Homemade IGBT MIG welder with ATX PSU parts. [blown yet again][blown count: 5]

[magic]

Are both FETs still driven from two opposite windings of the same GDT? This oscillation waveform makes no sense to me

[Refrigerator]

Are both FETs still driven from two opposite windings of the same GDT? This oscillation waveform makes no sense to me

The GDT has two isolated outputs, both of which are inverted.

[Refrigerator]

I got it to make a few snotty welds before it died (i died a little inside too) but all is not lost.
Turns out the low side gate discharge transistor shorted out and prevented the low side MOSFETs from turning on.
Since i'm all out of suitable PNP transistors i redesigned the gate discharge to be handled with a NPN transistor.
I basically flipped the PNP circuit over and put an NPN transistor instead of the PNP transistor, confirmed that they both work identically with Ltspice.
Now all that is left to do is to make a new MOSFET board.
So far all the testing was done without gas, because i don't have any at the moment and also my current sensing circuit needs some tweaking.

[tautech]

Get some gasless wire.

[Refrigerator]

Previously i had a pair of SS8550 transistors, one per side.
The collector of the SS8550 shorted to the emitter but the base seems fine so i assume the collector current was too high.
The MOSFET board was made for one MOSFET per side but i added two more so i guess discharging both gates at once was a little too much to ask, even if the transistor is rated for 1.5A.
All MOSFETs had a 10Ω resistor on the gate.

[T3sl4co1l]

Yeah, when MOSFETs die, they fail three-way shorted.  Expect high voltages on all three pins.  Standard procedure is to check gate resistors and replace drivers when transistors fail.  Even stuff on the other side of the transformer may be affected.

Instead of attempting to weld -- which is clearly not working, and only serving to rack up a larger pile of failed components -- why not stage some tests, measuring peak voltages and currents on the transistors under given input and output conditions?  Example, low input voltage and full output load current; shorted load; full input voltage and no load or light load; etc.

The measurements taken in such tests can be extrapolated to the full operating input voltage and output current, and you can determine whether it is safe to operate at full power, or if the design needs further refinement.

Tim

[Refrigerator]

Yeah, when MOSFETs die, they fail three-way shorted.  Expect high voltages on all three pins.  Standard procedure is to check gate resistors and replace drivers when transistors fail.  Even stuff on the other side of the transformer may be affected.

Instead of attempting to weld -- which is clearly not working, and only serving to rack up a larger pile of failed components -- why not stage some tests, measuring peak voltages and currents on the transistors under given input and output conditions?  Example, low input voltage and full output load current; shorted load; full input voltage and no load or light load; etc.

The measurements taken in such tests can be extrapolated to the full operating input voltage and output current, and you can determine whether it is safe to operate at full power, or if the design needs further refinement.

Tim

Thanks for the advice but the pile of blown silicon has grown once more.
The MOSFETs were not blown when the gate discharge transistor failed, but i got lazy and inpatient and thought that a bigger transistor would solve the problem, which was not the case.
The result: two more blown MOSFETs, both low side MOSFETs in this case (high side MOSFETs are ok). Seems the welder wasn't switching correctly.
Also the load resistor wasn't there just to be burned, while the welder was busy making toast i poked around and measured everything i could with the limited amount of equipment i have (one DMM, 1985 USSR clamp meter and a USB scope).
I'll put this project on hold because there's other projects i want to try out (2 switch forward circuits, current transformers and some other stuff), also i'm out of parts.
I think i might also rebuild the welder with a new controller and a better layout. I don't like the GDT and i think photocouplers would be the better way to go, the GDT is finicky and rewinding it is a pain in the arse. With an external power supply for the gate drive circuits i could adjust many more things.
So far i'm about 30€ in and i think this project was worth it even if it doesn't work just yet.