Thermal Analysis Of A Linear Voltage Regulator

Thermal Analysis of a Linear Voltage Regulator

1 din 4

http://dev.emcelettronica.com/print/52072

Your Electronics Open Source (http://dev.emcelettronica.com) Home > Blog > brumbarchris's blog > Contenuti

Thermal Analysis of a Linear Voltage Regulator By brumbarchris Created 02/25/2009 - 07:09

Technology voltage regulator Taking into account the amount of time electronic products have been around us, one would think that choosing the correct power supply would be only a trivial task, just an easy off-the-shelf exercise. However, given that a poorly constructed power supply can easily lead to the demise of the entire product (or just parts of it) it is still one of the most important parts on the schematic. It is exactly the large variety of linear and switching regulators that are available, that makes choosing the correct IC for you application so complicated; and for products that require a reasonable amount of power to function correctly, the heat dissipation and thermal management of the power supply is of prime importance. The purpose of this article is to give an example of how you should determine through computation whether your linear voltage regulator is safe from a thermal point of view. Our team has recently been assigned the task of designing a small electronic module for the automotive industry. Given the harsh environment that a car represents for any electronic module, the requirements on the power supply were a little more than ordinary. A summary of these requirements (for the power supply) is given below: - It should provide an output voltage of 3.3V +/-5%, and a maximum current of 150mA - It should function (performance) between 8V and 16V, for the automotive temperature range: -40 Celsius to +85 Celsius. - It should withstand input voltages of up 24V for 60 seconds without functionality (to simulate a jump start of the car); once input voltage returns to the normal 8V-16V range, functionality should be restored automatically - It should withstand reverse battery voltage of -14V (obviously without functionality); once input voltage returns to the normal 8V-16V range, functionality should be restored automatically - It should withstand start-up pulses ranging from -450V up to +40V (these are rather short pulses or groups of pulses, sometimes lasting hundreds of microseconds; they are thoroughly detailed in automotive specifications) At a first glance, it did not sound like much of a task; to make it even easier, our options were narrowed to linear voltage regulators only, as the nature of the application would not allow the use of a switching power supply. Given the situation, one of our junior engineers was assigned the task, and shortly he came to us with the following schematic (I stripped it down of any ESD protection components or anything else which is not of any interest for the thermal management):

25.02.2009 15:31

Thermal Analysis of a Linear Voltage Regulator

2 din 4

http://dev.emcelettronica.com/print/52072

The components involved are fairly common automotive components. The diode is rated 1A and 600V repeatable peak reversed voltage while the voltage regulator can withstand input voltages of over 40V while giving 400mA at 3.3V with a better accuracy than the one required. These are obvious and clearly stated in the datasheets of the two components (S1J, TLE4274). Go to Download section Datasheet_Thermal_Analysis_Linear_Voltage_Regulator.rar Moreover, the TLE4274 regulator has a smart way of shutting down at over-temperature, which would thus render it safe in case an increase of the input voltage would determine an increase in the dissipated power. So the IC would not be destroyed, but it would safely shut down until it would cool down enough for it to start working again. Thus, most of the requirements are fulfilled at a first glance. Since this was mainly a “copypaste” schematic from a previous product (with different requirement, though) everybody assumed it would easily work without any problems. And we all continued to do so until one of us proceeded to perform the theoretical thermal analysis of this power supply. The aim of the analysis was to prove that the regulator would be able to provide the 150mA at a worst case temperature, that of +85 Celsius; and it would have to work under a 16V battery input voltage. Given the requirements, the circuit should have been able to handle this situation without getting damaged or without turning off due to over-temperature. The flow of the calculation would have to be: 1) determine the maximum voltage drop on the regulator 2) determine the maximum dissipated power 3) determine the maximum temperature of the regulator die taking into account an ambient temperature of +85 Celsius 4) compare the computed die temperature against the maximum allowed (specified in the regulator datasheet; if the computed value is greater than the one specified, it means that at some point, the regulator overt-temperature protection would kick in and turn off the 3.3V output. The official tool to perform such calculations is Mathcad, but due to lack of licenses at home, I performed these calculations in MathSuga, a freely available math calculations tool available under: http://www.gahum.com/ First of all, we need to get from the datasheet all the variables involved in the calculations. The maximum battery voltage is (taken from the requirements):

The minimum voltage drop on the protection diode is (taken from the diode datasheet):

25.02.2009 15:31

Thermal Analysis of a Linear Voltage Regulator

3 din 4

http://dev.emcelettronica.com/print/52072

The minimum output voltage of the regulator is (taken from the regulator datasheet):

With these, applying Kirchoff’s second law, it is easy to compute the maximum voltage drop on the regulator itself:

The maximum current that is to be supplied by the regulator is (taken from requirements):

Knowing these values, it is easy to compute the power dissipated on the voltage

There are two factors which are contributing to the increase of temperature on the regulator die. One of them is the ambient temperature; the other one is self-heating due to the power dissipated by the component itself. In order to determine the effect of the dissipated power, the parameter required is the thermal resistance of the package. This is measured in degrees/watt and is generally specified in the datasheet of a component; for the given regulator, taking into account the fact that we use a TO252 package, it is easy to find this parameter in the datasheet:

This value is valid for the minimum footprint designed for this package. A larger footprint, which dissipates the heat on a copper plane and on internal copper planes through smart use of vias would allow us to use a smaller value for this parameter. However, given the space constraints for this project, it was not possible to ensure anything more than a minimal footprint for this component. The effect of the self heating on the die temperature increase (in degrees) may be obtained directly through the multiplication of the dissipated power by times the thermal resistance of the package. However, this effect will have to be cumulated to the ambient temperature in order to determine the real temperature of the die. Thus, taking into account a temperature of:

We can determine the maximum temperature that would be developed in the silicon:

25.02.2009 15:31

Thermal Analysis of a Linear Voltage Regulator

4 din 4

http://dev.emcelettronica.com/print/52072

Unfortunately, the result is much higher then the 150 Celsius degrees mentioned in the datasheet as being the maximum accepted junction temperature. So what would happened in reality is that once turned on, this regulator would start heating up, and when the die would reach about 150 Celsius, the over-temperature protection would step in to turn it down. Given the fact that we proved through calculation a value of more than 200 Celsius, it became obvious that the proposed solution is was not appropriate. We obviously had to change the concept. Buy now from

Farnell (24h delivery). Trademarks

Source URL: http://dev.emcelettronica.com/thermal-analysis-linear-voltage-regulator

25.02.2009 15:31