Proper use of an IGBT (1 destroyed)

[katzohki]

Hi all, I'm working on a project and my intent is to use an IGBT to dump about 4kV through an inductive load (a simple inductor). I'm pretty experienced with MOSFETs and BJTs, but not immune to making mistakes.

During our last testing step we were attempting to dump the load from our capacitor bank into a 1 Meg resistor. Math says the IGBT should be able to handle this easily. Didn't see much so we adjusted to a 100k resistor at which point we blew the IGBT after a few tries.

I had the IGBT on the high side of the load, which now that I've had time to rethink things I realized was probably a big part of our issues (since Vge would not be higher than threshold  ). Trying to trigger it by clipping an aligator onto the 12V power supply. Heard a snap and now CE is close to zero ohms, so the IGBT is dead.

Here's my question: Is the damage purely due to the IGBT being on the high side of the load? Or is it possible we're missing something else? We're a 2 man team working late at night and our budget is low, so I can only afford to replace this $100 IGBT so many times... If you have drivers you recommend for this app, I'd like to hear them as well.

IGBT: IXYH30N450HV
Cap: 10kpf x10

[MariuszD]

You didn't write anything that would allow us to know the facts, assess the parameters the transistor is working with, show us the schematic.

Math says the IGBT should be able to handle this easily.

I won't believe it until I see it.

Trying to trigger it by clipping an aligator onto the 12V power supply.

It must have it must have ended badly.

We're a 2 man team working late at night and our budget is low, so I can only afford to replace this $100 IGBT so many times...

What you described doesn't look like engineering work, but rather like blind experiments.
There are two paths to the goal: the first is to learn a lot before you start building anything, the second is to connect and check what comes out. You chose the second one, here you can skip the learning time - that's your gain, but the cost is the number of burned transistors. On HV hobbyist websites, you can often read that they burned out a bucket of transistors.

[moffy]

Perhaps a high voltage relay would be both cheaper and more robust: https://au.mouser.com/ProductDetail/MEDER-electronic/HE24-1A83-02?qs=sGAEpiMZZMtGt%252Bn33CgIP5nj8dCC9rHCFD6H0YQrxOg%3D
P.S. make sure your discharge resistor can withstand the full voltage without breaking down, resistors have max voltage ratings.

[dietert1]

HV work requires experience and/or a profound understanding of physics, like electric field strength near "peaked" parts.
Apart from that you may need special equipment. What is the isolation spec of your test leads?
Switching the high side means a fast HV transient on E terminal, which certainly can cause excess GE voltage when using test leads.
A gate protection circuit (like a close-by zener) might have avoided the expensive failure. You also need a GE pull-down resistor.
Apart from that i would first make a gate drive circuit that avoids contact bouncing, but delivers a clean pulse instead. Use a scope to check that.
You can also use a scope to check IGBT switching, of course at lower test voltages, within the voltage spec of your scope probes.
I'd share the recommendation to check the voltage spec of the discharge resistor, too.

Regards, Dieter

Edit:
You need cooling for the IGBT. As far as i understand the datasheet specs 250 uA leakage at 4500 V. That is about 1 W and may be sufficient to overheat the IGBT if without cooling. I didn't find its isolated case voltage spec.