High power, short time power pulses in FETs

[treez]

Hi,
Do you think that this FET (STD3NK90Z) could conduct 0.7A (with a Vds of 440V) for 50us?
Then also for 500us?
STD3NK90Z FET:
https://www.st.com/resource/en/datasheet/stp3nk90z.pdf

 

[asdf44]

Treez I see the answer in figure 1, Safe operating area for DPAK, TO-220.

I see it able to conduct more than 10A for 1mS with 500Vds.

 

[treez]

Thanks, this fet can dissipate 1600w for 100us and no more....imagine it dissipating 1600W for 200us, then nothing for 0.999800 seconds...repeatedly.....thats only 0.32W on average, but this TO220 FET wouldnt be able to do that...seems incredible.
I cant see how it can heat up seriously in 100us....even with 1600W

 

[barry]

Thanks, this fet can dissipate 1600w for 100us and no more....imagine it dissipating 1600W for 200us, then nothing for 0.999800 seconds...repeatedly.....thats only 0.32W on average, but this TO220 FET wouldnt be able to do that...seems incredible.
I cant see how it can heat up seriously in 100us....even with 1600W

That's a naive view. There will be local hotspots in the device that can cause damage. Do you think you can withstand 100000 volts across your body with a 1:100000 duty cycle? It's only an "average" of 1 volt...

 

[c_mitra]

this fet can dissipate 1600w for 100us and no more....imagine it dissipating 1600W for 200us, then nothing for 0.999800 seconds...repeatedly.....thats only 0.32W on average

It depends on the anatomy: the inner constructions.

1600W for 100us results in 0.16J of energy. How efficiently this is spread over inside the the die?

During the rest period, the heat will spread- it moves by diffusion. The die will cool down but not to the ambient temp.

When the next pulse comes with another 0.16J of heat, the temp will rise further (because the die is already pre-heated). I do not know the exact details, but the steady state dissipation will be the one when the input heat is balanced by the heat lost by the package.

The graphs are there for guide only; you need to be careful while working close to the borderline.

- - - Updated - - -

I see it able to conduct more than 10A for 1mS with 500Vds.

Did you mean 1.0A?

 

[asdf44]

Did you mean 1.0A?

Yes you're right, I misread.