UD2.7 output stage keeps dying

[davekni]

Any capacitor that has been in backwards needs to be discarded and replaced with a new capacitor.  The old one will be damaged - high ESR and likely to leak in the future corroding your circuitry.  If any other electrolytic caps are bulging on top, they need replacing too.  Bulging is a sign of either backwards or a bad cap.

[Miki_407]

What should i do to prevent any chance of my transistors dying after i plug in the gdt and replace C32 with a new one? How do i verify that my gdt is made good?

[davekni]

It looked like C1 was in backwards in the picture.  That needs replacing, unless I'm seeing wrong.  Check for any other caps with any bulge on top.

Once it is all working correctly without a GDT, connect the GDT in series with that 51-ohm resistor.  Without IGBTs connected (no GDT secondary load), GDT voltage should look like a sagging square wave, not short spikes.

[Miki_407]

So this is a update on my progress in making my output stage work as it should.
1.I tested my GDT on my function generator.
input waveform:

output waveform:

Input waveform is blue. Yellow is voltage drop over 51ohm resistor in series with GDT:

2.I tested the output stage of UD2.7 with GDT and this is the output
Input waveform is blue. Yellow is voltage drop over 51ohm resistor in series with GDT:

but I don't like this oscillation at the turn off:



Now to the questions.
Should I use a second GDT for a full bridge with 2 parallel 40n60n igbts?

[davekni]

How are you measuring GDT input?  Do you have a differential scope probe?  The off-state is not showing 0V as it should.  Could be a scoping issue, if you are connecting a scope ground lead to one of the driver outputs (one of the GDT inputs).  Unless you have a differential probe, it's better to scope each driver output with a separate probe, grounded to driver ground.

[Miki_407]

Sorry for not replying quicker.
I didn't use a differential probe. I did it like this. Oscilloscope gnd is insolated from gnd on the board.


After replacing 51ohm resistor with 10ohm one it died but this time only source and drain were shorted.


And for my unanswered question:

Now to the questions.
Should I use a second GDT for a full bridge with 2 parallel 40n60n igbts?

[davekni]

Connecting the scope ground to one of the driver's outputs may be part of the FET frying issue.  Even if the scope's line-ground is disconnected from the line, at 100kHz+ there is significant load to scope power line connections through EMI filtering on scope line input.  Ground the scope probes to the driver board ground.

Concerning 1 vs 2 GDTs, either works.  1 is more common and usually easier.  Skip-cycle drivers generally require two GDTs connected to two different outputs.

[Miki_407]

Oscilloscope was connected to the output of gdt after changing to 10ohm resistor and it died then as well. I will rewind a gdt in hopes it is a problem.

[Miki_407]

Finally problem solved. Rewinding gdt third time on a new core and replacing all passive components around it solved the problem. Now i can continue with my build. I fell little bad for wasting time for my mistake with bad gdt.
Thanks to davekni for his help.

[davekni]

You are welcome.  Glad to hear that it is working now!

[Paultesla]

If the GDT had a turn to turn short or input to output winding short that would cause the transistors to die. Good luck with your project.

Paul