Thermal Modeling of 0805 Thick Film Chip Resistor

Question

I recently got confronted with the question what peak/pulse power a standard 0805 thick film resistor can withstand.

Background: We are designing a 2kW bidirectional DCDC converter. The Primary side is fed by a 3-phase generator/motor (rectified and smoothed) with 180V DC, the secondary side is connected to a 28V battery. The generator/motor is attached to the drive shaft of a combustion engine. There are two operational modes:

1. Combustion engine is off, power is drawn from the battery and stepped up to start the combustion engine by using the generator as motor.
2. Combustion engine is running and driving the generator à power is drawn from the generator and used to charge the battery (and supply some other stuff)

On the secondary side (28V battery) an LC output filter is implemented with and parallel RC damping circuit for the filter resonant frequency (see figure below).

Problem Statement: When the battery is first connected to the secondary side the damping resistor has to handle a large current pulse to charge the 120µ capacitor. At first we used a standard 0805 resistor but during review the power capability of this resistor was questioned. Of course there are resistors which have specified pulse power capabilities mentioned in their datasheets but I was curious about what a standard thick film could handle and don’t just hit it with some big resistors. The design is also very limited in physical volume and weight therefore I want to find the best solution for the problem. That’s why I tried to build a thermal model of a 0805 thick film resistor in LTSpice.

The Model: The heat power in a thick film resistor is naturally generated in the resistive element (the resistive film). The film is attached to the substrate with a specific thermal resistance. The film also has a specific thermal capacity, as has the substrate. I found this (table 1 page 6) document which states that the thermal resistance between film a substrate of a 0805 resistor is approximately 38°C/W. Also I found in this document that the film is 10µm thick and can be made of eg. silver or palladium. The isobaric volumetric heat capacitance of silver is 2.44 J cm-3 K-1. The footprint of a 0805 resistor is 2mm x 1.25mm. The volume of the thick film is in best case 2mm x 1.25mm x 10µm = 0,000025cm³ à therefore the heat capacity of the film is 0.061mJ/K and the heat capacity of the aluminum substrate approximately 1.5mJ/K respectively.

Considering the above I came up with the LTSpice simulation simulation is here on dropbox

Question/Results: According to the Spice simulation the resistor film will peak with ~670°C which is mostly dependent on the heat capacitance of the film because the thermal resistance to the substrate is too big. I also made a lab test setup which resulted in a cracked 0805 chip resistor.

My questions are:

1. Is the approach of the model acceptable or have i made some fundamental errors?
2. I could not find information about the maximum temperature the film is allowed to reach (datasheets dont differ between film, substrate and leads temeratures as they do with eg. ICs where you can find junction and case temperatures etc.)
3. Is this generally worth to be considered or do you normally just take a few more resistors and don’t waste time how I did?

Thanks in advance

Answer 1

I took a different approach to solving your problem. Looking at the data sheet for a Stackpole thick film resistor I noticed that their current sensing resistors had an overload rating, that is based on an industry standard test. Googling "thick film resistor short term overload specification" I found this document https://xdevs.com/doc/VPG/VFR-TN105.pdf "The Effects of Short-Time Overload on Resistive Elements in Precision Resistors" published by Vishay. Didn't read it, but the short answer is that it looks like the industry signs up for 2.5 times the rated current for 5 seconds as the short term overload specification. For a 0.25W .2 Ohm resistor that works out to be about 2.8 amps for 5 seconds. Presumably you could read the above doc and perhaps others to justify why in your case you can use a different, larger specification, if you have to.

Answer 2

Silicon has 1.6 picoJoules per cubic micron per degree Centigrade thermal capacity (specific heat). Assume the entire resistor is silicon/ceramic/clay, Inject a heat pulse, and compute the rise.