Hi. I'm making a PWM (~20kHz) motor controller for my tablesaw but the IGBT keeps blowing. I'm UK based so mains is 240VRMS->330Vp. The IGBT is rated 20A continuous and 600Vmax C-E voltage. I have a free-wheeling diode do I don't understand how the 600V is exceeded. I have a 15V zener protecting the base (20Vmax B-E voltage). Nothing else on the PCB is damaged. The IGBT seems to fail when the motor goes open circuit by releasing the latching switch on the tablesaw (not shown below).

Any help much appreciated. At the mo in desperation I'm thinking using ~400V varistor across the C-E, and using a 1000V IGBT.

 

Not shown on the schematic, I assume the gate driver has a good connection GND to Q1 Emitter. That loop needs to be small.

 

Not shown on the schematic, I assume the gate driver has a good connection GND to Q1 Emitter. That loop needs to be small.

They are connected thru the ground plane on the underside of the pcb, ~10mm apart

 

Is Q1 running hot? Watch out the voltage is high.
What is the current rating on the motor? Often there is a current for no load, one for full load which is much higher, then a startup current (stall current) which is much higher. Give us all the data you have.

 

Q1 is not running hot. It gets to ~50c. It's a universal motor so would have high starting current but I've added a soft-start so it typically draws 3A on start up and 1.5A at operating speed. I have been testing this circuit for a couple of weeks on another motor I had (similar specs) with no issues. The problem has only started when I'm using a tablesaw with a latching switch that breaks the circuit when the red button is hit.

 

The problem has only started when I'm using a tablesaw with a latching switch that breaks the circuit when the red button is hit.

Does the IGBT brake at turn off of the switch? That might be a clue. Some saws have a safety switch that not only turns off the power but also breaks the motor.

 

Does the IGBT brake at turn off of the switch? That might be a clue. Some saws have a safety switch that not only turns off the power but also breaks the motor.

tbh I don't know if it happens at turn off or turn on. do you mean "brakes the motor" ie. shorts it, although that couldn't short my side of the circuit else it'd be shorting the mains. That's in an interesting avenue of investigation. I'll take a look at the switch and come back. Thanks a lot for your help so far.
ps. a quick check and the switch has a helpful schematic showing that the motor should appear open-circuit when the stop button is pressed

 

What is the part number?

The IGBT is rated 20A continuous and 600Vmax C-E voltage.

I think it is a 3A continuous @100C and 6A at 25C. Try adding a heatsink. Does not have to be large but try reducing the heat.

 

The TVS ( P6KE400(C)A ) was added in desperation soldered close to Q1. It seems underrated (548Vc @ 1.1A), however I don't see how it's being called on with D1 close on the board.

Sorry, I should have updated the schematic (ps. I've now updated the OP). I upgraded Q1 to STGF20H60DF. It's well within it's continuous operating range. Only gets warm. The issue is transient. Either current or voltage surge. An idea I had last night is to add an LED+resistor across B-E which will be OFF the moment Q1 blows (bc all the failures so far have shorted C-E and B-E), so at least I can see if it blows at turn off or turn on.

 

Without showing us the whole circuit, the very best you can get is good guesses. Quite likely the cause of the problem is in the portion we are not shown.
This same thing does happen a lot: We really need to see the whole circuit.

 

I take your point. However there's nothing innovative in the rest of the circuit. I posted only what I thought was the relevant part to make it easier for people to help. And it has been a help bc in lieu of an explanation how there could be a voltage spike at Q1 I've stopped thinking this is the cause. I think a transient power dump is blowing the IGBT. So the problem is in my code on the PICMX not regulating peak current. I'm making changes to the code to fix this and I'll report back.