My IGBT bridge is BAD and needs redoing

Actually IIRC saturating inductors were another source of little bits of FET cases scattered in the lab (I should mention this wasn't my project, which just means that I learned a bit less from the other guy's mistakes with _much_ less embarrassment).

When an inductor saturates it's apparent inductance goes way down, so if it's got drive behind it and it's looking into a short the current goes _way_ up. You can get some pretty dramatic views on a scope screen before the current set of final drive transistors scatter bits of themselves over the bench.

I am working on a high power inverter bridge to make square wave AC from DC supplied by my welding machine. I got ot to kind of work.

I tested my bridge with about 85 volts, without anything connected to the AC circuit. Using a big power supply.

The bridge is of shorting variety, I tried to make sure that it never is open.

Unfortunately, the current leak was about 5 amps, without the inductor in series. The snubber capacitor that I had, seriously overheated its supply wires.

That means that the short time is too much and needs to be reduced. I will need to replace turn off resistors leading to every gate, with a lower resistor value.

His intent is to drive this with a current source, specifically a TIG/stick welder. These machines typically have 80 volt OCV, short circuit current varies with machine setting from 1 amp to maybe 200 amps in Ig's case. Operating voltage (arc established) is typically in the region of 24 volts. Saturation is not an issue (!)

With no load, energy stored in the driving-point inductance during short has nowhere to go when the short is removed except for the load-side snubber.

I will try to do that over the weekend.

Absolutely, I am a complete newbie.

Got it. I will retry my test with an inductor made of 2/3 of a standard 500 ft home depot spool of 12 gauge wire.

(this is a cool experience to watch that inductor. When it is powered, the slightly loose coils of wire start moving about)

I would prefer to avoid having any dead time. I think that you are right indeed and I should first redo my test _with_ the inductor in series. And then, I should make sure that short time is small enough, by adjusting the off resistors, all 8 of them.

Regarding having settable dead time, yes, my Semikron SKHI 23 allows me to do that very easily with a resistor. I can set dead time to, say, 0.9 uS (one of the values in the datasheet) But, like I said, I would prefer not to go there.

My rough calculations suggest that I simply have too much of short time and need to adjust that. Having an inductor may make it less of an issue. By the way, I have little understanding of "saturated inductors" and I am not sure if they act like simple inductors. If the inductor in my welder is saturated and if saturated inductors do not act like inductors, then I may have surprises.

The datasheet for my Semikron SKHI 23, with schematics for paralleled bridges, is here:

The picture of my current test setup is here:

Yep, hence my question.

I think that I will try to reduce short time first and will see what happens, on the scope. I am able to see the signs of the short time on the scope signal.

If that does not help, I may switch to dead time. I will experiment a little bit, I have a relatively simple testing setup that actually works.

It is an interesting project and I am still playing with low amperages (under 20A, vs 400A max capacity) and (relative to capacities of my bridge) peak transient voltages.

All correct, my OCV is 85 volt.

Correct.

So, would it be correct to conclude that the source for my bridge will be highly inductive?

That's 100% correct.

So, Don, do you think that my bridge should be shorting, or not shorting? Is there some hope that by using this bridge after the welder's inductor, and by reducing short time as much as practicable, I would be able to run this bridge?

Or do you think that I should use a non shorting bridge with interlock (dead time)?

I got some hard data from my oscilloscope. My bridge shorts for 5.7 microseconds each time it switches.

That's way too much. I think that if I can make it short for less time, like under 1 uS, it would be just fine.

The cause of the problem is a too high value of Rgoff.

small inductor ?

last year I found a web site that had downloadable binaries from newsgroups. there was a limited free service (I think 2Mb daily)

This year I have an ISP with a full (if slow) news server.

it may have been this site:

Bye. Jasen

By the way, my IGBT bridge is now better. :)