.3. Real-Time JTAG and Analysis .................................................. 51 3.1.3.4. SARAMs .................................................................................... 51 3.1.3.5. Flash ........................................................................................... 53 3.1.3.6. Security....................................................................................... 53 3.1.3.7. Peripheral Interrupt Expansion (PIE) Block .............................. 54 3.1.3.8. Oscillator and PLL ..................................................................... 54 3.1.4.
DSC Periphery Description ............................................................. 55
3.1.4.1. Event Manager ........................................................................... 55 3.1.4.1.1. GP timers ............................................................................. 57 3.1.4.1.2. Capture unit ......................................................................... 57 3.1.4.1.3. Starting the A/D Converter with a Timer Event ................. 58 3.1.4.2. ADC ........................................................................................... 58 3.1.4.3. General purpose I/O ................................................................... 61 3.2. Board Description.................................................................................... 62 3.2.1. Signal interfacing and conditioning ................................................ 63 3.2.1.1. Measurement transformers ......................................................... 63 3.2.1.2. Analog filters .............................................................................. 64 3.2.1.3. Level shifter................................................................................ 66 3.2.2.
Current and voltage values at load controller inputs ....................... 69
3.2.3.
Power Supply .................................................................................. 71
3.2.4.
eZDSP F2812 .................................................................................. 73
4 Software.......................................................................................................... 75 4.1. IDE .......................................................................................................... 75 4.2. User Manual ............................................................................................ 76 4.2.1. Developing board connection .......................................................... 77
viii
4.2.2.
The project ....................................................................................... 78
4.2.3.
Configurations ................................................................................. 81
4.2.4.
Modify, build and load a project ..................................................... 82
4.2.5.
Debug project .................................................................................. 85
4.2.6.
Gel File ............................................................................................ 86
Contents 4.3. The load controller program ................................................................... 88 4.3.1. File structure ................................................................................... 88 4.3.2.
Memory mapping ............................................................................ 89
4.3.3.
Program structure ............................................................................ 92
4.3.3.1. Headers ...................................................................................... 93 4.3.3.1.1. Inclusions ............................................................................ 93 4.3.3.1.2. Constant definition and Location of attributes ................... 93 4.3.3.1.3. Functions definition ............................................................ 96 4.3.3.1.4. Variables definition ............................................................ 98 4.3.3.2. Device initialization ................................................................... 99 4.3.3.2.1. System Control Initialization ............................................ 100 4.3.3.2.2. Peripheral High speed clock configuration....................... 100 4.3.3.2.3. GPIO configuration .......................................................... 101 4.3.3.2.4. PIE initialization ............................................................... 101 4.3.3.2.5. ADC initialization............................................................. 102 4.3.3.2.6. EV configuration, timers and capture unit ........................ 104 4.3.3.2.7. Relay initialization ............................................................ 105 4.3.3.2.8. Enable CPU interruptions ................................................. 105 4.3.3.2.9. Average frequency buffer initialization ............................ 106 4.3.4.
Capture unit subroutine: Capture1_srt() ....................................... 107
4.3.5.
ADC interruption subroutine: ADC_srt() ..................................... 112
4.3.6.
Other subroutines .......................................................................... 114
5 functionalities and performance ................................................................ 117 6 Future work ................................................................................................. 119 6.1. Flash memory ....................................................................................... 119 6.2. Communication part ............................................................................. 119 6.3. Persistent storage .................................................................................. 120 7 Conclusions .................................................................................................. 121 8 Bibliography ................................................................................................ 123 A
Source code .............................................................................................. 125 A.1 Main program source code ................................................................... 125 A.2 Memory mapping .................................................................................. 139
ix
Contents B
Board Schemes ......................................................................................... 140
C
Other plots................................................................................................ 143
D
Tables........................................................................................................ 145 D.1 ADC Measurements .............................................................................. 145 D.2 Frequency measures .............................................................................. 148 D.3 Frequency decay .................................................................................... 156
x
LIST OF FIGURES Figure 1-I – Load controller Block diagram ..................................................................... 4 Figure 2-I - period of a signal ........................................................................................... 9 Figure 2-II - Schmitt-triggered output ............................................................................ 10 Figure 2-III - Frequency measurement block diagram ................................................... 11 Figure 2-IV - Overflow in frequency counting ............................................................... 12 Figure 2-V – Capture unit measurements ...................................................................... 14 Figure 2-VI - Instant Frequency measurements ............................................................. 15 Figure 2-VII - Error spread using average ...................................................................... 16 Figure 2-VIII - Averaged frequency measurements in integer values............................ 17 Figure 2-IX - Averaged frequency measurements in Hz ................................................ 17 Figure 2-X - Instant and averaged measurements .......................................................... 18 Figure 2-XI - frequency counting error .......................................................................... 19 Figure 2-XII - Software frequency filter block diagram................................................. 22 Figure 2-XIII - Frequency decay .................................................................................... 24 Figure 2-XIV – Frequency decay ................................................................................... 26 Figure 2-XV – Averaged frequency measurements vs frequency thresholds ................ 27 Figure 2-XVI Frequency decay Vs Frequency decay threshold ..................................... 28 Figure 2-XVII- Threshold checking flow diagram ......................................................... 29 Figure 2-XVIII - ADC curret measures .......................................................................... 32 Figure 2-XIX - ADC Voltage measures ......................................................................... 33 Figure 2-XX - Integer measures of current without offset ............................................. 33 Figure 2-XXI - Integer measures of voltage without offset............................................ 34 Figure 2-XXII - Measured Current ................................................................................. 34 Figure 2-XXIII - Measured Voltage ............................................................................... 35 Figure 2-XXIV - Instant power in integer units ............................................................. 35 Figure 2-XXV - Calculated instant Power...................................................................... 36
xi
List of Figures Figure 2-XXVI - Measured RMS current (int) ............................................................... 37 Figure 2-XXVII - Measured RMS Voltage (int) ............................................................. 38 Figure 2-XXVIII - Measured RMS power (int) .............................................................. 38 Figure 2-XXIX - Measured RMS current (A) ................................................................. 40 Figure 2-XXX - Measured RMS Voltage (V) ................................................................. 41 Figure 2-XXXI - Measured RMS power (W) ................................................................ 41 Figure 3-I - Texas instruments Logo ............................................................................... 49 Figure 3-II - DSC Memory map ...................................................................................... 52 Figure 3-III - Event Manager A Functional Block Diagram ........................................... 56 Figure 3-IV - ADC block diagram .................................................................................. 59 Figure 3-V – ADC example Voltage output ................................................................... 60 Figure 3-VI - Board distribution ..................................................................................... 62 Figure 3-VII - Measuring Transformers .......................................................................... 63 Figure 3-VIII - Filter transfer function ............................................................................ 65 Figure 3-IX – Filter photo ............................................................................................... 65 Figure 3-X- Filter Scheme............................................................................................... 66 Figure 3-XI - Voltage level Shifter Scheme .................................................................... 67 Figure 3-XII - Current Level Shifter Scheme ................................................................. 68 Figure 3-XIII - Level shifter result .................................................................................. 69 Figure 3-XIV - Single power supply scheme .................................................................. 71 Figure 3-XV - Measure at the bridge output, point C (Without capacitors) ................... 72 Figure 3-XVI – Power supply output .............................................................................. 72 Figure 3-XVII - Double Power Supply scheme .............................................................. 73 Figure 3-XVIII - eZdsp F2812 photo .............................................................................. 74 Figure 4-I - Code Composer Studio main window ......................................................... 76 Figure 4-II – Connect ...................................................................................................... 77 Figure 4-III - Board connected message ......................................................................... 77 Figure 4-IV - Open existing project ................................................................................ 78 Figure 4-V - Open Project dialog .................................................................................... 79 Figure 4-VI - Opened project .......................................................................................... 79 Figure 4-VII - Main file source Code .............................................................................. 80 Figure 4-VIII - Linker configuration ............................................................................... 81 Figure 4-IX – Select Configurationt ................................................................................ 81
xii
Contents Figure 4-X - Build a project ............................................................................................ 82 Figure 4-XI - Build report............................................................................................... 82 Figure 4-XII - Load Program .......................................................................................... 83 Figure 4-XIII - Load project output ................................................................................ 83 Figure 4-XIV - Reload Project........................................................................................ 84 Figure 4-XVRun a program ............................................................................................ 84 Figure 4-XVI - watch window ........................................................................................ 85 Figure 4-XVII - Break point ........................................................................................... 86 Figure 4-XVIII - GEL file .............................................................................................. 86 Figure 4-XIX - GEL Menu ............................................................................................. 87 Figure 4-XX - Example code of GEL language ............................................................. 87 Figure 4-XXI - Simplified program flow ....................................................................... 92 Figure 8-I - Board Scheme............................................................................................ 141 Figure 8-II - DSC connectors Detail ............................................................................. 142 Figure 8-III – Averaged frequency measurements (Integer units) ............................... 143 Figure 8-IV – Averaged frequency measurements (Hz)............................................... 144
xiii
LIST OF TABLES
Table 7-I – ADC samples table..................................................................................... 145 Table 7-II - Instant frequency measurements ............................................................... 148 Table 7-III - Averaged frequency measurements ......................................................... 149 Table 7-IV - RMS measurements ................................................................................. 154 Table 7-V - Frequency decay........................................................................................ 156
xv
LIST OF SYMBOLS Symbol
Unit
Integer
Output integer value of ADC
Integer
First averaged measure in a pair of period measurements
Integer
Second averaged measure in a pair of period measurements
Hz
_
!
Frequency decay
N/A
Factor that converts integer current measurements in S.I units
N/A
Factor that converts integer power measurements in S.I units
Hz
Frequency at power grid
N/A
Factor that converts integer measurements in S.I units
V/V
Total current gain in current channel
V/V
Current gain in current level shifter
V/V
Current gain in current transformer
V/V
Total voltage gain in voltage channel
V/V
Voltage gain in voltage level shifter
V/V
Voltage gain in voltage transformer
A
Current consumed by the load
A
Current present at ADC input
Integer N/A
"#
%$Integer
!
&_ '
W !
Frequency of capture unit counter
Hz/s
N
&
Definition
RMS Current in integer units at ADC Number of event manager cycles within a power grid cycle or number of ADC measures used in RMS calculations Integer value of offset sensed by ADC Power grid power in RMS
Integer
RMS power in integer units at ADC
Ω
Resistance of ADC current resistor
xvii
List of Symbols Symbol
Unit
% %
% % )
)
)
!
)_ ∆ ∆(
∆%
xviii
S
First period value in a pair of period measurements
S
Second period value in a pair of period measurements
Integer
(
!
Definition
Period measured by capture unit
S
Time of a time base count unit
S
Period at power grid
V
Voltage at load controller input
V
Voltage value sensed by ADC in Volts
V
Power grid Voltage in RMS
Integer
RMS voltage in integer units at ADC
Hz
Increment of frequency in two following frequency measures
S
Increment of time in two following frequency measures
S
Relative period error in frequency measurements
ACRONYMS
# µC
Microcontroller
A ADC
Analog to Digital Converter
C CPU
Central Processing Unit
CSM
Code Security Module
CUPS
Capture Unit input clock PreScaler
D dB
Decibel
DC
Direct Current
DSC
Digital Signal Controller
DSP
Digital Signal Processor
E EV
EVent manager
EVA
EVent manager A
EVB
EVent manager B
F FCF
Frequency Conversion Factor
G GP
General Purpose
GPIO
General Purpose Input/Output
xix
Acronyms
I IDE
Integrated Development Environment
J JTAG
Joint Test Action Group
M MAC
Multiplication And aCcumulation
P PC
Personal Computer
PIE
Peripheral Interrupt Expansion
PLL
Phase Locked Loop
R RAM
Random Access Memory
RMS
Root Mean Square
ROM
Read Only Memory
S SI
Le Système International d'unités (International System of units)
SRAM
Single-access Random Access Memory
U UART
Universal Asynchronous Receiver-Transmitter
xx
1
INTRODUCTION
1.1. The Power grid Electrical systems in the form of power grids are one of the most efficient ways of transferring energy over large distances directly to the consumers, hereby limiting the need for production of energy in the closeness of consumers. Electrical power in the quantities that modern countries consume cannot be stored in any currently available storage solution, and can therefore not be used as a solution to preserve the power grid. The energy Balance between consuming and producing units need to be maintained continuously, or otherwise the system will became unstable ultimately resulting in a total system collapse. The responsibility of maintaining the system balance is placed with the system operators. These operators have to monitor the power system continuously and the most effective way of doing so, is to measure the frequency. When the production and consumption of active electrical power is in balance the frequency will be 50Hz. The reason being, that electrical energy taken out of the system matches the mechanical power grid into the system, in fact in this state the rotational speed of generators will remain fixed. As the system starts to lose synchronization, the rotations speed of the synchronous generators will begin to change, which make us know that the balance is beginning to degrade by changing the frequency of the grid. In instance of frequency rise, it indicates that the production starts to exceed the consumption. The generators will begin to rotate faster because of the excess energy delivered from the mechanical side of the generation unit. If the frequency drops below 50Hz then the consumption exceeds the production.
1
Introduction The electrical systems are thought up to solve this last problem by activating the reserve systems that will inject or consume energy until the consumption-generation balance is recover.
1.2. How the power grid can be helped As explained in previous section when the consumption exceeds the production the frequency starts to drop. Although the electrical system is able to solve this problem it is possible to help to solve it from the consumer side, reducing the need of reserve systems. By doing so, many benefits can be achieved, low costs, the possibility to use the reserve system as primary production and also energy savings while de loads are disconnected. Most importantly, it is to facilitate increased integration of fluctuating renewable energy, in example wind generators, into the power system in the future. The renewable energy sources have an unstable production. Therefore an unexpected drop of power production in this kind of sources can also cause the excess of consumption. The easier possibility to introduce renewable energy sources can reduce the contaminant emissions can be reduced. The consumers have at home several loads that can be switched on and off without noticing it, these loads are named FCLs. A control device that can be able to sense the power grid status and under frequency changes could react will help to preserve the power grid integrity, for example by switching off this FCLs in an excess of consumption, or increasing the thermostat temperature in the particular cases of a fridge or decreasing it in a water heater.
2
Introduction
1.3. The aim and objectives of the project The objective of this project is to provide a fast-responding control device that reacts in response to frequency events. It also has to be able to measure the Root Mean Square (RMS) power, voltage and current. The controller must sense the local grid frequency at the wall outlet and deenergize the load, based in two control logics. The load will be tripped if the frequency goes below a specified threshold, or if the rate of frequency decay is lower than a user specified threshold for the logic Type I. Tripping criteria: Maximum delay 0,5s (in can be reduced by compromising the frequency accuracy) C1: Low frequency criteria: <49.90 Hz and C2: Fast frequency decay criteria <-0.13 Hz/sec Reconnection criteria: R1: Tripped by C1, frequency recovers to a margin above tripping set point (49.98 Hz) and C2 not met R2: Tripped by C2, C2 not met and after a time delay R3: Both subject to a random time lag 0~30 Secs It is possible to change the values of C1, C2, R1, and R3
The control logic type II is specially designed for those thermostically controlled loads with cyclic on/off durations. In this type, instead of tripping the loads according to the frequency directly, their temperature set points are dynamically controlled according to the frequency For type II, the set points of the appliance (fridge or air-con) can be normal T high = T high − kf * ( f − f 0 )
(1.1)
normal Tlow = Tlow − kf * ( f − f 0 )
This control logic is not completely development in this load controller prototype due to the miss of the interface that can change the thermostat temperature of a FLC of this type. From this point a frequency event is understood as the frequency change that fulfills any of the control logics described above.
3
Introduction
The load is re-energized after the recovery criteria fulfill, utilizing a random waiting period between a selectable amount of time (30 seconds as default) providing a smooth transition for the grid. The random time lag prevents all loads from turning back on all at the very same time after a frequency event, which could cause a rebound effect, potentially tripping other overload relays on transmission or generation equipment
1.4. Block Diagram of Load Controller
Power Outlet
Voltage channel
Current channel
ADC&EM
Voltage Divider
DSC
Transimpedance amplifier
I/O
LOAD
Relay
Figure 1-I – Load controller Block diagram The block diagram above, (Figure 1-I – Load controller Block diagram), describes the basic operation of the control device. The load represents the controlled device such as a fridge, a water heater or any other FCL appliance. The relay represents can energize or de-energize the device according to control logics, in the case that the power grid is under a frequency event. It can also be a control module able to change the thermostat temperature or any other action that can reduce the consumption of the controlled load. The voltage divider or voltage transformer reduces the voltage of the power grid to a range that can be sensed by the Analog to Digital Converter (ADC) integrated on the Digital Signal Controller (DSC) chip.
4
Introduction The transimpedance transformer outputs a voltage proportional to the input current. It is also connected with the ADC input which allows measuring the current and consequently the power of consumed by the controller device. The transimpedance transformer is compound by a current transformer and a know resistor that drops a know amount of voltage according to the input current. The output signal of both transformers is conditioned before being connected with the ADC inputs; the conditioning hardware is included in the transformers blocks in order to simplify the diagram.
1.5. Frequency measurement overview The frequency measurement method used is the zero-crossing detection. This method measures the time between zero-crossings, which is the period of the power grid signal. The frequency can be directly obtained madding use of the period, This algorithm is explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement. The frequency measurement is used to implement the control logic Type I, which is based on the frequency changes.
1.6. Frequency decay measurement overview The frequency decay measurement is realized using the frequency measurements of consecutive cycles. It is calculated the decay between selectable amounts of pairs of measures in a row. Immediately after it, is calculated the average of these frequency decays and finally it is compared with the frequency decay threshold. In the case that the threshold is surpassed, the control device proceeds to start the frequency event response. More details about the algorithm used are covered in section 2.2 Frequency decay measurement. This type of measurement is used to implements the control logic Type II, which is based on the frequency decay changes.
5
Introduction
1.7. RMS Current, voltage and power measurement overview The instant voltage, current and power varies in the time from negative to positive values. Current, voltage and power should be expressed in Root Mean Square (RMS) values. The RMS measurements can contribute in the subsequent studies more than instant values. A microcontroller will acquire only discrete values, therefore in the RMS calculations it is used a discrete method. RMS measurements can be calculated for a series of discrete values varying function. The name, RMS, comes from the fact that it is the square root of the mean of the squares of the values. It is a power mean with the power p = 2. The RMS for a collection of n values is:
(1.2)
In the project, it is important to monitor all the time the power consumption of the FCL appliances, in order to know exactly the state of the load when a frequency event happens . In addition, the accuracy of the power measures is not as important as the time they are made. For example it is more important to have a power measure with accurate time stamp than the accurate power measurement itself. The RMS values are obtained from the ADC samples. The ADC has to sample the channels at a fast speed to have sufficient accuracy in measurements. Experimentally a sample rate of every 1ms is found enough for our purposes. This sample rate allows us to calculate the RMS values from 20 measures within a cycle. It is also possible to select the number of cycles used in the calculation of RMS values, by default are used two cycles so 40 measures are involved in the RMS calculation process. The more cycles used in RMS values the more accurate is the result.
6
Introduction
In the worst case the power measure is displaced 1ms from the frequency measurement which is affordable according to the project specifications. The maximum displacement is produced when the frequency measurement is made just before a new sample acquisition. As it is explained in following sections, frequency measurements are obtained every zero-crossing and RMS measurements every one millisecond.
7
2
ALGORITHMS, RESULTS
PROCEDURES
AND
This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the
2.1. Frequency measurement algorithm This section describes the frequency measurement algorithm. It is based in measuring the time between two consecutive rising zero-crossings.
Rising zero-crossing
Rising zero-crossing
T
t
Figure 2-I - period of a signal
The frequency measurement algorithm is part of the Control logic Type I and Control logic Type II. In control logic Type I the frequency is used as a tripping criteria if it goes below the frequency event threshold. In control logic Type II it is used to regulate the temperature set points according to the control logic Type II equations, equation (1.1).
9
Algorithms, procedures and results First of all is needed to detect when the power grid signal crosses by zero, then it is needed to measure the time between two consecutive crossings. The DSC has a specific module that can carry out these actions, the capture unit (3.1.4.1.2 Capture unit), the capture unit is part of the periphery Event Manager (3.1.4.1 Event Manager). The capture unit is able to detect level changes at the signal connected with its inputs. It can detect four different events, rising or falling edges and constant positive or negative level. The capture unit can be configured to detect any of them. In order to simplify the level changing detection the capture unit has and Schmitttriggered gate at the input, this gate converts the power grid sinusoidal signal in a square signal with the same frequency. The square signal makes easy to detect the level changes in the input signal.
Figure 2-II - Schmitt-triggered output The red plot in the figure above represents the output of the Schmitt triggered gate. The square signal period Tsq is the same as the sinusoidal one Tsin. The capture unit has a timer associated, it is used as time base, when the capture unit is enabled the timer stats to count. A signal event is detected when the level is maintained at least for two time base units. Every time it is detected any of the selected signal events the capture unit carries out the following actions. It stores the current count value of the timer in the specific two-level FIFO stack. It is also called the interruption subroutine configured in the capture unit interruption vector (4.3.3.2.4 PIE initialization).
10
Algorithms, procedures and results
The FIFO stack, which contains two consecutive signal events, can be read from the interruption subroutine. In the case of the load controller the capture unit is configured to detect the rising edges, which coincide with the beginning and end of a single cycle. The power grid period is calculated in the interruption subroutine madding use of capture time stored in the FIFO stack. The first capture time is subtracted to the second one, this subtraction results on the number of time base count units between the two rising edges. Knowing the time that corresponds to a count unit is easy to get the period of the power grid signal in seconds, whose inverse is the frequency.
Level-Shifter output
S chmitt-trigger
Time base clock Stores the Counter value CAP1FIFO.1
CAP1FIFO.1
Shifts previous value CAP1FIFO.2
CAP1FIFO.2
-
CAP1FIFO - CAP2FIFO
Cycle period Figure 2-III - Frequency measurement block diagram
This algorithm fails only when the counter is overflowed, in other words, when the counter reach its maximum count value and starts again from zero. In this case, the second capture has a lower value because the counter has reach the overflow value between the two captures. This is solved checking if the value of last capture is lower than the previous. If it is lower, a counter overflow is considered. In this case the period is obtained by adding the time from the first capture to the overflow
11
Algorithms, procedures and results and the time between the overflow and the second. The Figure 2-IV - Overflow in frequency counting illustrates the cases of a measurement with and without an overflow.
Without overflow
p-2 p-1 p p+1 p+2 p+3 …
p+n p+n+1 p+n+2 p+n+3 …
With overflow
k k+1 … OF 0 1 2 ….
n
p p+1 p+2 p+3
n
temp1_1
temp1_1 temp1_2
Temp1_2>temp1_1 Period = (n+p)-p= = temp1_2 – temp1_1 =n
temp1_2 Assuming OF as maximun value of counter Temp1_2
Figure 2-IV - Overflow in frequency counting The timer base clock frequency cannot be selected arbitrarily. As is going to be explained in a following section, the higher is the frequency the better is the accuracy, which is because the count unit is smaller and the quantification error is smaller too. However if the time base frequency is too much high it can be possible that more than one overflow happens between two consecutive zero crossings, which is impossible to detected it and in consequence the obtained measurement results wrong. The interruption subroutine configured as the capture unit interruption subroutine is capture1_srt().The code analysis of this subroutine is done in section 4.3.4 Capture unit subroutine: Capture1_srt(). As explained above the number of time base count units is proportional to the period in seconds, it is only needed to know the time that corresponds to a single time base count unit and multiply it by the total number of time base count units. The time base timer is fed by the High Speed Periphery Clock (HSCLK)
12
Algorithms, procedures and results +,-./ 0 12345
(2.1)
The time base timer has a count register of 16 bits. Therefore the maximum count value is 65535. In order to avoid more than one overflow within a power grid cycle the clock speed has to be reduced. A regular power grid cycle has a period of 0,02 seconds, so that the clock speed has to be reduced to the speed that no more than 65535 count units can fit in 0,02 seconds. The way of reducing the input clock is using the prescaler included in the capture unit. The prescaler divides the speed by a fixed value. The capture unit prescaler (CUPS) can take the values of the powers of two from 1 to 256. Some easy calculations conclude that the optimal prescaler is 8. +,-./ 12345 (2.2) 6-7 0 0 0 :. <12345 -78, 9 Finally this clock speed is used to know the time in seconds that corresponds to a single count unit. < (2.3) =-7 0 0 >. :1?@ :. <12345 For example, if the Counter of the Capture unit gives the period value of 62500 count units, the calculations are the following: A8B 0 C12>> D >. :1?@ 0 >. >1@ 68B 0
< 0 2>+5 >. >1@
(2.4) (2.5)
Where TPG is the period and fPG is the frequency at power grid. It can be referred to a period expressed in the number as being in integer units. Since the number of count units is always an integer number. The DSC works internally with the integer values. It will store and compare only the integer values. It is attempted to use the integer values as much as possible, since the DSC do not have a float coprocessor. It only uses the values in Hz, which are expressed with float numbers, in the frequency decay measurement whose algorithm do not allows to do it with integer values due to an excessive lost of precision.
13
Algorithms, procedures and results In order to make easy the conversion from the integer measures to the Herzl or Seconds measures it is defined a conversion factor FCF. This factor is not more than the frequency of the time base clock. E-E 0 :<12>>> F
E-E F+5I A-7
(2.7)
A-7 F@I E-E
(2.8)
Where TCU is the period measured by the capture unit in integer units and fPG is the frequency at power grid in Hz as well as TPG is its period in seconds. The following plot shows the measures of period directly returned by the Capture unit interruption subroutine without averaging. The period is represented in integer units.
Instant Measures 62540 62520 62500 62480 62460 62440 62420 1
6
11
16
21
Instant Measures
Figure 2-V – Capture unit measurements
14
26
Algorithms, procedures and results The next figure shows the instant frequency measurements again, but It is divided into the FCF. Therefore the frequency is represented in Hz.
Instant Measures 50,040 50,030 50,020 50,010 50,000 49,990 49,980 49,970 49,960 49,950 1
6
11
16
21
26
Series1
Figure 2-VI - Instant Frequency measurements As seen in the plots, the frequency has some fluctuations. These fluctuations are reduced by averaging the output frequency measurements in order to improve the accuracy. The instant frequency measurements are not stored after being used in the averaging calculations; that is because the plots within this section shows only 30 measurements in a row, in following sections and in the appendix can be found other plots with more consecutive measurements.
15
Algorithms, procedures and results
2.1.1. Averaging measures As seen in the previous section, the instant frequency measurements presents some fluctuations, in order to improve the accuracy and stabilize the measurements it is use the average of a selectable amount of instant measurements. These averaged measurements are the ones used in the threshold checking. The more instant measures used in the average the more accurate is the measure, but the more measurements are used the more increases the delay of a measure. A frequency change will be assuredly detected after the number of cycles used in the average. This is because the frequency event measures are spread within the cycles in the average and is not assured that the measures within a frequency event will change the average enough to fulfill the threshold, at least with the early measures. The error is cancelled in the inner averaged measurements and the error in the first and the last cycle is spread in all the cycles.
time
Real zero-crossings Sensed zero-crossings Individual error
Figure 2-VII - Error spread using average
The positive error in one cycle is the opposite in the adjacent cycles. The only error that is not cancelled is the error produced by the first and the last measure. This error is divided according with the cycles used in the average. In instance of using the average of 25, the default in the project, the measurement delay is 0.5 Seconds. A the detection of a frequency event is not assured until 0.5 seconds.
16
Algorithms, procedures and results The instant measurements averaged can be observed in the following plot.
Average of 25 values 62505 62504 62503 62502 62501 62500 62499 62498 1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
Average of 25 values
Figure 2-VIII - Averaged frequency measurements in integer values
The next plot shows the same values as the previous one but in Hz units.
Average of 25 values (Hz) 50,001 50,000 49,999 49,998 49,997 49,996 49,995 1
3
5
7
9
11 13
15 17 19 21 23 25
27 29
Average of 25 values (Hz)
Figure 2-IX - Averaged frequency measurements in Hz The figure above shows the averaged frequency measurements in Hz, comparing these plots with the previous instant measurements in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the
17
Algorithms, procedures and results Frequency measurement the measures are smother. In the following figure is possible to compare in the same plot the instant measures with the averaged measurements.
instant Vs average measures 50,040 50,030 50,020 50,010 50,000 49,990 49,980 49,970 49,960 49,950 1
3
5
7
9
11 13 15 17 19 21 23 25 27 29
Instant values (Hz)
Average, 25 values (Hz)
Figure 2-X - Instant and averaged measurements It can be found more plots of frequency measurements in appendix C Other plots.
2.1.2. Maximum theoretical Accuracy This section explains the maximum theoretical accuracy that could be reached using the frequency algorithm explained in the previous sections. In this section is considered that the input power grid signal is clean of noise and the only source of error is the quantification error of the capture unit. As obvious, the accuracy calculated in this section is not the accuracy of the load controller but it is the maximum reachable accuracy. The frequency measurement depends on the frequency of Capture unit clock. It has an error of ±1 count units every measure. But as explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement algorithm the capture unit clock frequency cannot be increased more than a limit.
18
Algorithms, procedures and results
Case of counting one unit less Starts just after checking
Ends just before checking
Lost time Real length of period signal
EV clock:
The period have not started yet
Here ends The counting
Here starts The counting Case of counting one unit more
Starts just before checking
Excess time Real length of period signal
Ends just after checking Excess time
EV clock:
Here starts The counting
Signal period have not finished yet
Here ends The counting
Figure 2-XI - frequency counting error The figure above explains the error of measurements. It shows how are measured cycles whose edges are just in the limits of the Capture unit cycles. These case produces the measurements with the error of +1 or -1 count units. In other cases the excess time compensates the lost time, always withing the range of the absolute error is ±1 time base count units. The absolute in error in period can be expressed with the following equation. ∆AJKL 0 M<
(2.9)
N represents the number of time base count unit, therefore the error does not depend on the frequency measured. The accuracy or relative error depends on the frequency measured. In lower frequencies, the number of time base count units is higher, for that reason the absolute error is proportional smaller than the measure itself. In higher frequencies the number of time base count units is smaller and the absolute error is proportional higher.
19
Algorithms, procedures and results
As explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement algorithm. The capture unit uses a GP timer to count the Capture Unit Cycles between rising edges. The timer is a 16-bit-timer so the maximum count is from 0 to 216-1 or from 0 to 65535 in decimal. The Capture unit is fed by the high speed periphery clock that runs at 25Mhz. Using directly this clock in a regular power grid cycle (1/50Hz=20ms) the counter should count 500000 that is higher that the overflow limit of the counter. A prescaler is used to reduce the clock speed. Using all the data above is possible to calculate the theoretic accuracy at 50Hz, the calculations are the following: JK M, :1μ@ 6OP 6OP So that the period of a signal at 50Hz will have values between
∆A 0
TPG(50Hz)=0.02s+∆Ts=0.02±32µ
(2.10)
(2.11)
These periods corresponds to these frequencies: 6STU 0
<
ASVW
0
< 0 XY, YYY1+5 1>. >>>:1S@
(2.12)
That is a relative error of: fmin-50=-0.8Hz And finally: 6SVW 0
<
ASTU
0
< 0 2>, >>>9+5
That is a relative error of:
20
(2.13)
Algorithms, procedures and results fmax-50=0.8Hz In summary, at 50Hz we have an accuracy of ∆f(50Hz)=±0.8Hz, or a relative error of 1,6%, The relative accuracy depends on the frequency measurement, at higher speeds the absolute error can be proportional higher and the accuracy is worse. The use of accuracy improves the accuracy. The error is divided into the number of instant frequency measurements.
2.1.3. Soft frequency filter Experimentally it is observed that some measures differ too much from expected measures. These measures are produced by external interferences such as the interferences produced by the physical movement of the extension cord. In order to avoid these fake measures it is implemented a frequency filter by software. This filter checks every measure if the frequency decay is higher than a limit. In the case the measure exceeds the threshold the measurement takes the value of the previous one. The frequency threshold in this filter is too much higher than the frequency decay in the power grid in normal state and within a frequency event.
21
Algorithms, procedures and results
Attribute: MAXIMUN_FREQUENCY_CHANGE
Prev. Frequency measure
-
+
Last frequency measure
Outs the higher
Outs the lower
Frequency out
Figure 2-XII - Software frequency filter block diagram The filter basically compares the last measure with the previous one, if the different between them is higher than the limit stored in the attribute MAXIMUN_FREQUENCY_CHANGE the output frequency is the last measure, otherwise it is used the unmodified frequency measure. By default it is used the value of 150 time base count units between two consecutive frequency measurements, it corresponds at 50Hz to a decay of ±6Hz/s, the control logic type II uses a decay of -0,16Hz/s
2.1.4. Frequency sensor range The speed of the clock that feeds the capture unit sets the frequency range The lower limit is when the period is enough long to have more than one overflow inside. The limit is equal to the overflow limit of the counter, 65535 EV-cycles →65535/3.125 MHz ≈ 0,021ms = 47,7Hz
22
Algorithms, procedures and results , that is much lower than the limit in the specifications. If the period is longer is not possible to be sure that more than one overflow happens within a cycle. The higher frequency hypothetically has the limit when the frequency is so fast to at least count one unit in but is not useful because the relative error is ±100%. Therefore the higher frequency limit is at the point where the accuracy drops below a desider limit.
2.1.5. Accuracy In the accuracy study it was not possible to obtain the signal to noise ratio of power grid line, and also it was not possible to obtain the Spectral Frequency Noise density present in the working frequency band, moreover the manufacturer of DSC don’t provide the hysteresis windows width and finally it was not available a frequency counter with enough accuracy to compare the measurements made by the load controller. Therefore it is not possible to do a theoretical study of accuracy. The study of section 2.1.2 Maximum theoretical Accuracy gives the maximum theoretical accuracy, but not the accuracy of the Load Controller. Assuming the frequency measurements were not taken within a frequency event. The measurements showed in this project are randomly chosen from several similar measurements in different times, therefore it not probable that the frequency measurements where done within a frequency event. The frequency measurements vary in the experimental measurements not more than 15 mHZ within a measurement session, that means that this fluctuations can be due to regular power grid fluctuations but it cannot be possible to ensure it. Therefore the accuracy of this load Controller is within the range of ±0.8mHz and ±7,5mHz. The minimum accuracy grated by the load controller is enough for its purposes. Minimum grated accuracy ±7,5mHz. @50Hz Maximum delay: 0,5seconds.
23
Algorithms, procedures and results
2.2. Frequency decay measurement The frequency decay is measured by the average of the frequency changes within a selectable number of consecutive power grid cycles. The frequency decay measurement is part of the control logic Type I. In the control device as explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement what is really measured is the period in number of cycles of EV clock, otherwise, the time between zero-crossings. The frequency of the sample is the inverse of the period and the time between frequency measures is the period itself.
Frequency
1/T0
Δf=1/T1-1/T0 1/T1
time Δt=T1
Figure 2-XIII - Frequency decay
As seen in the figure above, the time between two correlative measures is the period of the second one and the frequency change is the subtraction of the inverses of correlative period measures. The frequency decay is defined as the difference of frequency in any amount of time. It is obtained from the period measures with the following equation:
24
Algorithms, procedures and results < < ∆6 A< N A> A> N A< EZ 0 0 0 ∆= A< A> D A1<
(2.14)
In section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement it was described that the measure of the periods is measured in number of EV cycles. The frequency decay measurement algorithm uses lots of resources. The calculations made in this algorithm needs more calculation power than others algorithms. The frequency and RMS calculations are made with the integer values from the ADC and the EV but this algorithm cannot be made using the integer values due to the rounding. The operations must to be float operations, not integer operations. The internal operations made by the CPU are specially selected to avoid as much as possible the float operations and use the integer and long operations. < < +5 ∆6 E-E D bc3< N c3> d E[\][U^_ Z[^V_ ` a 0 0 0 c3< @ ∆= E-E 1 N c3 N c3 c3 E-E c3 E-E1 c3> > < > < 1 0 E-E D 0 D 0 De N c3< 1 D c3> c3< 1 c3< 1 c3<
(2.15)
Where FCF is the Frequency conversion factor that relates the integer values with the frequency values in Hz. See 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement. And AM1 represents the last frequency measurement value and AMs0 represents the previous one. Note that AM are the integer values given by the Capture unit and they are period measurements.
25
Algorithms, procedures and results g is a long multiplication, the rest of operations are float operations, one division, one subtraction and one multiplication. The frequency decay measurements are averaged in order to have more accurate results. The average can be done with a selectable number of frequency decay results. By default it is done the average of last 5 values.
Frequency decay (Hz/s) 0,1 0,05 0 1
29
57
85
113
141
169
197
-0,05 -0,1 frequency decay
Figure 2-XIV – Frequency decay
2.3. Control logic Type I: Frequency Thresholds The load tripping according to the frequency and frequency decay measured implement the control logic Type I. The load controller can be in three different states depending on the frequency sensed. The default sate is the CLOSE state. In this state the frequency is higher than the lower frequency threshold. Therefore the power grid is not within a frequency event. The load is connected with the power grid. If a frequency event is detected, when the frequency goes below the frequency threshold, the state is changed to OPEN state. In this state the load is disconnected. The load will remain always disconnected while the frequency is below the recover frequency event, which is higher than the frequency threshold.
26
Algorithms, procedures and results Once the frequency is recovered, so the recover threshold is fulfilled, the state is changed to WAIT state. When the load controller enters in this state, it is set a random time to attachment. The load controller remain in this state while the delay time is over, when the delay time is up the load controller changes to CLOSE state, being the load reconnected. In this state it is also checked if the frequency returns below the frequency threshold cancelling the reconnection and changing the state to OPEN again. The default thresholds used are the following. The frequency values in integer units is obtained by using the conversion equation (2.7): DEFAULT_HIGHER_LIMIT = 62525 = 49,98Hz DEFAULT_LOWER_LIMIT = 62626 = 49,89Hz The default lower threshold can be also called default frequency event threshold. In the case of the frequency below this threshold it is considered the power grid under a frequency event. The default higher threshold it can be called also the recover threshold, this threshold defines the frequency threshold which in the case of being surpassed within a frequency event the frequency event is considered to be recovered.
Averaged Vs Thresholds 50,1 50,05 50 49,95 49,9 49,85 49,8 1
29
57
Average measurements
85
113
recover threshold
141
169
197
frequency event threshold
Figure 2-XV – Averaged frequency measurements vs frequency thresholds The frequency decay threshold works in a similar way as frequency threshold in exception that the frequency decay event has a single threshold, it is used the same
27
Algorithms, procedures and results threshold for detecting a frequency decay event and checking the frequency decay event recover.
Frequency decay Vs Threshold 0,1 0,05 0 1
29
57
85
113
141
169
197
-0,05 -0,1 -0,15 -0,2 frequency decay
decay threshold
Figure 2-XVI Frequency decay Vs Frequency decay threshold The next figure shows the state flow diagram, the load controller starts by default in the state CLOSE. The circles represent the states. The “f” inside the decision blocks is the last frequency measure acquired. Lower_limit is the frequency threshold and higher_limit is the recover threshold. Decay represents the last frequency decay measurement and limit represents the frequency decay threshold.
28
Algorithms, procedures and results
Figure 2-XVII- Threshold checking flow diagram
29
Algorithms, procedures and results
2.4.
Control logic Type II: Temperature Set points change
The dynamic temperature set points regulation implements the control Logic Type II. As explained in the introduction, this control logic is not completely implemented. This control logic is based on regulate the load temperature set points dynamically, according to the frequency measured in the power grid. This control logic is applicable only to the loads that has a thermostat with two set points, and the load is only started when it reach one of the set points and it is stopped when the temperature reach the other set point. In this loads, a fridge or an electric heater can change the normal set points by the user. The load controller should modify this normal set points according to the frequency measured. The load controller assumes that the input normal temperature set points are introduced in the load controller thought the ADC. The thermostat should have two analog outputs which vary from 0 to 3 Volts depending on the temperature set point. This normal temperature set point reading method is only a proposal that could be changed depending on the load. The load controller has to calculate a new high and low set points depending on the power grid frequency. The output Temperature set points are calculated with the following equation: normal T high = T high − kf * ( f − f 0 )
(2.16)
normal Tlow = Tlow − kf * ( f − f 0 )
The input set points are assumed to be an integer number, the output temperature is an integer too, the Kf factor has to be regulated to use the Temperature in integer units, as well as use the frequency measurements in integer units too. The load controller stores the new temperature set points in a variable, the variable should be used to change the temperature set points according to the interface used to set the temperatures in the load thermostat.
30
Algorithms, procedures and results The normal temperature set points that should be read by the ADC are read in the ADC interruption subroutine (4.3.5 ADC interruption subroutine: ADC_srt()) the output temperature set points are calculated in the capture unit interruption subroutine (4.3.4 Capture unit subroutine: Capture1_srt()), the set points are calculated every frequency measurement, every 0,02 seconds
2.5. Voltage, current and Power measurement This section describes the process of RMS voltage, current and power measurement. First of all it is described the sampling method, and finally the process of obtaining the RMS values from the instant measurements, it is also show some sample plots of the measurements made by the ADC.
2.5.1. Sample acquisition The sample acquisition is made by the ADC. As explained in section 3.1.4.2 ADC the ADC is configured to sample the voltage and current channel every one millisecond The sample acquisition is made by the subroutine adc_isr(void). The EV is configured to start the ADC conversion every one millisecond, once the ADC has the sampling results available the interruption subroutine is called. Every time this subroutine is executed it is ensured that the ADC registers has ready to use the last measures stored inside. At this point the offset of each channel is subtracted to the measure of the input signals. The measures of voltage and current, with the offset already subtracted, are stored in an array, one array for each channel. These arrays contain by default the last 40 measures of voltage and current those are used to calculate the power and RMS values. Note that this subroutine stores the integer values of these measures and not the floating point values corresponding for the equivalent in Volts or Amperes. The DSC works every time with these integer values simplifying the calculations. As seen in section 3.1.4.2 ADC the values in volts or in amperes are obtained directly with fixed operations that can be made in the analysis time, that mean, when the measurements are retrieved into the server.
31
Algorithms, procedures and results This subroutine doesn’t calculate the RMS power, voltage and current but it calculates the square of the measures. The capture unit interruption subroutine is the responsible of the RMS calculations and it uses the last 40 measures from the ADC. In the RMS calculations is needed the square of these measures and doing it in the ADC subroutine it is only needed to do it once every new measure, otherwise the capture unit subroutine should do the square of last 40 ADC measures, which would result in not efficient DSC calculation power usage.
2.5.2. Acquired measures This section shows the plots of the measures made by the ADC. The first two plots shows the integer measures given by the ADC. These measures are within the range of the ADC, between 0 and 4096. Next two figures show the integer values given by the ADC channels of current and voltage.
Input Current 3000
Current Values
2500 2000 1500 1000 500
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
0 samples
Figure 2-XVIII - ADC curret measures
32
Algorithms, procedures and results
4500 4000 3500 3000 2500 2000 1500 1000 500 0 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Voltage values
Input Voltage
samples
Figure 2-XIX - ADC Voltage measures The offset has been measured without connecting any signal to the ADC input. The offset is so stable that it can be considered as a constant value. The offset in current and voltage is estimated in 1904 and 1921 ADC integer units. The signals used in all the load controller internal calculations are integer values given by the ADC subtracted with the corresponding offset. These measurements can be observed in the two following figures.
ADC Current integer measures (without offset) 3000
Current Values
2000 1000 0 -1000
1
11
21
31
41
51
61
71
81
-2000 -3000
samples
Figure 2-XX - Integer measures of current without offset
33
Algorithms, procedures and results
ADC Voltage integer measures (without offset) 3000
Voltage (int)
2000 1000 0 -1000
1
11
21
31
41
51
61
71
81
91
-2000 -3000
samples
Figure 2-XXI - Integer measures of voltage without offset The measures above can be also observed in I.S. units of measure in the next two figures. These values are obtained using the method described in section 3.2.2 Current and voltage values at load controller inputs.
Input Current 2 1,5
0,5 0 -0,5
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Current (Int)
1
-1 -1,5 -2
samples
Figure 2-XXII - Measured Current
34
Algorithms, procedures and results
Input Voltage 400 300
100 0 -100
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Voltage (V)
200
-200 -300 -400
samples
Figure 2-XXIII - Measured Voltage With the measurements of voltage and current it is possible to calculate the RMS power. The RMS power is calculated using the individual values of current and voltage. Therefore it is first calculated the instant power and then is calculated the RMS power using the same method used in voltage and current. The next figure shows the instant power values in integer units.
Instant power (integer untis) 1200000
Power (W)
1000000 800000 600000 400000 200000
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
0 samples
Figure 2-XXIV - Instant power in integer units The figure below shows the instant power in S.I units.
35
Algorithms, procedures and results
1000 900 800 700 600 500 400 300 200 100 0 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Power (W)
Input power
samples
Figure 2-XXV - Calculated instant Power In order to improve the accuracy the channels are oversampled, therefore the channels are sampled many times in a row and it is use as final measurement the average of all of them. In this prototype it is use the oversampling of two values, the load controller hardware allows up to oversample the channel eight times. The time between consecutives samples is 250ns which is enough small compared with the signal period to permit the oversampling.
2.5.3. RMS Current, calculation
voltage
and
power
The Current, voltage and power in RMS values are calculated in the same subroutine where the frequency is measured (Capture1_isr). RMS Voltage, current and power is calculated using this equation: 1 ∑K Tl> cZ-S[V@]k[ T i h3, 0 K
36
(2.17)
Algorithms, procedures and results Where ADC_MEASUREi represents the integer sample made by the ADC or the multiplication of the voltage and current samples in the case of the power RMS measurement. N represents the number of measures used in the calculation, by default this value is 40, therefore, the RMS value is calculated according to the last two cycles. In order to simplify the RMS calculations, the square of the instant samples are stored in specific arrays, this avoids recalculating the instant sample squares every time a RMS calculation is done. It is only calculated the square of the new samples. In summary the load controller does the following calculations in order to calculate the RMS values: h3, 0 i
∑K Tl> adc_square_sampleT K
(2.18)
Where adc_square_sample represents the respective squared samples array (voltage, current or power) The square samples are calculated in the ADC sampling subroutine (4.3.5 ADC interruption subroutine: ADC_srt()) in the array adc_square_voltage[] for voltage, adc_squeare_current[] for current and adc_squeare_power[] for power. The following plots show the measurements made by the load controller in integer units. RMS current measurement:
RMS Current (Int) Current RMS (Int)
363,5 363 362,5 362 361,5 361 360,5 360 1
11
21
31
41
51
61
71
81
Figure 2-XXVI - Measured RMS current (int)
37
Algorithms, procedures and results RMS voltage measurements:
RMS Voltage (Int) 1410,5 Voltage (INT)
1410 1409,5 1409 1408,5 1408 1407,5 1407 1
11
21
31
41
51
61
71
81
Figure 2-XXVII - Measured RMS Voltage (int) And finally the RMS power measurements:
Power RMS (Int)
RMS power (int) 1000 900 800 700 600 500 400 300 200 100 0 1
11
21
31
41
51
61
71
81
Figure 2-XXVIII - Measured RMS power (int)
It is used all the time Integer values that simplify the calculations and the subroutine can be executed faster. The integer values have a direct relation with the values in Volts, Amperes or Watts.
38
Algorithms, procedures and results In the case of the Voltage this relation is: Ph3, 0
P_cZ- h3, Ex
(2.19)
Where Fv is the conversion factor from ADC values to Real values and it can be calculated by the following way: Ex 0
X>YC D yx :
(2.20)
Where gv is the total gain of the Voltage channel.
In the case of the Current this relation is: zh3, 0
z_cZ- h3, ET
(2.21)
Where Fi is the conversion factor from ADC values to Real values and it can be calculated by the following way: ET 0
X>YC D yT D <> :
(2.22)
Where gi is the total gain of the current channel. And 10 is the resitance of the resistor connected at the ADC input
The power RMS is also calculated using the integer values given by the ADC. The RMS power can be obtained from the RMS power calculated with the integer values using the following equation:
8h3, 0
8_cZ- h3, E{
(2.23)
Where P_ADCRMS is the power calculated from the integer ADC measures and Fp is the power conversion factor that is defined in the following equation:
39
Algorithms, procedures and results ET 0 ET D Ex 0
X>YC D h D yT X>YC D yx X>YC1 D 0 D yT D yx : : :1
(2.24)
Where R is the resistor connected to the ADC current channel (see 3.1.4.2 ADC) They are some reasons because the calculation is made this routine instead of be done in the ADC subroutine, the sampling routine. ADC subroutine is executed every 1ms, 20 times every power grid cycle, but the RMS values are only needed one time every power grid cycle, which is the measurements rate specified in the specifications. By doing it in the ADC subroutine results in 20 measurements of RMS values per cycle, which means calculating it 20 times per cycle. That requires more performance that the DSC can deliver. The other reason is that the RMS values have to be of the same cycle of the frequency measure. By doing the RMS calculations in the same subroutine as the frequency measurement it is ensured that the RMS results correspond to the same cycle as the frequency value.
Current RMS (A)
RMS Current (A) 2,218 2,216 2,214 2,212 2,21 2,208 2,206 2,204 2,202 2,2 2,198 2,196 1
11
21
31
41
51
61
Figure 2-XXIX - Measured RMS current (A) RMS voltage measurements:
40
71
81
Algorithms, procedures and results
RMS Voltage (V) 224,6
Voltage (V)
224,5 224,4 224,3 224,2 224,1 224 1
11
21
31
41
51
61
71
81
71
81
Figure 2-XXX - Measured RMS Voltage (V) And finaly the RMS power measurements:
RMS power (W) 623000
Power RMS (W)
622000 621000 620000 619000 618000 617000 616000 1
11
21
31
41
51
61
Figure 2-XXXI - Measured RMS power (W)
The three figure above shows the RMS values of current, voltage and power in the international system of units but the DSC works internally with the integer values, the plots of the RMS results in integer values are omitted in this section and it can be found in the appendix C Other plots. In the appendix it can also be found the tables of values of these plots.
41
Algorithms, procedures and results As well as explained in the frequency accuracy it is not enough available data to calculate and exact voltage, current or power accuracy. The maximum variation in RMS voltage measurements is 0,6 V in all the voltage measurement sessions, therefore the voltage accuracy should be better than ±0,3V. This corresponds to an accuracy of ±0,13% Using the same reasoning as before the maximum variation in Current measurement is 0,03A, therefore the current accuracy should be better than ±0,015A. This corresponds to an accuracy of ±0,05%. Finally referring to the power measurements the maximum variation observed is 10W in the test load. Therefore the power accuracy should be better than ±5W. This corresponds to an accuracy of ±0,8%. All the results described above are only estimated results that could change depending on the test load stability. It should be considered a detailed accuracy study. Accuracy Voltage
±0,3V
Current
±0,015A
Power
±5W
Note that this result are done using the test load of nominal 600W. The current accuracy could be improved by expanding the current signal in the ADC conversion range, it can be done by changing the number of loops in the current transformer. The maximum RMS measurement delay is 1ms.
2.6. Real-time stamp The real time stamp is used in order to locate the measurements in time. The time stamp is a long integer value that is incremented every one millisecond. The Real-time system also offers the possibility to get the current time and also to synchronize the time stamp with a external time source.
42
Algorithms, procedures and results
The time stamp register is incremented in the ADC interruption subroutine that is periodically executed exactly every one millisecond. The time stamp uses an arbitrary date as origin. For example choosing January 1st of 2008, the register will contain the millisecond elapsed from that date. The time stamp is valid until the register doesn’t overflow. Long integer variables stores 64 bits so that the total time while this load controller can work is:
|}~ 0 2 0 18446 D 10 Therefore the time stamp will remain valid until around 584942417 years. What is more than enough. The function time_synchronization(long time) sets the current time to the time passed in the argument. The current time is obtained by reading the current_time variable.
43
3
HARDWARE
This chapter explains everything related with the hardware, the selection of DSC, design of the board and description of the developing board.
3.1. DSC The DSC (Digital signal controller) is one of the most important parts of the hardware. It is the core of the control device. It manages the hardware and made the calculations. As a key part of the project, a deep research had been made. The conclusion of the research is show within this section. The function of the DSC can be made also by microcontrollers (µC) and Digital Signal Controllers (DSC), they are some differences between then. In the following paragraphs the different proposals are compared.
The microcontroller is designed for a wide field of uses but not for any specific function. Therefore, they are capable of virtually carry out any application, it is designed as a generic device. Microcontrollers are also easy to use and cheap. Referring to this project, microcontrollers in budget do not have enough signal processing features. A more specific solution is needed. Regular microcontrollers are not able to carry out all the requirements of the control device. In instance they are used to be lack of ADC inputs and other signal processing specific hardware, such as a 32 bit MAC (Multiplication And aCcumulation) module needed to carry out the numerous multiplications needed in the RMS calculations. The DSPs solves the typical microcontroller problems described above. They have specific instructions for signal processing, the used to have enough ADCs and they
45
Hardware are quite powerful. But like the microcontrollers, DSPs have some problems, they are more expensive and difficult to program and they use lack on general purpose peripheries like Universal Asynchronous Receiver-Transmitters (UART) or Serial Periphery Interfaces (SPI), being needed to use a microcontroller as co-processor. Therefore it will be very effective in Signal processing but not so good in other stuff. The solution is a DSC. The DSC is basically a Microcontroller with DSP capabilities. They have a microcontroller as core with the common peripheries in microcontrollers (UART, SPI, Timers, GPIO and so on) and they have also peripheries common in DSPs (ADCs, MAC, EV). It is as simple to program as a microcontroller, even having some signal specific instructions. For that reason is decided to use a DSC that is more flexible than a DSP and has more Signal processing capabilities than a microcontroller.
3.1.1. DSC prerequisites The project specifications determine some features that the DSC must have in order to fulfill all the specifications The aim of the project is measure the frequency of the power grid as well as the power consumption of the controlled appliances. It should be able to measure in the future tri-phased signals if the project is continued by that way. Power measurement specification requires that the microcontroller must have an ADC with multiple input channels. It must have at least two channels for a mono-phase power grid source, one for voltage and other for current. These two channels must also be sampled at simultaneous time, in other words, it must have at least one dual channel able to sample both of the inputs at the same time. Looking for the lower complexity the best option is to choose a DSC that has the ADC on-chip with at least one dual channel, for a single phase of Current and Voltage. The DSC is hast to do some complex calculations periodically, so it is required that it is able perform all the calculations within the period of available time. It has to calculate the average of last frequency measures, frequency decay and also voltage, current and power in RMS values, all of them every power grid cycle.
46
Hardware
The DSC must have a MAC module able to calculate 16x16 (used in RMS voltage and current calculations) and 32x32 (used in RMS power Calculation) integer multiplications. The RMS calculations made an intensive use of multiplications every a short amount of time (1ms), these MAC unit is essential to ensure that all of the calculations are made before the next RMS calculation. The DSC has to support the connection of external memory. All the previous measurements before a frequency event should be stored in order to send them to the server. The external memory connection can be done using the on-chip external memory interface (cable select outputs and memory address outputs) or using a seral flash memory (using the SPI periphery). Finally according the DSC should have a UART port in order to connect the GPRS module for data communications between device and a Server. Other matters not related with the technical features had been considered in order to take the best decision. It has been considered the availability of a developing board, a board with the DSC integrated and with the PC connection ports, in order to start quickly the developing of the hardware. The availability of a good Integrated Development Environment (IDE), more details in section 4.1 IDE, had also been considered, as well as the support and documentation of the manufacturer and obviously, the final price. A detailed research drops as result that the best option is the TSM320F2812 manufactured by Texas instruments. Its technical features are showed in the next section.
47
Hardware
3.1.2. DSC Technical features The features of the selected DSC are the following. High-Performance Static CMOS Technology 150 MHz (6.67-ns Cycle Time) Low-Power (1.8-V Core @135 MHz, 1.9-V Core @150 MHz, 3.3-V I/O) Design 3.3-V Flash Programming Voltage JTAG Boundary Scan Support High-Performance 32-Bit CPU (TMS320C28x) 16 x 16 and 32 x 32 MAC Operations 16 x 16 Dual MAC Harvard Bus Architecture Atomic Operations Fast Interrupt Response and Processing Unified Memory Programming Model 4M Linear Program Address Reach 4M Linear Data Address Reach Code-Efficient (in C/C++ and Assembly) TMS320F24x/LF240x Processor Source Code Compatible On-Chip Memory Flash Devices: Up to 128K x 16 Flash (Four 8K x 16 and Six 16K x 16 Sectors) ROM Devices: Up to 128K x 16 ROM 1K x 16 OTP ROM L0 and L1: 2 Blocks of 4K x 16 Each Single-Access RAM (SARAM) H0: 1 Block of 8K x 16 SARAM M0 and M1: 2 Blocks of 1K x 16 Each SARAM Boot ROM (4K x 16) With Software Boot Modes Standard Math Tables External Interface (F2812) Up to 1M Total Memory Programmable Wait States Programmable Read/Write Strobe Timing Three Individual Chip Selects Clock and System Control Dynamic PLL Ratio Changes Supported On-Chip Oscillator Watchdog Timer Module Three External Interrupts Peripheral Interrupt Expansion (PIE) Block That Supports 45 Peripheral Interrupts 128-Bit Security Key/Lock Protects Flash/ROM/OTP and L0/L1 SARAM Prevents Firmware Reverse Engineering Three 32-Bit CPU-Timers Motor Control Peripherals Two Event Managers (EVA, EVB) Compatible to 240x Devices Serial Port Peripherals Serial Peripheral Interface (SPI) Two Serial Communications Interfaces (SCIs), Standard UART Enhanced Controller Area Network (eCAN) Multichannel Buffered Serial Port (McBSP) With SPI Mode
48
Hardware 12-Bit ADC, 16 Channels 2 x 8 Channel Input Multiplexer Two Sample-and-Hold Single/Simultaneous Conversions Fast Conversion Rate: 80 ns/12.5 MSPS Up to 56 Individually Programmable, Multiplexed General-Purpose Input/Output (GPIO) Pins Advanced Emulation Features Analysis and Breakpoint Functions Real-Time Debug via Hardware Development Tools Include ANSI C/C++ Compiler/Assembler/Linker Supports TMS320C24x™/240x Instructions Code Composer Studio™ IDE DSP/BIOS™ JTAG Scan Controllers [Texas Instruments (TI) or Third-Party] Evaluation Modules Broad Third-Party Digital Motor Control Support Low-Power Modes and Power Savings IDLE, STANDBY, HALT Modes Supported Disable Individual Peripheral Clocks Package Options 179-Ball MicroStar BGA™ With External Memory Interface (GHH), (ZHH) (2812) 176-Pin Low-Profile Quad Flatpack (LQFP) With External Memory Interface (PGF) (2812) 128-Pin LQFP Without External Memory Interface (PBK) (2810, 2811) Temperature Options: A: -40°C to 85°C (GHH, ZHH, PGF, PBK) S: -40°C to 125°C (GHH, ZHH, PGF, PBK) Q: -40°C to 125°C (PGF, PBK)
This DSC fulfills all the prerequisites. It has enough ADC inputs, it has all the external interfaces required, it is not expensive and it has enough calculation power. It also has a specific periphery that carries out the frequency measurement, the Event Manager (EV).
Figure 3-I - Texas instruments Logo The final price of a single DSC unit is about 75DKK, depending on the amount purchased the price can be lower.
49
Hardware
3.1.3. DSC Hardware Description This section describes the DSC hardware and the peripheries on-chip.
3.1.3.1.
C28x CPU
The C28x DSP generation is the newest member of the TMS320C2000 DSP platform. The C28x is an efficient C/C++ engine, enabling users to develop their system control software in a high-level language using C/C++. The 32 x 32-bit MAC capabilities of the C28x and its 64-bit processing capabilities, enable the C28x to efficiently handle higher numerical resolution problems that would otherwise demand a more expensive floating-point processor solution. This is perfect for the calculation of power from voltage and current in real time which requires lots of long multiplications. The C28x has an 8-level-deep protected pipeline with pipelined memory accesses. This pipelining enables the C28x to execute at high speeds without resorting to expensive high-speed memories. Therefore this CPU is able to execute up to 8 instructions at the same time. Special branch-look-ahead hardware minimizes the latency for conditional discontinuities, so in branches the execution taken up again more efficiently. This feature is very useful in the project that is based in asynchronous and synchronous interruptions, shouted by the power grid cycles and ADC every sample. That ensures a good performance on the interruption execution against the continuous changes of context. Every time that an interruption is called, the microcontroller must save the context of execution, load the specific context of the interruption and when it ends the interruption execution it must restore the previous context. This change of context happens around 20ms in asynchronous interruption (every power grid zero-crossing) and every 1ms in periodical interruption (sample rate of ADC).
3.1.3.2.
Memory Bus (Harvard Bus Architecture)
As with many DSP type devices, multiple busses are used to move data between the memories and peripherals and the CPU. The C28x memory bus architecture contains a program read bus, data read bus and data write bus. The program read bus consists of
50
Hardware 22 address lines and 32 data lines. The data read and write busses consist of 32 address lines and 32 data lines each. The 32-bit-wide data busses enable single cycle 32-bit operations. The multiple bus architecture, commonly termed “Harvard Bus”, enables the C28x to fetch an instruction, read a data value and write a data value in a single cycle. This feature increases the execution speed being able to retrieve the results of the peripheries at the same time that the next instruction is read.
3.1.3.3.
Real-Time JTAG and Analysis
The F281x implement the standard IEEE 1149.1 JTAG interface. Additionally, the F281x support real-time mode of operation whereby the contents of memory, peripheral, and register locations can be modified while the processor is running and executing code and servicing interrupts. This is very useful in the debug stage where it is necessary to trace the running of the program and in also useful to modify the register to emulate external events. The F281x implement the real-time mode in hardware within the CPU. This is a unique feature to the F281x, no software monitor is required. Additionally, special analysis hardware is provided that allows the user to set hardware breakpoint or data/address watch-points and generate various user selectable break events when a match occurs. The features above can be done using the parallel connection with the Personal computer (PC) A specific JTAG emulator hardware also allows the user to write inside the onchip flash memory unfortunately it cost about 1300$ (6 857DKK), but in this prototype there is enough memory with only the on-chip and external RAM memory. So that it is not used in this project stage.
3.1.3.4.
SARAMs
The DSC contains two blocks of single access memory, each 1K x 16 in size. The stack pointer points to the beginning of one of this block on reset. The F281x also contains an additional 16K x 16 of single-access RAM, divided into 3 blocks (4K + 4K + 8K). Each block can be independently accessed hence minimizing pipeline stalls.
51
Hardware
Figure 3-II - DSC Memory map The RAM memory described in this section is not all the memory available in the load controller; it is only the on-chip RAM. External RAM from the Developing Boar is also available.
52
Hardware
3.1.3.5.
Flash
The F2812 contain 128K x 16 of embedded flash memory, segregated into four 8K X 16 sectors, and six 16K X 16 sectors The user can individually erase, program, and validate a flash sector while leaving other sectors untouched. However, it is not possible to use one sector of the flash to execute flash algorithms that erase/program other sectors. Special memory pipelining is provided to enable the flash module to achieve higher performance. The flash is mapped to both program and data space; therefore, it can be used to execute code or store data information. The use of flash is slower in program execution speed than the internal RAM. In order to use the flash memory is mandatory to use the JTAG emulator. The JTAG was not purchased and the final program is inserted it in RAM memory. Anyway, the interruption functions developed in this prototype are time critical function and they must be on the RAM memory.
3.1.3.6.
Security
The F281x and C281x support high levels of security to protect the user firmware from being reverse-engineered. The security features a 128-bit password (hardcoded for 16 wait states), which the user programs into the flash. One code security module (CSM) is used to protect the flash/ROM and the SARAM blocks. The security feature prevents unauthorized users from examining the memory contents via the JTAG port, executing code from external memory or trying to boot-load some undesirable software that would export the secure memory contents. To enable access to the secure blocks, the user must write the correct 128-bit ”KEY” value, which matches the value stored in the password locations within the Flash/ROM. In this project I will not use this feature that will be only useful in a final product in order to avoid the copy of the source. If using this feature it is possible under some circumstances to leave the microcontroller locked, so that the memory is unprotected all the prototyping time.
53
Hardware
3.1.3.7.
Peripheral Interrupt Expansion (PIE) Block
The PIE block serves to multiplex numerous interrupt sources into a smaller set of interrupt inputs. The PIE block can support up to 96 peripheral interrupts. On the DSC, 45 of the possible 96 interrupts are used by peripherals. The 96 interrupts are grouped into blocks of 8 and each group is fed into 1 of 12 CPU interrupt lines (INT1 to INT12). Each of the 96 interrupts is supported by its own vector stored in a dedicated RAM block that can be overwritten by the user. The vector is automatically fetched by the CPU on servicing the interrupt. It takes 8 CPU clock cycles to fetch the vector and save critical CPU registers. Hence the CPU can quickly respond to interrupt events. Prioritization of interrupts is controlled in hardware and software. Each individual interrupt can be enabled/disabled within the PIE block. In this project the interruptions are essential. The ADC interruption has higher priority because it is executed every 1ms versus the frequency measurement interruption that is executed every 20ms. The remaining time for the frequency interruption (time where the ADC interruption is not executing) is more than enough to complete the desired calculations.
3.1.3.8.
Oscillator and PLL
The DSC can be clocked by an external oscillator or by a crystal attached to the on-chip oscillator circuit. A Phase Locked Loop (PLL) is provided supporting up to 10input clock-scaling ratios. The PLL ratios can be changed on-the-fly in software, enabling the user to scale back on operating frequency if lower power operation is desired. The PLL block can be set in bypass mode in order to use the external source. This PLL feeds the main clock of the DSC, in this project it is set up to work at maximum speed, 150MHz. Using a lower speed results in a lower consume of energy but the time critical and calculation power exigencies of this project should not allow using a lower clock frequency. The calculation in the frequency decay algorithm uses lots of available calculation power in the CPU as well as RMS calculations. Using a lower speed could result in inaccurate frequency measurements in the case that the next frequency interruption is called before the previous one finish its execution.
54
Hardware
3.1.4. DSC Periphery Description This DSC has several peripheries but not all of them are used. In the following sections, the peripheries used in this project are described.
3.1.4.1.
Event Manager
The event manager will be the periphery entrusted with the frequency measurement. The DSC counts with two Event Manager Modules. The event-manager (EV) modules include general-purpose (GP) timers, fullcompare/PWM units, capture units, and quadrature-encoder pulse (QEP) circuits. Every EV module count with two GP timers and three capture units. The two EV modules are similar, EVA and EVB. EVA and EVB functions, timers, compare units, capture units and so on are identical. Event managers A and B have similar peripheral register sets except for different addressing. The capture unit need to be associated with a GP timer within the same module, the associated timer is used as time base of the capture unit. The capture units can use any of the EV timers with the only restriction that the capture units 2 and 3 must use the same timer as time base. The load controller uses two timers, one of each EV module and the capture unit 1 of the EV module A (EVA). One GP timer is used to control the ADC sample rate, this timer is also used to increase the real-time stamp. The capture unit also uses another timer in the measurement of the power grid signal period. More information about the timers usage can be found in following sections.
55
Hardware
Figure 3-III - Event Manager A Functional Block Diagram
56
Hardware
3.1.4.1.1.
GP timers
Each Event Manager (EV) has two General Purpose timers. These timers can be used for any purpose. They are not reserved for DSC internal functions. Every timer has individual configurations registers, so that it can be used for different purposes at the same time. In this project, it is used one timer in frequency measurement. It is used as time base of the capture unit. It will be explained in the section 3.1.4.1.2 Capture unit. Another timer is in charge of the ADC sample. It will be explained in section 3.1.4.1.3 Starting the A/D Converter with a Timer Event, this last timer is also used in the real-time stamp, explained in section 2.6 Real-time stamp.
3.1.4.1.2.
Capture unit
This is a module of the event manager that reacts with external events. It is configurable to detect any change in its inputs. The capture module is configured to detect rising edges in its inputs. It has Schmitt-triggered inputs, so that, when connecting the voltage signal with the input, the corresponding interruption subroutine is called every power grid cycle. This module basically starts the associated general purpose counter used as time base and in every rising-edge, it stores the value of the timer in a register and throws an interruption. When the timer overflows the count starts at zero again. In the interruption subroutine (capture1_srt()) the program reads the registers where the last two measures are stored, by subtracting them the period of the input signal is obtained. The Timer speed is directly related with frequency accuracy. The faster runs the timer the better accuracy is obtained. If the timer counter is increase at too much speed, the counter that measures the frequency can be overflowed within two zero-crossings and the measurements would be wrong. The frequency measurement process deeply explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement.
57
Hardware
3.1.4.1.3.
Starting the A/D Converter with a Timer Event
The event manager can be used to control the sample rate of the ADC. It is possible to configure a Timer to define the period of sampling. The ADC has four special registers called sequence registers, in this registers is stored the sequence of inputs measured. In this project EV is configured to react with timer 1 period interruption. This interruption is generated every 1 millisecond, when the timer count is equal to its compare register. This results in a sample rate of 1kHz, therefore, it samples the input twenty times per cycle.
3.1.4.2.
ADC
The ADC module has 16 channels, configurable as two independent 8-channel modules to service event managers A and B. The two independent 8-channel modules can be cascaded to form a 16-channel module. Although there are multiple input channels and two sequencers, there is only one converter in the ADC module. Figure 3-IV - ADC block diagram shows the block diagram of the F2812 ADC module. The two 8-channel modules have the capability to auto sequence a series of conversions; each module has the choice of selecting any one of the respective eight channels available through an analog MUX. The two 8-channel modules can be configured in a simultaneous sample mode, where the pairs of inputs (one of each 8channel module) can be sampled at the same time. On each sequencer, once the conversion is complete, the selected channel value is stored in its respective ADCRESULT register (ADCRESULT1, ADCRESULT2…). In the project the sequencers are set to measure the instant current and voltage simultaneously two times. The result used is the average of the two measurements, this over sampling will increase the SNR of the meassurments. Although the ADC only has one ADC converter it is possible to sample dual channels using the sample and hold circuit. This circuit samples the channels at exactly the same time and keeps the value until the single ADC converter converts both of the channels. This circuit is necessary in order to have instantaneous measures of current and voltage.
58
Hardware
Figure 3-IV - ADC block diagram Below is a detailed list of the ADC module features: 12-bit ADC core with built-in dual sample-and-hold (S/H) Simultaneous sampling or sequential sampling modes Analog input: 0 V to 3 V Fast conversion time runs at 25 MHz, ADC clock, or 12.5 MSPS 16-channel, multiplexed inputs Autosequencing capability provides up to 16 “autoconversions” in a single session. Each conversion can be programmed to select any 1 of 16 input channels Sequencer can be operated as two independent 8-state sequencers or as one large 16-state sequencer (i.e., two cascaded 8-state sequencers) Sixteen result registers (individually addressable) to store conversion values Multiple triggers as sources for the start-of-conversion (SOC) sequence S/W − software immediate start EVA − Event manager A (multiple event sources within EVA) EVB − Event manager B (multiple event sources within EVB) External pin Flexible interrupt control allows interrupt request on every end-of-sequence (EOS) or every other EOS Sequencer can operate in “start/stop” mode, allowing multiple “time-sequenced triggers” to Synchronize conversions EVA and EVB triggers can operate independently in dual-sequencer mode Sample-and-hold (S/H) acquisition time window has separate prescale control
59
Hardware
4000 3500 3000 2500 2000 1500
Input Voltage Offset
1000 500 0 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99
ADC output (integer Units)
Input Voltage
Number of sample
Figure 3-V – ADC example Voltage output
The figure above shows the output of the ADC in the voltage channel and the respective offset. The offset is fixed to the value measured in absence of other input signal. The ADC output is a series of integer values between 0 and 4096. The integer value corresponds with the range from 0 to 3 volts. The integer results are convertible to voltages present at ADC input with the following equation. PcZ- 0
JcZ-S[V@]k[ N EE,OAL D : FPI X>YC
(3.1)
ADCmeasure represents the value returned by the ADC, in instance of the voltage channel, the OFFSET has the value of 1921. A similar equation can be used to obtain the current at the input of the current ADC channel.
zcZ- 0
60
JcZ-S[V@]k[ N EE,OAL D : FPI X>YC D h
(3.2)
Hardware As well as in the previous equation, ADCmeasure represents the value returned by the ADC, in the case of current, the OFFSET has the value of 1904. The R in the denominator represents the resistance of the resistor connected to the current transformer. The ADC is only be able to measure the voltage, so that, the ADC measures the voltage dropped in a know resistor, being direct to obtain the current thought it. It is used a 10Ω resistor. The OFFSET is the integer value corresponding to the level shifter input. This offset is added in order to sample the input signal that varies from negative to positive values. The ADC is only able to work with positive signals. The offset cannot be measured every signal measurement, that is because the level shifter inverts the input signals, so that, the offset is regulated to the invert value. In order to improve the accuracy of the measurements, the offset could be measured every time the signal is sampled. The offset should be obtained from the level-shifter reference input but it has to be inverted, due to the stability of the offset reference value and the fact that the offset has to be inverted, it is used a fixed offset value. There are more details about the level shifter in section 3.2.1.3 Level shifter.
3.1.4.3.
General purpose I/O
Most of the periphery pins are configurable to be used as a General purpose input/output (GPIO). In this project, one pin is used as I/O. This pin controls the relay that will disconnect or reconnect the load. It could be possible to use more than one pin in the case that more reactions were needed against a frequency event. There is more details about the General purpose Inputs/outputs in section 4.3.3.2.3 GPIO configuration.
61
Hardware
3.2. Board Description The load controller needs some special hardware specially designed to achieve its purposes. The included hardware in the DSC developing board (more details in 3.2.4 eZDSP F2812) is not enough. It only has RAM memory connected to the on-chip external memory interface, direct connections with the chip pins, the PC connection (thought JTAG port) and the usual ADC external reference hardware (two capacitors). The load controller has to measure the frequency and the power of the power grid. The wall outlet output has to be transformed in order to be measured. First of all the AC power grid at 230V cannot be connected to the 3V inputs of the microcontroller’s ADC. It is needed to reduce the voltage to the allowed range. The DSC has only ADCs that convert voltage to digital values but in order to calculate the power we also need to know the Current. Therefore, a current to voltage converter is required. It has as well to be reduced to a compatible range. The power grid is affected by noise, external interferences, and also by harmonics. In order to obtain the cleanest possible measurements the input signal must be filtered. Once the input signal is reduced and filtered, It has to be adapted to the DSC voltage input range. The input signal varies between positive and negative values. The DSC, and in consequence the ADC, only works with voltages between 0 and 3 Volts, so it is also needed to shift the signal to only positive values. It is done by adding to the input signal and offset that centers it in the middle of the ADC range. Regarding to the power supply, all parts of this hardware need to be powered so that, a single positive power supply for the DSC and a double power supply for the filters are included in this board.
Figure 3-VI - Board distribution
62
Hardware
3.2.1. Signal interfacing and conditioning This section it is describes how the signal coming from the power grid is conditioned in order to obtain a proper signal for the DSC hardware.
3.2.1.1.
Measurement transformers
These transformers are in charge of reducing the power grid signal at 230V to a range that can be used by the microcontroller. These transformers have lots of repercussion in the final measures. Therefore, high-precision transformers are used at this point.
Figure 3-VII - Measuring Transformers The voltage Transformer is a high-precision toroidal transformer. It has a conversion ratio of 325:1.65, its gain is fixed and cannot be changed. It gain is calculated in the following equation (3.3):
63
Hardware yx= 0
:12 0 >, >>2 FP/PI <, C2
(3.3)
The current transformer ratio is variable, the installed transformer has 50 loops in the primary coil and the ratio can be change according to the number of loops thought the transformer. The gain of the current transformer is the following (3.4) : K Fc/cI 2> Where N represents the number of loops thought the primary coil.
y^= 0
(3.4)
The number of loops cannot be set arbitrary. The current at the ADC input cannot have amplitude higher than 150mA. Higher amplitude causes voltages with amplitude higher than 1.5V which saturates the range of the ADC. In order to measure the test load (A heater) it is used two loops. Therefore the current transformer ratio is 50:2 which results in this special case in a current gain of: . K 1 (3.5) y^= 0 0 0 >, >X Fc/cI 2> Kl1 2> The gain of the transformers is calibrated in the last stage of the signal conditioning hardware, the level shifters, it is explained in section 3.2.1.3 Level shifter.
3.2.1.2.
Analog filters
The power grid signal is affected by noise and harmonics. In order to obtain accurate measurements active analog filters are used. In both of the channels, voltage and current, it is used butterworth second order filters. The signals outside the working band will be reduced 40dB per decade, the signal inside the band is not affected, it has a gain of 0dB. The measurement band is centered in 50Hz and has a band-width of 1Hz. This band is too small to be implemented with real components without compromise other
64
Hardware filter characteristic like the phase spread. Therefore, it is used a real pass-band filter with a band-width of 20 Hz. It fulfills the purposes of the project because it sufficiently reduces the noise outside the working band and the first harmonic at 100Hz.
Figure 3-VIII - Filter transfer function
In the figure above it is show the transfer function of the filters, it is a second order filter so it has a gain of -40dB/decade outside the pass band. The band-width is exactly 20.96 Hz at -3dB. The label at 100Hz corresponds to the first harmonic, that it is reduced in -22.46dB.
Figure 3-IX – Filter photo
65
Hardware The scheme of the filters is the following:
Figure 3-X- Filter Scheme
3.2.1.3.
Level shifter
The level shifter is the last stage of the signal conditioning hardware. The ADC of the DSC works with voltages between 0 and 3 Volts but the signals at the output of filters is in the range of ±1.5 Volts. The level shifter consists on one Operational Amplifier (OA) in the configuration of an inverter mixer, whose inputs are connected with the voltage or current signal and with the offset reference. The level shifter has to add an offset of 1.5 Volts. This offset centers the signal in 1,5Volts and removes the negative voltages. Due to the inverter configuration, the offset reference has a value of -1,5V which is converted to 1,5V at the level shifter output. The level shifters is as well used as gain stage (The filter stage has no gain in the working band). The gain of the level shifters is calibrated to expand as much as possible the input signals into the ADC input range. Therefore, in the voltage channel the level shifter is calibrated to output the power grid signal with 3 Volts in peak-to-peak and with a DC offset of 1.5 Volts.
66
Hardware The current channel calibration cannot be done without knowing the controlled load, depending on the load the current amplitude can change. As seen in section 3.2.1.1 Measurement transformers the current signal range can be calibrated in the current transformer.
Figure 3-XI - Voltage level Shifter Scheme
The gain of the voltage level shifter is calculated with the followings equations: yx@ 0 N
h1 <>> D <>: P 0N 0 N>, Y ` a : h< <<> D <> P
(3.6)
It is also added an offset of -1,5V Offset, with a gain of -1. Finally the transfer function of this level shifter is the following: 0 N>, Y D <, 2 FI
(3.7)
The current channel is similar in exception of the resistor R1. It can be observed in the following scheme.
67
Hardware
Figure 3-XII - Current Level Shifter Scheme
The gain of the current level shifter is calculated using the following equation: y^@
h1 <>> D <>: P 0N 0N 0 N>, :1 ` a :<< D <>: h< P
(3.8)
Note that the units of this gain are [V/V] because at this point the current is yet converted to voltage. As well as in the voltage channel it is added an offset of -1,5V Offset, with a gain of -1. Finally the transfer function of this level shifter is the following: P 0 N>, : D PT <, 2 FPI
68
(3.9)
Hardware
Figure 3-XIII - Level shifter result The input signal inversion do not affect in the final calculations, the RMS calculations uses the square of the ADC samples, therefore, a change of sign is not representative.
3.2.2. Current and voltage controller inputs
values
at
load
Section 3.1.4.2 ADC described how to obtain the values of current and voltage in the ADC inputs but these values are not the same as the values present in the power grid. This section describes how to obtain the values at the wall outlet the values acquired by the ADC. In order to know the real values is needed to consider the voltage and current transformers, as well as the gain of the level shifter. Considering them, the current and voltage in the load controller inputs can be obtained with theses equations: P 0 PcZ- D
< C2> 0 PcZ- D yx :
(3.10)
Where gv is the total gain of the voltage channel:
69
Hardware P : yx 0 yx= D yx@ 0 >. >>2 D >. Y< 0 >, >>XC ` a 0 P C2>
(3.11)
Where gvt is the total gain of the voltage transformer (3.2.1.1 Measurement transformers), and gvls (equation (3.3)) is the total gain of the voltage level shifter. The instant peak-to-peak power grid amplitude is 650 V and the 3V is the peak-topeak amplitude in the ADC input. The level shifter gain is calibrated to obtain a 3V peak-to-peak signal from the transformer output that has a fixed transfer ratio. The voltage don’t depend on the load, therefore the equation above is suitable with any load. The current calculations are similar to the voltage calculations. The equation for obtaining the current at the load controller inputs is the following: z 0 zcZ- D
< y^
(3.12)
Where gc is the total gain of the current channel, the current measured depends on the load connected with the load controller inputs. The current transformer gain can be configured according to the load connected. P y^ 0 y^= D y^@ ` a P
(3.13)
Where gct is the total gain of the current transformer, and gcls the total gain of the current level shifter.. The total gain in the current channel in the particular case of the test load, two loops through the primary coil, is the following: P
y^ 0 y^= D y^@ 0 >. >X D >. : 0 >, ><1 P
70
(3.14)
Hardware
3.2.3. Power Supply The board has two different powers supplies. A 5 Volts power supply to power the DSC and a double ±5Volts power supply to power the filters and the signal drivers. It is used two power supplies because the power supply used in the filters introduces interferences in the DSC which needs s supply as stable as possible. The energy is obtained from the power grid but the output of this power supplies will not be used to measure the frequency or the power of the line. These supplies are used only to power the system. It is used a simple power supply based on a bridge rectifier and on a voltage regulator. In the Figure 3-XIV - Single power supply scheme it is show the scheme of a single power supply. The double power supply will be exactly the same with exception on the voltage regulator that will be a negative voltage regulator (LM7905). A detailed scheme of both of the power supplies can be found in annex B Board Schemes. In the rest of this section it is explained the function of the power supply. In the Figure 3-XIV, which is below this paragraph, the letters from A to D represents voltage measurers or oscilloscopes and in following figures the measurements at these points is showed.
Figure 3-XIV - Single power supply scheme At point A its present a typical sinusoidal signal at 230 Volts (RMS) from the power grid. In point B it is found the signal with the voltage reduced, its amplitude is around 10Volts. The output of the transformer is connected with a bridge rectifier (DB101G). This component is composed of 4 diodes that invert the negative part of the signal. The output result, at point C, without connecting anything more is the following.
71
Hardware
Figure 3-XV - Measure at the bridge output, point C (Without capacitors) At this point the power supply offers only positive voltages, but it is not the constant value in a DC power supply. In order to “fill up” the valleys it is used two capacitors in parallel, they works as a very low low-pass filter. The results of at this point, point C with the capacitors installed ,can be watch at next figure, in the magenta plot.
Figure 3-XVI – Power supply output As observed in the graphic above the output starts to be similar to a DC power supply but the curl is too much big to be useable (magenta plot).
72
Hardware In order to reduce this curl we used a voltage regulator. This component “cuts” the input signal to a fixed value, the regulator used regulates to the voltage to 5V (cyan plot). It consumes the excess voltage, so it gets warm and sometimes is needed a heat sink. Finally it is used a capacitor in order to stabilize the DC level with the ground. The double power supply is similar to the single one, in exception that it uses each transformer output for each voltage level. It also uses other voltage regulator (LM7905C) for the negative output that regulates the voltage to -5 Volts.
Figure 3-XVII - Double Power Supply scheme
3.2.4. eZDSP F2812 In order to start the prototyping of any hardware based on the DSC F2812 it is used the developing board eZdsp manufactured by Spectrum digital. The developing board is a board with the DSC integrated and the minimum additional hardware needed to develop any project. Te hardware included in eZdsp is the following: • One tms32F2812 DSC •
PC parallel communications port, This port is used to connect the device to the PC, thought this port the programs are loaded into the DSC memory. It is also possible to control the execution and watch the contents of the memory in real time in order to debug the developed software
73
Hardware •
64Kx16 of extra external RAM. The board has this memory attached to the external memory interface, it is used to extend the code and data memory.
•
ADC analog interface, This port is used to connect the external reference of ADC and the inputs to sample, the board also includes the extra hardware needed in the case of using the internal reference.
•
I/O interface, This port are the friendly pins connected with the on-chip DSC peripheries, these pins are multiplexed with the GPIO pins.
eZdsp is fully supported by the IDE developed by Texas instruments, which is included in the developing board kit, as a reduced version only functional with this DSC.
Figure 3-XVIII - eZdsp F2812 photo More details can be found in the technical reference provided by Spectrum digital. The reference can be found in the Bibliography section
74
4
SOFTWARE
This section describes all the software process. Once the hardware is designed this is next step in the load controller development, where the processor is programmed to carry out its duty. This section uses two different types of font styles. The regular type, Times new Roman, used in regular text, and console type used in source code Example source code text
4.1. IDE The Integrated Development Environment (IDE) is the software installed in the PC that allows the user to develop, debug and load a program into the DSC. The quality of the IDE is decisive in order to obtain a good quality program. The IDE used in this project is Code Composer Studio (CCS) which is included in the developing board. This IDE is one of the best IDEs in the market. The provided CCS has the only limitation that it can only be used with the included DSC. It cannot be used with other DSC, DSP or microcontroller. The IDE has all the featured usually present in a microcontroller environment. It has a text editor with a syntax coloring engine. It manages all the source code and header files. The IDE is able to compile and link the code and obtain output binary program that is loaded into the DSC.
75
Software The IDE also provides the user debug functions like the possibility of adding software beaks and follow the execution step by step, as well as watch the values stored in any memory location at any time.
4.2. User Manual This section contains the basics to use the CCS IDE, for more details it should be referred to the CCS included tutorial. The first step is opening the application. The main window looks like this:
Figure 4-I - Code Composer Studio main window It has the look-and-feel as most of the applications available under Windows operative system. In this brief tutorial it is supposed that the user is familiarized with the Windows operative system.
76
Software
4.2.1. Developing board connection It is not needed to change any configuration in Windows neither in CCS. It has to be connected the parallel port wire from the ezDSP board to the PC parallel input and finally it is needed to connect the power socket to the power grid. Once all the connections are done it hast to connect the CCS with the board. It is done by clicking the option Debug->Connect in the main window.
Figure 4-II – Connect If the connection succeeds The picture of a socket in the bottom left corner goes from red to green, it appears a balloon message advising that the connection is done and it appears a new window that shows the current contents of the DSC memory.
Figure 4-III - Board connected message
77
Software
4.2.2. The project All the CCS developments are formed by a CCS project. The CCS project is folder that contains a file with the extension .pjt, this file contains all the IDE configurations made for the development, as well as the location of all the other files included in the project. These files should be located in the same folder where the pjt file is located. Otherwise it should be possible to miss files if the application is ported to other computer.
In the menu Project->Open… is possible to open a existing project, it can be selected new… to create a new one.
Figure 4-IV - Open existing project
78
Software When the open option is selected it appears a regular opening windows dialog.
Figure 4-V - Open Project dialog It has to be selected the .pjt file, after clicking the Open button (Abrir) the project is open and the main window changes.
Figure 4-VI - Opened project
79
Software The File view tab shows all the files involved in the load controller development. By double clicking any file it is opened to be modified. For example if it is double-clicked the main source code file proyectoMain.c it is showed on the right part of main window.
Figure 4-VII - Main file source Code
80
Software
4.2.3. Configurations The general, compiler and linker configurations can be changed in the build menu. For example the following figure shows the linker configuration window.
Figure 4-VIII - Linker configuration It is possible to have simultaneous profiles with different configurations.
Figure 4-IX – Select Configurationt It is possible to create a new profile in the project menu option configurations…. More details about the configuration can be found in the CCS included tutorial.
81
Software
4.2.4. Modify, build and load a project If it is desired to modify the source code of any file it has to be opened as explained in the previous version. Once the changes are made the files have to be build in order to be loaded into the DSC memory.
Figure 4-X - Build a project This option compiles all the files modified, if wanted it is possible to rebuild all the files modified or not by clicking on rebuild all. After clicking the build option it is showed a report of the building on the botton part of the main window.
Figure 4-XI - Build report If the compilation ends without any problem the report should look like the figure above. Once the project is correctly compiled the next step is to load it into the DSC memory in order be executed. The program is loaded by clicking on the menu option Load Program inside the menu File.
82
Software
Figure 4-XII - Load Program It appears a dialog similar to the open project one. In this dialog it has to be selected the file .out contained in the folder debug that is inside the main folder, where the pjt file is contained. The debug folder corresponds to the output generated by the configuration named debug, it is possible to create various configurations and use any of them at any time. In this project it is only used one configuration.
Figure 4-XIII - Load project output
83
Software Once the building output is loaded one time, it is possible to reload it bypassing the open dialog, the project could be modified and rebuilding and the reload option will reload the last built project.
Figure 4-XIV - Reload Project The loading process is the last step to run a program, if the loading process finish successfully the program can be run. A loaded program is run by pressing the option run in the menu debug. The execution of a program is stopped by clicking on the option halt, also in the menu debug. It is not possible to reload a program while a program is running.
Figure 4-XVRun a program
84
Software
4.2.5. Debug project The IDE CCS offers lots of debug tools. One of the tools is the watch window, by right-clicking on any variable in the source code and selecting the option add to watch window… it is possible to watch the contents of the variables in real time.
Figure 4-XVI - watch window
Other typical debug tool is the control of execution, CCS offers the possibility to set software and hardware break points, the execution stops in these points and can be controlled. For example it is possible to run to the next break point, execute just one assembler instruction or one c instruction.
85
Software A software break point can be easily set by double-clicking on the left of any line of code. It can be watch in the next figure, where the red circle is the break point and the yellow arrow is the point of execution.
Figure 4-XVII - Break point
4.2.6. Gel File The CCS has the possibility to be programmed and personalized madding use of the programming language GEL. It is possible to load a GEL file in CCS, only one for all the projects opened, but in can be changed at any time.
Figure 4-XVIII - GEL file
86
Software A possible personalization is the possibility to add a options menu with personalized actions.
Figure 4-XIX - GEL Menu In this project the GEL file has been modified to add in the watch window the relevant variables and arrays every time the program is loaded.
Figure 4-XX - Example code of GEL language It is encouraged do the getting started tutorial included in the CCS help center.
87
Software
4.3. The load controller program This section describes the structure of the files that compound the project and the relevant source code.
4.3.1. File structure A project developed with CCS is form by many files all of them grouped in a CCS (Code Composer Studio) project. The project root folder contains six files and four folders, not all of them are relevant in the project, some of them are automatically generated in the compilation process and they are logging files or temporal files. his section only describes the relevant files used in the project. The project root folder contains the following files: alex_project.pjt This is the CCS main project file. It stores the configuration of the compiler and linker, as well as the list of the files that compound the project. This file is essential in order to work with the IDE. proyectoMain.c This file contains the most of the source code of the project. It is deeply described in following sections. The other files in the root folder are automatically generated by the IDE. One of them contents information about the CPU (alex_project.sbl). The IDE uses it to generate the specific assembler code for the CPU. Other file contains temporal information about the project loaded (alex_project.paf2), It allows the IDE not to phrase the project file every time a change is made, only when opening an saving the project. The other files with the extension “log” are loggings about the internal process of the compiler. The root folder contains the following three folders:
88
Software Debug This folder contains the output files of the compiler, binary code generated from the source code. The content of this file is the program that is loaded into the DSC memory. DSP281x_common and DSP281x_headers These folders contain the files provided by Texas instruments needed to develop a program for the DSC. They contain the CPU initialization subroutines and the registers structures. The register structures relates the register memory address with user-friendly names which allows the user not to use the memory address when writing of reading a register, only the mnemonic assigned names. alex_project.CS_ This folder is a temporal automatically generated folder. It is used by the IDE in internal process, for example, to store the CPU variables that are wanted to be showed in the screen.
4.3.2. Memory mapping The memory available is scarce. Although this DSC has the possibility of use flash memory it is only used the RAM memory. In order to use the flash memory a JTAG emulator is needed. However, in the developing of this prototype the RAM memory is enough.to archive the prototype purposes. There are two different types of RAM, the integrated in the DSC core and the external RAM included in the eZDSP board (developing board). The RAM has to be shared between data and code, due to the DSC Harvard structure. The memory mapping has to be made before the compilation and it is not possible to change it after compilation. Memory mapping is defined in the command file “F2812_EzDSP_RAM_Link.cmd”. It is stored inside the DSP281x_common folder, it has to be modified specifically for the project. The main difference between the internal and external RAM is the speed. The internal RAM is a single access RAM (SRAM), very fast, and the external RAM is slower, it uses several CPU cycles for a single access.
89
Software
The faster RAM is used to store time critical subroutines code, the other subroutines and in the initialization process uses the external RAM. The appendix Memory mapping shows the code to distribute the memory. The memory mapping file is divided into two sections, MEMORY and SECTIONS. The memory section defines the addresses and lengths of memories; it divides physical memory into zones. The memory is divided in two pages; each page has different functions. PAGE 0 is used to store the code of the program and de PAGE 1 to store the data. The DSC has Harvard architecture and it has CODE and DATA moved a part. The second part of the memory mapping file is the SECTIONS part. In this part the sections of the program are defined and it is assigned a memory zone to them. Any program has different memory sections, like the initialization code, the stack, the user code and so on. It is possible to assign several sections to a single memory zone; the memory is shared between all sections assigned to it. In the case that a memory zone has not enough memory for all sections designed the program cannot be compiled. By default the initialization sections are assigned to external ram, they are executed once and they are not time critical. The stack and other internal sections are allocated to internal RAM. These sections will access to the memory very frequently and the location of these sections affects too much to the global speed. The user code is assigned to external ram and user data to internal ram. There are some exceptions, user code has to be executed as fast as possible, in instance of frequency measurement subroutine, is stored into the internal SRAM. In order to select where data of functions code are stored there are two special compiler directives. Used to choose the memory section where a variable or a function is allocated. They are only used when it is needed to change the location of the function from the original section defined in the memory mapping file. The directive to allocate the date in a desired section is:.
90
Software #pragma DATA_SECTION(variable, ".section"); To allocate a specific function in a desired section it is used the directive: #pragma CODE_SECTION(function, ".section"); The data directive can only allocate data within the data page sections and code directive within code page sections. By default the user code is allocated in external ram which is slow. But there are two time critical subroutines in the project, the one that measures the frequency and the one that samples de input signals. These subroutines are periodically executed in a short period of time, they have to be executed as fast as possible in order to avoid that the next execution of that subroutine were before the previous execution has finished. They are allocated in the code section “ramfuncs” which is defined into the internal RAM memory zone.
91
Software
4.3.3. Program structure This section describes the structure of the program. The figure below shows the program flow.
Program start
MAIN FUNCTION
Capture1_isr Frequency measurement
Initialization process
Average calculation RMS calculation
Main loop
Create record If ZeroCrossi ng
Threshold checking
If ADC Ready
ADC_isr
idle
Get ADC results Store ADC results
. Figure 4-XXI - Simplified program flow The start point is the red circle on the top of the figure, the programs stars its execution at that point. The initialization of the load controller is done in the first lines of the main function. The initialization process consists on configuring the peripheries implied in the functions of the load controller. It is done the initialization of the internal peripheries such as the PLL, the watchdog and ADC. Then they are filled up the ADC and EV configuration registers. The initialization process ends by setting up the interruption vectors of the ADC and EV interruption subroutines and enabling all the interruptions.
92
Software Once the initialization is finished the main function enters into an infinite loop without any function. At this point the program waits for the interruption events and calls the interruption subroutines. The most of the work it is done in the interruptions. Other functions are used in the program but they are not relevant in the program flow.
4.3.3.1.
Headers
This section describes the header of the program. The headers define the library inclusion, the special directives to map functions or variables to a specific part of memory, the definition of the function prototypes and the definition of constants and variables.
4.3.3.1.1.
Inclusions
The load control program includes three files. #include "DSP281x_Device.h" #include "DSP281x_Examples.h" #include <math.h>
// DSP281x Headerfile Include File // DSP281x Examples Include File //mathematical functions and constants
The first and the second file included are the libraries provided with the microcontroller, these libraries contain the CPU initialization subroutines and the userfriendly names of registers as well as the interruption vector. The last file is the mathematics library that is provided by the compiler. This librarie contains the mathematical function such as sin, cos, sqrt or constants as π or e.
4.3.3.1.2.
Constant definition and Location of attributes
As explained in section 4.3.2 Memory mapping, it is possible to map any data structure or variable to the internal or external RAM depending on its time requirements. The mapping directives take place in this part of the header section. They are also defined in this part of the header the constant used in the rest of the program.
93
Software According to the memory mapping, explained by the section 4.3.2 Memory mapping, by default, the variables are allocated in the on-chip RAM. The directive #pragma DATA_SECTION(variable_name,“memory_zone”) allocates a variable or data structure to the specified zone of memory (in instance of memory zone “.extRAM” that is allocated in the external RAM memory). The constants are defined with the directive #define name value. In the compilation process the compiler replaces the constants names with its value. frequency measurements constants : number of frequency measurements used in average #define AVERAGE_FREQUENCY_VALUES 25
Recover frequency threshold in integer units (Running at 25MHz) define DEFAULT_HIGHER_LIMIT 62525
Frequency event threshold in integer units (Running at 25MHz) #define DEFAULT_LOWER_LIMIT 62626
length of averaged frequency measurements array #define AVERAGE_MEASURES 5
Frequency decay filter constants: Maximum allowable frequency change between frequency measurements.,150 integer units/cycle= 6Hz/s #define MAXIMUN_FREQUENCY_CHANGE 150
frequency decay measurement constants: //square of FCF, It is needed in frequency decay calculations; it is defined with one decimal value to specify that the number is a float number. #define FCF2 976562500000.0
94
Software frequency decay event threshold #define DECAY_THRESHOLD -0.16
number of decay measurements used in the average #define AVERAGE_MEASURES 5
RMS constants: Number of samples used in RMS calculations #define ADC_MEASURES_PER_CYCLE 40 Offset present in current and voltage channels. #define VOLTAGE_OFFSET 1904 #define CURRENT_OFFSET 1904
Realy constants:
The different load controller states are the following: 0=OPEN CIRCUIT 1=CLOSED CIRCUIT RECONNECTION
2=WAITING
FOR
Define the initial state #define DEFAULT_RELAY_STATUS 0x1 Define the states. #define OPEN 0x0 #define CLOSE 0x1 #define WAIT 0x2 Define the 1500*0.02s=30s
maximum
//relay status //relay status //relay status time
to
re-attachment
power
line
cycles.
#define MAXIMUN_WAIT_TO_ATTACHMENT 1500
Control logic type II Constants Kf value in Control logic type II equation #define KF 1
Nominal frequency in Nordic systme, integer units, 50Hz #define FO 62500
95
Software Data recording constants: This constant defines the number of records stored in external memory simulating the external memory #define STUB_EXT_MEMORY_LEGTH 50
The value of these constants cannot be changed once the program is compiled. The constants are named by collective agreement in uppercase letters, it makes easy to distinguish them from variables. Some of the values that could be changed in running time are stored in the initialization process in a variable, from that point it can be modified while the program is running. These constants define the initial configuration of the load controller. As it is explained in the chapter future work, it should be better to define these values in the external flash memory that it is not erased if the hardware is powered down, and if they are modified the modifications remains after a reboot.
4.3.3.1.3.
Functions definition
This part of the header defines the prototypes of the functions. It also allocates some functions to and specified part of memory. As default the functions are allocated in the external RAM memory, which is slower than the internal memory. Using the directive: #pragma CODE_SECTION(name_of_function, memory_zone); it is possible to allocate the time critical into the faster internal RAM, in instace of code section “ramfuncs” that is allocated in the internal ram. As in the data allocation directive, the memory zone must be defined in the memory mapping file and it must to be in the code page.
/*** frequency measurement functions ***/ //definition of the interruption subroutine of capture unit 1 //and allocation in internal RAM interrupt void Capture1_isr(void); #pragma CODE_SECTION(Capture1_isr, "ramfuncs"); /*** data acquisition functions ***/ //definition of the interruption subroutine of ADC //and allocation in internal RAM
96
Software interrupt void adc_isr(void); #pragma CODE_SECTION(adc_isr, "ramfuncs");
/*** relay control functions ***/ void close_relay(void); //opens the load relay void open_relay(void); //closes the load relay //sets the frequency recover threshold void set_higher_limit(unsigned int); //sets the frequency event threshold void set_lower_limit(unsigned int); /*** time stamp ***/ //synchronizes the real time counter void time_synchronization(long);
/*** decay functions ***/ //sets the frequency decay threshold void set_decay_limit(float);
Control logic type II function. This function changes the temperature set points. void set_points_update(unsigned int,unsigned int); /*** storing stub ***/ //Function that should store the recors in external memory void recod(long time,unsigned int period,unsigned int voltage,unsigned int current,unsigned long power, float decay,short relay);
The functions are defined in the same way as is done in ANSI C. First of all it is defined the type of value that the function returns (void means nothing), next it is defined the name of the function, and finally, in brackets, the parameters and its types. All the functions are explained in following sections.
97
Software
4.3.3.1.4.
Variables definition
This part of the header defines the variables of the project. All the variables and arrays in C have to be defined before being used. In this part of the header it is assigned the type (short, integer, unsigned integer…) to the attributes and if necessary, some of them are initialized to a specified value.
//generic variables short i; //generic index short j; //generic index /*** frequency measurement ***/ // Temp integers variables in capture unit subroutine unsigned int temp1_1; unsigned int temp1_2; […] //table of average frequency measurements unsigned int average_measures[AVERAGE_MEASURES]; //index used in frequency measurement array unsigned int measure_index=0;
unsigned int average_measure; //average period measure
/*** Frequency decay measurement ***/ //table of frequency decay measurements float decay[AVERAGE_MEASURES]; //last frequency decay measurement float average_decay=0; //frequency decay threshold definition and initialization float decay_limit= DECAY_THRESHOLD; /*** relay control ***/ //frequency recover threshold initialization unsigned int higher_limit=DEFAULT_HIGHER_LIMIT; […] /*** data adquisition ***/
98
Software
//ADC current measures without offset unsigned int adc_samples_current[ADC_MEASURES_PER_CYCLE]; […] /*** RMS calculations ***/ //square of ADC measures in RMS calculations unsigned long adc_square_current[ADC_MEASURES_PER_CYCLE]; unsigned long adc_square_voltage[ADC_MEASURES_PER_CYCLE]; […] unsigned int voltage_RMS; //voltage RMS calculation result […] //temporal variables in RMS calculations long temp_results; long temp2_1; […] /*** Time stamp ***/ //time variable definition and initialization long current_time=0;
/*** external memory stub ***/ unsigned int ext_index=0; […] short ext_relay[STUB_EXT_MEMORY_LEGTH]; //relay s.
Not all the variables and arrays used are showed in the piece of code above, the whole variable definition section can be found in the appendix A.1 Main program source code.
4.3.3.2.
Device initialization
The main function is the first function executed by the DSC. This function does the initialization task. Once time the control device is initialized the main function ends and all the actions are done by the interruption subroutines.
99
Software This section describes the functions and the registers that are involved in the initialization process. Some of the registers used in the initialization are ELLAOW protected registers. The directives ELLAOW and EDIS are used to write in protected registers. The EALLOW protected registers are protected against spurious writes after configuration. It can only be modified between these two directives. These directives are ignored in the description of the code in the following sections.
4.3.3.2.1.
System Control Initialization
InitSysCtrl(); // Init the system control
This function initializes the system control. This function is provided by Texas Instruments. This function initializes the internal peripheries, the PLL, peripheral clocks and so on. The PLL sets the speed of the CPU . It is initialized with the default speed of 150Mhz.
4.3.3.2.2.
Peripheral High speed clock configuration
This code configures the speed of the High Speed Clock (HSCLK). The peripheries are clocks are fed by this global clock, subsequently the peripheries uses a prescaler to adapt the speed to its requirements. //set HSPCLK to SYSCLKOUT / 6 (25Mhz assuming 150Mhz SYSCLKOUT) EALLOW; SysCtrlRegs.HISPCP.all = 0x3; // HSPCLK = SYSCLKOUT/6 EDIS;
The configuration of the HSCLK is done by setting up the prescaler applied to the main clock , SYSCKOUT, that runs at 150Hz. It is set to obtain the clock frequency of 25MHz that is the maximum that the peripheries support. The register of the High speed peripheries prescaler (HISPCP) is EALLOW protected.
100
Software
4.3.3.2.3.
GPIO configuration
The next code sets up the DSC pins that are used as General purpose Input/Output (GPIO). The GPIO multiplexer registers are EALLOW protected. EALLOW; // Enable write protected registers GpioMuxRegs.GPAMUX.bit.CAP1Q1_GPIOA8 = 1; // Sets pin as Caputure unit GpioMuxRegs.GPFMUX.bit.MCLKXA_GPIOF8=0x0; //sets as I/O GpioMuxRegs.GPFDIR.bit.GPIOF8 = 0x1; //sets as output EDIS; // Disables write protected registers modification
Most of the DSC GPIO pins are multiplexed with peripheries. By default the pins are used by the peripheries. The pins of a periphery which is not used can be used as GPIO. Each periphery has a register associated where is possible to set the pins are GPIOs or as periphery pin, it is also possible to chose the direction of the GPIO pins. The first line, ignoring the directives, sets the first pin of the capture unit to work as the capture input. It is configured in the capture unit GPIO register number 8. The second line sets the pin, which is going to be used to control the load relay, as a general purpose IO. It is set in the register of the McBSP, this periphery is not used in the project, that is because their pins are used as GPIO. The last line specifies the direction of the GPIO. It sets the pin as an output.
4.3.3.2.4.
PIE initialization
The next directive disables all interruptions. The initialization process is not desired to be interrupted, particularly in the initialization of the PIE vectors. // Disable CPU interrupts DINT;
This function starts the interruption system. // Initialize the PIE control registers to their default state.
101
Software // The default state is all PIE interrupts disabled and flags // are cleared. InitPieCtrl();
The next step is clear all the interruption flags, it sets all the interruptions as attended and cleared. They are also disabled all the interruptions individually. // Disable CPU interrupts and clear all CPU interrupt flags: IER = 0x0000; // Disables al interrputions IFR = 0x0000; // Clear al interruption flags
Then it is initialized the vector table. It uses the default interruption subroutines even if they are not used. // Initialize the PIE vector table with pointers to the shell Interrupt // Service Routines (ISR). // This will populate the entire table, even if the interrupt // is not used. InitPieVectTable(); Finally they are set up the interruption vectors used in the load controller program. The ADC vector is used for the sampling subroutine and the Capture unit vector is used for the frequency measurement subroutine. EALLOW; // This is needed to write to EALLOW protected register PieVectTable.CAPINT1 = &Capture1_isr;// sets capture 1 unit interruption routine as Capture1_isr PieVectTable.ADCINT = &adc_isr; //sets adc_isr as interruption routine of ADC EDIS; // This is needed to disable write to EALLOW protected registers
This piece of code stores the address of the interruption function in the interruption vector. Every time that the interruptions requirements are fulfill, it is called the function stored in its interruption vector.
4.3.3.2.5.
ADC initialization
The next initializes the ADC internals hardware, it is provided by Texas Instruments
102
Software
InitAdc(); //init ADC.
The ADC measures buffer index is initialized to 0. ConversionCount = 0;
The next lines of code configure the ADC sequencer to sample only the desired channels. Setup 3 conversions on SEQ1 AdcRegs.ADCMAXCONV.all = 0x0002;
Setup ADCINA0 & ADCINB0 as 1st SEQ1 conv. Current and voltage channels AdcRegs.ADCCHSELSEQ1.bit.CONV00 = 0x0;
Setup ADCINA1 & ADCINB1 as 2n SEQ1 conv. Current and voltage, second sample AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x1;
Setup ADCINA2 & ADCINB2 as 2n SEQ1 conv. Read User normal high and low.temperature AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x2;
The next lines of code enables the event manager to start the ADC’s sequencer one it is also enabled the interruption of the sequencer one. The end of sequence interruption calls the ADC interruption subroutine, the sampling subroutine, once al the conversions configured in the sequencer are done. AdcRegs.ADCTRL2.bit.EVA_SOC_SEQ1 = 1; AdcRegs.ADCTRL2.bit.INT_ENA_SEQ1 = 1;
// Enable EVASOC to start SEQ1 // Enable SEQ1 interrupt (every EOS)
The next line activates the simultaneous mode. Therefore the sample and hold circuits are activated and the pair of channels are sampled at the same time and held until both of the conversions are done. AdcRegs.ADCTRL3.bit.SMODE_SEL = 1; //active the simultaneous mode in order to measure current and voltage at the same time
103
Software
4.3.3.2.6.
EV configuration, timers and capture unit
First of all the general purpose timers configuration registers are cleared. /* EVA GP timers configuration */ EvaRegs.GPTCONA.all = 0; // Clears configuration
The next lines configure the timer 1 that controls the ADC sample rate and real time stamp. /* Configures GP timer 1 with ADC sample rate */ EvaRegs.T1PR = 0x61A8;
// Setup period register
The value of this register is 25000 in decimal base, the prescaler is disabled, so that, the timer runs at the High Speed periphery clock, at 25Mhz. The timer required to reach the count of 25000 takes 1 millisecond, which is the desired sample rate. Enable timer to start ADC EvaRegs.GPTCONA.bit.T1TOADC = 0x2;
// Enable EVASOC in EVA
Finally the timer is configured in up-count mode and in compare mode. The compare mode allows the timer to generate an interruption when the count goes up to one millisecond. It is also disabled the prescaler. EvaRegs.T1CON.all = 0x1040; // Enable timer 1 compare (upcount mode)
The next lines configure the timer 2. This timer is used as time base by the capture unit, this counter counts the time between zero-crossings. The frequency of the timer is set in order to not to be overflowed within two zero-crossing, as explained in previous sections. Configures GP timer 2 with capture unit EvaRegs.T2PR = 0xffff; // Sets timer periode EvaRegs.T2CNT = 0x0000; // Conpare value,not needed The next line sets the prescaler to 1/8, reducing the High speed periphery clock to 3.125Mhz. EvaRegs.T2CON.all = 0x1340; // Configures timer 2
104
Software All the timers belonging to an EV module have some common configuration registers. The Global configuration of timers 1 and 2 is show in the next lines. EvaRegs.CAPCONA.all = 0x0000; // Clears configuration EvaRegs.CAPCONA.bit.CAP12EN = 1; // Enables capture 1 and 2 interruption EvaRegs.CAPCONA.bit.CAP12TSEL = 0; // Chooses timer 2 as timer for capture unit 1 EvaRegs.CAPCONA.bit.CAP1EDGE = 1; // Configures capture unit to detect rising edges EvaRegs.CAPFIFOA.all = 0x0000; // clears FIFO A EvaRegs.CAP1FBOT = 0x0000; // clears capture 1 FIFO top EvaRegs.CAP1FIFO = 0x0000; // clears capture 1 FIFO bottom
Configures interrptions of EVA EvaRegs.EVAIFRC.all = 0xFFFF; EvaRegs.EVAIMRC.bit.CAP1INT = 1;
4.3.3.2.7.
// Sets capture unit 1 flags // Enables captured unit 1 interruption
Relay initialization
This piece of code sets the relay to the default status defined in the constant section of source code. //initialize Relay if(relay_status==CLOSE){ close_relay(); }else{ open_relay(); }
4.3.3.2.8.
Enable CPU interruptions
The PIE block can support up to 96 peripheral interrupts. The 96 interrupts are grouped into blocks of 8 and each group is fed into 1 of 12 CPU interrupt lines (INT1 to INT12). Therefore it is needed to enable the whole interruption line where the single interruption belongs.
105
Software
// Enable CAP1 in PIE PieCtrlRegs.PIEIER3.bit.INTx5 = 1; IER |= M_INT3; // Enable CPU Interrupt 3
The code above enables the individual interruption of the capture unit in the interruption group number 3 and activates the whole group in the CPU interruption register (IER). // Enable ADCINT in PIE PieCtrlRegs.PIEIER1.bit.INTx6 = 1; IER |= M_INT1; // Enable CPU Interrupt 1
The code above enables the individual interruption of the ADC in the group number 1 and activates the whole group in the CPU interruption register (IER).
EINT; ERTM;
// Enable Global interrupt INTM // Enable Global realtime interrupt DBGM
The groups where an interruption belongs cannot be changed. By default the lower number of group the higher priority has the interruption. Therefore, the ADC has more priority than the other subroutine as desired.
4.3.3.2.9.
Average frequency buffer initialization
This code initializes the frequency buffers to the initial values. for(i=0;i0;i--){ average_measures[i]=62500; } average_decay = 0;
106
Software
4.3.4. Capture unit subroutine: Capture1_srt() This function is the interruption subroutine configured in the interruption vector of the capture unit, therefore, it is going to be called when the capture unit requires it, every time it detects a rising edge. This function does more task than frequency measurements, such as threshold checking and RMS calculation. Most of the code is already explained in previous sections. The next lines of code reads the values stored in the capture unit FIFO, These reads returns the time of the last and the previous zero-crossings. temp1_1 = EvaRegs.CAP1FIFO; // Reads value in fifo stack temp1_2 = EvaRegs.CAP1FIFO; // Reads value in fifo stack
Once the values of last frequency measured are retrieved the frequency averaging is started. Resets the measurement accumulator used in the average. measures_addition=0;
Measurements are shifted to leave space for the new measurements, the and average calculation, the for loop is also used to add the previous measurements, the new one is added later. for(i=AVERAGE_FREQUENCY_VALUES-1;i>0;i--){ measures[i]=measures[i-1]; //adds last instant measures except the last one. measures_addition=measures_addition+measures[i]; }
Calculate the new measure checking if the capture unit had been overflowed. //gets the last measure if(temp1_2>temp1_1){ //No overflow measures[0] = temp1_2 - temp1_1; }else{ //overflow measures[0] = temp1_2 +(0xffff- temp1_1); }
107
Software The next lines describe the frequency soft filter. It is described in section 2.1.3 Soft frequency filter. if(measures[0]>average_measures[0] + MAXIMUN_FREQUENCY_CHANGE || measures[0]
The averaging calculation is continued after the frequency decay soft filter. The adding of instant frequency measurements started some lines before in order to improve the efficiency of the code, avoiding two loops that use the same variables. //adds the last measure measures_addition=measures_addition+measures[0]; //calculates the average average_measure= measures_addition/AVERAGE_FREQUENCY_VALUES;
Finally the table of average measures is shifted and the new value is inserted. //Shift the average values array and insert the new one. for(i=AVERAGE_MEASURES-1;i rel="nofollow">0;i--){ average_measures[i]=average_measures[i-1]; } average_measures[0]=average_measure;
The next lines implements the decay measurement algorithm described in section 2.2 Frequency decay measurement.
The process of averaging is the same as the used in frequency measurement. First of all it is clear the average decay accumulator. Then the previous decays are shifted, the loop is also used for adding the decay measurements. average_decay=0; //clear average decay addition
for(i=AVERAGE_MEASURES-1;i>0;i--){ //freency decay shift average_decay=average_decay+decay[i-1]; decay[i]= decay[i-1]; }
108
Software It is calculated the frequency decay using the described equation in the frequency decay section. The only different is that the next line uses the square of FCF divided into one million and next multiplied by the same amount; it is done in order to avoid the integer rounding. //frequency decay calculation decay[0]= (FCF2/(float)((long)average_measures[0]*(long)average_measures[0])) *(((float)average_measures[1]/(float)average_measures[0])-1);
Finally it is calculated the decay average adding the new decay measurement and the decay accumulator.
//average_decay Calculation average_decay=(average_decay+decay[0])/AVERAGE_DECAY_MEASURES; //
The next lines calculates the RMS values, it is showed only the algorithm used to calculate the voltage RMS, the code is exactly the same for current and power in exception of the input data tables used. The RMS algorithm uses the square measures calculated in the sampling subroutine. It is explained in the next section. //RMS power voltage and current //voltage
The accumulator is cleared temp_results=0;
Main loop, it adds all the square measures. for(i=0;i
The total adding is divided into the number of measurements used in RMS calculation. temp_results=temp_results/ADC_MEASURES_PER_CYCLE;
Finally it is stored in the respective result variable, this variable is passed to the record function, that simulates the external memory data recording.. voltage_RMS=sqrt( temp_results);
The result of this calculations gives RMS values in integer units used internally by the microcontroller, in section 2.5.3 RMS Current, voltage and power calculation it is described the process to convert this results into the international system of units (Volts,
109
Software Amps or Watts). The other RMS calculations are done in a similar way in exception that is used the respective squared samples array. The Control logic Type II Source code is very simple. Measures[0] contains the last frequency measurement and T_high and T_low should contain the Normal temperature set points measurements. It is used the algorithm explained in section 2.4 Control logic Type II: Temperature Set points change.
T_high_out=T_high*KF*(measures[0] - FO); T_low_out=T_low*KF*(measures[0] - FO);
Finally this interruption subroutine checks the frequency thresholds that implements the tripping criteria of control logic Type I.. If frequency is lower than the frequency threshold or the frequency decay is lower than the frequency decay threshold is considered that a frequency event has happened. It has to be considered that the frequency checking are made comparing the period of the signal, therefore the frequency event happens when the measure is higher than the lower frequency limit. if(average_measure rel="nofollow"> lower_limit || average_decay < decay_limit){ //there is a frequency event.
The relay is opened open_relay(); }
If the frequency and the decay is recover else if(average_measure < higher_limit && average_decay > decay_limit){ //frequency event is recovered
It is set an initial random waiting time and the state is changed to waiting for attachment. In this state the relay remains opened, it is used to distinguish the open state when the frequency event is recovered or not. if(relay_status == OPEN){ //only The first time
It is set the new state and it is get the random waiting time measured in power grid cycles. relay_status=WAIT; waiting_cycles= temp1_2 % MAXIMUN_WAIT_TO_ATTACHMENT; }else{
110
Software If the load controller is in the waiting state it is subtracted one unit to the waiting counter. //from the second time waiting_cycles--; }
If the waiting time ends it is closed the relay. if(waiting_cycles <= 0){ //the waiting time is up. close_relay(); } } else if(relay_status == WAIT){
If the frequency goes below the recover threshold and it is in the waiting state it is get a new waiting time. // if frequency and frequency decay do not fulfill the event thresholds but they don’t fulfils the recover criteria thresholds. waiting_cycles= temp1_2 % MAXIMUN_WAIT_TO_ATTACHMENT; }
It is called the stub function that should store the record into the external memory for subsequent sending. recod(current_time,average_measure, decay[0],relay_status);
voltage_RMS,current_RMS,
power_RMS,
The next line calls the function that should change the temperature set points, it is passed the calculated high and low set points. set_points_update(T_high_out,T_low_out); Finally it is cleared the interruption in the capture unit and in the respective acknowledge group. This prepares the interruption for the next executions. EvaRegs.EVAIFRC.bit.CAP1INT = 1; // Clears capture unit 1 interruption flag PieCtrlRegs.PIEACK.all = PIEACK_GROUP3; // PIE 3 ack
111
Software
4.3.5. ADC interruption subroutine: ADC_srt() This function gets the current and voltage measurements and calculates its squares that are used in the RMS calculations. It also increases the real time stamp counter.
First of all the function increments the real time counter current_time++;
The next action retrieves the samples gotten by the ADC. If this function is called, the ADC guaranties that the samples are ready in the results register. The results are allocated in the set of registers AdcRegs.ADCRESULTn where n is the number of sample. The results are allocated according to the acquisition order. In this prototype are used the dual simultaneous sampling, in this case it is firstly stored the channel A which corresponds to the Current channel, and next the channel B which corresponds to the voltage. Once the signal channels are sampled they are sampled the offset channels, which are allocated in the next two result registers in the same way as the signal channels. In the following lines it is used ConversionCount as index of arrays, this index in an integer variable that is incremented every execution of this function. //Current measurements adc_samples_current[ConversionCount] = >>4)+(AdcRegs.ADCRESULT2 >>4))/2-CURRENT_OFFSET;
((AdcRegs.ADCRESULT0
////Voltage measurements adc_samples_voltage[ConversionCount] = >>4)+(AdcRegs.ADCRESULT3 >>4))/2-VOLTAGE_OFFSET;
((AdcRegs.ADCRESULT1
The pair of samples of each channel are added and divided into two, finally it is subtracted the offset. The result of these subtractions is stored in the samples array, type integer, it is not used an unsigned integer because the values varies from negative to positive values. The normal temperature set points are read from the ADC inputs, The load controller assumes that the normal temperature set points are introduced thought the ADC but it could be done though any other way as the GP inputs. T_high = AdcRegs.ADCRESULT4 >>4; T_low = AdcRegs.ADCRESULT5 >>4;
112
Software
Now it is calculated the square values for the RMS calculations, it use long type variables because the square of the maximum sample is higher than the maximum number represented by the integer type. It uses the measurements made in the previous lines of code. adc_square_current[ConversionCount]=(long)adc_samples_current[ConversionCount ]*(long)adc_samples_current[ConversionCount]; adc_square_voltage[ConversionCount]=(long)adc_samples_voltage[ConversionCount ]*(long)adc_samples_voltage[ConversionCount]; adc_square_power[ConversionCount]=adc_square_current[ConversionCount]*adc_squ are_voltage[ConversionCount];
If ADC_MEASURES_PER_CYCLE conversions have been logged, start over, where ADC_MEASURES_PER_CYCLE is the selected amount of measures used for the RMS calculations in the constants definition. if(ConversionCount == ADC_MEASURES_PER_CYCLE-1) { ConversionCount = 0; } else ConversionCount++;
Reinitialize interruption for next ADC sequence Reset SEQ1, so the ADC sequence interruption is marked as attended. AdcRegs.ADCTRL2.bit.RST_SEQ1 = 1;
Clear INT SEQ1 bit, so the interruption is activated to be called the next time. AdcRegs.ADCST.bit.INT_SEQ1_CLR = 1;
Acknowledge interrupt to PIE. So the whole interruption vector is marked as attended, it is possible to acknowledge the whole group because this interruption is the only one of this group which can be called in this project.
PieCtrlRegs.PIEACK.all = PIEACK_GROUP1;
113
Software
4.3.6. Other subroutines This sections describe the others function present in the load controller source code which do not fit in the previous categories.
This function sets the time stamp to a desired value. It should be called by an external source of synchronization, for example the time given by the GPRS network. void time_synchronization(long time){ current_time = time; }
The next functions controls the relay. They change the state flag and open or close the relay according to the function. In this prototype the relay is simulated by a LED diode which is turned on when the load should be connected and is turned off in the opposite case. The way of activating the LED is the same to activate an external relay. Inside this fuctions can be added any other instructions required in the final design in order to do any other task like changing the thermostat themperature of a heat pump or a fridge. Sets the state of load controller to close and activates the LED. void close_relay(){ relay_status=CLOSE; GpioDataRegs.GPFSET.bit.GPIOF8 =0x1; }
Sets the state of load controller to open an disconnects the LED. void open_relay(){ relay_status=OPEN; GpioDataRegs.GPFCLEAR.bit.GPIOF8
=0x1;
//IT SHOULD BE STARTED HERE THE DATA SUBMISION }
This function changes the frequency and decay thresholds. They should be used in order to change them once the program is compiled and running by future extensions. For example for change them remotely though the GPRS network.
114
Software
void set_higher_limit(unsigned int new_limit){ higher_limit=new_limit; }
void set_lower_limit(unsigned int new_limit){ lower_limit=new_limit; } void set_decay_threshold(float new_limit){ decay_limit=new_limit; //set new frequency decay threshold }
Finally, the last functions are stubs, which are not implemented function. The first stub corresponds to the frequency that should change the temperature set points; it should be modified depending on the selected interface. void set_points_update(unsigned int T_high,unsigned int T_low){ //this function should change the temperature set points if applicable }
The second stub should be implemented to store the data in the external memory. In the case of using a serial flash memory, as recommended, is should store all the acquired data in the last free memory address. Another function would be needed to read the acquired data when it is needed to delivery it. In order to simulate the data storage it is used the external RAM, it obviously cannot fulfill the memory capacity required. //this stub should store last measurements in external flash memory //store measurements in internal RAM ext_time[ext_index]=time; ext_period[ext_index]=period; ext_voltage[ext_index]=voltage; ext_current[ext_index]=current; ext_power[ext_index]=power; ext_decay[ext_index]=decay; ext_relay[ext_index]=relay;
115
Software
//increment the index and reset to zero if out of bounds ext_index++; ext_index=ext_index%STUB_EXT_MEMORY_LEGTH; }
116
5
FUNCTIONALITIES AND PERFORMANCE
The load controller is able to carry out the two control logics, Type I and Type II. It can reconnect the load within a random selectable amount of time, 30 seconds by default. The load controller is prepared to be extended in the future. It supports the time synchronization and changing the configuration values in running time, in the case of adding an external synchronization source and a communication module. It is also prepared to store the acquired data in a external memory in the case of adding an external data storage module. Accuracy Frequency
±7,5mHz. @50Hz
Voltage
±0,3V
Current
±0,015A
Power
±5W
Maximum measurement delay: 0,5s (Average of 25 values) Maximum power measurement delay: 1ms Note that this result corresponds to a load of nominal 600W. The current accuracy could be improved by expanding the current signal in the ADC conversion range; it can be done by changing the number of loops in the current transformer. In consequence the power accuracy could be improved to.
117
6
FUTURE WORK
This section describes the next logic steps in following stages of this project, if it is continued in the future.
6.1. Flash memory This program is programmed using only RAM memory. Using this memory entails three problems. The first problem is the Memory limitation, the RAM memory is not enough to expand too much the program, in order to follow the development of this project is recommendable to use the extra memory space given by Flash Memory. The second problem is about the persistence of the program. The RAM memory does not allow maintaining the stored data when it is powered off, therefore, in a power cut, the internal data could be erased. The last problem is the remote software update, it is not possible to update the program stored in RAM memory from a external source in exception of the PC, the onchip flash memory is possible to be updated from other sources, like a GPRS module that sends the new program by the internet.
6.2. Communication part In order to make the load controller fully useful, is needed the communication with remote locations. It has no sense to store the measurements if it is not possible to retrieve it in order to be studied. A solution can be to use a GPRS module. This module through the RS-232 port can get the acquired data and send it to a desired location. This module can be the GM862 QUAD manufactured by the corporation Round Solutions. This module
119
Future Work supports the TCP/IP stack as well as for example the FTP protocol. TCP/IP allows sending though the GPRS network the data directly to a computer that could record the data from every load controller.
This module also has the possibility of being programmed, with the high level language Python and it also has two RS-232 ports, one can be used for receiving the measurements and the other to update the internal software. The capability of being programmed is essential because during the update the DSC cannot execute any code, therefore the GSPS module can take the control.
6.3. Persistent storage The load controller measurement has a pretty high throughput of data. It has to store some measures in short periods of time. It has to store and send the measures ten minutes before a frequency event. The measures have to be stored somewhere, this place can be and external persistent storage based on Serial Flash memory. This serial flash has enough capacity to store the measurements, their capacities goes from 1Mb to 128Mb. They are connected with the SPI interface of the DSC. The serial nature of the measurements made by the load controller is suitable for this kind of memory. This memory can be used also to preserve configuration changes made remotely.
120
7
CONCLUSIONS
In this section I going to explain all the objectives achieved in the developing of this project, as well as all the problems solved. The development of a real project evolves lots of mishaps and problems which before the begging of this project I was not able to imaging. I thought that these problems are common and the experience acquired in this project allows me to surpass them easier than before. The first point solved is to search the hardware needed to achieve the specifications. The market is full of manufacturers which has lots of processors. The research process is a critical point, choosing a wrong processor could result in a complete fail. Changing the processor, which is the main part of these kind of project, in an advance stage could imply in starting the development again. Now I know how to carry out these researches and how to make them. The next point is the design of the additional hardware that completes the capabilities of the microcontroller. The microcontroller is the core of the hardware but it cannot fulfill all the hardware requirements, it is needed other essential hardware, such as the power supplies, or other needed hardware like the filters or the signal conditioning hardware, whose functions cannot be made by any microcontroller. I designed a simple board with making use of my knowledge about electronics but it could not completely fulfill all the requirements, but fortunately my co-tutor Tonny improved the board until the point to be suitable for my purposes. Once the hardware part is completed the next logical step is the software development which is the “mind” of the any electronically device based on a microcontroller. The software is also a key part of any electrical development. I have learned to use the Code Composer IDE, which is one of the best IDEs in the market. I think that know I am able to develop any other microcontroller software based on any other platform and also using any other advanced IDE.
121
Conclusions
Referring to the product obtained. I have developed a prototype of a load controller based on the specifications given by my Tutor Zhao Xu. The prototype developed is a frequency meter. The load controller is also able to measure voltage, current and power. It is also capable of react to frequency events, frequency drops and excess of frequency decay. The project has being developed with the expandability in mind, because this prototype has to be continued in order to be fully working product. In resume, I have developed a prototype able to help the power grid integrity against changes of power consumption which can involve in the imbalance of the consumption-production power grid equilibrium. The continuity of this project is leaved open to be followed by any other student in order to complete a fully working load controller as described in the future work section. With these words I consider this Master thesis ended.
Thank you for reading, Alejandro Nieto.
122
8
BIBLIOGRAPHY
1. Grastrup, Martin Skov Johansen & Lars Henrik. Consumer Based Freqeuncy Controlled Reserve in the Danish Power Grid. Lyngby : s.n., 2006. 2. Agilent Technologies. Understanding Frequency. agilent.com. [Online] http://cp.literature.agilent.com/litweb/pdf/5965-7664E.pdf. 3. Texas Instruments. TMS320x281x Event Manager (EV) Reference Guide (Rev. E). Texas instruments official web site. [Online] June 27, 2007. [Cited: September 1, 2007.] http://www.ti.com/litv/pdf/spru065e. 4. —. TMS320x281x Analog-to-Digital Converter (ADC) Reference Guide (Rev. D). [Online] July 2005, 14. [Cited: October 2007, 2.] http://focus.ti.com/lit/ug/spru060d/spru060d.pdf. 5. —. TMS320F2810, TMS320F2811, TMS320F2812, TMS320C2810, TMS320C2811,TMS320C2812 DSPs (Rev. O). Texas Instruments official web site. [Online] July 17, 2007. [Cited: September 1, 2007.] http://www.ti.com/lit/gpn/tms320f2812. 6. wikipedia. Root mean square. Wikipedia. [Online] wikipedia. http://en.wikipedia.org/wiki/Root_mean_square. 7. Spectrum digital. eZdspTM F2812 Technical reference. 2003. 8. California Energy Comission. Smart Load Control and Grid-Friendly Appliances. October 2003. P-500-03-096-A10. 9. Z. Xu, Member, IEEE, J. Østergaard, Member IEEE, M. Togeby, C. Marcus-Møller. Design and Modelling of Thermostatically Controlled Loads as Frequency Controlled Reserve. 10. Ning Lu, Member, IEEE and Donald J. Hammerstrom, Member, IEEE. Design Considerations for Frequency Responsive Grid Friendly Appliances. 11. Rasmussen, Tonny W. Additional hardware design and layout. Lyngby : oersted DTU, 2007.
123
A
SOURCE CODE
A.1
Main program source code
/***************************************************************** * * File inclusion * *****************************************************************/ #include "DSP281x_Device.h" #include "DSP281x_Examples.h"
// DSP281x Headerfile Include File // DSP281x Examples Include File
/***************************************************************** * * constants definition * *****************************************************************/
/*** frequency measurements constants ***/ //number of frequency measurements used in average #define AVERAGE_FREQUENCY_VALUES 25 //Recover frequency threshold in integer units (Running at 25MHz) #define DEFAULT_HIGHER_LIMIT 62525 //Frequency event threshold in integer units (Running at 25MHz) #define DEFAULT_LOWER_LIMIT 62626
/*** frequency decay filter constants ***/ //maximun allowable frequency change between frequencies //measurements. 150 integer units/cycle= 6Hz/s #define MAXIMUN_FREQUENCY_CHANGE 150
/*** frequency decay measurement constants ***/ //square of FCF, It is nedeed in frequency decay calculations, //see thesis for more details. #define FCF2 976562500000.0 //frequency decay event threshold #define DECAY_THRESHOLD -0.16 //length of averaged frequency measurements, used in frequency
125
Board Schemes //decay calculations #define AVERAGE_MEASURES 5 //Measures used in RMS calculations #define ADC_MEASURES_PER_CYCLE 40 /*** relay constants ***/ //0=OPEN CIRCUIT 1=CLOSED CIRCUIT 2=WAITING FOR RECONNECTION #define DEFAULT_RELAY_STATUS 0x1 #define OPEN 0x0 #define CLOSE 0x1 #define WAIT 0x2
//relay status //relay status //relay status
//in number of powerline cycles. 1500*0.02s=30s #define MAXIMUN_WAIT_TO_ATTACHMENT 1500
/*** control logic type II ***/ #define KF 1 //KF value in Control logic type II #define FO 62500 //Nominal frequency in Nordic systme, integer units.
/*** data recording constants ***/ //This constan defines the number of records stored in //external memory simulating the external memory #define STUB_EXT_MEMORY_LEGTH 5
/***************************************************************** * * Special memory allocations * *****************************************************************/
#pragma DATA_SECTION(measures, ".extRAM"); //stores this array in external RAM #pragma DATA_SECTION(average_measures, ".extRAM"); //stores this array in external RAM
/***************************************************************** * * Function prototype definition * *****************************************************************/ /*** frequency measurement functions ***/ //definition of the interruption subroutine of capture unit 1 //and allocation in internal RAM interrupt void Capture1_isr(void); #pragma CODE_SECTION(Capture1_isr, "ramfuncs");
126
Board Schemes /*** data adquisition functions ***/ //definition of the interruption subroutine of ADC //and allocation in internal RAM interrupt void adc_isr(void); #pragma CODE_SECTION(adc_isr, "ramfuncs");
/*** relay control functions ***/ void close_relay(void); void open_relay(void);
//opens the load relay //closes the load relay
//sets the frequency recover threshold void set_higher_limit(unsigned int); //sets the frequency event threshold void set_lower_limit(unsigned int); /*** time stamp ***/ //synchronices the real time counter void time_synchronization(long);
/*** decay functions ***/ //sets the frequency decay threshold void set_decay_limit(float);
/*** Control logic type II ***/ void set_points_update(unsigned int,unsigned int);
/*** storing stub ***/ //Function that should store the recors in external memory void recod(long time,unsigned int period,unsigned int voltage,unsigned int current,unsigned long power, float decay,short relay);
/***************************************************************** * * variables and array definition and initialization * *****************************************************************/
//generic variables short i; //generic index short j; //generic index /*** frequency measurement ***/ // Temp integers variables in capture unit subroutine unsigned int temp1_1; unsigned int temp1_2;
127
Board Schemes //addition of al the last AVERAGE_FREQUENCY_VALUES periods. unsigned long measures_addition; //table of instant frequency measurements unsigned int measures[AVERAGE_FREQUENCY_VALUES]; //table of average frequency measurements unsigned int average_measures[AVERAGE_MEASURES]; //index used in frequency measurement array unsigned int measure_index=0;
unsigned int average_measure; //average period measure
/*** Frequency decay measurement ***/ //table of frequency decay measurements float decay[AVERAGE_MEASURES]; //last frequency decay measurement float average_decay=0; //frequency decay threshold definition and initialization float decay_limit= DECAY_THRESHOLD; /*** relay control ***/ //frequency recover threshold initialization unsigned int higher_limit=DEFAULT_HIGHER_LIMIT; //frequency event threshold initialization unsigned int lower_limit=DEFAULT_LOWER_LIMIT; //defalut relay state initialization short relay_status=DEFAULT_RELAY_STATUS; //count of waited cycles on reconection int waiting_cycles;
/*** data adquisition ***/ //ADC current measures without offset unsigned int adc_samples_current[ADC_MEASURES_PER_CYCLE]; //ADC voltage measures without offset unsigned int adc_samples_voltage[ADC_MEASURES_PER_CYCLE]; //Index of ADC samples table Uint16 ConversionCount=0; /*** Control logic II ***/ //variables to store the user high and low temperature. unsigned int T_high=0; undigned int T_low=0; //output control logic II temperature unsigned int T_high_out=0; unsigned int T_low_out=0;
128
Board Schemes
/*** RMS calculations ***/ //square unsigned unsigned unsigned
of ADC measures in RMS calculations long adc_square_current[ADC_MEASURES_PER_CYCLE]; long adc_square_voltage[ADC_MEASURES_PER_CYCLE]; long adc_square_power[ADC_MEASURES_PER_CYCLE];
unsigned int voltage_RMS; //voltage RMS calculation result unsigned int current_RMS; //current RMS calculation result unsigned int power_RMS; //power RMS calculation result //temporal variables in RMS calculations long temp_results; long temp2_1; long temp2_2; long temp2_3;
/*** Time stamp ***/ //time variable definition and initialization long current_time=0;
/*** external memory stub ***/ unsigned int ext_index=0; long ext_time[STUB_EXT_MEMORY_LEGTH]; unsigned int ext_period[STUB_EXT_MEMORY_LEGTH]; unsigned int ext_voltage[STUB_EXT_MEMORY_LEGTH]; unsigned int ext_current[STUB_EXT_MEMORY_LEGTH]; unsigned long ext_power[STUB_EXT_MEMORY_LEGTH]; float ext_decay[STUB_EXT_MEMORY_LEGTH]; short ext_relay[STUB_EXT_MEMORY_LEGTH];
//index //time //period //voltage //current //power //f. decay //relay s.
/***************************************************************** * * Main function, hardware initialization * *****************************************************************/
main() { // // Step 1. Initialize System Control: // PLL, WatchDog, enable Peripheral Clocks // InitSysCtrl(); // Init the system control
//set HSPCLK to SYSCLKOUT / 6
129
Board Schemes //(25Mhz assuming 150Mhz SYSCLKOUT) EALLOW; SysCtrlRegs.HISPCP.all = 0x3; // HSPCLK = SYSCLKOUT/6 EDIS;
// Step 2. Initialize GPIO: EALLOW; // Enable writing on write protected registers // Sets pin as Caputure unit GpioMuxRegs.GPAMUX.bit.CAP1Q1_GPIOA8 = 1; //Sets relay pin as GPIO GpioMuxRegs.GPFMUX.bit.MCLKXA_GPIOF8=0x0; //sets as I/O GpioMuxRegs.GPFDIR.bit.GPIOF8 = 0x1; //sets as output
EDIS; // Disables writing on write protected registers
// // Step 3. Clear all interrupts and initialize PIE vector // table: // // Disable CPU interrupts DINT;
// Initialize the PIE control registers to their default state. // The default state is all PIE interrupts disabled and flags // are cleared. InitPieCtrl(); // Disable CPU interrupts and clear all CPU interrupt flags: IER = 0x0000; // Disables al interrputions IFR = 0x0000; // Clear al interruption flags // Initialize the PIE vector table with pointers to the shell // Interrupt Service Routines (ISR). // This will populate the entire table, even if the interrupt // is not used. InitPieVectTable(); // Interrupts that are used in this example are re-mapped to // ISR functions found within this file. EALLOW; // enable write in EALLOW protected registerd // sets capture 1 unit interruption routine as Capture1_isr PieVectTable.CAPINT1 = &Capture1_isr; //sets adc_isr as interruption routine of ADC PieVectTable.ADCINT = &adc_isr; EDIS; // disable write in EALLOW protected registers
// // // // Step 4. Initialize all the Device Peripherals:
130
Board Schemes // // InitAdc(); //init ADC.
// Step 5. User specific code, enable interrupts: // Configure ADC AdcRegs.ADCTRL1.bit.RESET = 1; //resets ADC asm(" RPT #10 || NOP"); // Setup 2 conv's on SEQ1 AdcRegs.ADCMAXCONV.all = 0x0002; // Setup ADCINA0 & ADCINB0 as 1st SEQ1 conv. // Current and voltage channels AdcRegs.ADCCHSELSEQ1.bit.CONV00 = 0x0; // Setup ADCINA1 & ADCINB1 as 2n SEQ1 conv. //Current and voltage, second sample AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x1;
// Setup ADCINA2 & ADCINB2 as 2n SEQ1 conv. //Read User normal temperature high and low. AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x2;
// Enable EVASOC to start SEQ1 AdcRegs.ADCTRL2.bit.EVA_SOC_SEQ1 = 1; // Enable SEQ1 interrupt (every EOS) AdcRegs.ADCTRL2.bit.INT_ENA_SEQ1 = 1; //set simultaneous mode to measure current and voltage //at the same time AdcRegs.ADCTRL3.bit.SMODE_SEL = 1; AdcRegs.ADCTRL3.bit.ADCEXTREF=0x0;
//use internal reference
/*** EVA GP timers configuration ***/ EvaRegs.GPTCONA.all = 0; // Clears configuration /*** Configures GP timer 1 with ADC sample rate ***/
EvaRegs.T1PR = 0x61A8; EvaRegs.GPTCONA.bit.T1TOADC = 0x2; EvaRegs.T1CON.all = 0x1040;
// Setup period register // Enable EVASOC in EVA // Enable timer (upcount mode)
/* Configures GP timer 2 with capture unit */ EvaRegs.T2PR = 0xffff; // Sets timer periode EvaRegs.T2CNT = 0x0000; // Initial counter value EvaRegs.T2CON.all = 0x1340; // Configures timer 2 /* Global configuration of timers 1 and 2 */ EvaRegs.CAPCONA.all = 0x0000; // Clears configuration
131
Board Schemes // Enables capture 1 and 2 interruption EvaRegs.CAPCONA.bit.CAP12EN = 1; // Chooses timer 2 as timer for capture unit 1 EvaRegs.CAPCONA.bit.CAP12TSEL = 0; // Configures caputure unit to ditect rising edges EvaRegs.CAPCONA.bit.CAP1EDGE = 1; EvaRegs.CAPFIFOA.all = 0x0000; EvaRegs.CAP1FBOT = 0x0000; EvaRegs.CAP1FIFO = 0x0000;
// clears FIFO A // clears capture 1 FIFO top // clears capture 1 FIFO bottom
/*** Configures interrptions of EVA ***/ EvaRegs.EVAIFRC.all = 0xFFFF;
// Sets caputre unit 1 flags
// Enables caputure unit 1 interruption EvaRegs.EVAIMRC.bit.CAP1INT = 1;
//initialize Relay if(relay_status==1){ close_relay(); }else{ open_relay(); }
//initialization of instant frequency table for(i=0;i0;i--){ average_measures[i]=62500; }
//enable CPU interruptions // Enable CAP1 in PIE, frequency measurement subroutine PieCtrlRegs.PIEIER3.bit.INTx5 = 1; IER |= M_INT3; // Enable CPU Interrupt 3
// Enable ADCINT in PIE, ADC sampling subroutine PieCtrlRegs.PIEIER1.bit.INTx6 = 1; IER |= M_INT1; // Enable CPU Interrupt 1 EINT; ERTM;
return 0;
132
// Enable Global interrupt INTM // Enable Global realtime interrupt DBGM
Board Schemes } /*****************************************/
/*****************************************
FREQUENCY MEASUREMENT FUNCTIONS
*****************************************/ interrupt void Capture1_isr(void) { temp1_1 = EvaRegs.CAP1FIFO; // Reads value in fifo stack temp1_2 = EvaRegs.CAP1FIFO; // Reads value in fifo stack //resets the measures accumulator measures_addition=0;
//Measures shift and average calculation for(i=AVERAGE_FREQUENCY_VALUES-1;i>0;i--){ measures[i]=measures[i-1]; // adds last instant measures except the last one. measures_addition=measures_addition+measures[i]; }
//gets the last measure if(temp1_2>temp1_1){ //No overflow measures[0] = temp1_2 - temp1_1; }else{ //overflow measures[0] = temp1_2 +(0xffff- temp1_1); } //Software frequency filter if(measures[0]>average_measures[0] + MAXIMUN_FREQUENCY_CHANGE || measures[0]
//average calculation //adds the last measure measures_addition=measures_addition+measures[0];
133
Board Schemes //calculates the average average_measure= measures_addition/AVERAGE_FREQUENCY_VALUES;
//Shift the averege values array and insert the new one. for(i=AVERAGE_MEASURES-1;i rel="nofollow">0;i--){ average_measures[i]=average_measures[i-1]; } average_measures[0]=average_measure;
/*** frequency decay measurement ***/ average_decay=0; //clear average decay addition
for(i=AVERAGE_MEASURES-1;i>0;i--){ //freency decay shift average_decay=average_decay+decay[i-1]; decay[i]= decay[i-1]; }
//frequency decay calculation decay[0]=(FCF2/(float)((long)average_measures[0]*(long)average_measures[0])) *(((float)average_measures[1]/(float)average_measures[0])-1); //average_decay Calculation average_decay=(average_decay+decay[0])/AVERAGE_MEASURES; //
//RMS power voltage and current //voltage temp_results=0; //clear temp variable for(i=0;i
//Store result
//Current temp_results=0; for(i=0;i
//Power temp_results=0; for(i=0;i
134
Board Schemes temp_results=temp_results/ADC_MEASURES_PER_CYCLE; power_RMS=temp_results;
//Control logic Type II, Temperature Control
T_high_out=T_high*KF*(measures[0] - FO); T_low_out=T_low*KF*(measures[0] - FO); //Control logic Type I, relay control if(average_measure rel="nofollow"> lower_limit || average_decay < decay_limit){ //there is a frequency event. open_relay(); } else if(average_measure < higher_limit && average_decay > decay_limit && relay_status!=CLOSE){ //frequency event is recovered //the relay status in changed to "wating for attachment" if(relay_status == OPEN){ //only once relay_status=WAIT; waiting_cycles= temp1_2
%
MAXIMUN_WAIT_TO_ATTACHMENT; }else{ //from the second time waiting_cycles--; } if(waiting_cycles <= 0){ //the waiting time is up. close_relay(); } }else if(relay_status == WAIT){ // if frequency and frequency decay do not fullfit the event // thresholds but they dont fullfits the recover // criteria thresholds. waiting_cycles= temp1_2 % MAXIMUN_WAIT_TO_ATTACHMENT; }
//it is call the stub function that should store the data adquired in the external memory. recod(current_time,average_measure, decay[0],relay_status);
voltage_RMS,current_RMS,
power_RMS,
//Temperature Set points change set_points_update(T_high_out,T_low_out); // Clears capture unit 1 interruption flag EvaRegs.EVAIFRC.bit.CAP1INT = 1; // PIE 3 ack PieCtrlRegs.PIEACK.all = PIEACK_GROUP3; }
135
Board Schemes
/*****************************************/
/*****************************************
DATA ADQUISITION FUNCTIONS
*****************************************/
interrupt void {
adc_isr(void)
//current time stamp increment current_time++; //Current measurements adc_samples_current[ConversionCount] (AdcRegs.ADCRESULT2 >>4))/2;
=
((AdcRegs.ADCRESULT0
>>4)
+
////Voltage measurements adc_samples_voltage[ConversionCount] (AdcRegs.ADCRESULT3 >>4))/2;
=
((AdcRegs.ADCRESULT1
>>4)
+
T_high = AdcRegs.ADCRESULT4 >>4; T_low = AdcRegs.ADCRESULT5 >>4; //Calculates the square values for the RMS calculations adc_square_current[ConversionCount]=(long)adc_samples_current[ConversionCount]*(l ong)adc_samples_current[ConversionCount]; adc_square_voltage[ConversionCount]=(long)adc_samples_voltage[ConversionCount]*(l ong)adc_samples_voltage[ConversionCount]; adc_square_power[ConversionCount]=adc_square_current[ConversionCount]*adc_square_ voltage[ConversionCount];
// If ADC_MEASURES_PER_CYCLE conversions have been logged, start over if(ConversionCount == ADC_MEASURES_PER_CYCLE-1) { ConversionCount = 0; } else ConversionCount++; // Reinitialize for next ADC sequence AdcRegs.ADCTRL2.bit.RST_SEQ1 = 1; AdcRegs.ADCST.bit.INT_SEQ1_CLR = 1; PieCtrlRegs.PIEACK.all = PIEACK_GROUP1;
136
// Reset SEQ1 // Clear INT SEQ1 bit // Acknowledge interrupt to PIE
Board Schemes return; } /*****************************************/
/*****************************************
RELAY CONTROL FUNCTIONS
*****************************************/ void time_synchronization(long time){ current_time = time; }
/*****************************************/
/*****************************************
RELAY CONTROL FUNCTIONS
*****************************************/
void close_relay(){ relay_status=CLOSE; //sets status close GpioDataRegs.GPFSET.bit.GPIOF8 =0x1; //close relay }
void open_relay(){ relay_status=OPEN; //sets status open GpioDataRegs.GPFCLEAR.bit.GPIOF8=0x1; //open relay }
void set_higher_limit(unsigned int new_limit){ higher_limit=new_limit;
//set new recover threshold
}
void set_lower_limit(unsigned int new_limit){ lower_limit=new_limit;
//Set new frequency event threshold
}
137
Board Schemes void set_decay_threshold(float new_limit){ decay_limit=new_limit;
//set new frequency decay threshold
}
/*****************************************/
void set_points_update(unsigned int T_high,unsigned int T_low){ //this function should change the temperature set points if applicable
}
void recod(long time,unsigned int period,unsigned int voltage,unsigned int current,unsigned long power, float decay,short relay){ //this stub should store last measurements in external flash memory //store measurements in internal RAM ext_time[ext_index]=time; ext_period[ext_index]=period; ext_voltage[ext_index]=voltage; ext_current[ext_index]=current; ext_power[ext_index]=power; ext_decay[ext_index]=decay; ext_relay[ext_index]=relay;
//increment the index and reset to zero if out of bounds ext_index++; ext_index=ext_index%STUB_EXT_MEMORY_LEGTH; }
138
Board Schemes
A.2
Memory mapping
MEMORY { PAGE 0 : /* For this example, H0 is split between PAGE 0 and PAGE 1 /* BEGIN is used for the "boot to HO" bootloader mode /* RESET is loaded with the reset vector only if /* the boot is from XINTF Zone 7. Otherwise reset vector /* is fetched from boot ROM. See .reset section below RAMM0 BEGIN PRAMH0 RESET XINTF60
: origin : origin : origin : origin : origin =
= 0x000000, length = 0x3F8000, length = 0x3F8002, length = 0x3FFFC0, length 0x100000, length =
*/ */ */ */ */
= 0x000400 = 0x000002 = 0x000FFE = 0x000002 0x008000
PAGE 1 : /* For this example, H0 is split between PAGE 0 and PAGE 1 */ RAMM1 DRAMH0 XINTF61
: origin = 0x000400, length = 0x000400 : origin = 0x3f9000, length = 0x001000 : origin = 0x108000, length = 0x008000
}
SECTIONS { /* Setup for "boot to H0" mode: The codestart section (found in DSP28_CodeStartBranch.asm) re-directs execution to the start of user code. Place this section at the start of H0 */ codestart ramfuncs .text .cinit .cio .pinit .switch .reset
: : : : : : : :
> > > > > > > >
BEGIN, PRAMH0 XINTF60, XINTF60, XINTF60, XINTF60, RAMM0, RESET,
.stack .ebss .econst .esysmem .sysmem .extRAM
: : : : :
> > > > >
RAMM1, DRAMH0, DRAMH0, DRAMH0, DRAMH0,
PAGE PAGE PAGE PAGE PAGE PAGE PAGE PAGE
= = = = = = = =
0 0 0 0 0 0 0 0, TYPE = DSECT /* not used, */
PAGE = 1 PAGE = 1 PAGE = 1 PAGE = 1 PAGE = 1 : > XINTF61,
PAGE = 1
}
139
Board Schemes
B
BOARD SCHEMES
This appendix contains the full schemes of the load controller. is understanded as an input, as well as the symbol as an The symbol output. This symbols are used to connect the intputs and outputs of the main scheme with the detail of the DSC connectors. They are also used to represent the power outlet connection.
140
Filter
Voltage Level Shifter
5V
100kΩ
5V 5V 100nF
Voltage Measuring Transformer 69,8kΩ
4,02kΩ
Wall-Outlet
100nF
196kΩ
100nF
100nF
100nF
-
100nF
91Ω
100nF
5,49kΩ
OPA277
+
220:1.1
5V 5V
261kΩ 100nF
-
100nF
-5V
110kΩ
5V
+
100nF
-
OPA277
100nF
Voltage 23,3kΩ Offset
-5V
Voltage Output
OPA277 100kΩ
+
100nF -5V
100nF
10kΩ
Filter 5V
5V 5V 100nF
100nF
196kΩ
4,02kΩ
100nF
100nF
100nF
-
50:1 69,8kΩ
5V
100nF
91Ω
+ Current Measuring Transformer
100nF
5,49kΩ
OPA277
261kΩ 100nF Current Level Shifter 100kΩ
OPA277
5V
+
100nF -5V
100nF 100nF
-5V 311kΩ
5V Double Power Supply
230:6
1N4001
-
Current
OPA277
Current 23,3kΩ Offset
100kΩ
+
10Ω 100nF -5V
Bridge
LM7805 INPUT
100nF
10kΩ
OUTPUT
-5V 100nF
100nF
470µF
Wall-Outlet 100nF
470µF
100nF 5V OUTPUT
LM7905
Bridge
5V
Stub Relay
Power On LED
1N4001 5V
Processor Single Power Supply
230:6
100kΩ
1N4001
Relay Wall-Outlet
Bridge
LM7805 INPUT
100nF
470µF
OUTPUT
100nF
+5V
Board Schemes
141
Figure 8-I - Board Scheme
INPUT
Board Schemes
ADCINA1
ADC Connection Socket 10x3 Port 5 and 9
ADCINA0
Connection Socket 20x3 Ports 4 and 8
GP I/O 19
22
25
28
14
17
20
23
26
29
15
18
21
24
27
30
13
16
10
7
4
1
58
55
52
49
46
43
40
37
34
31
28
25
22
19
13
16
10
7
4
1
Voltage
Voltage Offset
Relay Control
Capture Unit
Capture Unit
9
12
60
11
57
8
5
2
59
54
45
56
42
53
44
39
48
41
36
51
38
33
50
35
47
32
27
24
30
21
29
23
26
20
18
14
17
11
8
5
2
Figure 8-II - DSC connectors Detail
142
6
3
15
9
12
6
3
Current
ADCINB1
ADCINB0
Voltage
Current Offset
C
OTHER PLOTS
The next plots show other averaged measurement plots, it shows 200 values instead of 30 as the one in the frequency measurement section. The data of these tables can be found in Table 8-III - Averaged frequency measurements.
Averaged Vs Thresholds 62650 62600 62550 62500 62450 62400 62350 1
30
59
Average measurements
88
117
recover threshold
146
175
Frequency event threshold
Figure 8-III – Averaged frequency measurements (Integer units)
143
Board Schemes
Average measurements 50,045 50,04 50,035 50,03 50,025 1
21
41
61
81
101
121
141
161
Average measurements
Figure 8-IV – Averaged frequency measurements (Hz)
144
181
D TABLES
D.1
ADC Measurements
Table 8-I – ADC samples table No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
145
V 2480 2996 3422 3714 3845 3794 3571 3204 2721 2170 1532 864 301 0 0 0 99 590 1226 1896 2476 2995 3421 3716 3844 3793 3567 3194 2707 2161 1523 849 294
I 2047 2182 2284 2370 2393 2382 2327 2230 2109 1969 1805 1635 1492 1413 1417 1416 1429 1557 1723 1898 2045 2183 2291 2365 2400 2382 2328 2227 2107 1966 1802 1628 1485
Tables
Table 8-I – ADC samples table No 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82
146
V 0 0 0 99 611 1243 1910 2485 3004 3428 3719 3842 3793 3570 3196 2710 2162 1526 853 293 0 0 0 100 606 1239 1905 2482 3001 3427 3719 3844 3796 3571 3198 2715 2165 1530 862 302 0 0 0 98 599 1238 1902 2479 3000
I 1403 1407 1411 1442 1570 1732 1901 2049 2184 2288 2370 2393 2384 2325 2235 2106 1967 1803 1629 1490 1407 1411 1413 1433 1569 1730 1900 2050 2182 2295 2370 2400 2386 2327 2235 2109 1966 1803 1632 1492 1412 1411 1405 1434 1567 1727 1899 2050 2180
Tables
Table 8-I – ADC samples table No 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
V 3425 3717 3844 3794 3573 3200 2716 2166 1534 863 294 0 0 0 102 596 1226 1899
I 2287 2365 2397 2389 2332 2230 2108 1967 1806 1635 1483 1416 1413 1416 1441 1562 1723 1898
147
Tables
D.2
Frequency measures
Note that the measurements made by the load controller are the ones that are in integer units and the values in Hz are calculated as explained in the respective section.
Table 8-II - Instant frequency measurements No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
148
Instant frequency (int) 62509 62492 62495 62510 62498 62498 62523 62509 62506 62505 62498 62519 62502 62520 62509 62460 62509 62495 62512 62521 62509 62491 62492 62491 62509 62498 62502 62495 62527 62500
Averaged freq (int) 62503 62502 62503 62503 62500 62500 62500 62500 62500 62500 62500 62501 62500 62501 62500 62501 62502 62503 62503 62503 62502 62502 62503 62503 62503 62503 62504 62504 62504 62504
Instant freq. (Hz) 49,993 50,006 50,004 49,992 50,002 50,002 49,982 49,993 49,995 49,996 50,002 49,985 49,998 49,984 49,993 50,032 49,993 50,004 49,990 49,983 49,993 50,007 50,006 50,007 49,993 50,002 49,998 50,004 49,978 50,000
averaged freq. (int) 49,998 49,998 49,998 49,998 50,000 50,000 50,000 50,000 50,000 50,000 50,000 49,999 50,000 49,999 50,000 49,999 49,998 49,998 49,998 49,998 49,998 49,998 49,998 49,998 49,998 49,998 49,997 49,997 49,997 49,997
Tables
Table 8-III - Averaged frequency measurements No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
Frequency (Int) 62455 62454 62454 62454 62454 62455 62456 62456 62457 62456 62456 62455 62455 62455 62455 62454 62454 62454 62455 62455 62455 62455 62454 62455 62455 62455 62454 62455 62454 62454 62454 62454 62454 62454 62453 62452 62452 62452 62453 62454 62456 62456 62457 62455 62456 62455 62455
Frequency (Hz) 50,0360259 50,0368271 50,0368271 50,0368271 50,0368271 50,0360259 50,0352248 50,0352248 50,0344237 50,0352248 50,0352248 50,0360259 50,0360259 50,0360259 50,0360259 50,0368271 50,0368271 50,0368271 50,0360259 50,0360259 50,0360259 50,0360259 50,0368271 50,0360259 50,0360259 50,0360259 50,0368271 50,0360259 50,0368271 50,0368271 50,0368271 50,0368271 50,0368271 50,0368271 50,0376283 50,0384295 50,0384295 50,0384295 50,0376283 50,0368271 50,0352248 50,0352248 50,0344237 50,0360259 50,0352248 50,0360259 50,0360259
149
Tables
Table 8-III - Averaged frequency measurements No. 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
150
Frequency (Int) 62456 62455 62456 62456 62456 62457 62457 62457 62457 62456 62456 62457 62457 62458 62458 62459 62459 62458 62457 62457 62456 62457 62457 62457 62458 62458 62457 62456 62456 62456 62456 62456 62456 62456 62456 62456 62455 62456 62455 62455 62455 62453 62452 62452 62451 62452 62451 62450 62450
Frequency (Hz) 50,0352248 50,0360259 50,0352248 50,0352248 50,0352248 50,0344237 50,0344237 50,0344237 50,0344237 50,0352248 50,0352248 50,0344237 50,0344237 50,0336226 50,0336226 50,0328215 50,0328215 50,0336226 50,0344237 50,0344237 50,0352248 50,0344237 50,0344237 50,0344237 50,0336226 50,0336226 50,0344237 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0360259 50,0352248 50,0360259 50,0360259 50,0360259 50,0376283 50,0384295 50,0384295 50,0392308 50,0384295 50,0392308 50,040032 50,040032
Tables
Table 8-III - Averaged frequency measurements No. 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
Frequency (Int) 62449 62449 62449 62449 62448 62447 62448 62449 62449 62448 62447 62447 62447 62446 62446 62446 62445 62445 62445 62446 62447 62448 62448 62449 62449 62450 62450 62450 62451 62451 62452 62451 62450 62450 62451 62452 62452 62453 62452 62453 62454 62454 62454 62455 62454 62453 62454 62453 62453
Frequency (Hz) 50,0408333 50,0408333 50,0408333 50,0408333 50,0416346 50,042436 50,0416346 50,0408333 50,0408333 50,0416346 50,042436 50,042436 50,042436 50,0432374 50,0432374 50,0432374 50,0440388 50,0440388 50,0440388 50,0432374 50,042436 50,0416346 50,0416346 50,0408333 50,0408333 50,040032 50,040032 50,040032 50,0392308 50,0392308 50,0384295 50,0392308 50,040032 50,040032 50,0392308 50,0384295 50,0384295 50,0376283 50,0384295 50,0376283 50,0368271 50,0368271 50,0368271 50,0360259 50,0368271 50,0376283 50,0368271 50,0376283 50,0376283
151
Tables
Table 8-III - Averaged frequency measurements No. 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
152
Frequency (Int) 62452 62451 62451 62451 62452 62452 62452 62451 62452 62451 62451 62451 62452 62451 62452 62451 62451 62452 62452 62452 62452 62452 62452 62451 62451 62452 62452 62453 62453 62453 62453 62453 62454 62453 62454 62454 62454 62453 62454 62454 62454 62453 62453 62452 62452 62452 62451 62451 62452
Frequency (Hz) 50,0384295 50,0392308 50,0392308 50,0392308 50,0384295 50,0384295 50,0384295 50,0392308 50,0384295 50,0392308 50,0392308 50,0392308 50,0384295 50,0392308 50,0384295 50,0392308 50,0392308 50,0384295 50,0384295 50,0384295 50,0384295 50,0384295 50,0384295 50,0392308 50,0392308 50,0384295 50,0384295 50,0376283 50,0376283 50,0376283 50,0376283 50,0376283 50,0368271 50,0376283 50,0368271 50,0368271 50,0368271 50,0376283 50,0368271 50,0368271 50,0368271 50,0376283 50,0376283 50,0384295 50,0384295 50,0384295 50,0392308 50,0392308 50,0384295
Tables
Table 8-III - Averaged frequency measurements No. 195 196 197 198 199 200
Frequency (Int) 62453 62452 62452 62451 62452 62452
Frequency (Hz) 50,0376283 50,0384295 50,0384295 50,0392308 50,0384295 50,0384295
153
Tables Note that the values in integer units are the measurements done by the load controller and the values in the S.I are calculated as explained in the respective section. Table 8-IV - RMS measurements voltage (int) 1410 1410 1410 1410 1410 1410 1410 1409 1409 1409 1409 1409 1409 1410 1410 1410 1410 1410 1409 1408 1408 1408 1409 1409 1409 1409 1409 1409 1409 1410 1410 1409 1409 1409 1409 1409 1409 1409 1410 1410 1410 1410 1409 1409 1409 1410
154
Voltage (V) 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,503227 224,503227 224,503227 224,344005 224,184783 224,184783 224,184783 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,503227 224,503227 224,344005 224,344005 224,344005 224,503227
current (int) 363 363 363 363 364 364 364 364 364 364 363 364 364 363 363 363 363 363 362 362 362 362 362 362 362 361 361 361 361 361 361 361 361 362 362 361 361 361 361 361 361 361 361 361 361 362
Current (A) 2,21557617 2,21557617 2,21557617 2,21557617 2,22167969 2,22167969 2,22167969 2,22167969 2,22167969 2,22167969 2,21557617 2,22167969 2,22167969 2,21557617 2,21557617 2,21557617 2,21557617 2,21557617 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20947266 2,20947266 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20947266
power (Int) 622406 622408 622424 622165 623789 623741 623300 623573 623138 623093 623091 623098 623224 622940 622699 622584 621658 621646 620754 620712 620712 620718 620820 621078 620208 618550 618746 619261 619377 619337 619337 619335 619405 619853 620191 618636 618278 617978 618528 618545 618545 618538 618436 618489 618439 620316
Power (W) 604,8634 604,865344 604,880893 604,629193 606,20742 606,160773 605,732203 605,997508 605,574769 605,531037 605,529093 605,535896 605,658345 605,382349 605,148142 605,036383 604,136483 604,124821 603,257962 603,217146 603,217146 603,222977 603,322102 603,57283 602,727351 601,116082 601,306558 601,807042 601,919773 601,8809 601,8809 601,878957 601,946984 602,382357 602,71083 601,199658 600,851748 600,560204 601,094702 601,111223 601,111223 601,10442 601,005295 601,056801 601,00821 602,832307
Tables
Table 8-IV - RMS measurements voltage (int) 1410 1410 1410 1410 1410 1410 1410 1409 1409 1409 1409 1409 1410 1410 1410 1410 1410 1409 1409 1409 1409 1409 1409 1409 1409 1409 1409 1410 1410 1410 1410 1410 1410 1410 1410
Voltage (V) 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,503227 224,503227 224,503227 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227
current (int) 362 361 361 361 361 361 361 360 361 361 361 362 362 362 362 362 361 361 361 361 361 361 362 362 362 362 362 362 363 362 362 362 362 362 362
Current (A) 2,20947266 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,19726563 2,20336914 2,20336914 2,20336914 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,21557617 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266
power (Int) 619299 618714 618733 618763 618763 618761 618500 617741 617917 618790 619367 620667 620396 620406 620406 620402 620345 620057 619835 619448 620337 620398 620589 620599 620599 620594 620710 621041 622379 621486 621092 620827 620902 620947 620947
Power (W) 601,843971 601,27546 601,293924 601,323079 601,323079 601,321135 601,067491 600,329884 600,500923 601,349318 601,910055 603,173414 602,910052 602,91977 602,91977 602,915883 602,86049 602,580607 602,364864 601,988772 602,852715 602,911996 603,097612 603,107331 603,107331 603,102472 603,215202 603,536873 604,837161 603,96933 603,586435 603,328904 603,40179 603,445522 603,445522
155
Tables
D.3
Frequency decay Table 8-V - Frequency decay No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
156
fequency decay 0,03970647 -0,03954634 0 0 0 0,03970647 0,03971905 0 0,03973164 -0,0395714 0 -0,03955887 0 0 0 -0,03954634 0 0 0,03970647 0 0 0 -0,03954634 0,03970647 0 0 -0,03954634 0,03970647 -0,03954634 0 0 0 0 0 -0,03953382 -0,03952131 0 0 0,03968133 0,0396939 0,07943811 0 0,03973164 -0,07911773 0,03971905 -0,03955887
Tables
Table 8-V - Frequency decay No 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
fequency decay 0 0,03971905 -0,03955887 0,03971905 0 0 0,03973164 0 0 0 -0,0395714 0 0,03973164 0 0,03974424 0 0,03975685 0 -0,03959649 -0,03958394 0 -0,0395714 0,03973164 0 0 0,03974424 0 -0,03958394 -0,0395714 0 0 0 0 0 0 0 0 -0,03955887 0,03971905 -0,03955887 0 0 -0,07906764 -0,03952131 0 -0,03950881 0,03966878 -0,03950881 -0,03949631
157
Tables
Table 8-V - Frequency decay No 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
158
fequency decay 0 -0,03948383 0 0 0 -0,03947135 -0,03945888 0,03961863 0,03963116 0 -0,03947135 -0,03945888 0 0 -0,03944642 0 0 -0,03943396 0 0 0,03959361 0,03960612 0,03961863 0 0,03963116 0 0,03964369 0 0 0,03965623 0 0,03966878 -0,03950881 -0,03949631 0 0,03965623 0,03966878 0 0,03968133 -0,03952131 0,03968133 0,0396939 0 0 0,03970647 -0,03954634 -0,03953382 0,0396939 -0,03953382
Tables
Table 8-V - Frequency decay No 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
fequency decay 0 -0,03952131 -0,03950881 0 0 0,03966878 0 0 -0,03950881 0,03966878 -0,03950881 0 0 0,03966878 -0,03950881 0,03966878 -0,03950881 0 0,03966878 0 0 0 0 0 -0,03950881 0 0,03966878 0 0,03968133 0 0 0 0 0,0396939 -0,03953382 0,0396939 0 0 -0,03953382 0,0396939 0 0 -0,03953382 0 -0,03952131 0 0 -0,03950881 0
159
Tables
Table 8-V - Frequency decay No 194 195 196 197 198 199 200
160
fequency decay 0,03966878 0,03968133 -0,03952131 0 -0,03950881 0,03966878 0
161
www.oersted.dtu.dk/cet Ørsted·DTU Centre for Electric Technology (CET) Technical University of Denmark Elektrovej 325 DK-2800 Kgs. Lyngby Denmark Tel: (+45) 45 25 35 00 Fax: (+45) 45 88 61 11 E-mail: [email protected]
Hardware design of demand as frequency controlled reserve Master’s Thesis, January 2008
Alejandro Nieto Gutiérrez
Hardware design of demand as frequency controlled reserve Master’s Thesis, January 2008
Hardware design of demand as frequency controlled reserve
This report was drawn up by: Alejandro Nieto Gutiérrez Supervisor(s): Zhao Xu Tonny W. Rasmussen
Ørsted•DTU Centre for Electric Technology (CET) Technical University of Denmark Elektrovej Building 325 2800 Kgs. Lyngby Denmark www.oersted.dtu.dk/cet Tel: (+45) 45 25 35 00 Fax: (+45) 45 88 61 11 E-mail: [email protected]
Release date:
05-01-2008
Category:
1 (offentlig)
Edition:
1st edition
Comments:
This report is part of the requirements to achieve the Erasmus Final Thesis Project at the Technical University of Denmark. This report represents 35 ECTS points or at least 6 full-time months of dedication.
Rights:
© Alejandro Nieto Gutiérrez, 2008
HOME UNIVERSITY INFORMATION
Titulo: Hardware design of demand as frequency controlled reserve Prototipo de controlador de consumo de cargas según la frecuencia de la red Autor: Alejandro Nieto Gutiérrez Supervisores: Zhao Xu Tonny W. Rasmussen Censor:
Nota:
Universidad: DTU, Lyngby.
Departamento: Ørsted•DTU
Duración, carga de trabajo: 35 ETCS. Six months full-time project
Tipo: Master Thesis project, Proyecto Final de Carrera.
Fecha de inicio: 1 de Febrero de 2007 Fecha de finalizacion: 5 de Enero de 2008
i
Home University Information Resumen en Español: El proyecto consiste en el desarrollo de un medidor de frecuencia de la red eléctrica desde cero. La red eléctrica ante desequilibrios de la producción y consumo de energía puede reaccionar de maneras indeseadas. El controlador desarrollado en este proyecto es capaz de medir los cambios en la frecuencia de la red eléctrica y a partir de los datos obtenidos reaccionar desconectado la carga y así disminuyendo el consumo eléctrico. Las cargas propicias para estos propósitos son las cargas, y más específicamente, electrodomésticos, que pueden ser desconectados de la red eléctrica sin que sea apreciado por parte de los usuarios. Estos electrodomésticos pueden ser por ejemplo calentadores de agua eléctricos, frigoríficos o también bombas de calor o climatizadores. Según la carga a controlar cortes de unos segundos serán inapreciables por parte del usuario pero si por parte de la red eléctrica. El uso generalizado de esta tecnología podrá reducir costes en el mantenimiento de centrales eléctricas de reserva, que en estos momentos, permanecen a la espera de que se produzca un exceso de consumo para así empezar su producción. El mantenimiento de estas centrales es caro y podrían ser aprovechadas para la producción regular de energía. Otra ventaja de este controlador puede ser su capacidad de reacción ante producciones de energía irregulares como pueden ser las fuentes de energía renovables. Estas fuentes de energía pueden mermar su producción sin dejar suficiente tiempo a la red para recuperarse, mediante el uso de esta tecnología, se podrá asumir gran parte de estas inestabilidades dejando un mayor margen de tiempo para la recuperación de la red. En este documento se explica el proceso completo del desarrollo del controlador de cargas según la frecuencia. Está contemplado desde los inicios del desarrollo, con la selección de componentes, hasta la programación del código fuente del micro controlador, pasando por el desarrollo y diseño del hardware adicional utilizado para completar los propósitos.
Palabras clave: Microcontroladores, medidor de frecuencia, programación C, DSP, DSC, Code Composer Studio, filtro paso banda, fuente de alimentación, ADC, hardware, software, medidor de potencia RMS, Event Manager, Reserva eléctrica, equilibrio consumoproducción, Energías renovables, fuente de alimentación.
ii
ABSTRACT This document is the final report for “Hardware design of demand as frequency controlled reserve” project, carried out by Alejandro Nieto Gutiérrez, Erasmus student from Spain. The aim of this project is to develop a load controller, which is able to monitor the frequency and power of the power grid, from the wall outlet, and react with frequency changes, as well as prepare for recording the acquired data for subsequent studies. The energy Balance between consuming and producing units need to be maintained continuously, or otherwise the system will became unstable ultimately resulting in a total system collapse. This project will react with the power grid imbalances according to the frequency measured at the wall outlet. If necessary it will disconnect the load reducing the consumption, which will help to avoid a power grid imbalance as well as recover the power grid balance within instability. The introduction and the theoretical overview about concepts present in the project process are included into the first chapter. The hardware part of the project, design and development is described in the second chapter. Software development process, in instance algorithms used and programming of the application and other related matters, is explained in the third chapter. The last chapters are the future work chapter, which describes a recommended way to follow the development of the load controller project, the conclusions chapter, which gathers all the conclusions obtained once this load controller prototype is finish, the bibliography and finally the appendix, which contains all the additional information that is not included in neither of previous sections, like the tables of acquired data, as well as the full source code of application and the hardware schemes.
iii
ACKNOWLEDGES
I specially want to acknowledge the invaluable help and understanding given by my tutor and co-tutor, Zhao Xu and Tonny W. Rasmussen, without their help the accomplishment of this project would not be possible. I want to thank Knud Ole Helgesen Pedersen for the help given and also for teaching me the working of microcontrollers in the January 2007 course. I also want to thank everybody in DTU University and in Denmark in general, who made me feel like at home, in spite of being in other country. Finally I want to thank to my family for their support and my home university for given me this opportunity.
v
CONTENTS
1 Introduction..................................................................................................... 1 1.1. The Power grid.......................................................................................... 1 1.2. How the power grid can be helped ........................................................... 2 1.3. The aim and objectives of the project ....................................................... 3 1.4. Block Diagram of Load Controller ........................................................... 4 1.5. Frequency measurement overview ........................................................... 5 1.6. Frequency decay measurement overview ................................................. 5 1.7. RMS Current, voltage and power measurement overview ....................... 6 2 Algorithms, procedures and results. ............................................................. 9 2.1. Frequency measurement algorithm ........................................................... 9 2.1.1. Averaging measures ........................................................................ 16 2.1.2.
Maximum theoretical Accuracy...................................................... 18
2.1.3.
Soft frequency filter ........................................................................ 21
2.1.4.
Frequency sensor range .................................................................. 22
2.1.5.
Accuracy ......................................................................................... 23
2.2. Frequency decay measurement ............................................................... 24 2.3. Control logic Type I: Frequency Thresholds .......................................... 26 2.4. Control logic Type II: Temperature Set points change........................... 30 2.5. Voltage, current and Power measurement .............................................. 31 2.5.1. Sample acquisition .......................................................................... 31 2.5.2.
Acquired measures .......................................................................... 32
2.5.3.
RMS Current, voltage and power calculation ................................. 36
2.6.
Real-time stamp ...................................................................................... 42
3 Hardware ....................................................................................................... 45 3.1. DSC......................................................................................................... 45 3.1.1. DSC prerequisites ........................................................................... 46 3.1.2.
DSC Technical features .................................................................. 48
3.1.3.
DSC Hardware Description ............................................................ 50
vii
Contents 3.1.3.1. C28x CPU .................................................................................. 50 3.1.3.2. Memory Bus (Harvard Bus Architecture) .................................. 50 3.1.3.3. Real-Time JTAG and Analysis .................................................. 51 3.1.3.4. SARAMs .................................................................................... 51 3.1.3.5. Flash ........................................................................................... 53 3.1.3.6. Security....................................................................................... 53 3.1.3.7. Peripheral Interrupt Expansion (PIE) Block .............................. 54 3.1.3.8. Oscillator and PLL ..................................................................... 54 3.1.4.
DSC Periphery Description ............................................................. 55
3.1.4.1. Event Manager ........................................................................... 55 3.1.4.1.1. GP timers ............................................................................. 57 3.1.4.1.2. Capture unit ......................................................................... 57 3.1.4.1.3. Starting the A/D Converter with a Timer Event ................. 58 3.1.4.2. ADC ........................................................................................... 58 3.1.4.3. General purpose I/O ................................................................... 61 3.2. Board Description.................................................................................... 62 3.2.1. Signal interfacing and conditioning ................................................ 63 3.2.1.1. Measurement transformers ......................................................... 63 3.2.1.2. Analog filters .............................................................................. 64 3.2.1.3. Level shifter................................................................................ 66 3.2.2.
Current and voltage values at load controller inputs ....................... 69
3.2.3.
Power Supply .................................................................................. 71
3.2.4.
eZDSP F2812 .................................................................................. 73
4 Software.......................................................................................................... 75 4.1. IDE .......................................................................................................... 75 4.2. User Manual ............................................................................................ 76 4.2.1. Developing board connection .......................................................... 77
viii
4.2.2.
The project ....................................................................................... 78
4.2.3.
Configurations ................................................................................. 81
4.2.4.
Modify, build and load a project ..................................................... 82
4.2.5.
Debug project .................................................................................. 85
4.2.6.
Gel File ............................................................................................ 86
Contents 4.3. The load controller program ................................................................... 88 4.3.1. File structure ................................................................................... 88 4.3.2.
Memory mapping ............................................................................ 89
4.3.3.
Program structure ............................................................................ 92
4.3.3.1. Headers ...................................................................................... 93 4.3.3.1.1. Inclusions ............................................................................ 93 4.3.3.1.2. Constant definition and Location of attributes ................... 93 4.3.3.1.3. Functions definition ............................................................ 96 4.3.3.1.4. Variables definition ............................................................ 98 4.3.3.2. Device initialization ................................................................... 99 4.3.3.2.1. System Control Initialization ............................................ 100 4.3.3.2.2. Peripheral High speed clock configuration....................... 100 4.3.3.2.3. GPIO configuration .......................................................... 101 4.3.3.2.4. PIE initialization ............................................................... 101 4.3.3.2.5. ADC initialization............................................................. 102 4.3.3.2.6. EV configuration, timers and capture unit ........................ 104 4.3.3.2.7. Relay initialization ............................................................ 105 4.3.3.2.8. Enable CPU interruptions ................................................. 105 4.3.3.2.9. Average frequency buffer initialization ............................ 106 4.3.4.
Capture unit subroutine: Capture1_srt() ....................................... 107
4.3.5.
ADC interruption subroutine: ADC_srt() ..................................... 112
4.3.6.
Other subroutines .......................................................................... 114
5 functionalities and performance ................................................................ 117 6 Future work ................................................................................................. 119 6.1. Flash memory ....................................................................................... 119 6.2. Communication part ............................................................................. 119 6.3. Persistent storage .................................................................................. 120 7 Conclusions .................................................................................................. 121 8 Bibliography ................................................................................................ 123 A
Source code .............................................................................................. 125 A.1 Main program source code ................................................................... 125 A.2 Memory mapping .................................................................................. 139
ix
Contents B
Board Schemes ......................................................................................... 140
C
Other plots................................................................................................ 143
D
Tables........................................................................................................ 145 D.1 ADC Measurements .............................................................................. 145 D.2 Frequency measures .............................................................................. 148 D.3 Frequency decay .................................................................................... 156
x
LIST OF FIGURES Figure 1-I – Load controller Block diagram ..................................................................... 4 Figure 2-I - period of a signal ........................................................................................... 9 Figure 2-II - Schmitt-triggered output ............................................................................ 10 Figure 2-III - Frequency measurement block diagram ................................................... 11 Figure 2-IV - Overflow in frequency counting ............................................................... 12 Figure 2-V – Capture unit measurements ...................................................................... 14 Figure 2-VI - Instant Frequency measurements ............................................................. 15 Figure 2-VII - Error spread using average ...................................................................... 16 Figure 2-VIII - Averaged frequency measurements in integer values............................ 17 Figure 2-IX - Averaged frequency measurements in Hz ................................................ 17 Figure 2-X - Instant and averaged measurements .......................................................... 18 Figure 2-XI - frequency counting error .......................................................................... 19 Figure 2-XII - Software frequency filter block diagram................................................. 22 Figure 2-XIII - Frequency decay .................................................................................... 24 Figure 2-XIV – Frequency decay ................................................................................... 26 Figure 2-XV – Averaged frequency measurements vs frequency thresholds ................ 27 Figure 2-XVI Frequency decay Vs Frequency decay threshold ..................................... 28 Figure 2-XVII- Threshold checking flow diagram ......................................................... 29 Figure 2-XVIII - ADC curret measures .......................................................................... 32 Figure 2-XIX - ADC Voltage measures ......................................................................... 33 Figure 2-XX - Integer measures of current without offset ............................................. 33 Figure 2-XXI - Integer measures of voltage without offset............................................ 34 Figure 2-XXII - Measured Current ................................................................................. 34 Figure 2-XXIII - Measured Voltage ............................................................................... 35 Figure 2-XXIV - Instant power in integer units ............................................................. 35 Figure 2-XXV - Calculated instant Power...................................................................... 36
xi
List of Figures Figure 2-XXVI - Measured RMS current (int) ............................................................... 37 Figure 2-XXVII - Measured RMS Voltage (int) ............................................................. 38 Figure 2-XXVIII - Measured RMS power (int) .............................................................. 38 Figure 2-XXIX - Measured RMS current (A) ................................................................. 40 Figure 2-XXX - Measured RMS Voltage (V) ................................................................. 41 Figure 2-XXXI - Measured RMS power (W) ................................................................ 41 Figure 3-I - Texas instruments Logo ............................................................................... 49 Figure 3-II - DSC Memory map ...................................................................................... 52 Figure 3-III - Event Manager A Functional Block Diagram ........................................... 56 Figure 3-IV - ADC block diagram .................................................................................. 59 Figure 3-V – ADC example Voltage output ................................................................... 60 Figure 3-VI - Board distribution ..................................................................................... 62 Figure 3-VII - Measuring Transformers .......................................................................... 63 Figure 3-VIII - Filter transfer function ............................................................................ 65 Figure 3-IX – Filter photo ............................................................................................... 65 Figure 3-X- Filter Scheme............................................................................................... 66 Figure 3-XI - Voltage level Shifter Scheme .................................................................... 67 Figure 3-XII - Current Level Shifter Scheme ................................................................. 68 Figure 3-XIII - Level shifter result .................................................................................. 69 Figure 3-XIV - Single power supply scheme .................................................................. 71 Figure 3-XV - Measure at the bridge output, point C (Without capacitors) ................... 72 Figure 3-XVI – Power supply output .............................................................................. 72 Figure 3-XVII - Double Power Supply scheme .............................................................. 73 Figure 3-XVIII - eZdsp F2812 photo .............................................................................. 74 Figure 4-I - Code Composer Studio main window ......................................................... 76 Figure 4-II – Connect ...................................................................................................... 77 Figure 4-III - Board connected message ......................................................................... 77 Figure 4-IV - Open existing project ................................................................................ 78 Figure 4-V - Open Project dialog .................................................................................... 79 Figure 4-VI - Opened project .......................................................................................... 79 Figure 4-VII - Main file source Code .............................................................................. 80 Figure 4-VIII - Linker configuration ............................................................................... 81 Figure 4-IX – Select Configurationt ................................................................................ 81
xii
Contents Figure 4-X - Build a project ............................................................................................ 82 Figure 4-XI - Build report............................................................................................... 82 Figure 4-XII - Load Program .......................................................................................... 83 Figure 4-XIII - Load project output ................................................................................ 83 Figure 4-XIV - Reload Project........................................................................................ 84 Figure 4-XVRun a program ............................................................................................ 84 Figure 4-XVI - watch window ........................................................................................ 85 Figure 4-XVII - Break point ........................................................................................... 86 Figure 4-XVIII - GEL file .............................................................................................. 86 Figure 4-XIX - GEL Menu ............................................................................................. 87 Figure 4-XX - Example code of GEL language ............................................................. 87 Figure 4-XXI - Simplified program flow ....................................................................... 92 Figure 8-I - Board Scheme............................................................................................ 141 Figure 8-II - DSC connectors Detail ............................................................................. 142 Figure 8-III – Averaged frequency measurements (Integer units) ............................... 143 Figure 8-IV – Averaged frequency measurements (Hz)............................................... 144
xiii
LIST OF TABLES
Table 7-I – ADC samples table..................................................................................... 145 Table 7-II - Instant frequency measurements ............................................................... 148 Table 7-III - Averaged frequency measurements ......................................................... 149 Table 7-IV - RMS measurements ................................................................................. 154 Table 7-V - Frequency decay........................................................................................ 156
xv
LIST OF SYMBOLS Symbol
Unit
Integer
Output integer value of ADC
Integer
First averaged measure in a pair of period measurements
Integer
Second averaged measure in a pair of period measurements
Hz
_
!
Frequency decay
N/A
Factor that converts integer current measurements in S.I units
N/A
Factor that converts integer power measurements in S.I units
Hz
Frequency at power grid
N/A
Factor that converts integer measurements in S.I units
V/V
Total current gain in current channel
V/V
Current gain in current level shifter
V/V
Current gain in current transformer
V/V
Total voltage gain in voltage channel
V/V
Voltage gain in voltage level shifter
V/V
Voltage gain in voltage transformer
A
Current consumed by the load
A
Current present at ADC input
Integer N/A
"#
%$Integer
!
&_ '
W !
Frequency of capture unit counter
Hz/s
N
&
Definition
RMS Current in integer units at ADC Number of event manager cycles within a power grid cycle or number of ADC measures used in RMS calculations Integer value of offset sensed by ADC Power grid power in RMS
Integer
RMS power in integer units at ADC
Ω
Resistance of ADC current resistor
xvii
List of Symbols Symbol
Unit
% %
% % )
)
)
!
)_ ∆ ∆(
∆%
xviii
S
First period value in a pair of period measurements
S
Second period value in a pair of period measurements
Integer
(
!
Definition
Period measured by capture unit
S
Time of a time base count unit
S
Period at power grid
V
Voltage at load controller input
V
Voltage value sensed by ADC in Volts
V
Power grid Voltage in RMS
Integer
RMS voltage in integer units at ADC
Hz
Increment of frequency in two following frequency measures
S
Increment of time in two following frequency measures
S
Relative period error in frequency measurements
ACRONYMS
# µC
Microcontroller
A ADC
Analog to Digital Converter
C CPU
Central Processing Unit
CSM
Code Security Module
CUPS
Capture Unit input clock PreScaler
D dB
Decibel
DC
Direct Current
DSC
Digital Signal Controller
DSP
Digital Signal Processor
E EV
EVent manager
EVA
EVent manager A
EVB
EVent manager B
F FCF
Frequency Conversion Factor
G GP
General Purpose
GPIO
General Purpose Input/Output
xix
Acronyms
I IDE
Integrated Development Environment
J JTAG
Joint Test Action Group
M MAC
Multiplication And aCcumulation
P PC
Personal Computer
PIE
Peripheral Interrupt Expansion
PLL
Phase Locked Loop
R RAM
Random Access Memory
RMS
Root Mean Square
ROM
Read Only Memory
S SI
Le Système International d'unités (International System of units)
SRAM
Single-access Random Access Memory
U UART
Universal Asynchronous Receiver-Transmitter
xx
1
INTRODUCTION
1.1. The Power grid Electrical systems in the form of power grids are one of the most efficient ways of transferring energy over large distances directly to the consumers, hereby limiting the need for production of energy in the closeness of consumers. Electrical power in the quantities that modern countries consume cannot be stored in any currently available storage solution, and can therefore not be used as a solution to preserve the power grid. The energy Balance between consuming and producing units need to be maintained continuously, or otherwise the system will became unstable ultimately resulting in a total system collapse. The responsibility of maintaining the system balance is placed with the system operators. These operators have to monitor the power system continuously and the most effective way of doing so, is to measure the frequency. When the production and consumption of active electrical power is in balance the frequency will be 50Hz. The reason being, that electrical energy taken out of the system matches the mechanical power grid into the system, in fact in this state the rotational speed of generators will remain fixed. As the system starts to lose synchronization, the rotations speed of the synchronous generators will begin to change, which make us know that the balance is beginning to degrade by changing the frequency of the grid. In instance of frequency rise, it indicates that the production starts to exceed the consumption. The generators will begin to rotate faster because of the excess energy delivered from the mechanical side of the generation unit. If the frequency drops below 50Hz then the consumption exceeds the production.
1
Introduction The electrical systems are thought up to solve this last problem by activating the reserve systems that will inject or consume energy until the consumption-generation balance is recover.
1.2. How the power grid can be helped As explained in previous section when the consumption exceeds the production the frequency starts to drop. Although the electrical system is able to solve this problem it is possible to help to solve it from the consumer side, reducing the need of reserve systems. By doing so, many benefits can be achieved, low costs, the possibility to use the reserve system as primary production and also energy savings while de loads are disconnected. Most importantly, it is to facilitate increased integration of fluctuating renewable energy, in example wind generators, into the power system in the future. The renewable energy sources have an unstable production. Therefore an unexpected drop of power production in this kind of sources can also cause the excess of consumption. The easier possibility to introduce renewable energy sources can reduce the contaminant emissions can be reduced. The consumers have at home several loads that can be switched on and off without noticing it, these loads are named FCLs. A control device that can be able to sense the power grid status and under frequency changes could react will help to preserve the power grid integrity, for example by switching off this FCLs in an excess of consumption, or increasing the thermostat temperature in the particular cases of a fridge or decreasing it in a water heater.
2
Introduction
1.3. The aim and objectives of the project The objective of this project is to provide a fast-responding control device that reacts in response to frequency events. It also has to be able to measure the Root Mean Square (RMS) power, voltage and current. The controller must sense the local grid frequency at the wall outlet and deenergize the load, based in two control logics. The load will be tripped if the frequency goes below a specified threshold, or if the rate of frequency decay is lower than a user specified threshold for the logic Type I. Tripping criteria: Maximum delay 0,5s (in can be reduced by compromising the frequency accuracy) C1: Low frequency criteria: <49.90 Hz and C2: Fast frequency decay criteria <-0.13 Hz/sec Reconnection criteria: R1: Tripped by C1, frequency recovers to a margin above tripping set point (49.98 Hz) and C2 not met R2: Tripped by C2, C2 not met and after a time delay R3: Both subject to a random time lag 0~30 Secs It is possible to change the values of C1, C2, R1, and R3
The control logic type II is specially designed for those thermostically controlled loads with cyclic on/off durations. In this type, instead of tripping the loads according to the frequency directly, their temperature set points are dynamically controlled according to the frequency For type II, the set points of the appliance (fridge or air-con) can be normal T high = T high − kf * ( f − f 0 )
(1.1)
normal Tlow = Tlow − kf * ( f − f 0 )
This control logic is not completely development in this load controller prototype due to the miss of the interface that can change the thermostat temperature of a FLC of this type. From this point a frequency event is understood as the frequency change that fulfills any of the control logics described above.
3
Introduction
The load is re-energized after the recovery criteria fulfill, utilizing a random waiting period between a selectable amount of time (30 seconds as default) providing a smooth transition for the grid. The random time lag prevents all loads from turning back on all at the very same time after a frequency event, which could cause a rebound effect, potentially tripping other overload relays on transmission or generation equipment
1.4. Block Diagram of Load Controller
Power Outlet
Voltage channel
Current channel
ADC&EM
Voltage Divider
DSC
Transimpedance amplifier
I/O
LOAD
Relay
Figure 1-I – Load controller Block diagram The block diagram above, (Figure 1-I – Load controller Block diagram), describes the basic operation of the control device. The load represents the controlled device such as a fridge, a water heater or any other FCL appliance. The relay represents can energize or de-energize the device according to control logics, in the case that the power grid is under a frequency event. It can also be a control module able to change the thermostat temperature or any other action that can reduce the consumption of the controlled load. The voltage divider or voltage transformer reduces the voltage of the power grid to a range that can be sensed by the Analog to Digital Converter (ADC) integrated on the Digital Signal Controller (DSC) chip.
4
Introduction The transimpedance transformer outputs a voltage proportional to the input current. It is also connected with the ADC input which allows measuring the current and consequently the power of consumed by the controller device. The transimpedance transformer is compound by a current transformer and a know resistor that drops a know amount of voltage according to the input current. The output signal of both transformers is conditioned before being connected with the ADC inputs; the conditioning hardware is included in the transformers blocks in order to simplify the diagram.
1.5. Frequency measurement overview The frequency measurement method used is the zero-crossing detection. This method measures the time between zero-crossings, which is the period of the power grid signal. The frequency can be directly obtained madding use of the period, This algorithm is explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement. The frequency measurement is used to implement the control logic Type I, which is based on the frequency changes.
1.6. Frequency decay measurement overview The frequency decay measurement is realized using the frequency measurements of consecutive cycles. It is calculated the decay between selectable amounts of pairs of measures in a row. Immediately after it, is calculated the average of these frequency decays and finally it is compared with the frequency decay threshold. In the case that the threshold is surpassed, the control device proceeds to start the frequency event response. More details about the algorithm used are covered in section 2.2 Frequency decay measurement. This type of measurement is used to implements the control logic Type II, which is based on the frequency decay changes.
5
Introduction
1.7. RMS Current, voltage and power measurement overview The instant voltage, current and power varies in the time from negative to positive values. Current, voltage and power should be expressed in Root Mean Square (RMS) values. The RMS measurements can contribute in the subsequent studies more than instant values. A microcontroller will acquire only discrete values, therefore in the RMS calculations it is used a discrete method. RMS measurements can be calculated for a series of discrete values varying function. The name, RMS, comes from the fact that it is the square root of the mean of the squares of the values. It is a power mean with the power p = 2. The RMS for a collection of n values is:
(1.2)
In the project, it is important to monitor all the time the power consumption of the FCL appliances, in order to know exactly the state of the load when a frequency event happens . In addition, the accuracy of the power measures is not as important as the time they are made. For example it is more important to have a power measure with accurate time stamp than the accurate power measurement itself. The RMS values are obtained from the ADC samples. The ADC has to sample the channels at a fast speed to have sufficient accuracy in measurements. Experimentally a sample rate of every 1ms is found enough for our purposes. This sample rate allows us to calculate the RMS values from 20 measures within a cycle. It is also possible to select the number of cycles used in the calculation of RMS values, by default are used two cycles so 40 measures are involved in the RMS calculation process. The more cycles used in RMS values the more accurate is the result.
6
Introduction
In the worst case the power measure is displaced 1ms from the frequency measurement which is affordable according to the project specifications. The maximum displacement is produced when the frequency measurement is made just before a new sample acquisition. As it is explained in following sections, frequency measurements are obtained every zero-crossing and RMS measurements every one millisecond.
7
2
ALGORITHMS, RESULTS
PROCEDURES
AND
This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the
2.1. Frequency measurement algorithm This section describes the frequency measurement algorithm. It is based in measuring the time between two consecutive rising zero-crossings.
Rising zero-crossing
Rising zero-crossing
T
t
Figure 2-I - period of a signal
The frequency measurement algorithm is part of the Control logic Type I and Control logic Type II. In control logic Type I the frequency is used as a tripping criteria if it goes below the frequency event threshold. In control logic Type II it is used to regulate the temperature set points according to the control logic Type II equations, equation (1.1).
9
Algorithms, procedures and results First of all is needed to detect when the power grid signal crosses by zero, then it is needed to measure the time between two consecutive crossings. The DSC has a specific module that can carry out these actions, the capture unit (3.1.4.1.2 Capture unit), the capture unit is part of the periphery Event Manager (3.1.4.1 Event Manager). The capture unit is able to detect level changes at the signal connected with its inputs. It can detect four different events, rising or falling edges and constant positive or negative level. The capture unit can be configured to detect any of them. In order to simplify the level changing detection the capture unit has and Schmitttriggered gate at the input, this gate converts the power grid sinusoidal signal in a square signal with the same frequency. The square signal makes easy to detect the level changes in the input signal.
Figure 2-II - Schmitt-triggered output The red plot in the figure above represents the output of the Schmitt triggered gate. The square signal period Tsq is the same as the sinusoidal one Tsin. The capture unit has a timer associated, it is used as time base, when the capture unit is enabled the timer stats to count. A signal event is detected when the level is maintained at least for two time base units. Every time it is detected any of the selected signal events the capture unit carries out the following actions. It stores the current count value of the timer in the specific two-level FIFO stack. It is also called the interruption subroutine configured in the capture unit interruption vector (4.3.3.2.4 PIE initialization).
10
Algorithms, procedures and results
The FIFO stack, which contains two consecutive signal events, can be read from the interruption subroutine. In the case of the load controller the capture unit is configured to detect the rising edges, which coincide with the beginning and end of a single cycle. The power grid period is calculated in the interruption subroutine madding use of capture time stored in the FIFO stack. The first capture time is subtracted to the second one, this subtraction results on the number of time base count units between the two rising edges. Knowing the time that corresponds to a count unit is easy to get the period of the power grid signal in seconds, whose inverse is the frequency.
Level-Shifter output
S chmitt-trigger
Time base clock Stores the Counter value CAP1FIFO.1
CAP1FIFO.1
Shifts previous value CAP1FIFO.2
CAP1FIFO.2
-
CAP1FIFO - CAP2FIFO
Cycle period Figure 2-III - Frequency measurement block diagram
This algorithm fails only when the counter is overflowed, in other words, when the counter reach its maximum count value and starts again from zero. In this case, the second capture has a lower value because the counter has reach the overflow value between the two captures. This is solved checking if the value of last capture is lower than the previous. If it is lower, a counter overflow is considered. In this case the period is obtained by adding the time from the first capture to the overflow
11
Algorithms, procedures and results and the time between the overflow and the second. The Figure 2-IV - Overflow in frequency counting illustrates the cases of a measurement with and without an overflow.
Without overflow
p-2 p-1 p p+1 p+2 p+3 …
p+n p+n+1 p+n+2 p+n+3 …
With overflow
k k+1 … OF 0 1 2 ….
n
p p+1 p+2 p+3
n
temp1_1
temp1_1 temp1_2
Temp1_2>temp1_1 Period = (n+p)-p= = temp1_2 – temp1_1 =n
temp1_2 Assuming OF as maximun value of counter Temp1_2
Figure 2-IV - Overflow in frequency counting The timer base clock frequency cannot be selected arbitrarily. As is going to be explained in a following section, the higher is the frequency the better is the accuracy, which is because the count unit is smaller and the quantification error is smaller too. However if the time base frequency is too much high it can be possible that more than one overflow happens between two consecutive zero crossings, which is impossible to detected it and in consequence the obtained measurement results wrong. The interruption subroutine configured as the capture unit interruption subroutine is capture1_srt().The code analysis of this subroutine is done in section 4.3.4 Capture unit subroutine: Capture1_srt(). As explained above the number of time base count units is proportional to the period in seconds, it is only needed to know the time that corresponds to a single time base count unit and multiply it by the total number of time base count units. The time base timer is fed by the High Speed Periphery Clock (HSCLK)
12
Algorithms, procedures and results +,-./ 0 12345
(2.1)
The time base timer has a count register of 16 bits. Therefore the maximum count value is 65535. In order to avoid more than one overflow within a power grid cycle the clock speed has to be reduced. A regular power grid cycle has a period of 0,02 seconds, so that the clock speed has to be reduced to the speed that no more than 65535 count units can fit in 0,02 seconds. The way of reducing the input clock is using the prescaler included in the capture unit. The prescaler divides the speed by a fixed value. The capture unit prescaler (CUPS) can take the values of the powers of two from 1 to 256. Some easy calculations conclude that the optimal prescaler is 8. +,-./ 12345 (2.2) 6-7 0 0 0 :. <12345 -78, 9 Finally this clock speed is used to know the time in seconds that corresponds to a single count unit. < (2.3) =-7 0 0 >. :1?@ :. <12345 For example, if the Counter of the Capture unit gives the period value of 62500 count units, the calculations are the following: A8B 0 C12>> D >. :1?@ 0 >. >1@ 68B 0
< 0 2>+5 >. >1@
(2.4) (2.5)
Where TPG is the period and fPG is the frequency at power grid. It can be referred to a period expressed in the number as being in integer units. Since the number of count units is always an integer number. The DSC works internally with the integer values. It will store and compare only the integer values. It is attempted to use the integer values as much as possible, since the DSC do not have a float coprocessor. It only uses the values in Hz, which are expressed with float numbers, in the frequency decay measurement whose algorithm do not allows to do it with integer values due to an excessive lost of precision.
13
Algorithms, procedures and results In order to make easy the conversion from the integer measures to the Herzl or Seconds measures it is defined a conversion factor FCF. This factor is not more than the frequency of the time base clock. E-E 0 :<12>>> F
E-E F+5I A-7
(2.7)
A-7 F@I E-E
(2.8)
Where TCU is the period measured by the capture unit in integer units and fPG is the frequency at power grid in Hz as well as TPG is its period in seconds. The following plot shows the measures of period directly returned by the Capture unit interruption subroutine without averaging. The period is represented in integer units.
Instant Measures 62540 62520 62500 62480 62460 62440 62420 1
6
11
16
21
Instant Measures
Figure 2-V – Capture unit measurements
14
26
Algorithms, procedures and results The next figure shows the instant frequency measurements again, but It is divided into the FCF. Therefore the frequency is represented in Hz.
Instant Measures 50,040 50,030 50,020 50,010 50,000 49,990 49,980 49,970 49,960 49,950 1
6
11
16
21
26
Series1
Figure 2-VI - Instant Frequency measurements As seen in the plots, the frequency has some fluctuations. These fluctuations are reduced by averaging the output frequency measurements in order to improve the accuracy. The instant frequency measurements are not stored after being used in the averaging calculations; that is because the plots within this section shows only 30 measurements in a row, in following sections and in the appendix can be found other plots with more consecutive measurements.
15
Algorithms, procedures and results
2.1.1. Averaging measures As seen in the previous section, the instant frequency measurements presents some fluctuations, in order to improve the accuracy and stabilize the measurements it is use the average of a selectable amount of instant measurements. These averaged measurements are the ones used in the threshold checking. The more instant measures used in the average the more accurate is the measure, but the more measurements are used the more increases the delay of a measure. A frequency change will be assuredly detected after the number of cycles used in the average. This is because the frequency event measures are spread within the cycles in the average and is not assured that the measures within a frequency event will change the average enough to fulfill the threshold, at least with the early measures. The error is cancelled in the inner averaged measurements and the error in the first and the last cycle is spread in all the cycles.
time
Real zero-crossings Sensed zero-crossings Individual error
Figure 2-VII - Error spread using average
The positive error in one cycle is the opposite in the adjacent cycles. The only error that is not cancelled is the error produced by the first and the last measure. This error is divided according with the cycles used in the average. In instance of using the average of 25, the default in the project, the measurement delay is 0.5 Seconds. A the detection of a frequency event is not assured until 0.5 seconds.
16
Algorithms, procedures and results The instant measurements averaged can be observed in the following plot.
Average of 25 values 62505 62504 62503 62502 62501 62500 62499 62498 1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
Average of 25 values
Figure 2-VIII - Averaged frequency measurements in integer values
The next plot shows the same values as the previous one but in Hz units.
Average of 25 values (Hz) 50,001 50,000 49,999 49,998 49,997 49,996 49,995 1
3
5
7
9
11 13
15 17 19 21 23 25
27 29
Average of 25 values (Hz)
Figure 2-IX - Averaged frequency measurements in Hz The figure above shows the averaged frequency measurements in Hz, comparing these plots with the previous instant measurements in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the
17
Algorithms, procedures and results Frequency measurement the measures are smother. In the following figure is possible to compare in the same plot the instant measures with the averaged measurements.
instant Vs average measures 50,040 50,030 50,020 50,010 50,000 49,990 49,980 49,970 49,960 49,950 1
3
5
7
9
11 13 15 17 19 21 23 25 27 29
Instant values (Hz)
Average, 25 values (Hz)
Figure 2-X - Instant and averaged measurements It can be found more plots of frequency measurements in appendix C Other plots.
2.1.2. Maximum theoretical Accuracy This section explains the maximum theoretical accuracy that could be reached using the frequency algorithm explained in the previous sections. In this section is considered that the input power grid signal is clean of noise and the only source of error is the quantification error of the capture unit. As obvious, the accuracy calculated in this section is not the accuracy of the load controller but it is the maximum reachable accuracy. The frequency measurement depends on the frequency of Capture unit clock. It has an error of ±1 count units every measure. But as explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement algorithm the capture unit clock frequency cannot be increased more than a limit.
18
Algorithms, procedures and results
Case of counting one unit less Starts just after checking
Ends just before checking
Lost time Real length of period signal
EV clock:
The period have not started yet
Here ends The counting
Here starts The counting Case of counting one unit more
Starts just before checking
Excess time Real length of period signal
Ends just after checking Excess time
EV clock:
Here starts The counting
Signal period have not finished yet
Here ends The counting
Figure 2-XI - frequency counting error The figure above explains the error of measurements. It shows how are measured cycles whose edges are just in the limits of the Capture unit cycles. These case produces the measurements with the error of +1 or -1 count units. In other cases the excess time compensates the lost time, always withing the range of the absolute error is ±1 time base count units. The absolute in error in period can be expressed with the following equation. ∆AJKL 0 M<
(2.9)
N represents the number of time base count unit, therefore the error does not depend on the frequency measured. The accuracy or relative error depends on the frequency measured. In lower frequencies, the number of time base count units is higher, for that reason the absolute error is proportional smaller than the measure itself. In higher frequencies the number of time base count units is smaller and the absolute error is proportional higher.
19
Algorithms, procedures and results
As explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement algorithm. The capture unit uses a GP timer to count the Capture Unit Cycles between rising edges. The timer is a 16-bit-timer so the maximum count is from 0 to 216-1 or from 0 to 65535 in decimal. The Capture unit is fed by the high speed periphery clock that runs at 25Mhz. Using directly this clock in a regular power grid cycle (1/50Hz=20ms) the counter should count 500000 that is higher that the overflow limit of the counter. A prescaler is used to reduce the clock speed. Using all the data above is possible to calculate the theoretic accuracy at 50Hz, the calculations are the following: JK M
∆A 0
TPG(50Hz)=0.02s+∆Ts=0.02±32µ
(2.10)
(2.11)
These periods corresponds to these frequencies: 6STU 0
<
ASVW
0
< 0 XY, YYY1+5 1>. >>>:1S@
(2.12)
That is a relative error of: fmin-50=-0.8Hz And finally: 6SVW 0
<
ASTU
0
< 0 2>, >>>9+5
That is a relative error of:
20
(2.13)
Algorithms, procedures and results fmax-50=0.8Hz In summary, at 50Hz we have an accuracy of ∆f(50Hz)=±0.8Hz, or a relative error of 1,6%, The relative accuracy depends on the frequency measurement, at higher speeds the absolute error can be proportional higher and the accuracy is worse. The use of accuracy improves the accuracy. The error is divided into the number of instant frequency measurements.
2.1.3. Soft frequency filter Experimentally it is observed that some measures differ too much from expected measures. These measures are produced by external interferences such as the interferences produced by the physical movement of the extension cord. In order to avoid these fake measures it is implemented a frequency filter by software. This filter checks every measure if the frequency decay is higher than a limit. In the case the measure exceeds the threshold the measurement takes the value of the previous one. The frequency threshold in this filter is too much higher than the frequency decay in the power grid in normal state and within a frequency event.
21
Algorithms, procedures and results
Attribute: MAXIMUN_FREQUENCY_CHANGE
Prev. Frequency measure
-
+
Last frequency measure
Outs the higher
Outs the lower
Frequency out
Figure 2-XII - Software frequency filter block diagram The filter basically compares the last measure with the previous one, if the different between them is higher than the limit stored in the attribute MAXIMUN_FREQUENCY_CHANGE the output frequency is the last measure, otherwise it is used the unmodified frequency measure. By default it is used the value of 150 time base count units between two consecutive frequency measurements, it corresponds at 50Hz to a decay of ±6Hz/s, the control logic type II uses a decay of -0,16Hz/s
2.1.4. Frequency sensor range The speed of the clock that feeds the capture unit sets the frequency range The lower limit is when the period is enough long to have more than one overflow inside. The limit is equal to the overflow limit of the counter, 65535 EV-cycles →65535/3.125 MHz ≈ 0,021ms = 47,7Hz
22
Algorithms, procedures and results , that is much lower than the limit in the specifications. If the period is longer is not possible to be sure that more than one overflow happens within a cycle. The higher frequency hypothetically has the limit when the frequency is so fast to at least count one unit in but is not useful because the relative error is ±100%. Therefore the higher frequency limit is at the point where the accuracy drops below a desider limit.
2.1.5. Accuracy In the accuracy study it was not possible to obtain the signal to noise ratio of power grid line, and also it was not possible to obtain the Spectral Frequency Noise density present in the working frequency band, moreover the manufacturer of DSC don’t provide the hysteresis windows width and finally it was not available a frequency counter with enough accuracy to compare the measurements made by the load controller. Therefore it is not possible to do a theoretical study of accuracy. The study of section 2.1.2 Maximum theoretical Accuracy gives the maximum theoretical accuracy, but not the accuracy of the Load Controller. Assuming the frequency measurements were not taken within a frequency event. The measurements showed in this project are randomly chosen from several similar measurements in different times, therefore it not probable that the frequency measurements where done within a frequency event. The frequency measurements vary in the experimental measurements not more than 15 mHZ within a measurement session, that means that this fluctuations can be due to regular power grid fluctuations but it cannot be possible to ensure it. Therefore the accuracy of this load Controller is within the range of ±0.8mHz and ±7,5mHz. The minimum accuracy grated by the load controller is enough for its purposes. Minimum grated accuracy ±7,5mHz. @50Hz Maximum delay: 0,5seconds.
23
Algorithms, procedures and results
2.2. Frequency decay measurement The frequency decay is measured by the average of the frequency changes within a selectable number of consecutive power grid cycles. The frequency decay measurement is part of the control logic Type I. In the control device as explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement what is really measured is the period in number of cycles of EV clock, otherwise, the time between zero-crossings. The frequency of the sample is the inverse of the period and the time between frequency measures is the period itself.
Frequency
1/T0
Δf=1/T1-1/T0 1/T1
time Δt=T1
Figure 2-XIII - Frequency decay
As seen in the figure above, the time between two correlative measures is the period of the second one and the frequency change is the subtraction of the inverses of correlative period measures. The frequency decay is defined as the difference of frequency in any amount of time. It is obtained from the period measures with the following equation:
24
Algorithms, procedures and results < < ∆6 A< N A> A> N A< EZ 0 0 0 ∆= A< A> D A1<
(2.14)
In section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement it was described that the measure of the periods is measured in number of EV cycles. The frequency decay measurement algorithm uses lots of resources. The calculations made in this algorithm needs more calculation power than others algorithms. The frequency and RMS calculations are made with the integer values from the ADC and the EV but this algorithm cannot be made using the integer values due to the rounding. The operations must to be float operations, not integer operations. The internal operations made by the CPU are specially selected to avoid as much as possible the float operations and use the integer and long operations. < < +5 ∆6 E-E D bc3< N c3> d E[\][U^_ Z[^V_ ` a 0 0 0 c3< @ ∆= E-E 1 N c3 N c3 c3 E-E c3 E-E1 c3> > < > < 1 0 E-E D 0 D 0 De N
(2.15)
Where FCF is the Frequency conversion factor that relates the integer values with the frequency values in Hz. See 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement. And AM1 represents the last frequency measurement value and AMs0 represents the previous one. Note that AM are the integer values given by the Capture unit and they are period measurements.
25
Algorithms, procedures and results g is a long multiplication, the rest of operations are float operations, one division, one subtraction and one multiplication. The frequency decay measurements are averaged in order to have more accurate results. The average can be done with a selectable number of frequency decay results. By default it is done the average of last 5 values.
Frequency decay (Hz/s) 0,1 0,05 0 1
29
57
85
113
141
169
197
-0,05 -0,1 frequency decay
Figure 2-XIV – Frequency decay
2.3. Control logic Type I: Frequency Thresholds The load tripping according to the frequency and frequency decay measured implement the control logic Type I. The load controller can be in three different states depending on the frequency sensed. The default sate is the CLOSE state. In this state the frequency is higher than the lower frequency threshold. Therefore the power grid is not within a frequency event. The load is connected with the power grid. If a frequency event is detected, when the frequency goes below the frequency threshold, the state is changed to OPEN state. In this state the load is disconnected. The load will remain always disconnected while the frequency is below the recover frequency event, which is higher than the frequency threshold.
26
Algorithms, procedures and results Once the frequency is recovered, so the recover threshold is fulfilled, the state is changed to WAIT state. When the load controller enters in this state, it is set a random time to attachment. The load controller remain in this state while the delay time is over, when the delay time is up the load controller changes to CLOSE state, being the load reconnected. In this state it is also checked if the frequency returns below the frequency threshold cancelling the reconnection and changing the state to OPEN again. The default thresholds used are the following. The frequency values in integer units is obtained by using the conversion equation (2.7): DEFAULT_HIGHER_LIMIT = 62525 = 49,98Hz DEFAULT_LOWER_LIMIT = 62626 = 49,89Hz The default lower threshold can be also called default frequency event threshold. In the case of the frequency below this threshold it is considered the power grid under a frequency event. The default higher threshold it can be called also the recover threshold, this threshold defines the frequency threshold which in the case of being surpassed within a frequency event the frequency event is considered to be recovered.
Averaged Vs Thresholds 50,1 50,05 50 49,95 49,9 49,85 49,8 1
29
57
Average measurements
85
113
recover threshold
141
169
197
frequency event threshold
Figure 2-XV – Averaged frequency measurements vs frequency thresholds The frequency decay threshold works in a similar way as frequency threshold in exception that the frequency decay event has a single threshold, it is used the same
27
Algorithms, procedures and results threshold for detecting a frequency decay event and checking the frequency decay event recover.
Frequency decay Vs Threshold 0,1 0,05 0 1
29
57
85
113
141
169
197
-0,05 -0,1 -0,15 -0,2 frequency decay
decay threshold
Figure 2-XVI Frequency decay Vs Frequency decay threshold The next figure shows the state flow diagram, the load controller starts by default in the state CLOSE. The circles represent the states. The “f” inside the decision blocks is the last frequency measure acquired. Lower_limit is the frequency threshold and higher_limit is the recover threshold. Decay represents the last frequency decay measurement and limit represents the frequency decay threshold.
28
Algorithms, procedures and results
Figure 2-XVII- Threshold checking flow diagram
29
Algorithms, procedures and results
2.4.
Control logic Type II: Temperature Set points change
The dynamic temperature set points regulation implements the control Logic Type II. As explained in the introduction, this control logic is not completely implemented. This control logic is based on regulate the load temperature set points dynamically, according to the frequency measured in the power grid. This control logic is applicable only to the loads that has a thermostat with two set points, and the load is only started when it reach one of the set points and it is stopped when the temperature reach the other set point. In this loads, a fridge or an electric heater can change the normal set points by the user. The load controller should modify this normal set points according to the frequency measured. The load controller assumes that the input normal temperature set points are introduced in the load controller thought the ADC. The thermostat should have two analog outputs which vary from 0 to 3 Volts depending on the temperature set point. This normal temperature set point reading method is only a proposal that could be changed depending on the load. The load controller has to calculate a new high and low set points depending on the power grid frequency. The output Temperature set points are calculated with the following equation: normal T high = T high − kf * ( f − f 0 )
(2.16)
normal Tlow = Tlow − kf * ( f − f 0 )
The input set points are assumed to be an integer number, the output temperature is an integer too, the Kf factor has to be regulated to use the Temperature in integer units, as well as use the frequency measurements in integer units too. The load controller stores the new temperature set points in a variable, the variable should be used to change the temperature set points according to the interface used to set the temperatures in the load thermostat.
30
Algorithms, procedures and results The normal temperature set points that should be read by the ADC are read in the ADC interruption subroutine (4.3.5 ADC interruption subroutine: ADC_srt()) the output temperature set points are calculated in the capture unit interruption subroutine (4.3.4 Capture unit subroutine: Capture1_srt()), the set points are calculated every frequency measurement, every 0,02 seconds
2.5. Voltage, current and Power measurement This section describes the process of RMS voltage, current and power measurement. First of all it is described the sampling method, and finally the process of obtaining the RMS values from the instant measurements, it is also show some sample plots of the measurements made by the ADC.
2.5.1. Sample acquisition The sample acquisition is made by the ADC. As explained in section 3.1.4.2 ADC the ADC is configured to sample the voltage and current channel every one millisecond The sample acquisition is made by the subroutine adc_isr(void). The EV is configured to start the ADC conversion every one millisecond, once the ADC has the sampling results available the interruption subroutine is called. Every time this subroutine is executed it is ensured that the ADC registers has ready to use the last measures stored inside. At this point the offset of each channel is subtracted to the measure of the input signals. The measures of voltage and current, with the offset already subtracted, are stored in an array, one array for each channel. These arrays contain by default the last 40 measures of voltage and current those are used to calculate the power and RMS values. Note that this subroutine stores the integer values of these measures and not the floating point values corresponding for the equivalent in Volts or Amperes. The DSC works every time with these integer values simplifying the calculations. As seen in section 3.1.4.2 ADC the values in volts or in amperes are obtained directly with fixed operations that can be made in the analysis time, that mean, when the measurements are retrieved into the server.
31
Algorithms, procedures and results This subroutine doesn’t calculate the RMS power, voltage and current but it calculates the square of the measures. The capture unit interruption subroutine is the responsible of the RMS calculations and it uses the last 40 measures from the ADC. In the RMS calculations is needed the square of these measures and doing it in the ADC subroutine it is only needed to do it once every new measure, otherwise the capture unit subroutine should do the square of last 40 ADC measures, which would result in not efficient DSC calculation power usage.
2.5.2. Acquired measures This section shows the plots of the measures made by the ADC. The first two plots shows the integer measures given by the ADC. These measures are within the range of the ADC, between 0 and 4096. Next two figures show the integer values given by the ADC channels of current and voltage.
Input Current 3000
Current Values
2500 2000 1500 1000 500
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
0 samples
Figure 2-XVIII - ADC curret measures
32
Algorithms, procedures and results
4500 4000 3500 3000 2500 2000 1500 1000 500 0 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Voltage values
Input Voltage
samples
Figure 2-XIX - ADC Voltage measures The offset has been measured without connecting any signal to the ADC input. The offset is so stable that it can be considered as a constant value. The offset in current and voltage is estimated in 1904 and 1921 ADC integer units. The signals used in all the load controller internal calculations are integer values given by the ADC subtracted with the corresponding offset. These measurements can be observed in the two following figures.
ADC Current integer measures (without offset) 3000
Current Values
2000 1000 0 -1000
1
11
21
31
41
51
61
71
81
-2000 -3000
samples
Figure 2-XX - Integer measures of current without offset
33
Algorithms, procedures and results
ADC Voltage integer measures (without offset) 3000
Voltage (int)
2000 1000 0 -1000
1
11
21
31
41
51
61
71
81
91
-2000 -3000
samples
Figure 2-XXI - Integer measures of voltage without offset The measures above can be also observed in I.S. units of measure in the next two figures. These values are obtained using the method described in section 3.2.2 Current and voltage values at load controller inputs.
Input Current 2 1,5
0,5 0 -0,5
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Current (Int)
1
-1 -1,5 -2
samples
Figure 2-XXII - Measured Current
34
Algorithms, procedures and results
Input Voltage 400 300
100 0 -100
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Voltage (V)
200
-200 -300 -400
samples
Figure 2-XXIII - Measured Voltage With the measurements of voltage and current it is possible to calculate the RMS power. The RMS power is calculated using the individual values of current and voltage. Therefore it is first calculated the instant power and then is calculated the RMS power using the same method used in voltage and current. The next figure shows the instant power values in integer units.
Instant power (integer untis) 1200000
Power (W)
1000000 800000 600000 400000 200000
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
0 samples
Figure 2-XXIV - Instant power in integer units The figure below shows the instant power in S.I units.
35
Algorithms, procedures and results
1000 900 800 700 600 500 400 300 200 100 0 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Power (W)
Input power
samples
Figure 2-XXV - Calculated instant Power In order to improve the accuracy the channels are oversampled, therefore the channels are sampled many times in a row and it is use as final measurement the average of all of them. In this prototype it is use the oversampling of two values, the load controller hardware allows up to oversample the channel eight times. The time between consecutives samples is 250ns which is enough small compared with the signal period to permit the oversampling.
2.5.3. RMS Current, calculation
voltage
and
power
The Current, voltage and power in RMS values are calculated in the same subroutine where the frequency is measured (Capture1_isr). RMS Voltage, current and power is calculated using this equation: 1 ∑K Tl> cZ-S[V@]k[ T i h3, 0 K
36
(2.17)
Algorithms, procedures and results Where ADC_MEASUREi represents the integer sample made by the ADC or the multiplication of the voltage and current samples in the case of the power RMS measurement. N represents the number of measures used in the calculation, by default this value is 40, therefore, the RMS value is calculated according to the last two cycles. In order to simplify the RMS calculations, the square of the instant samples are stored in specific arrays, this avoids recalculating the instant sample squares every time a RMS calculation is done. It is only calculated the square of the new samples. In summary the load controller does the following calculations in order to calculate the RMS values: h3, 0 i
∑K Tl> adc_square_sampleT K
(2.18)
Where adc_square_sample represents the respective squared samples array (voltage, current or power) The square samples are calculated in the ADC sampling subroutine (4.3.5 ADC interruption subroutine: ADC_srt()) in the array adc_square_voltage[] for voltage, adc_squeare_current[] for current and adc_squeare_power[] for power. The following plots show the measurements made by the load controller in integer units. RMS current measurement:
RMS Current (Int) Current RMS (Int)
363,5 363 362,5 362 361,5 361 360,5 360 1
11
21
31
41
51
61
71
81
Figure 2-XXVI - Measured RMS current (int)
37
Algorithms, procedures and results RMS voltage measurements:
RMS Voltage (Int) 1410,5 Voltage (INT)
1410 1409,5 1409 1408,5 1408 1407,5 1407 1
11
21
31
41
51
61
71
81
Figure 2-XXVII - Measured RMS Voltage (int) And finally the RMS power measurements:
Power RMS (Int)
RMS power (int) 1000 900 800 700 600 500 400 300 200 100 0 1
11
21
31
41
51
61
71
81
Figure 2-XXVIII - Measured RMS power (int)
It is used all the time Integer values that simplify the calculations and the subroutine can be executed faster. The integer values have a direct relation with the values in Volts, Amperes or Watts.
38
Algorithms, procedures and results In the case of the Voltage this relation is: Ph3, 0
P_cZ- h3, Ex
(2.19)
Where Fv is the conversion factor from ADC values to Real values and it can be calculated by the following way: Ex 0
X>YC D yx :
(2.20)
Where gv is the total gain of the Voltage channel.
In the case of the Current this relation is: zh3, 0
z_cZ- h3, ET
(2.21)
Where Fi is the conversion factor from ADC values to Real values and it can be calculated by the following way: ET 0
X>YC D yT D <> :
(2.22)
Where gi is the total gain of the current channel. And 10 is the resitance of the resistor connected at the ADC input
The power RMS is also calculated using the integer values given by the ADC. The RMS power can be obtained from the RMS power calculated with the integer values using the following equation:
8h3, 0
8_cZ- h3, E{
(2.23)
Where P_ADCRMS is the power calculated from the integer ADC measures and Fp is the power conversion factor that is defined in the following equation:
39
Algorithms, procedures and results ET 0 ET D Ex 0
X>YC D h D yT X>YC D yx X>YC1 D 0 D yT D yx : : :1
(2.24)
Where R is the resistor connected to the ADC current channel (see 3.1.4.2 ADC) They are some reasons because the calculation is made this routine instead of be done in the ADC subroutine, the sampling routine. ADC subroutine is executed every 1ms, 20 times every power grid cycle, but the RMS values are only needed one time every power grid cycle, which is the measurements rate specified in the specifications. By doing it in the ADC subroutine results in 20 measurements of RMS values per cycle, which means calculating it 20 times per cycle. That requires more performance that the DSC can deliver. The other reason is that the RMS values have to be of the same cycle of the frequency measure. By doing the RMS calculations in the same subroutine as the frequency measurement it is ensured that the RMS results correspond to the same cycle as the frequency value.
Current RMS (A)
RMS Current (A) 2,218 2,216 2,214 2,212 2,21 2,208 2,206 2,204 2,202 2,2 2,198 2,196 1
11
21
31
41
51
61
Figure 2-XXIX - Measured RMS current (A) RMS voltage measurements:
40
71
81
Algorithms, procedures and results
RMS Voltage (V) 224,6
Voltage (V)
224,5 224,4 224,3 224,2 224,1 224 1
11
21
31
41
51
61
71
81
71
81
Figure 2-XXX - Measured RMS Voltage (V) And finaly the RMS power measurements:
RMS power (W) 623000
Power RMS (W)
622000 621000 620000 619000 618000 617000 616000 1
11
21
31
41
51
61
Figure 2-XXXI - Measured RMS power (W)
The three figure above shows the RMS values of current, voltage and power in the international system of units but the DSC works internally with the integer values, the plots of the RMS results in integer values are omitted in this section and it can be found in the appendix C Other plots. In the appendix it can also be found the tables of values of these plots.
41
Algorithms, procedures and results As well as explained in the frequency accuracy it is not enough available data to calculate and exact voltage, current or power accuracy. The maximum variation in RMS voltage measurements is 0,6 V in all the voltage measurement sessions, therefore the voltage accuracy should be better than ±0,3V. This corresponds to an accuracy of ±0,13% Using the same reasoning as before the maximum variation in Current measurement is 0,03A, therefore the current accuracy should be better than ±0,015A. This corresponds to an accuracy of ±0,05%. Finally referring to the power measurements the maximum variation observed is 10W in the test load. Therefore the power accuracy should be better than ±5W. This corresponds to an accuracy of ±0,8%. All the results described above are only estimated results that could change depending on the test load stability. It should be considered a detailed accuracy study. Accuracy Voltage
±0,3V
Current
±0,015A
Power
±5W
Note that this result are done using the test load of nominal 600W. The current accuracy could be improved by expanding the current signal in the ADC conversion range, it can be done by changing the number of loops in the current transformer. The maximum RMS measurement delay is 1ms.
2.6. Real-time stamp The real time stamp is used in order to locate the measurements in time. The time stamp is a long integer value that is incremented every one millisecond. The Real-time system also offers the possibility to get the current time and also to synchronize the time stamp with a external time source.
42
Algorithms, procedures and results
The time stamp register is incremented in the ADC interruption subroutine that is periodically executed exactly every one millisecond. The time stamp uses an arbitrary date as origin. For example choosing January 1st of 2008, the register will contain the millisecond elapsed from that date. The time stamp is valid until the register doesn’t overflow. Long integer variables stores 64 bits so that the total time while this load controller can work is:
|}~ 0 2 0 18446 D 10 Therefore the time stamp will remain valid until around 584942417 years. What is more than enough. The function time_synchronization(long time) sets the current time to the time passed in the argument. The current time is obtained by reading the current_time variable.
43
3
HARDWARE
This chapter explains everything related with the hardware, the selection of DSC, design of the board and description of the developing board.
3.1. DSC The DSC (Digital signal controller) is one of the most important parts of the hardware. It is the core of the control device. It manages the hardware and made the calculations. As a key part of the project, a deep research had been made. The conclusion of the research is show within this section. The function of the DSC can be made also by microcontrollers (µC) and Digital Signal Controllers (DSC), they are some differences between then. In the following paragraphs the different proposals are compared.
The microcontroller is designed for a wide field of uses but not for any specific function. Therefore, they are capable of virtually carry out any application, it is designed as a generic device. Microcontrollers are also easy to use and cheap. Referring to this project, microcontrollers in budget do not have enough signal processing features. A more specific solution is needed. Regular microcontrollers are not able to carry out all the requirements of the control device. In instance they are used to be lack of ADC inputs and other signal processing specific hardware, such as a 32 bit MAC (Multiplication And aCcumulation) module needed to carry out the numerous multiplications needed in the RMS calculations. The DSPs solves the typical microcontroller problems described above. They have specific instructions for signal processing, the used to have enough ADCs and they
45
Hardware are quite powerful. But like the microcontrollers, DSPs have some problems, they are more expensive and difficult to program and they use lack on general purpose peripheries like Universal Asynchronous Receiver-Transmitters (UART) or Serial Periphery Interfaces (SPI), being needed to use a microcontroller as co-processor. Therefore it will be very effective in Signal processing but not so good in other stuff. The solution is a DSC. The DSC is basically a Microcontroller with DSP capabilities. They have a microcontroller as core with the common peripheries in microcontrollers (UART, SPI, Timers, GPIO and so on) and they have also peripheries common in DSPs (ADCs, MAC, EV). It is as simple to program as a microcontroller, even having some signal specific instructions. For that reason is decided to use a DSC that is more flexible than a DSP and has more Signal processing capabilities than a microcontroller.
3.1.1. DSC prerequisites The project specifications determine some features that the DSC must have in order to fulfill all the specifications The aim of the project is measure the frequency of the power grid as well as the power consumption of the controlled appliances. It should be able to measure in the future tri-phased signals if the project is continued by that way. Power measurement specification requires that the microcontroller must have an ADC with multiple input channels. It must have at least two channels for a mono-phase power grid source, one for voltage and other for current. These two channels must also be sampled at simultaneous time, in other words, it must have at least one dual channel able to sample both of the inputs at the same time. Looking for the lower complexity the best option is to choose a DSC that has the ADC on-chip with at least one dual channel, for a single phase of Current and Voltage. The DSC is hast to do some complex calculations periodically, so it is required that it is able perform all the calculations within the period of available time. It has to calculate the average of last frequency measures, frequency decay and also voltage, current and power in RMS values, all of them every power grid cycle.
46
Hardware
The DSC must have a MAC module able to calculate 16x16 (used in RMS voltage and current calculations) and 32x32 (used in RMS power Calculation) integer multiplications. The RMS calculations made an intensive use of multiplications every a short amount of time (1ms), these MAC unit is essential to ensure that all of the calculations are made before the next RMS calculation. The DSC has to support the connection of external memory. All the previous measurements before a frequency event should be stored in order to send them to the server. The external memory connection can be done using the on-chip external memory interface (cable select outputs and memory address outputs) or using a seral flash memory (using the SPI periphery). Finally according the DSC should have a UART port in order to connect the GPRS module for data communications between device and a Server. Other matters not related with the technical features had been considered in order to take the best decision. It has been considered the availability of a developing board, a board with the DSC integrated and with the PC connection ports, in order to start quickly the developing of the hardware. The availability of a good Integrated Development Environment (IDE), more details in section 4.1 IDE, had also been considered, as well as the support and documentation of the manufacturer and obviously, the final price. A detailed research drops as result that the best option is the TSM320F2812 manufactured by Texas instruments. Its technical features are showed in the next section.
47
Hardware
3.1.2. DSC Technical features The features of the selected DSC are the following. High-Performance Static CMOS Technology 150 MHz (6.67-ns Cycle Time) Low-Power (1.8-V Core @135 MHz, 1.9-V Core @150 MHz, 3.3-V I/O) Design 3.3-V Flash Programming Voltage JTAG Boundary Scan Support High-Performance 32-Bit CPU (TMS320C28x) 16 x 16 and 32 x 32 MAC Operations 16 x 16 Dual MAC Harvard Bus Architecture Atomic Operations Fast Interrupt Response and Processing Unified Memory Programming Model 4M Linear Program Address Reach 4M Linear Data Address Reach Code-Efficient (in C/C++ and Assembly) TMS320F24x/LF240x Processor Source Code Compatible On-Chip Memory Flash Devices: Up to 128K x 16 Flash (Four 8K x 16 and Six 16K x 16 Sectors) ROM Devices: Up to 128K x 16 ROM 1K x 16 OTP ROM L0 and L1: 2 Blocks of 4K x 16 Each Single-Access RAM (SARAM) H0: 1 Block of 8K x 16 SARAM M0 and M1: 2 Blocks of 1K x 16 Each SARAM Boot ROM (4K x 16) With Software Boot Modes Standard Math Tables External Interface (F2812) Up to 1M Total Memory Programmable Wait States Programmable Read/Write Strobe Timing Three Individual Chip Selects Clock and System Control Dynamic PLL Ratio Changes Supported On-Chip Oscillator Watchdog Timer Module Three External Interrupts Peripheral Interrupt Expansion (PIE) Block That Supports 45 Peripheral Interrupts 128-Bit Security Key/Lock Protects Flash/ROM/OTP and L0/L1 SARAM Prevents Firmware Reverse Engineering Three 32-Bit CPU-Timers Motor Control Peripherals Two Event Managers (EVA, EVB) Compatible to 240x Devices Serial Port Peripherals Serial Peripheral Interface (SPI) Two Serial Communications Interfaces (SCIs), Standard UART Enhanced Controller Area Network (eCAN) Multichannel Buffered Serial Port (McBSP) With SPI Mode
48
Hardware 12-Bit ADC, 16 Channels 2 x 8 Channel Input Multiplexer Two Sample-and-Hold Single/Simultaneous Conversions Fast Conversion Rate: 80 ns/12.5 MSPS Up to 56 Individually Programmable, Multiplexed General-Purpose Input/Output (GPIO) Pins Advanced Emulation Features Analysis and Breakpoint Functions Real-Time Debug via Hardware Development Tools Include ANSI C/C++ Compiler/Assembler/Linker Supports TMS320C24x™/240x Instructions Code Composer Studio™ IDE DSP/BIOS™ JTAG Scan Controllers [Texas Instruments (TI) or Third-Party] Evaluation Modules Broad Third-Party Digital Motor Control Support Low-Power Modes and Power Savings IDLE, STANDBY, HALT Modes Supported Disable Individual Peripheral Clocks Package Options 179-Ball MicroStar BGA™ With External Memory Interface (GHH), (ZHH) (2812) 176-Pin Low-Profile Quad Flatpack (LQFP) With External Memory Interface (PGF) (2812) 128-Pin LQFP Without External Memory Interface (PBK) (2810, 2811) Temperature Options: A: -40°C to 85°C (GHH, ZHH, PGF, PBK) S: -40°C to 125°C (GHH, ZHH, PGF, PBK) Q: -40°C to 125°C (PGF, PBK)
This DSC fulfills all the prerequisites. It has enough ADC inputs, it has all the external interfaces required, it is not expensive and it has enough calculation power. It also has a specific periphery that carries out the frequency measurement, the Event Manager (EV).
Figure 3-I - Texas instruments Logo The final price of a single DSC unit is about 75DKK, depending on the amount purchased the price can be lower.
49
Hardware
3.1.3. DSC Hardware Description This section describes the DSC hardware and the peripheries on-chip.
3.1.3.1.
C28x CPU
The C28x DSP generation is the newest member of the TMS320C2000 DSP platform. The C28x is an efficient C/C++ engine, enabling users to develop their system control software in a high-level language using C/C++. The 32 x 32-bit MAC capabilities of the C28x and its 64-bit processing capabilities, enable the C28x to efficiently handle higher numerical resolution problems that would otherwise demand a more expensive floating-point processor solution. This is perfect for the calculation of power from voltage and current in real time which requires lots of long multiplications. The C28x has an 8-level-deep protected pipeline with pipelined memory accesses. This pipelining enables the C28x to execute at high speeds without resorting to expensive high-speed memories. Therefore this CPU is able to execute up to 8 instructions at the same time. Special branch-look-ahead hardware minimizes the latency for conditional discontinuities, so in branches the execution taken up again more efficiently. This feature is very useful in the project that is based in asynchronous and synchronous interruptions, shouted by the power grid cycles and ADC every sample. That ensures a good performance on the interruption execution against the continuous changes of context. Every time that an interruption is called, the microcontroller must save the context of execution, load the specific context of the interruption and when it ends the interruption execution it must restore the previous context. This change of context happens around 20ms in asynchronous interruption (every power grid zero-crossing) and every 1ms in periodical interruption (sample rate of ADC).
3.1.3.2.
Memory Bus (Harvard Bus Architecture)
As with many DSP type devices, multiple busses are used to move data between the memories and peripherals and the CPU. The C28x memory bus architecture contains a program read bus, data read bus and data write bus. The program read bus consists of
50
Hardware 22 address lines and 32 data lines. The data read and write busses consist of 32 address lines and 32 data lines each. The 32-bit-wide data busses enable single cycle 32-bit operations. The multiple bus architecture, commonly termed “Harvard Bus”, enables the C28x to fetch an instruction, read a data value and write a data value in a single cycle. This feature increases the execution speed being able to retrieve the results of the peripheries at the same time that the next instruction is read.
3.1.3.3.
Real-Time JTAG and Analysis
The F281x implement the standard IEEE 1149.1 JTAG interface. Additionally, the F281x support real-time mode of operation whereby the contents of memory, peripheral, and register locations can be modified while the processor is running and executing code and servicing interrupts. This is very useful in the debug stage where it is necessary to trace the running of the program and in also useful to modify the register to emulate external events. The F281x implement the real-time mode in hardware within the CPU. This is a unique feature to the F281x, no software monitor is required. Additionally, special analysis hardware is provided that allows the user to set hardware breakpoint or data/address watch-points and generate various user selectable break events when a match occurs. The features above can be done using the parallel connection with the Personal computer (PC) A specific JTAG emulator hardware also allows the user to write inside the onchip flash memory unfortunately it cost about 1300$ (6 857DKK), but in this prototype there is enough memory with only the on-chip and external RAM memory. So that it is not used in this project stage.
3.1.3.4.
SARAMs
The DSC contains two blocks of single access memory, each 1K x 16 in size. The stack pointer points to the beginning of one of this block on reset. The F281x also contains an additional 16K x 16 of single-access RAM, divided into 3 blocks (4K + 4K + 8K). Each block can be independently accessed hence minimizing pipeline stalls.
51
Hardware
Figure 3-II - DSC Memory map The RAM memory described in this section is not all the memory available in the load controller; it is only the on-chip RAM. External RAM from the Developing Boar is also available.
52
Hardware
3.1.3.5.
Flash
The F2812 contain 128K x 16 of embedded flash memory, segregated into four 8K X 16 sectors, and six 16K X 16 sectors The user can individually erase, program, and validate a flash sector while leaving other sectors untouched. However, it is not possible to use one sector of the flash to execute flash algorithms that erase/program other sectors. Special memory pipelining is provided to enable the flash module to achieve higher performance. The flash is mapped to both program and data space; therefore, it can be used to execute code or store data information. The use of flash is slower in program execution speed than the internal RAM. In order to use the flash memory is mandatory to use the JTAG emulator. The JTAG was not purchased and the final program is inserted it in RAM memory. Anyway, the interruption functions developed in this prototype are time critical function and they must be on the RAM memory.
3.1.3.6.
Security
The F281x and C281x support high levels of security to protect the user firmware from being reverse-engineered. The security features a 128-bit password (hardcoded for 16 wait states), which the user programs into the flash. One code security module (CSM) is used to protect the flash/ROM and the SARAM blocks. The security feature prevents unauthorized users from examining the memory contents via the JTAG port, executing code from external memory or trying to boot-load some undesirable software that would export the secure memory contents. To enable access to the secure blocks, the user must write the correct 128-bit ”KEY” value, which matches the value stored in the password locations within the Flash/ROM. In this project I will not use this feature that will be only useful in a final product in order to avoid the copy of the source. If using this feature it is possible under some circumstances to leave the microcontroller locked, so that the memory is unprotected all the prototyping time.
53
Hardware
3.1.3.7.
Peripheral Interrupt Expansion (PIE) Block
The PIE block serves to multiplex numerous interrupt sources into a smaller set of interrupt inputs. The PIE block can support up to 96 peripheral interrupts. On the DSC, 45 of the possible 96 interrupts are used by peripherals. The 96 interrupts are grouped into blocks of 8 and each group is fed into 1 of 12 CPU interrupt lines (INT1 to INT12). Each of the 96 interrupts is supported by its own vector stored in a dedicated RAM block that can be overwritten by the user. The vector is automatically fetched by the CPU on servicing the interrupt. It takes 8 CPU clock cycles to fetch the vector and save critical CPU registers. Hence the CPU can quickly respond to interrupt events. Prioritization of interrupts is controlled in hardware and software. Each individual interrupt can be enabled/disabled within the PIE block. In this project the interruptions are essential. The ADC interruption has higher priority because it is executed every 1ms versus the frequency measurement interruption that is executed every 20ms. The remaining time for the frequency interruption (time where the ADC interruption is not executing) is more than enough to complete the desired calculations.
3.1.3.8.
Oscillator and PLL
The DSC can be clocked by an external oscillator or by a crystal attached to the on-chip oscillator circuit. A Phase Locked Loop (PLL) is provided supporting up to 10input clock-scaling ratios. The PLL ratios can be changed on-the-fly in software, enabling the user to scale back on operating frequency if lower power operation is desired. The PLL block can be set in bypass mode in order to use the external source. This PLL feeds the main clock of the DSC, in this project it is set up to work at maximum speed, 150MHz. Using a lower speed results in a lower consume of energy but the time critical and calculation power exigencies of this project should not allow using a lower clock frequency. The calculation in the frequency decay algorithm uses lots of available calculation power in the CPU as well as RMS calculations. Using a lower speed could result in inaccurate frequency measurements in the case that the next frequency interruption is called before the previous one finish its execution.
54
Hardware
3.1.4. DSC Periphery Description This DSC has several peripheries but not all of them are used. In the following sections, the peripheries used in this project are described.
3.1.4.1.
Event Manager
The event manager will be the periphery entrusted with the frequency measurement. The DSC counts with two Event Manager Modules. The event-manager (EV) modules include general-purpose (GP) timers, fullcompare/PWM units, capture units, and quadrature-encoder pulse (QEP) circuits. Every EV module count with two GP timers and three capture units. The two EV modules are similar, EVA and EVB. EVA and EVB functions, timers, compare units, capture units and so on are identical. Event managers A and B have similar peripheral register sets except for different addressing. The capture unit need to be associated with a GP timer within the same module, the associated timer is used as time base of the capture unit. The capture units can use any of the EV timers with the only restriction that the capture units 2 and 3 must use the same timer as time base. The load controller uses two timers, one of each EV module and the capture unit 1 of the EV module A (EVA). One GP timer is used to control the ADC sample rate, this timer is also used to increase the real-time stamp. The capture unit also uses another timer in the measurement of the power grid signal period. More information about the timers usage can be found in following sections.
55
Hardware
Figure 3-III - Event Manager A Functional Block Diagram
56
Hardware
3.1.4.1.1.
GP timers
Each Event Manager (EV) has two General Purpose timers. These timers can be used for any purpose. They are not reserved for DSC internal functions. Every timer has individual configurations registers, so that it can be used for different purposes at the same time. In this project, it is used one timer in frequency measurement. It is used as time base of the capture unit. It will be explained in the section 3.1.4.1.2 Capture unit. Another timer is in charge of the ADC sample. It will be explained in section 3.1.4.1.3 Starting the A/D Converter with a Timer Event, this last timer is also used in the real-time stamp, explained in section 2.6 Real-time stamp.
3.1.4.1.2.
Capture unit
This is a module of the event manager that reacts with external events. It is configurable to detect any change in its inputs. The capture module is configured to detect rising edges in its inputs. It has Schmitt-triggered inputs, so that, when connecting the voltage signal with the input, the corresponding interruption subroutine is called every power grid cycle. This module basically starts the associated general purpose counter used as time base and in every rising-edge, it stores the value of the timer in a register and throws an interruption. When the timer overflows the count starts at zero again. In the interruption subroutine (capture1_srt()) the program reads the registers where the last two measures are stored, by subtracting them the period of the input signal is obtained. The Timer speed is directly related with frequency accuracy. The faster runs the timer the better accuracy is obtained. If the timer counter is increase at too much speed, the counter that measures the frequency can be overflowed within two zero-crossings and the measurements would be wrong. The frequency measurement process deeply explained in section 0 This chapter describes the algorithms and methods used to develop the load controller. In this chapter it is also show the results obtained by the Frequency measurement.
57
Hardware
3.1.4.1.3.
Starting the A/D Converter with a Timer Event
The event manager can be used to control the sample rate of the ADC. It is possible to configure a Timer to define the period of sampling. The ADC has four special registers called sequence registers, in this registers is stored the sequence of inputs measured. In this project EV is configured to react with timer 1 period interruption. This interruption is generated every 1 millisecond, when the timer count is equal to its compare register. This results in a sample rate of 1kHz, therefore, it samples the input twenty times per cycle.
3.1.4.2.
ADC
The ADC module has 16 channels, configurable as two independent 8-channel modules to service event managers A and B. The two independent 8-channel modules can be cascaded to form a 16-channel module. Although there are multiple input channels and two sequencers, there is only one converter in the ADC module. Figure 3-IV - ADC block diagram shows the block diagram of the F2812 ADC module. The two 8-channel modules have the capability to auto sequence a series of conversions; each module has the choice of selecting any one of the respective eight channels available through an analog MUX. The two 8-channel modules can be configured in a simultaneous sample mode, where the pairs of inputs (one of each 8channel module) can be sampled at the same time. On each sequencer, once the conversion is complete, the selected channel value is stored in its respective ADCRESULT register (ADCRESULT1, ADCRESULT2…). In the project the sequencers are set to measure the instant current and voltage simultaneously two times. The result used is the average of the two measurements, this over sampling will increase the SNR of the meassurments. Although the ADC only has one ADC converter it is possible to sample dual channels using the sample and hold circuit. This circuit samples the channels at exactly the same time and keeps the value until the single ADC converter converts both of the channels. This circuit is necessary in order to have instantaneous measures of current and voltage.
58
Hardware
Figure 3-IV - ADC block diagram Below is a detailed list of the ADC module features: 12-bit ADC core with built-in dual sample-and-hold (S/H) Simultaneous sampling or sequential sampling modes Analog input: 0 V to 3 V Fast conversion time runs at 25 MHz, ADC clock, or 12.5 MSPS 16-channel, multiplexed inputs Autosequencing capability provides up to 16 “autoconversions” in a single session. Each conversion can be programmed to select any 1 of 16 input channels Sequencer can be operated as two independent 8-state sequencers or as one large 16-state sequencer (i.e., two cascaded 8-state sequencers) Sixteen result registers (individually addressable) to store conversion values Multiple triggers as sources for the start-of-conversion (SOC) sequence S/W − software immediate start EVA − Event manager A (multiple event sources within EVA) EVB − Event manager B (multiple event sources within EVB) External pin Flexible interrupt control allows interrupt request on every end-of-sequence (EOS) or every other EOS Sequencer can operate in “start/stop” mode, allowing multiple “time-sequenced triggers” to Synchronize conversions EVA and EVB triggers can operate independently in dual-sequencer mode Sample-and-hold (S/H) acquisition time window has separate prescale control
59
Hardware
4000 3500 3000 2500 2000 1500
Input Voltage Offset
1000 500 0 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99
ADC output (integer Units)
Input Voltage
Number of sample
Figure 3-V – ADC example Voltage output
The figure above shows the output of the ADC in the voltage channel and the respective offset. The offset is fixed to the value measured in absence of other input signal. The ADC output is a series of integer values between 0 and 4096. The integer value corresponds with the range from 0 to 3 volts. The integer results are convertible to voltages present at ADC input with the following equation. PcZ- 0
JcZ-S[V@]k[ N EE,OAL D : FPI X>YC
(3.1)
ADCmeasure represents the value returned by the ADC, in instance of the voltage channel, the OFFSET has the value of 1921. A similar equation can be used to obtain the current at the input of the current ADC channel.
zcZ- 0
60
JcZ-S[V@]k[ N EE,OAL D : FPI X>YC D h
(3.2)
Hardware As well as in the previous equation, ADCmeasure represents the value returned by the ADC, in the case of current, the OFFSET has the value of 1904. The R in the denominator represents the resistance of the resistor connected to the current transformer. The ADC is only be able to measure the voltage, so that, the ADC measures the voltage dropped in a know resistor, being direct to obtain the current thought it. It is used a 10Ω resistor. The OFFSET is the integer value corresponding to the level shifter input. This offset is added in order to sample the input signal that varies from negative to positive values. The ADC is only able to work with positive signals. The offset cannot be measured every signal measurement, that is because the level shifter inverts the input signals, so that, the offset is regulated to the invert value. In order to improve the accuracy of the measurements, the offset could be measured every time the signal is sampled. The offset should be obtained from the level-shifter reference input but it has to be inverted, due to the stability of the offset reference value and the fact that the offset has to be inverted, it is used a fixed offset value. There are more details about the level shifter in section 3.2.1.3 Level shifter.
3.1.4.3.
General purpose I/O
Most of the periphery pins are configurable to be used as a General purpose input/output (GPIO). In this project, one pin is used as I/O. This pin controls the relay that will disconnect or reconnect the load. It could be possible to use more than one pin in the case that more reactions were needed against a frequency event. There is more details about the General purpose Inputs/outputs in section 4.3.3.2.3 GPIO configuration.
61
Hardware
3.2. Board Description The load controller needs some special hardware specially designed to achieve its purposes. The included hardware in the DSC developing board (more details in 3.2.4 eZDSP F2812) is not enough. It only has RAM memory connected to the on-chip external memory interface, direct connections with the chip pins, the PC connection (thought JTAG port) and the usual ADC external reference hardware (two capacitors). The load controller has to measure the frequency and the power of the power grid. The wall outlet output has to be transformed in order to be measured. First of all the AC power grid at 230V cannot be connected to the 3V inputs of the microcontroller’s ADC. It is needed to reduce the voltage to the allowed range. The DSC has only ADCs that convert voltage to digital values but in order to calculate the power we also need to know the Current. Therefore, a current to voltage converter is required. It has as well to be reduced to a compatible range. The power grid is affected by noise, external interferences, and also by harmonics. In order to obtain the cleanest possible measurements the input signal must be filtered. Once the input signal is reduced and filtered, It has to be adapted to the DSC voltage input range. The input signal varies between positive and negative values. The DSC, and in consequence the ADC, only works with voltages between 0 and 3 Volts, so it is also needed to shift the signal to only positive values. It is done by adding to the input signal and offset that centers it in the middle of the ADC range. Regarding to the power supply, all parts of this hardware need to be powered so that, a single positive power supply for the DSC and a double power supply for the filters are included in this board.
Figure 3-VI - Board distribution
62
Hardware
3.2.1. Signal interfacing and conditioning This section it is describes how the signal coming from the power grid is conditioned in order to obtain a proper signal for the DSC hardware.
3.2.1.1.
Measurement transformers
These transformers are in charge of reducing the power grid signal at 230V to a range that can be used by the microcontroller. These transformers have lots of repercussion in the final measures. Therefore, high-precision transformers are used at this point.
Figure 3-VII - Measuring Transformers The voltage Transformer is a high-precision toroidal transformer. It has a conversion ratio of 325:1.65, its gain is fixed and cannot be changed. It gain is calculated in the following equation (3.3):
63
Hardware yx= 0
:12 0 >, >>2 FP/PI <, C2
(3.3)
The current transformer ratio is variable, the installed transformer has 50 loops in the primary coil and the ratio can be change according to the number of loops thought the transformer. The gain of the current transformer is the following (3.4) : K Fc/cI 2> Where N represents the number of loops thought the primary coil.
y^= 0
(3.4)
The number of loops cannot be set arbitrary. The current at the ADC input cannot have amplitude higher than 150mA. Higher amplitude causes voltages with amplitude higher than 1.5V which saturates the range of the ADC. In order to measure the test load (A heater) it is used two loops. Therefore the current transformer ratio is 50:2 which results in this special case in a current gain of: . K 1 (3.5) y^= 0 0 0 >, >X Fc/cI 2> Kl1 2> The gain of the transformers is calibrated in the last stage of the signal conditioning hardware, the level shifters, it is explained in section 3.2.1.3 Level shifter.
3.2.1.2.
Analog filters
The power grid signal is affected by noise and harmonics. In order to obtain accurate measurements active analog filters are used. In both of the channels, voltage and current, it is used butterworth second order filters. The signals outside the working band will be reduced 40dB per decade, the signal inside the band is not affected, it has a gain of 0dB. The measurement band is centered in 50Hz and has a band-width of 1Hz. This band is too small to be implemented with real components without compromise other
64
Hardware filter characteristic like the phase spread. Therefore, it is used a real pass-band filter with a band-width of 20 Hz. It fulfills the purposes of the project because it sufficiently reduces the noise outside the working band and the first harmonic at 100Hz.
Figure 3-VIII - Filter transfer function
In the figure above it is show the transfer function of the filters, it is a second order filter so it has a gain of -40dB/decade outside the pass band. The band-width is exactly 20.96 Hz at -3dB. The label at 100Hz corresponds to the first harmonic, that it is reduced in -22.46dB.
Figure 3-IX – Filter photo
65
Hardware The scheme of the filters is the following:
Figure 3-X- Filter Scheme
3.2.1.3.
Level shifter
The level shifter is the last stage of the signal conditioning hardware. The ADC of the DSC works with voltages between 0 and 3 Volts but the signals at the output of filters is in the range of ±1.5 Volts. The level shifter consists on one Operational Amplifier (OA) in the configuration of an inverter mixer, whose inputs are connected with the voltage or current signal and with the offset reference. The level shifter has to add an offset of 1.5 Volts. This offset centers the signal in 1,5Volts and removes the negative voltages. Due to the inverter configuration, the offset reference has a value of -1,5V which is converted to 1,5V at the level shifter output. The level shifters is as well used as gain stage (The filter stage has no gain in the working band). The gain of the level shifters is calibrated to expand as much as possible the input signals into the ADC input range. Therefore, in the voltage channel the level shifter is calibrated to output the power grid signal with 3 Volts in peak-to-peak and with a DC offset of 1.5 Volts.
66
Hardware The current channel calibration cannot be done without knowing the controlled load, depending on the load the current amplitude can change. As seen in section 3.2.1.1 Measurement transformers the current signal range can be calibrated in the current transformer.
Figure 3-XI - Voltage level Shifter Scheme
The gain of the voltage level shifter is calculated with the followings equations: yx@ 0 N
h1 <>> D <>: P 0N 0 N>, Y ` a : h< <<> D <> P
(3.6)
It is also added an offset of -1,5V Offset, with a gain of -1. Finally the transfer function of this level shifter is the following: 0 N>, Y D <, 2 FI
(3.7)
The current channel is similar in exception of the resistor R1. It can be observed in the following scheme.
67
Hardware
Figure 3-XII - Current Level Shifter Scheme
The gain of the current level shifter is calculated using the following equation: y^@
h1 <>> D <>: P 0N 0N 0 N>, :1 ` a :<< D <>: h< P
(3.8)
Note that the units of this gain are [V/V] because at this point the current is yet converted to voltage. As well as in the voltage channel it is added an offset of -1,5V Offset, with a gain of -1. Finally the transfer function of this level shifter is the following: P 0 N>, : D PT <, 2 FPI
68
(3.9)
Hardware
Figure 3-XIII - Level shifter result The input signal inversion do not affect in the final calculations, the RMS calculations uses the square of the ADC samples, therefore, a change of sign is not representative.
3.2.2. Current and voltage controller inputs
values
at
load
Section 3.1.4.2 ADC described how to obtain the values of current and voltage in the ADC inputs but these values are not the same as the values present in the power grid. This section describes how to obtain the values at the wall outlet the values acquired by the ADC. In order to know the real values is needed to consider the voltage and current transformers, as well as the gain of the level shifter. Considering them, the current and voltage in the load controller inputs can be obtained with theses equations: P 0 PcZ- D
< C2> 0 PcZ- D yx :
(3.10)
Where gv is the total gain of the voltage channel:
69
Hardware P : yx 0 yx= D yx@ 0 >. >>2 D >. Y< 0 >, >>XC ` a 0 P C2>
(3.11)
Where gvt is the total gain of the voltage transformer (3.2.1.1 Measurement transformers), and gvls (equation (3.3)) is the total gain of the voltage level shifter. The instant peak-to-peak power grid amplitude is 650 V and the 3V is the peak-topeak amplitude in the ADC input. The level shifter gain is calibrated to obtain a 3V peak-to-peak signal from the transformer output that has a fixed transfer ratio. The voltage don’t depend on the load, therefore the equation above is suitable with any load. The current calculations are similar to the voltage calculations. The equation for obtaining the current at the load controller inputs is the following: z 0 zcZ- D
< y^
(3.12)
Where gc is the total gain of the current channel, the current measured depends on the load connected with the load controller inputs. The current transformer gain can be configured according to the load connected. P y^ 0 y^= D y^@ ` a P
(3.13)
Where gct is the total gain of the current transformer, and gcls the total gain of the current level shifter.. The total gain in the current channel in the particular case of the test load, two loops through the primary coil, is the following: P
y^ 0 y^= D y^@ 0 >. >X D >. : 0 >, ><1 P
70
(3.14)
Hardware
3.2.3. Power Supply The board has two different powers supplies. A 5 Volts power supply to power the DSC and a double ±5Volts power supply to power the filters and the signal drivers. It is used two power supplies because the power supply used in the filters introduces interferences in the DSC which needs s supply as stable as possible. The energy is obtained from the power grid but the output of this power supplies will not be used to measure the frequency or the power of the line. These supplies are used only to power the system. It is used a simple power supply based on a bridge rectifier and on a voltage regulator. In the Figure 3-XIV - Single power supply scheme it is show the scheme of a single power supply. The double power supply will be exactly the same with exception on the voltage regulator that will be a negative voltage regulator (LM7905). A detailed scheme of both of the power supplies can be found in annex B Board Schemes. In the rest of this section it is explained the function of the power supply. In the Figure 3-XIV, which is below this paragraph, the letters from A to D represents voltage measurers or oscilloscopes and in following figures the measurements at these points is showed.
Figure 3-XIV - Single power supply scheme At point A its present a typical sinusoidal signal at 230 Volts (RMS) from the power grid. In point B it is found the signal with the voltage reduced, its amplitude is around 10Volts. The output of the transformer is connected with a bridge rectifier (DB101G). This component is composed of 4 diodes that invert the negative part of the signal. The output result, at point C, without connecting anything more is the following.
71
Hardware
Figure 3-XV - Measure at the bridge output, point C (Without capacitors) At this point the power supply offers only positive voltages, but it is not the constant value in a DC power supply. In order to “fill up” the valleys it is used two capacitors in parallel, they works as a very low low-pass filter. The results of at this point, point C with the capacitors installed ,can be watch at next figure, in the magenta plot.
Figure 3-XVI – Power supply output As observed in the graphic above the output starts to be similar to a DC power supply but the curl is too much big to be useable (magenta plot).
72
Hardware In order to reduce this curl we used a voltage regulator. This component “cuts” the input signal to a fixed value, the regulator used regulates to the voltage to 5V (cyan plot). It consumes the excess voltage, so it gets warm and sometimes is needed a heat sink. Finally it is used a capacitor in order to stabilize the DC level with the ground. The double power supply is similar to the single one, in exception that it uses each transformer output for each voltage level. It also uses other voltage regulator (LM7905C) for the negative output that regulates the voltage to -5 Volts.
Figure 3-XVII - Double Power Supply scheme
3.2.4. eZDSP F2812 In order to start the prototyping of any hardware based on the DSC F2812 it is used the developing board eZdsp manufactured by Spectrum digital. The developing board is a board with the DSC integrated and the minimum additional hardware needed to develop any project. Te hardware included in eZdsp is the following: • One tms32F2812 DSC •
PC parallel communications port, This port is used to connect the device to the PC, thought this port the programs are loaded into the DSC memory. It is also possible to control the execution and watch the contents of the memory in real time in order to debug the developed software
73
Hardware •
64Kx16 of extra external RAM. The board has this memory attached to the external memory interface, it is used to extend the code and data memory.
•
ADC analog interface, This port is used to connect the external reference of ADC and the inputs to sample, the board also includes the extra hardware needed in the case of using the internal reference.
•
I/O interface, This port are the friendly pins connected with the on-chip DSC peripheries, these pins are multiplexed with the GPIO pins.
eZdsp is fully supported by the IDE developed by Texas instruments, which is included in the developing board kit, as a reduced version only functional with this DSC.
Figure 3-XVIII - eZdsp F2812 photo More details can be found in the technical reference provided by Spectrum digital. The reference can be found in the Bibliography section
74
4
SOFTWARE
This section describes all the software process. Once the hardware is designed this is next step in the load controller development, where the processor is programmed to carry out its duty. This section uses two different types of font styles. The regular type, Times new Roman, used in regular text, and console type used in source code Example source code text
4.1. IDE The Integrated Development Environment (IDE) is the software installed in the PC that allows the user to develop, debug and load a program into the DSC. The quality of the IDE is decisive in order to obtain a good quality program. The IDE used in this project is Code Composer Studio (CCS) which is included in the developing board. This IDE is one of the best IDEs in the market. The provided CCS has the only limitation that it can only be used with the included DSC. It cannot be used with other DSC, DSP or microcontroller. The IDE has all the featured usually present in a microcontroller environment. It has a text editor with a syntax coloring engine. It manages all the source code and header files. The IDE is able to compile and link the code and obtain output binary program that is loaded into the DSC.
75
Software The IDE also provides the user debug functions like the possibility of adding software beaks and follow the execution step by step, as well as watch the values stored in any memory location at any time.
4.2. User Manual This section contains the basics to use the CCS IDE, for more details it should be referred to the CCS included tutorial. The first step is opening the application. The main window looks like this:
Figure 4-I - Code Composer Studio main window It has the look-and-feel as most of the applications available under Windows operative system. In this brief tutorial it is supposed that the user is familiarized with the Windows operative system.
76
Software
4.2.1. Developing board connection It is not needed to change any configuration in Windows neither in CCS. It has to be connected the parallel port wire from the ezDSP board to the PC parallel input and finally it is needed to connect the power socket to the power grid. Once all the connections are done it hast to connect the CCS with the board. It is done by clicking the option Debug->Connect in the main window.
Figure 4-II – Connect If the connection succeeds The picture of a socket in the bottom left corner goes from red to green, it appears a balloon message advising that the connection is done and it appears a new window that shows the current contents of the DSC memory.
Figure 4-III - Board connected message
77
Software
4.2.2. The project All the CCS developments are formed by a CCS project. The CCS project is folder that contains a file with the extension .pjt, this file contains all the IDE configurations made for the development, as well as the location of all the other files included in the project. These files should be located in the same folder where the pjt file is located. Otherwise it should be possible to miss files if the application is ported to other computer.
In the menu Project->Open… is possible to open a existing project, it can be selected new… to create a new one.
Figure 4-IV - Open existing project
78
Software When the open option is selected it appears a regular opening windows dialog.
Figure 4-V - Open Project dialog It has to be selected the .pjt file, after clicking the Open button (Abrir) the project is open and the main window changes.
Figure 4-VI - Opened project
79
Software The File view tab shows all the files involved in the load controller development. By double clicking any file it is opened to be modified. For example if it is double-clicked the main source code file proyectoMain.c it is showed on the right part of main window.
Figure 4-VII - Main file source Code
80
Software
4.2.3. Configurations The general, compiler and linker configurations can be changed in the build menu. For example the following figure shows the linker configuration window.
Figure 4-VIII - Linker configuration It is possible to have simultaneous profiles with different configurations.
Figure 4-IX – Select Configurationt It is possible to create a new profile in the project menu option configurations…. More details about the configuration can be found in the CCS included tutorial.
81
Software
4.2.4. Modify, build and load a project If it is desired to modify the source code of any file it has to be opened as explained in the previous version. Once the changes are made the files have to be build in order to be loaded into the DSC memory.
Figure 4-X - Build a project This option compiles all the files modified, if wanted it is possible to rebuild all the files modified or not by clicking on rebuild all. After clicking the build option it is showed a report of the building on the botton part of the main window.
Figure 4-XI - Build report If the compilation ends without any problem the report should look like the figure above. Once the project is correctly compiled the next step is to load it into the DSC memory in order be executed. The program is loaded by clicking on the menu option Load Program inside the menu File.
82
Software
Figure 4-XII - Load Program It appears a dialog similar to the open project one. In this dialog it has to be selected the file .out contained in the folder debug that is inside the main folder, where the pjt file is contained. The debug folder corresponds to the output generated by the configuration named debug, it is possible to create various configurations and use any of them at any time. In this project it is only used one configuration.
Figure 4-XIII - Load project output
83
Software Once the building output is loaded one time, it is possible to reload it bypassing the open dialog, the project could be modified and rebuilding and the reload option will reload the last built project.
Figure 4-XIV - Reload Project The loading process is the last step to run a program, if the loading process finish successfully the program can be run. A loaded program is run by pressing the option run in the menu debug. The execution of a program is stopped by clicking on the option halt, also in the menu debug. It is not possible to reload a program while a program is running.
Figure 4-XVRun a program
84
Software
4.2.5. Debug project The IDE CCS offers lots of debug tools. One of the tools is the watch window, by right-clicking on any variable in the source code and selecting the option add to watch window… it is possible to watch the contents of the variables in real time.
Figure 4-XVI - watch window
Other typical debug tool is the control of execution, CCS offers the possibility to set software and hardware break points, the execution stops in these points and can be controlled. For example it is possible to run to the next break point, execute just one assembler instruction or one c instruction.
85
Software A software break point can be easily set by double-clicking on the left of any line of code. It can be watch in the next figure, where the red circle is the break point and the yellow arrow is the point of execution.
Figure 4-XVII - Break point
4.2.6. Gel File The CCS has the possibility to be programmed and personalized madding use of the programming language GEL. It is possible to load a GEL file in CCS, only one for all the projects opened, but in can be changed at any time.
Figure 4-XVIII - GEL file
86
Software A possible personalization is the possibility to add a options menu with personalized actions.
Figure 4-XIX - GEL Menu In this project the GEL file has been modified to add in the watch window the relevant variables and arrays every time the program is loaded.
Figure 4-XX - Example code of GEL language It is encouraged do the getting started tutorial included in the CCS help center.
87
Software
4.3. The load controller program This section describes the structure of the files that compound the project and the relevant source code.
4.3.1. File structure A project developed with CCS is form by many files all of them grouped in a CCS (Code Composer Studio) project. The project root folder contains six files and four folders, not all of them are relevant in the project, some of them are automatically generated in the compilation process and they are logging files or temporal files. his section only describes the relevant files used in the project. The project root folder contains the following files: alex_project.pjt This is the CCS main project file. It stores the configuration of the compiler and linker, as well as the list of the files that compound the project. This file is essential in order to work with the IDE. proyectoMain.c This file contains the most of the source code of the project. It is deeply described in following sections. The other files in the root folder are automatically generated by the IDE. One of them contents information about the CPU (alex_project.sbl). The IDE uses it to generate the specific assembler code for the CPU. Other file contains temporal information about the project loaded (alex_project.paf2), It allows the IDE not to phrase the project file every time a change is made, only when opening an saving the project. The other files with the extension “log” are loggings about the internal process of the compiler. The root folder contains the following three folders:
88
Software Debug This folder contains the output files of the compiler, binary code generated from the source code. The content of this file is the program that is loaded into the DSC memory. DSP281x_common and DSP281x_headers These folders contain the files provided by Texas instruments needed to develop a program for the DSC. They contain the CPU initialization subroutines and the registers structures. The register structures relates the register memory address with user-friendly names which allows the user not to use the memory address when writing of reading a register, only the mnemonic assigned names. alex_project.CS_ This folder is a temporal automatically generated folder. It is used by the IDE in internal process, for example, to store the CPU variables that are wanted to be showed in the screen.
4.3.2. Memory mapping The memory available is scarce. Although this DSC has the possibility of use flash memory it is only used the RAM memory. In order to use the flash memory a JTAG emulator is needed. However, in the developing of this prototype the RAM memory is enough.to archive the prototype purposes. There are two different types of RAM, the integrated in the DSC core and the external RAM included in the eZDSP board (developing board). The RAM has to be shared between data and code, due to the DSC Harvard structure. The memory mapping has to be made before the compilation and it is not possible to change it after compilation. Memory mapping is defined in the command file “F2812_EzDSP_RAM_Link.cmd”. It is stored inside the DSP281x_common folder, it has to be modified specifically for the project. The main difference between the internal and external RAM is the speed. The internal RAM is a single access RAM (SRAM), very fast, and the external RAM is slower, it uses several CPU cycles for a single access.
89
Software
The faster RAM is used to store time critical subroutines code, the other subroutines and in the initialization process uses the external RAM. The appendix Memory mapping shows the code to distribute the memory. The memory mapping file is divided into two sections, MEMORY and SECTIONS. The memory section defines the addresses and lengths of memories; it divides physical memory into zones. The memory is divided in two pages; each page has different functions. PAGE 0 is used to store the code of the program and de PAGE 1 to store the data. The DSC has Harvard architecture and it has CODE and DATA moved a part. The second part of the memory mapping file is the SECTIONS part. In this part the sections of the program are defined and it is assigned a memory zone to them. Any program has different memory sections, like the initialization code, the stack, the user code and so on. It is possible to assign several sections to a single memory zone; the memory is shared between all sections assigned to it. In the case that a memory zone has not enough memory for all sections designed the program cannot be compiled. By default the initialization sections are assigned to external ram, they are executed once and they are not time critical. The stack and other internal sections are allocated to internal RAM. These sections will access to the memory very frequently and the location of these sections affects too much to the global speed. The user code is assigned to external ram and user data to internal ram. There are some exceptions, user code has to be executed as fast as possible, in instance of frequency measurement subroutine, is stored into the internal SRAM. In order to select where data of functions code are stored there are two special compiler directives. Used to choose the memory section where a variable or a function is allocated. They are only used when it is needed to change the location of the function from the original section defined in the memory mapping file. The directive to allocate the date in a desired section is:.
90
Software #pragma DATA_SECTION(variable, ".section"); To allocate a specific function in a desired section it is used the directive: #pragma CODE_SECTION(function, ".section"); The data directive can only allocate data within the data page sections and code directive within code page sections. By default the user code is allocated in external ram which is slow. But there are two time critical subroutines in the project, the one that measures the frequency and the one that samples de input signals. These subroutines are periodically executed in a short period of time, they have to be executed as fast as possible in order to avoid that the next execution of that subroutine were before the previous execution has finished. They are allocated in the code section “ramfuncs” which is defined into the internal RAM memory zone.
91
Software
4.3.3. Program structure This section describes the structure of the program. The figure below shows the program flow.
Program start
MAIN FUNCTION
Capture1_isr Frequency measurement
Initialization process
Average calculation RMS calculation
Main loop
Create record If ZeroCrossi ng
Threshold checking
If ADC Ready
ADC_isr
idle
Get ADC results Store ADC results
. Figure 4-XXI - Simplified program flow The start point is the red circle on the top of the figure, the programs stars its execution at that point. The initialization of the load controller is done in the first lines of the main function. The initialization process consists on configuring the peripheries implied in the functions of the load controller. It is done the initialization of the internal peripheries such as the PLL, the watchdog and ADC. Then they are filled up the ADC and EV configuration registers. The initialization process ends by setting up the interruption vectors of the ADC and EV interruption subroutines and enabling all the interruptions.
92
Software Once the initialization is finished the main function enters into an infinite loop without any function. At this point the program waits for the interruption events and calls the interruption subroutines. The most of the work it is done in the interruptions. Other functions are used in the program but they are not relevant in the program flow.
4.3.3.1.
Headers
This section describes the header of the program. The headers define the library inclusion, the special directives to map functions or variables to a specific part of memory, the definition of the function prototypes and the definition of constants and variables.
4.3.3.1.1.
Inclusions
The load control program includes three files. #include "DSP281x_Device.h" #include "DSP281x_Examples.h" #include <math.h>
// DSP281x Headerfile Include File // DSP281x Examples Include File //mathematical functions and constants
The first and the second file included are the libraries provided with the microcontroller, these libraries contain the CPU initialization subroutines and the userfriendly names of registers as well as the interruption vector. The last file is the mathematics library that is provided by the compiler. This librarie contains the mathematical function such as sin, cos, sqrt or constants as π or e.
4.3.3.1.2.
Constant definition and Location of attributes
As explained in section 4.3.2 Memory mapping, it is possible to map any data structure or variable to the internal or external RAM depending on its time requirements. The mapping directives take place in this part of the header section. They are also defined in this part of the header the constant used in the rest of the program.
93
Software According to the memory mapping, explained by the section 4.3.2 Memory mapping, by default, the variables are allocated in the on-chip RAM. The directive #pragma DATA_SECTION(variable_name,“memory_zone”) allocates a variable or data structure to the specified zone of memory (in instance of memory zone “.extRAM” that is allocated in the external RAM memory). The constants are defined with the directive #define name value. In the compilation process the compiler replaces the constants names with its value. frequency measurements constants : number of frequency measurements used in average #define AVERAGE_FREQUENCY_VALUES 25
Recover frequency threshold in integer units (Running at 25MHz) define DEFAULT_HIGHER_LIMIT 62525
Frequency event threshold in integer units (Running at 25MHz) #define DEFAULT_LOWER_LIMIT 62626
length of averaged frequency measurements array #define AVERAGE_MEASURES 5
Frequency decay filter constants: Maximum allowable frequency change between frequency measurements.,150 integer units/cycle= 6Hz/s #define MAXIMUN_FREQUENCY_CHANGE 150
frequency decay measurement constants: //square of FCF, It is needed in frequency decay calculations; it is defined with one decimal value to specify that the number is a float number. #define FCF2 976562500000.0
94
Software frequency decay event threshold #define DECAY_THRESHOLD -0.16
number of decay measurements used in the average #define AVERAGE_MEASURES 5
RMS constants: Number of samples used in RMS calculations #define ADC_MEASURES_PER_CYCLE 40 Offset present in current and voltage channels. #define VOLTAGE_OFFSET 1904 #define CURRENT_OFFSET 1904
Realy constants:
The different load controller states are the following: 0=OPEN CIRCUIT 1=CLOSED CIRCUIT RECONNECTION
2=WAITING
FOR
Define the initial state #define DEFAULT_RELAY_STATUS 0x1 Define the states. #define OPEN 0x0 #define CLOSE 0x1 #define WAIT 0x2 Define the 1500*0.02s=30s
maximum
//relay status //relay status //relay status time
to
re-attachment
power
line
cycles.
#define MAXIMUN_WAIT_TO_ATTACHMENT 1500
Control logic type II Constants Kf value in Control logic type II equation #define KF 1
Nominal frequency in Nordic systme, integer units, 50Hz #define FO 62500
95
Software Data recording constants: This constant defines the number of records stored in external memory simulating the external memory #define STUB_EXT_MEMORY_LEGTH 50
The value of these constants cannot be changed once the program is compiled. The constants are named by collective agreement in uppercase letters, it makes easy to distinguish them from variables. Some of the values that could be changed in running time are stored in the initialization process in a variable, from that point it can be modified while the program is running. These constants define the initial configuration of the load controller. As it is explained in the chapter future work, it should be better to define these values in the external flash memory that it is not erased if the hardware is powered down, and if they are modified the modifications remains after a reboot.
4.3.3.1.3.
Functions definition
This part of the header defines the prototypes of the functions. It also allocates some functions to and specified part of memory. As default the functions are allocated in the external RAM memory, which is slower than the internal memory. Using the directive: #pragma CODE_SECTION(name_of_function, memory_zone); it is possible to allocate the time critical into the faster internal RAM, in instace of code section “ramfuncs” that is allocated in the internal ram. As in the data allocation directive, the memory zone must be defined in the memory mapping file and it must to be in the code page.
/*** frequency measurement functions ***/ //definition of the interruption subroutine of capture unit 1 //and allocation in internal RAM interrupt void Capture1_isr(void); #pragma CODE_SECTION(Capture1_isr, "ramfuncs"); /*** data acquisition functions ***/ //definition of the interruption subroutine of ADC //and allocation in internal RAM
96
Software interrupt void adc_isr(void); #pragma CODE_SECTION(adc_isr, "ramfuncs");
/*** relay control functions ***/ void close_relay(void); //opens the load relay void open_relay(void); //closes the load relay //sets the frequency recover threshold void set_higher_limit(unsigned int); //sets the frequency event threshold void set_lower_limit(unsigned int); /*** time stamp ***/ //synchronizes the real time counter void time_synchronization(long);
/*** decay functions ***/ //sets the frequency decay threshold void set_decay_limit(float);
Control logic type II function. This function changes the temperature set points. void set_points_update(unsigned int,unsigned int); /*** storing stub ***/ //Function that should store the recors in external memory void recod(long time,unsigned int period,unsigned int voltage,unsigned int current,unsigned long power, float decay,short relay);
The functions are defined in the same way as is done in ANSI C. First of all it is defined the type of value that the function returns (void means nothing), next it is defined the name of the function, and finally, in brackets, the parameters and its types. All the functions are explained in following sections.
97
Software
4.3.3.1.4.
Variables definition
This part of the header defines the variables of the project. All the variables and arrays in C have to be defined before being used. In this part of the header it is assigned the type (short, integer, unsigned integer…) to the attributes and if necessary, some of them are initialized to a specified value.
//generic variables short i; //generic index short j; //generic index /*** frequency measurement ***/ // Temp integers variables in capture unit subroutine unsigned int temp1_1; unsigned int temp1_2; […] //table of average frequency measurements unsigned int average_measures[AVERAGE_MEASURES]; //index used in frequency measurement array unsigned int measure_index=0;
unsigned int average_measure; //average period measure
/*** Frequency decay measurement ***/ //table of frequency decay measurements float decay[AVERAGE_MEASURES]; //last frequency decay measurement float average_decay=0; //frequency decay threshold definition and initialization float decay_limit= DECAY_THRESHOLD; /*** relay control ***/ //frequency recover threshold initialization unsigned int higher_limit=DEFAULT_HIGHER_LIMIT; […] /*** data adquisition ***/
98
Software
//ADC current measures without offset unsigned int adc_samples_current[ADC_MEASURES_PER_CYCLE]; […] /*** RMS calculations ***/ //square of ADC measures in RMS calculations unsigned long adc_square_current[ADC_MEASURES_PER_CYCLE]; unsigned long adc_square_voltage[ADC_MEASURES_PER_CYCLE]; […] unsigned int voltage_RMS; //voltage RMS calculation result […] //temporal variables in RMS calculations long temp_results; long temp2_1; […] /*** Time stamp ***/ //time variable definition and initialization long current_time=0;
/*** external memory stub ***/ unsigned int ext_index=0; […] short ext_relay[STUB_EXT_MEMORY_LEGTH]; //relay s.
Not all the variables and arrays used are showed in the piece of code above, the whole variable definition section can be found in the appendix A.1 Main program source code.
4.3.3.2.
Device initialization
The main function is the first function executed by the DSC. This function does the initialization task. Once time the control device is initialized the main function ends and all the actions are done by the interruption subroutines.
99
Software This section describes the functions and the registers that are involved in the initialization process. Some of the registers used in the initialization are ELLAOW protected registers. The directives ELLAOW and EDIS are used to write in protected registers. The EALLOW protected registers are protected against spurious writes after configuration. It can only be modified between these two directives. These directives are ignored in the description of the code in the following sections.
4.3.3.2.1.
System Control Initialization
InitSysCtrl(); // Init the system control
This function initializes the system control. This function is provided by Texas Instruments. This function initializes the internal peripheries, the PLL, peripheral clocks and so on. The PLL sets the speed of the CPU . It is initialized with the default speed of 150Mhz.
4.3.3.2.2.
Peripheral High speed clock configuration
This code configures the speed of the High Speed Clock (HSCLK). The peripheries are clocks are fed by this global clock, subsequently the peripheries uses a prescaler to adapt the speed to its requirements. //set HSPCLK to SYSCLKOUT / 6 (25Mhz assuming 150Mhz SYSCLKOUT) EALLOW; SysCtrlRegs.HISPCP.all = 0x3; // HSPCLK = SYSCLKOUT/6 EDIS;
The configuration of the HSCLK is done by setting up the prescaler applied to the main clock , SYSCKOUT, that runs at 150Hz. It is set to obtain the clock frequency of 25MHz that is the maximum that the peripheries support. The register of the High speed peripheries prescaler (HISPCP) is EALLOW protected.
100
Software
4.3.3.2.3.
GPIO configuration
The next code sets up the DSC pins that are used as General purpose Input/Output (GPIO). The GPIO multiplexer registers are EALLOW protected. EALLOW; // Enable write protected registers GpioMuxRegs.GPAMUX.bit.CAP1Q1_GPIOA8 = 1; // Sets pin as Caputure unit GpioMuxRegs.GPFMUX.bit.MCLKXA_GPIOF8=0x0; //sets as I/O GpioMuxRegs.GPFDIR.bit.GPIOF8 = 0x1; //sets as output EDIS; // Disables write protected registers modification
Most of the DSC GPIO pins are multiplexed with peripheries. By default the pins are used by the peripheries. The pins of a periphery which is not used can be used as GPIO. Each periphery has a register associated where is possible to set the pins are GPIOs or as periphery pin, it is also possible to chose the direction of the GPIO pins. The first line, ignoring the directives, sets the first pin of the capture unit to work as the capture input. It is configured in the capture unit GPIO register number 8. The second line sets the pin, which is going to be used to control the load relay, as a general purpose IO. It is set in the register of the McBSP, this periphery is not used in the project, that is because their pins are used as GPIO. The last line specifies the direction of the GPIO. It sets the pin as an output.
4.3.3.2.4.
PIE initialization
The next directive disables all interruptions. The initialization process is not desired to be interrupted, particularly in the initialization of the PIE vectors. // Disable CPU interrupts DINT;
This function starts the interruption system. // Initialize the PIE control registers to their default state.
101
Software // The default state is all PIE interrupts disabled and flags // are cleared. InitPieCtrl();
The next step is clear all the interruption flags, it sets all the interruptions as attended and cleared. They are also disabled all the interruptions individually. // Disable CPU interrupts and clear all CPU interrupt flags: IER = 0x0000; // Disables al interrputions IFR = 0x0000; // Clear al interruption flags
Then it is initialized the vector table. It uses the default interruption subroutines even if they are not used. // Initialize the PIE vector table with pointers to the shell Interrupt // Service Routines (ISR). // This will populate the entire table, even if the interrupt // is not used. InitPieVectTable(); Finally they are set up the interruption vectors used in the load controller program. The ADC vector is used for the sampling subroutine and the Capture unit vector is used for the frequency measurement subroutine. EALLOW; // This is needed to write to EALLOW protected register PieVectTable.CAPINT1 = &Capture1_isr;// sets capture 1 unit interruption routine as Capture1_isr PieVectTable.ADCINT = &adc_isr; //sets adc_isr as interruption routine of ADC EDIS; // This is needed to disable write to EALLOW protected registers
This piece of code stores the address of the interruption function in the interruption vector. Every time that the interruptions requirements are fulfill, it is called the function stored in its interruption vector.
4.3.3.2.5.
ADC initialization
The next initializes the ADC internals hardware, it is provided by Texas Instruments
102
Software
InitAdc(); //init ADC.
The ADC measures buffer index is initialized to 0. ConversionCount = 0;
The next lines of code configure the ADC sequencer to sample only the desired channels. Setup 3 conversions on SEQ1 AdcRegs.ADCMAXCONV.all = 0x0002;
Setup ADCINA0 & ADCINB0 as 1st SEQ1 conv. Current and voltage channels AdcRegs.ADCCHSELSEQ1.bit.CONV00 = 0x0;
Setup ADCINA1 & ADCINB1 as 2n SEQ1 conv. Current and voltage, second sample AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x1;
Setup ADCINA2 & ADCINB2 as 2n SEQ1 conv. Read User normal high and low.temperature AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x2;
The next lines of code enables the event manager to start the ADC’s sequencer one it is also enabled the interruption of the sequencer one. The end of sequence interruption calls the ADC interruption subroutine, the sampling subroutine, once al the conversions configured in the sequencer are done. AdcRegs.ADCTRL2.bit.EVA_SOC_SEQ1 = 1; AdcRegs.ADCTRL2.bit.INT_ENA_SEQ1 = 1;
// Enable EVASOC to start SEQ1 // Enable SEQ1 interrupt (every EOS)
The next line activates the simultaneous mode. Therefore the sample and hold circuits are activated and the pair of channels are sampled at the same time and held until both of the conversions are done. AdcRegs.ADCTRL3.bit.SMODE_SEL = 1; //active the simultaneous mode in order to measure current and voltage at the same time
103
Software
4.3.3.2.6.
EV configuration, timers and capture unit
First of all the general purpose timers configuration registers are cleared. /* EVA GP timers configuration */ EvaRegs.GPTCONA.all = 0; // Clears configuration
The next lines configure the timer 1 that controls the ADC sample rate and real time stamp. /* Configures GP timer 1 with ADC sample rate */ EvaRegs.T1PR = 0x61A8;
// Setup period register
The value of this register is 25000 in decimal base, the prescaler is disabled, so that, the timer runs at the High Speed periphery clock, at 25Mhz. The timer required to reach the count of 25000 takes 1 millisecond, which is the desired sample rate. Enable timer to start ADC EvaRegs.GPTCONA.bit.T1TOADC = 0x2;
// Enable EVASOC in EVA
Finally the timer is configured in up-count mode and in compare mode. The compare mode allows the timer to generate an interruption when the count goes up to one millisecond. It is also disabled the prescaler. EvaRegs.T1CON.all = 0x1040; // Enable timer 1 compare (upcount mode)
The next lines configure the timer 2. This timer is used as time base by the capture unit, this counter counts the time between zero-crossings. The frequency of the timer is set in order to not to be overflowed within two zero-crossing, as explained in previous sections. Configures GP timer 2 with capture unit EvaRegs.T2PR = 0xffff; // Sets timer periode EvaRegs.T2CNT = 0x0000; // Conpare value,not needed The next line sets the prescaler to 1/8, reducing the High speed periphery clock to 3.125Mhz. EvaRegs.T2CON.all = 0x1340; // Configures timer 2
104
Software All the timers belonging to an EV module have some common configuration registers. The Global configuration of timers 1 and 2 is show in the next lines. EvaRegs.CAPCONA.all = 0x0000; // Clears configuration EvaRegs.CAPCONA.bit.CAP12EN = 1; // Enables capture 1 and 2 interruption EvaRegs.CAPCONA.bit.CAP12TSEL = 0; // Chooses timer 2 as timer for capture unit 1 EvaRegs.CAPCONA.bit.CAP1EDGE = 1; // Configures capture unit to detect rising edges EvaRegs.CAPFIFOA.all = 0x0000; // clears FIFO A EvaRegs.CAP1FBOT = 0x0000; // clears capture 1 FIFO top EvaRegs.CAP1FIFO = 0x0000; // clears capture 1 FIFO bottom
Configures interrptions of EVA EvaRegs.EVAIFRC.all = 0xFFFF; EvaRegs.EVAIMRC.bit.CAP1INT = 1;
4.3.3.2.7.
// Sets capture unit 1 flags // Enables captured unit 1 interruption
Relay initialization
This piece of code sets the relay to the default status defined in the constant section of source code. //initialize Relay if(relay_status==CLOSE){ close_relay(); }else{ open_relay(); }
4.3.3.2.8.
Enable CPU interruptions
The PIE block can support up to 96 peripheral interrupts. The 96 interrupts are grouped into blocks of 8 and each group is fed into 1 of 12 CPU interrupt lines (INT1 to INT12). Therefore it is needed to enable the whole interruption line where the single interruption belongs.
105
Software
// Enable CAP1 in PIE PieCtrlRegs.PIEIER3.bit.INTx5 = 1; IER |= M_INT3; // Enable CPU Interrupt 3
The code above enables the individual interruption of the capture unit in the interruption group number 3 and activates the whole group in the CPU interruption register (IER). // Enable ADCINT in PIE PieCtrlRegs.PIEIER1.bit.INTx6 = 1; IER |= M_INT1; // Enable CPU Interrupt 1
The code above enables the individual interruption of the ADC in the group number 1 and activates the whole group in the CPU interruption register (IER).
EINT; ERTM;
// Enable Global interrupt INTM // Enable Global realtime interrupt DBGM
The groups where an interruption belongs cannot be changed. By default the lower number of group the higher priority has the interruption. Therefore, the ADC has more priority than the other subroutine as desired.
4.3.3.2.9.
Average frequency buffer initialization
This code initializes the frequency buffers to the initial values. for(i=0;i
106
Software
4.3.4. Capture unit subroutine: Capture1_srt() This function is the interruption subroutine configured in the interruption vector of the capture unit, therefore, it is going to be called when the capture unit requires it, every time it detects a rising edge. This function does more task than frequency measurements, such as threshold checking and RMS calculation. Most of the code is already explained in previous sections. The next lines of code reads the values stored in the capture unit FIFO, These reads returns the time of the last and the previous zero-crossings. temp1_1 = EvaRegs.CAP1FIFO; // Reads value in fifo stack temp1_2 = EvaRegs.CAP1FIFO; // Reads value in fifo stack
Once the values of last frequency measured are retrieved the frequency averaging is started. Resets the measurement accumulator used in the average. measures_addition=0;
Measurements are shifted to leave space for the new measurements, the and average calculation, the for loop is also used to add the previous measurements, the new one is added later. for(i=AVERAGE_FREQUENCY_VALUES-1;i>0;i--){ measures[i]=measures[i-1]; //adds last instant measures except the last one. measures_addition=measures_addition+measures[i]; }
Calculate the new measure checking if the capture unit had been overflowed. //gets the last measure if(temp1_2>temp1_1){ //No overflow measures[0] = temp1_2 - temp1_1; }else{ //overflow measures[0] = temp1_2 +(0xffff- temp1_1); }
107
Software The next lines describe the frequency soft filter. It is described in section 2.1.3 Soft frequency filter. if(measures[0]>average_measures[0] + MAXIMUN_FREQUENCY_CHANGE || measures[0]
The averaging calculation is continued after the frequency decay soft filter. The adding of instant frequency measurements started some lines before in order to improve the efficiency of the code, avoiding two loops that use the same variables. //adds the last measure measures_addition=measures_addition+measures[0]; //calculates the average average_measure= measures_addition/AVERAGE_FREQUENCY_VALUES;
Finally the table of average measures is shifted and the new value is inserted. //Shift the average values array and insert the new one. for(i=AVERAGE_MEASURES-1;i rel="nofollow">0;i--){ average_measures[i]=average_measures[i-1]; } average_measures[0]=average_measure;
The next lines implements the decay measurement algorithm described in section 2.2 Frequency decay measurement.
The process of averaging is the same as the used in frequency measurement. First of all it is clear the average decay accumulator. Then the previous decays are shifted, the loop is also used for adding the decay measurements. average_decay=0; //clear average decay addition
for(i=AVERAGE_MEASURES-1;i>0;i--){ //freency decay shift average_decay=average_decay+decay[i-1]; decay[i]= decay[i-1]; }
108
Software It is calculated the frequency decay using the described equation in the frequency decay section. The only different is that the next line uses the square of FCF divided into one million and next multiplied by the same amount; it is done in order to avoid the integer rounding. //frequency decay calculation decay[0]= (FCF2/(float)((long)average_measures[0]*(long)average_measures[0])) *(((float)average_measures[1]/(float)average_measures[0])-1);
Finally it is calculated the decay average adding the new decay measurement and the decay accumulator.
//average_decay Calculation average_decay=(average_decay+decay[0])/AVERAGE_DECAY_MEASURES; //
The next lines calculates the RMS values, it is showed only the algorithm used to calculate the voltage RMS, the code is exactly the same for current and power in exception of the input data tables used. The RMS algorithm uses the square measures calculated in the sampling subroutine. It is explained in the next section. //RMS power voltage and current //voltage
The accumulator is cleared temp_results=0;
Main loop, it adds all the square measures. for(i=0;i
The total adding is divided into the number of measurements used in RMS calculation. temp_results=temp_results/ADC_MEASURES_PER_CYCLE;
Finally it is stored in the respective result variable, this variable is passed to the record function, that simulates the external memory data recording.. voltage_RMS=sqrt( temp_results);
The result of this calculations gives RMS values in integer units used internally by the microcontroller, in section 2.5.3 RMS Current, voltage and power calculation it is described the process to convert this results into the international system of units (Volts,
109
Software Amps or Watts). The other RMS calculations are done in a similar way in exception that is used the respective squared samples array. The Control logic Type II Source code is very simple. Measures[0] contains the last frequency measurement and T_high and T_low should contain the Normal temperature set points measurements. It is used the algorithm explained in section 2.4 Control logic Type II: Temperature Set points change.
T_high_out=T_high*KF*(measures[0] - FO); T_low_out=T_low*KF*(measures[0] - FO);
Finally this interruption subroutine checks the frequency thresholds that implements the tripping criteria of control logic Type I.. If frequency is lower than the frequency threshold or the frequency decay is lower than the frequency decay threshold is considered that a frequency event has happened. It has to be considered that the frequency checking are made comparing the period of the signal, therefore the frequency event happens when the measure is higher than the lower frequency limit. if(average_measure rel="nofollow"> lower_limit || average_decay < decay_limit){ //there is a frequency event.
The relay is opened open_relay(); }
If the frequency and the decay is recover else if(average_measure < higher_limit && average_decay > decay_limit){ //frequency event is recovered
It is set an initial random waiting time and the state is changed to waiting for attachment. In this state the relay remains opened, it is used to distinguish the open state when the frequency event is recovered or not. if(relay_status == OPEN){ //only The first time
It is set the new state and it is get the random waiting time measured in power grid cycles. relay_status=WAIT; waiting_cycles= temp1_2 % MAXIMUN_WAIT_TO_ATTACHMENT; }else{
110
Software If the load controller is in the waiting state it is subtracted one unit to the waiting counter. //from the second time waiting_cycles--; }
If the waiting time ends it is closed the relay. if(waiting_cycles <= 0){ //the waiting time is up. close_relay(); } } else if(relay_status == WAIT){
If the frequency goes below the recover threshold and it is in the waiting state it is get a new waiting time. // if frequency and frequency decay do not fulfill the event thresholds but they don’t fulfils the recover criteria thresholds. waiting_cycles= temp1_2 % MAXIMUN_WAIT_TO_ATTACHMENT; }
It is called the stub function that should store the record into the external memory for subsequent sending. recod(current_time,average_measure, decay[0],relay_status);
voltage_RMS,current_RMS,
power_RMS,
The next line calls the function that should change the temperature set points, it is passed the calculated high and low set points. set_points_update(T_high_out,T_low_out); Finally it is cleared the interruption in the capture unit and in the respective acknowledge group. This prepares the interruption for the next executions. EvaRegs.EVAIFRC.bit.CAP1INT = 1; // Clears capture unit 1 interruption flag PieCtrlRegs.PIEACK.all = PIEACK_GROUP3; // PIE 3 ack
111
Software
4.3.5. ADC interruption subroutine: ADC_srt() This function gets the current and voltage measurements and calculates its squares that are used in the RMS calculations. It also increases the real time stamp counter.
First of all the function increments the real time counter current_time++;
The next action retrieves the samples gotten by the ADC. If this function is called, the ADC guaranties that the samples are ready in the results register. The results are allocated in the set of registers AdcRegs.ADCRESULTn where n is the number of sample. The results are allocated according to the acquisition order. In this prototype are used the dual simultaneous sampling, in this case it is firstly stored the channel A which corresponds to the Current channel, and next the channel B which corresponds to the voltage. Once the signal channels are sampled they are sampled the offset channels, which are allocated in the next two result registers in the same way as the signal channels. In the following lines it is used ConversionCount as index of arrays, this index in an integer variable that is incremented every execution of this function. //Current measurements adc_samples_current[ConversionCount] = >>4)+(AdcRegs.ADCRESULT2 >>4))/2-CURRENT_OFFSET;
((AdcRegs.ADCRESULT0
////Voltage measurements adc_samples_voltage[ConversionCount] = >>4)+(AdcRegs.ADCRESULT3 >>4))/2-VOLTAGE_OFFSET;
((AdcRegs.ADCRESULT1
The pair of samples of each channel are added and divided into two, finally it is subtracted the offset. The result of these subtractions is stored in the samples array, type integer, it is not used an unsigned integer because the values varies from negative to positive values. The normal temperature set points are read from the ADC inputs, The load controller assumes that the normal temperature set points are introduced thought the ADC but it could be done though any other way as the GP inputs. T_high = AdcRegs.ADCRESULT4 >>4; T_low = AdcRegs.ADCRESULT5 >>4;
112
Software
Now it is calculated the square values for the RMS calculations, it use long type variables because the square of the maximum sample is higher than the maximum number represented by the integer type. It uses the measurements made in the previous lines of code. adc_square_current[ConversionCount]=(long)adc_samples_current[ConversionCount ]*(long)adc_samples_current[ConversionCount]; adc_square_voltage[ConversionCount]=(long)adc_samples_voltage[ConversionCount ]*(long)adc_samples_voltage[ConversionCount]; adc_square_power[ConversionCount]=adc_square_current[ConversionCount]*adc_squ are_voltage[ConversionCount];
If ADC_MEASURES_PER_CYCLE conversions have been logged, start over, where ADC_MEASURES_PER_CYCLE is the selected amount of measures used for the RMS calculations in the constants definition. if(ConversionCount == ADC_MEASURES_PER_CYCLE-1) { ConversionCount = 0; } else ConversionCount++;
Reinitialize interruption for next ADC sequence Reset SEQ1, so the ADC sequence interruption is marked as attended. AdcRegs.ADCTRL2.bit.RST_SEQ1 = 1;
Clear INT SEQ1 bit, so the interruption is activated to be called the next time. AdcRegs.ADCST.bit.INT_SEQ1_CLR = 1;
Acknowledge interrupt to PIE. So the whole interruption vector is marked as attended, it is possible to acknowledge the whole group because this interruption is the only one of this group which can be called in this project.
PieCtrlRegs.PIEACK.all = PIEACK_GROUP1;
113
Software
4.3.6. Other subroutines This sections describe the others function present in the load controller source code which do not fit in the previous categories.
This function sets the time stamp to a desired value. It should be called by an external source of synchronization, for example the time given by the GPRS network. void time_synchronization(long time){ current_time = time; }
The next functions controls the relay. They change the state flag and open or close the relay according to the function. In this prototype the relay is simulated by a LED diode which is turned on when the load should be connected and is turned off in the opposite case. The way of activating the LED is the same to activate an external relay. Inside this fuctions can be added any other instructions required in the final design in order to do any other task like changing the thermostat themperature of a heat pump or a fridge. Sets the state of load controller to close and activates the LED. void close_relay(){ relay_status=CLOSE; GpioDataRegs.GPFSET.bit.GPIOF8 =0x1; }
Sets the state of load controller to open an disconnects the LED. void open_relay(){ relay_status=OPEN; GpioDataRegs.GPFCLEAR.bit.GPIOF8
=0x1;
//IT SHOULD BE STARTED HERE THE DATA SUBMISION }
This function changes the frequency and decay thresholds. They should be used in order to change them once the program is compiled and running by future extensions. For example for change them remotely though the GPRS network.
114
Software
void set_higher_limit(unsigned int new_limit){ higher_limit=new_limit; }
void set_lower_limit(unsigned int new_limit){ lower_limit=new_limit; } void set_decay_threshold(float new_limit){ decay_limit=new_limit; //set new frequency decay threshold }
Finally, the last functions are stubs, which are not implemented function. The first stub corresponds to the frequency that should change the temperature set points; it should be modified depending on the selected interface. void set_points_update(unsigned int T_high,unsigned int T_low){ //this function should change the temperature set points if applicable }
The second stub should be implemented to store the data in the external memory. In the case of using a serial flash memory, as recommended, is should store all the acquired data in the last free memory address. Another function would be needed to read the acquired data when it is needed to delivery it. In order to simulate the data storage it is used the external RAM, it obviously cannot fulfill the memory capacity required. //this stub should store last measurements in external flash memory //store measurements in internal RAM ext_time[ext_index]=time; ext_period[ext_index]=period; ext_voltage[ext_index]=voltage; ext_current[ext_index]=current; ext_power[ext_index]=power; ext_decay[ext_index]=decay; ext_relay[ext_index]=relay;
115
Software
//increment the index and reset to zero if out of bounds ext_index++; ext_index=ext_index%STUB_EXT_MEMORY_LEGTH; }
116
5
FUNCTIONALITIES AND PERFORMANCE
The load controller is able to carry out the two control logics, Type I and Type II. It can reconnect the load within a random selectable amount of time, 30 seconds by default. The load controller is prepared to be extended in the future. It supports the time synchronization and changing the configuration values in running time, in the case of adding an external synchronization source and a communication module. It is also prepared to store the acquired data in a external memory in the case of adding an external data storage module. Accuracy Frequency
±7,5mHz. @50Hz
Voltage
±0,3V
Current
±0,015A
Power
±5W
Maximum measurement delay: 0,5s (Average of 25 values) Maximum power measurement delay: 1ms Note that this result corresponds to a load of nominal 600W. The current accuracy could be improved by expanding the current signal in the ADC conversion range; it can be done by changing the number of loops in the current transformer. In consequence the power accuracy could be improved to.
117
6
FUTURE WORK
This section describes the next logic steps in following stages of this project, if it is continued in the future.
6.1. Flash memory This program is programmed using only RAM memory. Using this memory entails three problems. The first problem is the Memory limitation, the RAM memory is not enough to expand too much the program, in order to follow the development of this project is recommendable to use the extra memory space given by Flash Memory. The second problem is about the persistence of the program. The RAM memory does not allow maintaining the stored data when it is powered off, therefore, in a power cut, the internal data could be erased. The last problem is the remote software update, it is not possible to update the program stored in RAM memory from a external source in exception of the PC, the onchip flash memory is possible to be updated from other sources, like a GPRS module that sends the new program by the internet.
6.2. Communication part In order to make the load controller fully useful, is needed the communication with remote locations. It has no sense to store the measurements if it is not possible to retrieve it in order to be studied. A solution can be to use a GPRS module. This module through the RS-232 port can get the acquired data and send it to a desired location. This module can be the GM862 QUAD manufactured by the corporation Round Solutions. This module
119
Future Work supports the TCP/IP stack as well as for example the FTP protocol. TCP/IP allows sending though the GPRS network the data directly to a computer that could record the data from every load controller.
This module also has the possibility of being programmed, with the high level language Python and it also has two RS-232 ports, one can be used for receiving the measurements and the other to update the internal software. The capability of being programmed is essential because during the update the DSC cannot execute any code, therefore the GSPS module can take the control.
6.3. Persistent storage The load controller measurement has a pretty high throughput of data. It has to store some measures in short periods of time. It has to store and send the measures ten minutes before a frequency event. The measures have to be stored somewhere, this place can be and external persistent storage based on Serial Flash memory. This serial flash has enough capacity to store the measurements, their capacities goes from 1Mb to 128Mb. They are connected with the SPI interface of the DSC. The serial nature of the measurements made by the load controller is suitable for this kind of memory. This memory can be used also to preserve configuration changes made remotely.
120
7
CONCLUSIONS
In this section I going to explain all the objectives achieved in the developing of this project, as well as all the problems solved. The development of a real project evolves lots of mishaps and problems which before the begging of this project I was not able to imaging. I thought that these problems are common and the experience acquired in this project allows me to surpass them easier than before. The first point solved is to search the hardware needed to achieve the specifications. The market is full of manufacturers which has lots of processors. The research process is a critical point, choosing a wrong processor could result in a complete fail. Changing the processor, which is the main part of these kind of project, in an advance stage could imply in starting the development again. Now I know how to carry out these researches and how to make them. The next point is the design of the additional hardware that completes the capabilities of the microcontroller. The microcontroller is the core of the hardware but it cannot fulfill all the hardware requirements, it is needed other essential hardware, such as the power supplies, or other needed hardware like the filters or the signal conditioning hardware, whose functions cannot be made by any microcontroller. I designed a simple board with making use of my knowledge about electronics but it could not completely fulfill all the requirements, but fortunately my co-tutor Tonny improved the board until the point to be suitable for my purposes. Once the hardware part is completed the next logical step is the software development which is the “mind” of the any electronically device based on a microcontroller. The software is also a key part of any electrical development. I have learned to use the Code Composer IDE, which is one of the best IDEs in the market. I think that know I am able to develop any other microcontroller software based on any other platform and also using any other advanced IDE.
121
Conclusions
Referring to the product obtained. I have developed a prototype of a load controller based on the specifications given by my Tutor Zhao Xu. The prototype developed is a frequency meter. The load controller is also able to measure voltage, current and power. It is also capable of react to frequency events, frequency drops and excess of frequency decay. The project has being developed with the expandability in mind, because this prototype has to be continued in order to be fully working product. In resume, I have developed a prototype able to help the power grid integrity against changes of power consumption which can involve in the imbalance of the consumption-production power grid equilibrium. The continuity of this project is leaved open to be followed by any other student in order to complete a fully working load controller as described in the future work section. With these words I consider this Master thesis ended.
Thank you for reading, Alejandro Nieto.
122
8
BIBLIOGRAPHY
1. Grastrup, Martin Skov Johansen & Lars Henrik. Consumer Based Freqeuncy Controlled Reserve in the Danish Power Grid. Lyngby : s.n., 2006. 2. Agilent Technologies. Understanding Frequency. agilent.com. [Online] http://cp.literature.agilent.com/litweb/pdf/5965-7664E.pdf. 3. Texas Instruments. TMS320x281x Event Manager (EV) Reference Guide (Rev. E). Texas instruments official web site. [Online] June 27, 2007. [Cited: September 1, 2007.] http://www.ti.com/litv/pdf/spru065e. 4. —. TMS320x281x Analog-to-Digital Converter (ADC) Reference Guide (Rev. D). [Online] July 2005, 14. [Cited: October 2007, 2.] http://focus.ti.com/lit/ug/spru060d/spru060d.pdf. 5. —. TMS320F2810, TMS320F2811, TMS320F2812, TMS320C2810, TMS320C2811,TMS320C2812 DSPs (Rev. O). Texas Instruments official web site. [Online] July 17, 2007. [Cited: September 1, 2007.] http://www.ti.com/lit/gpn/tms320f2812. 6. wikipedia. Root mean square. Wikipedia. [Online] wikipedia. http://en.wikipedia.org/wiki/Root_mean_square. 7. Spectrum digital. eZdspTM F2812 Technical reference. 2003. 8. California Energy Comission. Smart Load Control and Grid-Friendly Appliances. October 2003. P-500-03-096-A10. 9. Z. Xu, Member, IEEE, J. Østergaard, Member IEEE, M. Togeby, C. Marcus-Møller. Design and Modelling of Thermostatically Controlled Loads as Frequency Controlled Reserve. 10. Ning Lu, Member, IEEE and Donald J. Hammerstrom, Member, IEEE. Design Considerations for Frequency Responsive Grid Friendly Appliances. 11. Rasmussen, Tonny W. Additional hardware design and layout. Lyngby : oersted DTU, 2007.
123
A
SOURCE CODE
A.1
Main program source code
/***************************************************************** * * File inclusion * *****************************************************************/ #include "DSP281x_Device.h" #include "DSP281x_Examples.h"
// DSP281x Headerfile Include File // DSP281x Examples Include File
/***************************************************************** * * constants definition * *****************************************************************/
/*** frequency measurements constants ***/ //number of frequency measurements used in average #define AVERAGE_FREQUENCY_VALUES 25 //Recover frequency threshold in integer units (Running at 25MHz) #define DEFAULT_HIGHER_LIMIT 62525 //Frequency event threshold in integer units (Running at 25MHz) #define DEFAULT_LOWER_LIMIT 62626
/*** frequency decay filter constants ***/ //maximun allowable frequency change between frequencies //measurements. 150 integer units/cycle= 6Hz/s #define MAXIMUN_FREQUENCY_CHANGE 150
/*** frequency decay measurement constants ***/ //square of FCF, It is nedeed in frequency decay calculations, //see thesis for more details. #define FCF2 976562500000.0 //frequency decay event threshold #define DECAY_THRESHOLD -0.16 //length of averaged frequency measurements, used in frequency
125
Board Schemes //decay calculations #define AVERAGE_MEASURES 5 //Measures used in RMS calculations #define ADC_MEASURES_PER_CYCLE 40 /*** relay constants ***/ //0=OPEN CIRCUIT 1=CLOSED CIRCUIT 2=WAITING FOR RECONNECTION #define DEFAULT_RELAY_STATUS 0x1 #define OPEN 0x0 #define CLOSE 0x1 #define WAIT 0x2
//relay status //relay status //relay status
//in number of powerline cycles. 1500*0.02s=30s #define MAXIMUN_WAIT_TO_ATTACHMENT 1500
/*** control logic type II ***/ #define KF 1 //KF value in Control logic type II #define FO 62500 //Nominal frequency in Nordic systme, integer units.
/*** data recording constants ***/ //This constan defines the number of records stored in //external memory simulating the external memory #define STUB_EXT_MEMORY_LEGTH 5
/***************************************************************** * * Special memory allocations * *****************************************************************/
#pragma DATA_SECTION(measures, ".extRAM"); //stores this array in external RAM #pragma DATA_SECTION(average_measures, ".extRAM"); //stores this array in external RAM
/***************************************************************** * * Function prototype definition * *****************************************************************/ /*** frequency measurement functions ***/ //definition of the interruption subroutine of capture unit 1 //and allocation in internal RAM interrupt void Capture1_isr(void); #pragma CODE_SECTION(Capture1_isr, "ramfuncs");
126
Board Schemes /*** data adquisition functions ***/ //definition of the interruption subroutine of ADC //and allocation in internal RAM interrupt void adc_isr(void); #pragma CODE_SECTION(adc_isr, "ramfuncs");
/*** relay control functions ***/ void close_relay(void); void open_relay(void);
//opens the load relay //closes the load relay
//sets the frequency recover threshold void set_higher_limit(unsigned int); //sets the frequency event threshold void set_lower_limit(unsigned int); /*** time stamp ***/ //synchronices the real time counter void time_synchronization(long);
/*** decay functions ***/ //sets the frequency decay threshold void set_decay_limit(float);
/*** Control logic type II ***/ void set_points_update(unsigned int,unsigned int);
/*** storing stub ***/ //Function that should store the recors in external memory void recod(long time,unsigned int period,unsigned int voltage,unsigned int current,unsigned long power, float decay,short relay);
/***************************************************************** * * variables and array definition and initialization * *****************************************************************/
//generic variables short i; //generic index short j; //generic index /*** frequency measurement ***/ // Temp integers variables in capture unit subroutine unsigned int temp1_1; unsigned int temp1_2;
127
Board Schemes //addition of al the last AVERAGE_FREQUENCY_VALUES periods. unsigned long measures_addition; //table of instant frequency measurements unsigned int measures[AVERAGE_FREQUENCY_VALUES]; //table of average frequency measurements unsigned int average_measures[AVERAGE_MEASURES]; //index used in frequency measurement array unsigned int measure_index=0;
unsigned int average_measure; //average period measure
/*** Frequency decay measurement ***/ //table of frequency decay measurements float decay[AVERAGE_MEASURES]; //last frequency decay measurement float average_decay=0; //frequency decay threshold definition and initialization float decay_limit= DECAY_THRESHOLD; /*** relay control ***/ //frequency recover threshold initialization unsigned int higher_limit=DEFAULT_HIGHER_LIMIT; //frequency event threshold initialization unsigned int lower_limit=DEFAULT_LOWER_LIMIT; //defalut relay state initialization short relay_status=DEFAULT_RELAY_STATUS; //count of waited cycles on reconection int waiting_cycles;
/*** data adquisition ***/ //ADC current measures without offset unsigned int adc_samples_current[ADC_MEASURES_PER_CYCLE]; //ADC voltage measures without offset unsigned int adc_samples_voltage[ADC_MEASURES_PER_CYCLE]; //Index of ADC samples table Uint16 ConversionCount=0; /*** Control logic II ***/ //variables to store the user high and low temperature. unsigned int T_high=0; undigned int T_low=0; //output control logic II temperature unsigned int T_high_out=0; unsigned int T_low_out=0;
128
Board Schemes
/*** RMS calculations ***/ //square unsigned unsigned unsigned
of ADC measures in RMS calculations long adc_square_current[ADC_MEASURES_PER_CYCLE]; long adc_square_voltage[ADC_MEASURES_PER_CYCLE]; long adc_square_power[ADC_MEASURES_PER_CYCLE];
unsigned int voltage_RMS; //voltage RMS calculation result unsigned int current_RMS; //current RMS calculation result unsigned int power_RMS; //power RMS calculation result //temporal variables in RMS calculations long temp_results; long temp2_1; long temp2_2; long temp2_3;
/*** Time stamp ***/ //time variable definition and initialization long current_time=0;
/*** external memory stub ***/ unsigned int ext_index=0; long ext_time[STUB_EXT_MEMORY_LEGTH]; unsigned int ext_period[STUB_EXT_MEMORY_LEGTH]; unsigned int ext_voltage[STUB_EXT_MEMORY_LEGTH]; unsigned int ext_current[STUB_EXT_MEMORY_LEGTH]; unsigned long ext_power[STUB_EXT_MEMORY_LEGTH]; float ext_decay[STUB_EXT_MEMORY_LEGTH]; short ext_relay[STUB_EXT_MEMORY_LEGTH];
//index //time //period //voltage //current //power //f. decay //relay s.
/***************************************************************** * * Main function, hardware initialization * *****************************************************************/
main() { // // Step 1. Initialize System Control: // PLL, WatchDog, enable Peripheral Clocks // InitSysCtrl(); // Init the system control
//set HSPCLK to SYSCLKOUT / 6
129
Board Schemes //(25Mhz assuming 150Mhz SYSCLKOUT) EALLOW; SysCtrlRegs.HISPCP.all = 0x3; // HSPCLK = SYSCLKOUT/6 EDIS;
// Step 2. Initialize GPIO: EALLOW; // Enable writing on write protected registers // Sets pin as Caputure unit GpioMuxRegs.GPAMUX.bit.CAP1Q1_GPIOA8 = 1; //Sets relay pin as GPIO GpioMuxRegs.GPFMUX.bit.MCLKXA_GPIOF8=0x0; //sets as I/O GpioMuxRegs.GPFDIR.bit.GPIOF8 = 0x1; //sets as output
EDIS; // Disables writing on write protected registers
// // Step 3. Clear all interrupts and initialize PIE vector // table: // // Disable CPU interrupts DINT;
// Initialize the PIE control registers to their default state. // The default state is all PIE interrupts disabled and flags // are cleared. InitPieCtrl(); // Disable CPU interrupts and clear all CPU interrupt flags: IER = 0x0000; // Disables al interrputions IFR = 0x0000; // Clear al interruption flags // Initialize the PIE vector table with pointers to the shell // Interrupt Service Routines (ISR). // This will populate the entire table, even if the interrupt // is not used. InitPieVectTable(); // Interrupts that are used in this example are re-mapped to // ISR functions found within this file. EALLOW; // enable write in EALLOW protected registerd // sets capture 1 unit interruption routine as Capture1_isr PieVectTable.CAPINT1 = &Capture1_isr; //sets adc_isr as interruption routine of ADC PieVectTable.ADCINT = &adc_isr; EDIS; // disable write in EALLOW protected registers
// // // // Step 4. Initialize all the Device Peripherals:
130
Board Schemes // // InitAdc(); //init ADC.
// Step 5. User specific code, enable interrupts: // Configure ADC AdcRegs.ADCTRL1.bit.RESET = 1; //resets ADC asm(" RPT #10 || NOP"); // Setup 2 conv's on SEQ1 AdcRegs.ADCMAXCONV.all = 0x0002; // Setup ADCINA0 & ADCINB0 as 1st SEQ1 conv. // Current and voltage channels AdcRegs.ADCCHSELSEQ1.bit.CONV00 = 0x0; // Setup ADCINA1 & ADCINB1 as 2n SEQ1 conv. //Current and voltage, second sample AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x1;
// Setup ADCINA2 & ADCINB2 as 2n SEQ1 conv. //Read User normal temperature high and low. AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x2;
// Enable EVASOC to start SEQ1 AdcRegs.ADCTRL2.bit.EVA_SOC_SEQ1 = 1; // Enable SEQ1 interrupt (every EOS) AdcRegs.ADCTRL2.bit.INT_ENA_SEQ1 = 1; //set simultaneous mode to measure current and voltage //at the same time AdcRegs.ADCTRL3.bit.SMODE_SEL = 1; AdcRegs.ADCTRL3.bit.ADCEXTREF=0x0;
//use internal reference
/*** EVA GP timers configuration ***/ EvaRegs.GPTCONA.all = 0; // Clears configuration /*** Configures GP timer 1 with ADC sample rate ***/
EvaRegs.T1PR = 0x61A8; EvaRegs.GPTCONA.bit.T1TOADC = 0x2; EvaRegs.T1CON.all = 0x1040;
// Setup period register // Enable EVASOC in EVA // Enable timer (upcount mode)
/* Configures GP timer 2 with capture unit */ EvaRegs.T2PR = 0xffff; // Sets timer periode EvaRegs.T2CNT = 0x0000; // Initial counter value EvaRegs.T2CON.all = 0x1340; // Configures timer 2 /* Global configuration of timers 1 and 2 */ EvaRegs.CAPCONA.all = 0x0000; // Clears configuration
131
Board Schemes // Enables capture 1 and 2 interruption EvaRegs.CAPCONA.bit.CAP12EN = 1; // Chooses timer 2 as timer for capture unit 1 EvaRegs.CAPCONA.bit.CAP12TSEL = 0; // Configures caputure unit to ditect rising edges EvaRegs.CAPCONA.bit.CAP1EDGE = 1; EvaRegs.CAPFIFOA.all = 0x0000; EvaRegs.CAP1FBOT = 0x0000; EvaRegs.CAP1FIFO = 0x0000;
// clears FIFO A // clears capture 1 FIFO top // clears capture 1 FIFO bottom
/*** Configures interrptions of EVA ***/ EvaRegs.EVAIFRC.all = 0xFFFF;
// Sets caputre unit 1 flags
// Enables caputure unit 1 interruption EvaRegs.EVAIMRC.bit.CAP1INT = 1;
//initialize Relay if(relay_status==1){ close_relay(); }else{ open_relay(); }
//initialization of instant frequency table for(i=0;i
//enable CPU interruptions // Enable CAP1 in PIE, frequency measurement subroutine PieCtrlRegs.PIEIER3.bit.INTx5 = 1; IER |= M_INT3; // Enable CPU Interrupt 3
// Enable ADCINT in PIE, ADC sampling subroutine PieCtrlRegs.PIEIER1.bit.INTx6 = 1; IER |= M_INT1; // Enable CPU Interrupt 1 EINT; ERTM;
return 0;
132
// Enable Global interrupt INTM // Enable Global realtime interrupt DBGM
Board Schemes } /*****************************************/
/*****************************************
FREQUENCY MEASUREMENT FUNCTIONS
*****************************************/ interrupt void Capture1_isr(void) { temp1_1 = EvaRegs.CAP1FIFO; // Reads value in fifo stack temp1_2 = EvaRegs.CAP1FIFO; // Reads value in fifo stack //resets the measures accumulator measures_addition=0;
//Measures shift and average calculation for(i=AVERAGE_FREQUENCY_VALUES-1;i>0;i--){ measures[i]=measures[i-1]; // adds last instant measures except the last one. measures_addition=measures_addition+measures[i]; }
//gets the last measure if(temp1_2>temp1_1){ //No overflow measures[0] = temp1_2 - temp1_1; }else{ //overflow measures[0] = temp1_2 +(0xffff- temp1_1); } //Software frequency filter if(measures[0]>average_measures[0] + MAXIMUN_FREQUENCY_CHANGE || measures[0]
//average calculation //adds the last measure measures_addition=measures_addition+measures[0];
133
Board Schemes //calculates the average average_measure= measures_addition/AVERAGE_FREQUENCY_VALUES;
//Shift the averege values array and insert the new one. for(i=AVERAGE_MEASURES-1;i rel="nofollow">0;i--){ average_measures[i]=average_measures[i-1]; } average_measures[0]=average_measure;
/*** frequency decay measurement ***/ average_decay=0; //clear average decay addition
for(i=AVERAGE_MEASURES-1;i>0;i--){ //freency decay shift average_decay=average_decay+decay[i-1]; decay[i]= decay[i-1]; }
//frequency decay calculation decay[0]=(FCF2/(float)((long)average_measures[0]*(long)average_measures[0])) *(((float)average_measures[1]/(float)average_measures[0])-1); //average_decay Calculation average_decay=(average_decay+decay[0])/AVERAGE_MEASURES; //
//RMS power voltage and current //voltage temp_results=0; //clear temp variable for(i=0;i
//Store result
//Current temp_results=0; for(i=0;i
//Power temp_results=0; for(i=0;i
134
Board Schemes temp_results=temp_results/ADC_MEASURES_PER_CYCLE; power_RMS=temp_results;
//Control logic Type II, Temperature Control
T_high_out=T_high*KF*(measures[0] - FO); T_low_out=T_low*KF*(measures[0] - FO); //Control logic Type I, relay control if(average_measure rel="nofollow"> lower_limit || average_decay < decay_limit){ //there is a frequency event. open_relay(); } else if(average_measure < higher_limit && average_decay > decay_limit && relay_status!=CLOSE){ //frequency event is recovered //the relay status in changed to "wating for attachment" if(relay_status == OPEN){ //only once relay_status=WAIT; waiting_cycles= temp1_2
%
MAXIMUN_WAIT_TO_ATTACHMENT; }else{ //from the second time waiting_cycles--; } if(waiting_cycles <= 0){ //the waiting time is up. close_relay(); } }else if(relay_status == WAIT){ // if frequency and frequency decay do not fullfit the event // thresholds but they dont fullfits the recover // criteria thresholds. waiting_cycles= temp1_2 % MAXIMUN_WAIT_TO_ATTACHMENT; }
//it is call the stub function that should store the data adquired in the external memory. recod(current_time,average_measure, decay[0],relay_status);
voltage_RMS,current_RMS,
power_RMS,
//Temperature Set points change set_points_update(T_high_out,T_low_out); // Clears capture unit 1 interruption flag EvaRegs.EVAIFRC.bit.CAP1INT = 1; // PIE 3 ack PieCtrlRegs.PIEACK.all = PIEACK_GROUP3; }
135
Board Schemes
/*****************************************/
/*****************************************
DATA ADQUISITION FUNCTIONS
*****************************************/
interrupt void {
adc_isr(void)
//current time stamp increment current_time++; //Current measurements adc_samples_current[ConversionCount] (AdcRegs.ADCRESULT2 >>4))/2;
=
((AdcRegs.ADCRESULT0
>>4)
+
////Voltage measurements adc_samples_voltage[ConversionCount] (AdcRegs.ADCRESULT3 >>4))/2;
=
((AdcRegs.ADCRESULT1
>>4)
+
T_high = AdcRegs.ADCRESULT4 >>4; T_low = AdcRegs.ADCRESULT5 >>4; //Calculates the square values for the RMS calculations adc_square_current[ConversionCount]=(long)adc_samples_current[ConversionCount]*(l ong)adc_samples_current[ConversionCount]; adc_square_voltage[ConversionCount]=(long)adc_samples_voltage[ConversionCount]*(l ong)adc_samples_voltage[ConversionCount]; adc_square_power[ConversionCount]=adc_square_current[ConversionCount]*adc_square_ voltage[ConversionCount];
// If ADC_MEASURES_PER_CYCLE conversions have been logged, start over if(ConversionCount == ADC_MEASURES_PER_CYCLE-1) { ConversionCount = 0; } else ConversionCount++; // Reinitialize for next ADC sequence AdcRegs.ADCTRL2.bit.RST_SEQ1 = 1; AdcRegs.ADCST.bit.INT_SEQ1_CLR = 1; PieCtrlRegs.PIEACK.all = PIEACK_GROUP1;
136
// Reset SEQ1 // Clear INT SEQ1 bit // Acknowledge interrupt to PIE
Board Schemes return; } /*****************************************/
/*****************************************
RELAY CONTROL FUNCTIONS
*****************************************/ void time_synchronization(long time){ current_time = time; }
/*****************************************/
/*****************************************
RELAY CONTROL FUNCTIONS
*****************************************/
void close_relay(){ relay_status=CLOSE; //sets status close GpioDataRegs.GPFSET.bit.GPIOF8 =0x1; //close relay }
void open_relay(){ relay_status=OPEN; //sets status open GpioDataRegs.GPFCLEAR.bit.GPIOF8=0x1; //open relay }
void set_higher_limit(unsigned int new_limit){ higher_limit=new_limit;
//set new recover threshold
}
void set_lower_limit(unsigned int new_limit){ lower_limit=new_limit;
//Set new frequency event threshold
}
137
Board Schemes void set_decay_threshold(float new_limit){ decay_limit=new_limit;
//set new frequency decay threshold
}
/*****************************************/
void set_points_update(unsigned int T_high,unsigned int T_low){ //this function should change the temperature set points if applicable
}
void recod(long time,unsigned int period,unsigned int voltage,unsigned int current,unsigned long power, float decay,short relay){ //this stub should store last measurements in external flash memory //store measurements in internal RAM ext_time[ext_index]=time; ext_period[ext_index]=period; ext_voltage[ext_index]=voltage; ext_current[ext_index]=current; ext_power[ext_index]=power; ext_decay[ext_index]=decay; ext_relay[ext_index]=relay;
//increment the index and reset to zero if out of bounds ext_index++; ext_index=ext_index%STUB_EXT_MEMORY_LEGTH; }
138
Board Schemes
A.2
Memory mapping
MEMORY { PAGE 0 : /* For this example, H0 is split between PAGE 0 and PAGE 1 /* BEGIN is used for the "boot to HO" bootloader mode /* RESET is loaded with the reset vector only if /* the boot is from XINTF Zone 7. Otherwise reset vector /* is fetched from boot ROM. See .reset section below RAMM0 BEGIN PRAMH0 RESET XINTF60
: origin : origin : origin : origin : origin =
= 0x000000, length = 0x3F8000, length = 0x3F8002, length = 0x3FFFC0, length 0x100000, length =
*/ */ */ */ */
= 0x000400 = 0x000002 = 0x000FFE = 0x000002 0x008000
PAGE 1 : /* For this example, H0 is split between PAGE 0 and PAGE 1 */ RAMM1 DRAMH0 XINTF61
: origin = 0x000400, length = 0x000400 : origin = 0x3f9000, length = 0x001000 : origin = 0x108000, length = 0x008000
}
SECTIONS { /* Setup for "boot to H0" mode: The codestart section (found in DSP28_CodeStartBranch.asm) re-directs execution to the start of user code. Place this section at the start of H0 */ codestart ramfuncs .text .cinit .cio .pinit .switch .reset
: : : : : : : :
> > > > > > > >
BEGIN, PRAMH0 XINTF60, XINTF60, XINTF60, XINTF60, RAMM0, RESET,
.stack .ebss .econst .esysmem .sysmem .extRAM
: : : : :
> > > > >
RAMM1, DRAMH0, DRAMH0, DRAMH0, DRAMH0,
PAGE PAGE PAGE PAGE PAGE PAGE PAGE PAGE
= = = = = = = =
0 0 0 0 0 0 0 0, TYPE = DSECT /* not used, */
PAGE = 1 PAGE = 1 PAGE = 1 PAGE = 1 PAGE = 1 : > XINTF61,
PAGE = 1
}
139
Board Schemes
B
BOARD SCHEMES
This appendix contains the full schemes of the load controller. is understanded as an input, as well as the symbol as an The symbol output. This symbols are used to connect the intputs and outputs of the main scheme with the detail of the DSC connectors. They are also used to represent the power outlet connection.
140
Filter
Voltage Level Shifter
5V
100kΩ
5V 5V 100nF
Voltage Measuring Transformer 69,8kΩ
4,02kΩ
Wall-Outlet
100nF
196kΩ
100nF
100nF
100nF
-
100nF
91Ω
100nF
5,49kΩ
OPA277
+
220:1.1
5V 5V
261kΩ 100nF
-
100nF
-5V
110kΩ
5V
+
100nF
-
OPA277
100nF
Voltage 23,3kΩ Offset
-5V
Voltage Output
OPA277 100kΩ
+
100nF -5V
100nF
10kΩ
Filter 5V
5V 5V 100nF
100nF
196kΩ
4,02kΩ
100nF
100nF
100nF
-
50:1 69,8kΩ
5V
100nF
91Ω
+ Current Measuring Transformer
100nF
5,49kΩ
OPA277
261kΩ 100nF Current Level Shifter 100kΩ
OPA277
5V
+
100nF -5V
100nF 100nF
-5V 311kΩ
5V Double Power Supply
230:6
1N4001
-
Current
OPA277
Current 23,3kΩ Offset
100kΩ
+
10Ω 100nF -5V
Bridge
LM7805 INPUT
100nF
10kΩ
OUTPUT
-5V 100nF
100nF
470µF
Wall-Outlet 100nF
470µF
100nF 5V OUTPUT
LM7905
Bridge
5V
Stub Relay
Power On LED
1N4001 5V
Processor Single Power Supply
230:6
100kΩ
1N4001
Relay Wall-Outlet
Bridge
LM7805 INPUT
100nF
470µF
OUTPUT
100nF
+5V
Board Schemes
141
Figure 8-I - Board Scheme
INPUT
Board Schemes
ADCINA1
ADC Connection Socket 10x3 Port 5 and 9
ADCINA0
Connection Socket 20x3 Ports 4 and 8
GP I/O 19
22
25
28
14
17
20
23
26
29
15
18
21
24
27
30
13
16
10
7
4
1
58
55
52
49
46
43
40
37
34
31
28
25
22
19
13
16
10
7
4
1
Voltage
Voltage Offset
Relay Control
Capture Unit
Capture Unit
9
12
60
11
57
8
5
2
59
54
45
56
42
53
44
39
48
41
36
51
38
33
50
35
47
32
27
24
30
21
29
23
26
20
18
14
17
11
8
5
2
Figure 8-II - DSC connectors Detail
142
6
3
15
9
12
6
3
Current
ADCINB1
ADCINB0
Voltage
Current Offset
C
OTHER PLOTS
The next plots show other averaged measurement plots, it shows 200 values instead of 30 as the one in the frequency measurement section. The data of these tables can be found in Table 8-III - Averaged frequency measurements.
Averaged Vs Thresholds 62650 62600 62550 62500 62450 62400 62350 1
30
59
Average measurements
88
117
recover threshold
146
175
Frequency event threshold
Figure 8-III – Averaged frequency measurements (Integer units)
143
Board Schemes
Average measurements 50,045 50,04 50,035 50,03 50,025 1
21
41
61
81
101
121
141
161
Average measurements
Figure 8-IV – Averaged frequency measurements (Hz)
144
181
D TABLES
D.1
ADC Measurements
Table 8-I – ADC samples table No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
145
V 2480 2996 3422 3714 3845 3794 3571 3204 2721 2170 1532 864 301 0 0 0 99 590 1226 1896 2476 2995 3421 3716 3844 3793 3567 3194 2707 2161 1523 849 294
I 2047 2182 2284 2370 2393 2382 2327 2230 2109 1969 1805 1635 1492 1413 1417 1416 1429 1557 1723 1898 2045 2183 2291 2365 2400 2382 2328 2227 2107 1966 1802 1628 1485
Tables
Table 8-I – ADC samples table No 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82
146
V 0 0 0 99 611 1243 1910 2485 3004 3428 3719 3842 3793 3570 3196 2710 2162 1526 853 293 0 0 0 100 606 1239 1905 2482 3001 3427 3719 3844 3796 3571 3198 2715 2165 1530 862 302 0 0 0 98 599 1238 1902 2479 3000
I 1403 1407 1411 1442 1570 1732 1901 2049 2184 2288 2370 2393 2384 2325 2235 2106 1967 1803 1629 1490 1407 1411 1413 1433 1569 1730 1900 2050 2182 2295 2370 2400 2386 2327 2235 2109 1966 1803 1632 1492 1412 1411 1405 1434 1567 1727 1899 2050 2180
Tables
Table 8-I – ADC samples table No 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
V 3425 3717 3844 3794 3573 3200 2716 2166 1534 863 294 0 0 0 102 596 1226 1899
I 2287 2365 2397 2389 2332 2230 2108 1967 1806 1635 1483 1416 1413 1416 1441 1562 1723 1898
147
Tables
D.2
Frequency measures
Note that the measurements made by the load controller are the ones that are in integer units and the values in Hz are calculated as explained in the respective section.
Table 8-II - Instant frequency measurements No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
148
Instant frequency (int) 62509 62492 62495 62510 62498 62498 62523 62509 62506 62505 62498 62519 62502 62520 62509 62460 62509 62495 62512 62521 62509 62491 62492 62491 62509 62498 62502 62495 62527 62500
Averaged freq (int) 62503 62502 62503 62503 62500 62500 62500 62500 62500 62500 62500 62501 62500 62501 62500 62501 62502 62503 62503 62503 62502 62502 62503 62503 62503 62503 62504 62504 62504 62504
Instant freq. (Hz) 49,993 50,006 50,004 49,992 50,002 50,002 49,982 49,993 49,995 49,996 50,002 49,985 49,998 49,984 49,993 50,032 49,993 50,004 49,990 49,983 49,993 50,007 50,006 50,007 49,993 50,002 49,998 50,004 49,978 50,000
averaged freq. (int) 49,998 49,998 49,998 49,998 50,000 50,000 50,000 50,000 50,000 50,000 50,000 49,999 50,000 49,999 50,000 49,999 49,998 49,998 49,998 49,998 49,998 49,998 49,998 49,998 49,998 49,998 49,997 49,997 49,997 49,997
Tables
Table 8-III - Averaged frequency measurements No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
Frequency (Int) 62455 62454 62454 62454 62454 62455 62456 62456 62457 62456 62456 62455 62455 62455 62455 62454 62454 62454 62455 62455 62455 62455 62454 62455 62455 62455 62454 62455 62454 62454 62454 62454 62454 62454 62453 62452 62452 62452 62453 62454 62456 62456 62457 62455 62456 62455 62455
Frequency (Hz) 50,0360259 50,0368271 50,0368271 50,0368271 50,0368271 50,0360259 50,0352248 50,0352248 50,0344237 50,0352248 50,0352248 50,0360259 50,0360259 50,0360259 50,0360259 50,0368271 50,0368271 50,0368271 50,0360259 50,0360259 50,0360259 50,0360259 50,0368271 50,0360259 50,0360259 50,0360259 50,0368271 50,0360259 50,0368271 50,0368271 50,0368271 50,0368271 50,0368271 50,0368271 50,0376283 50,0384295 50,0384295 50,0384295 50,0376283 50,0368271 50,0352248 50,0352248 50,0344237 50,0360259 50,0352248 50,0360259 50,0360259
149
Tables
Table 8-III - Averaged frequency measurements No. 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
150
Frequency (Int) 62456 62455 62456 62456 62456 62457 62457 62457 62457 62456 62456 62457 62457 62458 62458 62459 62459 62458 62457 62457 62456 62457 62457 62457 62458 62458 62457 62456 62456 62456 62456 62456 62456 62456 62456 62456 62455 62456 62455 62455 62455 62453 62452 62452 62451 62452 62451 62450 62450
Frequency (Hz) 50,0352248 50,0360259 50,0352248 50,0352248 50,0352248 50,0344237 50,0344237 50,0344237 50,0344237 50,0352248 50,0352248 50,0344237 50,0344237 50,0336226 50,0336226 50,0328215 50,0328215 50,0336226 50,0344237 50,0344237 50,0352248 50,0344237 50,0344237 50,0344237 50,0336226 50,0336226 50,0344237 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0352248 50,0360259 50,0352248 50,0360259 50,0360259 50,0360259 50,0376283 50,0384295 50,0384295 50,0392308 50,0384295 50,0392308 50,040032 50,040032
Tables
Table 8-III - Averaged frequency measurements No. 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
Frequency (Int) 62449 62449 62449 62449 62448 62447 62448 62449 62449 62448 62447 62447 62447 62446 62446 62446 62445 62445 62445 62446 62447 62448 62448 62449 62449 62450 62450 62450 62451 62451 62452 62451 62450 62450 62451 62452 62452 62453 62452 62453 62454 62454 62454 62455 62454 62453 62454 62453 62453
Frequency (Hz) 50,0408333 50,0408333 50,0408333 50,0408333 50,0416346 50,042436 50,0416346 50,0408333 50,0408333 50,0416346 50,042436 50,042436 50,042436 50,0432374 50,0432374 50,0432374 50,0440388 50,0440388 50,0440388 50,0432374 50,042436 50,0416346 50,0416346 50,0408333 50,0408333 50,040032 50,040032 50,040032 50,0392308 50,0392308 50,0384295 50,0392308 50,040032 50,040032 50,0392308 50,0384295 50,0384295 50,0376283 50,0384295 50,0376283 50,0368271 50,0368271 50,0368271 50,0360259 50,0368271 50,0376283 50,0368271 50,0376283 50,0376283
151
Tables
Table 8-III - Averaged frequency measurements No. 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
152
Frequency (Int) 62452 62451 62451 62451 62452 62452 62452 62451 62452 62451 62451 62451 62452 62451 62452 62451 62451 62452 62452 62452 62452 62452 62452 62451 62451 62452 62452 62453 62453 62453 62453 62453 62454 62453 62454 62454 62454 62453 62454 62454 62454 62453 62453 62452 62452 62452 62451 62451 62452
Frequency (Hz) 50,0384295 50,0392308 50,0392308 50,0392308 50,0384295 50,0384295 50,0384295 50,0392308 50,0384295 50,0392308 50,0392308 50,0392308 50,0384295 50,0392308 50,0384295 50,0392308 50,0392308 50,0384295 50,0384295 50,0384295 50,0384295 50,0384295 50,0384295 50,0392308 50,0392308 50,0384295 50,0384295 50,0376283 50,0376283 50,0376283 50,0376283 50,0376283 50,0368271 50,0376283 50,0368271 50,0368271 50,0368271 50,0376283 50,0368271 50,0368271 50,0368271 50,0376283 50,0376283 50,0384295 50,0384295 50,0384295 50,0392308 50,0392308 50,0384295
Tables
Table 8-III - Averaged frequency measurements No. 195 196 197 198 199 200
Frequency (Int) 62453 62452 62452 62451 62452 62452
Frequency (Hz) 50,0376283 50,0384295 50,0384295 50,0392308 50,0384295 50,0384295
153
Tables Note that the values in integer units are the measurements done by the load controller and the values in the S.I are calculated as explained in the respective section. Table 8-IV - RMS measurements voltage (int) 1410 1410 1410 1410 1410 1410 1410 1409 1409 1409 1409 1409 1409 1410 1410 1410 1410 1410 1409 1408 1408 1408 1409 1409 1409 1409 1409 1409 1409 1410 1410 1409 1409 1409 1409 1409 1409 1409 1410 1410 1410 1410 1409 1409 1409 1410
154
Voltage (V) 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,503227 224,503227 224,503227 224,344005 224,184783 224,184783 224,184783 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,503227 224,503227 224,344005 224,344005 224,344005 224,503227
current (int) 363 363 363 363 364 364 364 364 364 364 363 364 364 363 363 363 363 363 362 362 362 362 362 362 362 361 361 361 361 361 361 361 361 362 362 361 361 361 361 361 361 361 361 361 361 362
Current (A) 2,21557617 2,21557617 2,21557617 2,21557617 2,22167969 2,22167969 2,22167969 2,22167969 2,22167969 2,22167969 2,21557617 2,22167969 2,22167969 2,21557617 2,21557617 2,21557617 2,21557617 2,21557617 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20947266 2,20947266 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20947266
power (Int) 622406 622408 622424 622165 623789 623741 623300 623573 623138 623093 623091 623098 623224 622940 622699 622584 621658 621646 620754 620712 620712 620718 620820 621078 620208 618550 618746 619261 619377 619337 619337 619335 619405 619853 620191 618636 618278 617978 618528 618545 618545 618538 618436 618489 618439 620316
Power (W) 604,8634 604,865344 604,880893 604,629193 606,20742 606,160773 605,732203 605,997508 605,574769 605,531037 605,529093 605,535896 605,658345 605,382349 605,148142 605,036383 604,136483 604,124821 603,257962 603,217146 603,217146 603,222977 603,322102 603,57283 602,727351 601,116082 601,306558 601,807042 601,919773 601,8809 601,8809 601,878957 601,946984 602,382357 602,71083 601,199658 600,851748 600,560204 601,094702 601,111223 601,111223 601,10442 601,005295 601,056801 601,00821 602,832307
Tables
Table 8-IV - RMS measurements voltage (int) 1410 1410 1410 1410 1410 1410 1410 1409 1409 1409 1409 1409 1410 1410 1410 1410 1410 1409 1409 1409 1409 1409 1409 1409 1409 1409 1409 1410 1410 1410 1410 1410 1410 1410 1410
Voltage (V) 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,503227 224,503227 224,503227 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,344005 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227 224,503227
current (int) 362 361 361 361 361 361 361 360 361 361 361 362 362 362 362 362 361 361 361 361 361 361 362 362 362 362 362 362 363 362 362 362 362 362 362
Current (A) 2,20947266 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,19726563 2,20336914 2,20336914 2,20336914 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20336914 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,21557617 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266 2,20947266
power (Int) 619299 618714 618733 618763 618763 618761 618500 617741 617917 618790 619367 620667 620396 620406 620406 620402 620345 620057 619835 619448 620337 620398 620589 620599 620599 620594 620710 621041 622379 621486 621092 620827 620902 620947 620947
Power (W) 601,843971 601,27546 601,293924 601,323079 601,323079 601,321135 601,067491 600,329884 600,500923 601,349318 601,910055 603,173414 602,910052 602,91977 602,91977 602,915883 602,86049 602,580607 602,364864 601,988772 602,852715 602,911996 603,097612 603,107331 603,107331 603,102472 603,215202 603,536873 604,837161 603,96933 603,586435 603,328904 603,40179 603,445522 603,445522
155
Tables
D.3
Frequency decay Table 8-V - Frequency decay No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
156
fequency decay 0,03970647 -0,03954634 0 0 0 0,03970647 0,03971905 0 0,03973164 -0,0395714 0 -0,03955887 0 0 0 -0,03954634 0 0 0,03970647 0 0 0 -0,03954634 0,03970647 0 0 -0,03954634 0,03970647 -0,03954634 0 0 0 0 0 -0,03953382 -0,03952131 0 0 0,03968133 0,0396939 0,07943811 0 0,03973164 -0,07911773 0,03971905 -0,03955887
Tables
Table 8-V - Frequency decay No 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
fequency decay 0 0,03971905 -0,03955887 0,03971905 0 0 0,03973164 0 0 0 -0,0395714 0 0,03973164 0 0,03974424 0 0,03975685 0 -0,03959649 -0,03958394 0 -0,0395714 0,03973164 0 0 0,03974424 0 -0,03958394 -0,0395714 0 0 0 0 0 0 0 0 -0,03955887 0,03971905 -0,03955887 0 0 -0,07906764 -0,03952131 0 -0,03950881 0,03966878 -0,03950881 -0,03949631
157
Tables
Table 8-V - Frequency decay No 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
158
fequency decay 0 -0,03948383 0 0 0 -0,03947135 -0,03945888 0,03961863 0,03963116 0 -0,03947135 -0,03945888 0 0 -0,03944642 0 0 -0,03943396 0 0 0,03959361 0,03960612 0,03961863 0 0,03963116 0 0,03964369 0 0 0,03965623 0 0,03966878 -0,03950881 -0,03949631 0 0,03965623 0,03966878 0 0,03968133 -0,03952131 0,03968133 0,0396939 0 0 0,03970647 -0,03954634 -0,03953382 0,0396939 -0,03953382
Tables
Table 8-V - Frequency decay No 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
fequency decay 0 -0,03952131 -0,03950881 0 0 0,03966878 0 0 -0,03950881 0,03966878 -0,03950881 0 0 0,03966878 -0,03950881 0,03966878 -0,03950881 0 0,03966878 0 0 0 0 0 -0,03950881 0 0,03966878 0 0,03968133 0 0 0 0 0,0396939 -0,03953382 0,0396939 0 0 -0,03953382 0,0396939 0 0 -0,03953382 0 -0,03952131 0 0 -0,03950881 0
159
Tables
Table 8-V - Frequency decay No 194 195 196 197 198 199 200
160
fequency decay 0,03966878 0,03968133 -0,03952131 0 -0,03950881 0,03966878 0
161
www.oersted.dtu.dk/cet Ørsted·DTU Centre for Electric Technology (CET) Technical University of Denmark Elektrovej 325 DK-2800 Kgs. Lyngby Denmark Tel: (+45) 45 25 35 00 Fax: (+45) 45 88 61 11 E-mail: [email protected]
Related Documents
Nieto
June 2020 8
Theses
November 2019 30
Theses
November 2019 44
Theses
June 2020 19
Alejandro
May 2020 26
Nieto Sobejano
June 2020 9More Documents from ""
Poetry 9 : Berdua Di Hotel
April 2020 22
Theses 10 : Aleksander M Derdowski
May 2020 23
