How to choose a heatsink to sustain MOSFETs peak currents

Heatsinks are required to lower the thermal resistance of power MOSFETs for keeping the junction temperature at a safe level. This presentation explains the role of the thermal impedance of the heat sink in protecting the transistor when exposed to high peak currents and proposes a simulation approach for choosing a heatsink for a given application.

[@paulpaulzadeh6172]

Hi Prof. Yaakov, it is One of the most common approaches to measuring thermal resistance is to mount a
metal-cased 50-W power resistor on the heat sink or other material in question and
apply a known number of watts to the resistor. At the same time, an IC temperature
sensor can be mounted to the heat sink to monitor the temperature rise. The thermal
resistance is then easily calculated. By observing the rate of rise of temperature, the
thermal mass can also be inferred. Alternatively, you can raise the heat sink to a known
high temperature, remove the heat source, and record the time it takes for the temperature
to fall to 63% of the difference between the ambient and the starting temperatures.
This time value is then the time constant of the thermal capacitance and the known
thermal resistance of the heat sink.
usually in practice if you have stationary air then become worse , which is difficult to simulate and other thing is slew rate of switching , if it is slow then transistor pass class A region . this will be different for lateral mosfet vs vertical mosfet , it can get thermal run away too. anyway thanks a lot.

[@sambenyaakov]

​Thanks​ for you​r input. Measuring heatsinks is a common practice​. I will be happy to send you some examples of results if you send me your email.
I do not understand what is the relevance of heatsink measuring to switching losses. ​With modern simulation models of MOSFETs you can get good estimates of switching losses. See ua-cam.com/video/kEl2n9jcrIM/v-deo.html

[@MishapHapmiss-lx4sl]

What if All Mosfets are on same heat sink( three phase inverter with parallel mosfets). Can you please do a similar analysis for the above problem Professor?

[@sambenyaakov]

You combine all the losses into one current source and put all the thermal networks of the transistors in parallel (Rth/n, Ct*n).

[@giacomosammuri3531]

Clear lesson as always. Could you make a video for dimensioning a fan for the heatsink? Often Is a procedure based on experience and not with analitycal procedure. I'm intrested in tour approch.

[@sambenyaakov]

Thanks for discovering the error. Originally, this was supposed to be the thermal capacity of the fins but as labeled is incorrect. Sorry. I will put a note on that in the description section.

[@sambenyaakov]

Good subject. Will try.

[@cladisla]

@@sambenyaakov this is a very tricky topic as the performance does not only depend on the losses, heatsink and fan but also on the system's geometry. Most of the time the fan is selected either with the charts provided by the manufacturers (rule of thumb) or with CFD simulations to take into consideration all variables.

[@ats89117]

Interesting study with results that agree with intuition but the "thermal pad" model is probably oversimplified. CPUs and graphics processors use complex heat spreaders which may involve heat pipes or 2D materials that spread heat with efficiencies approaching diamond to make better use of the heat sink. These technologies tend to push the results toward even more emphasis on surface area rather than thermal mass. Without the improved heat spreading, you end up with a large temperature gradient in the heat sink itself as you move away from the die contact area, leading to obvious inefficiencies...

[@sambenyaakov]

​The video is an introduction to the subject t​o explain the basics. I would love to see your video describing models of the more advanced approaches.

[@ats89117]

@@sambenyaakov Coupled systems of partial differential equations always scared the hell out me, but companies like Amsys, Comsol and Agilent sell horribly expensive FEA based simulation packages that seem to do a reasonable job on thermodynamics...

[@ats89117]

I noted the comment below asking why this video appeared in his suggestions. I had the opposite question, i.e. why this video took 4 weeks to appear in my suggestions, when your videos are the first ones I look at. I'm guessing that your videos are getting lower priority because they generate less money for UA-cam, but maybe I am being too cynical...

[@sambenyaakov]

Any suggestions on how to improve exposure?

[@satiric_]

Thank you for another clear explanation. For the 5 stage heatsink model, why does it have six capacitors? It seems to me like the 5 stage heatsink model has 6/5ths of the capacitance of the 1 stage heatsink model. Am I misunderstanding something?

[@sambenyaakov]

Thanks for pointing to the error. Originally, the sixth cap was supposed to represent the thermal capacity of the fins but it is incorrect as shown. I will put a note on that in the description section.

[@satiric_]

@@sambenyaakov Ah I see. Interesting, I never thought about simulating the fins separately from the rest of the heatsink. That's a neat idea

[@sambenyaakov]

👍🙏

[@peterbrowner4687]

How do you extract C1 and C2 from mosfet datasheet?

[@sambenyaakov]

Fitting the ladder to the one pulse curve of the Thermal Impedance given in the datasheet, Hold on, I am planning to post a video on that and on the extraction of the heatsink model.

[@peterbrowner4687]

Sorry, I have just found one of your videos that explains my doubt. Congratulations for your channel!!

[@tamaseduard5145]

👍🙏❤️

[@sambenyaakov]

👍🙏😊

[@jensbuchholz5766]

👍

[@sambenyaakov]

Thanks

[@dayanidhi9391]

Why on earth did this video appear in my suggestions?

[@sambenyaakov]

Perhaps you need to select a heatsink to reamove your extra heat?

[@dayanidhi9391]

@@sambenyaakov