Telwin Force 125 inverter welder keeps blowing its IGBT

[ComfyChair]

Hi,

I'm trying to repair my modest inverter welder, a Telwin Force 125.

While I was using it, I heard a "pop" an it was gone. I found the schematic for it (attached for reference) and started my repair adventure.

Once I opened it up I found the AC input resistor (R2 - 47R 7W-) completely charred. It goes without saying that it measured open circuit. Then, also the IGBT (Q1) was completely shorted. I mean all three terminals connected together.

What I did was replacing IGBT and resistor. I probed around the IGBT section a bit and found no issues.

With the new IGBT the welder turned on for ~10s then BANG! it was gone again...

Not wanting to burn a 3€ IGBT again, I investigated further in the control circuitry. What follows is the list of things I tried:

• I was suspecting a EOS on the IGBT gate so I measured the zeners near Q1 (D8 & D9) they measure fine
• I started suspecting the peak current control circuitry so I measured all the resistors, transistors and diodes along the path from TA+ and TA- to U2. Again all looks good
• I was suspecting something wrong with the supply to U2 and the opams so I powered the welder without the IGBT. VCC reads 22V
• I thought U2 was not switching, keeping Q1 in some strange state, but I must admit the design is quite nice. There is C5 that is protecting Q1 from any DC voltage and R9 keeps G & E shorted. I measured anyway there with my meter and see the voltage going up and down.

Am I missing something?

I have to admit that for the time being, I'm just using my meter. I do not have a differential probe for my oscilloscope and all the signals are ~100V above ground level.

My last suspect is Q6. Can it be that perhaps the sink transistor is busted/weak for some reason and Q1 is not fully turned off?

I was wondering wether I could arrange a poor man's differential probe by subtracting two channels of my scope and measuring the gate voltage over time. The common mode voltage is high (~100V as I said) while the gate voltage should go from 0 to ~20V. With my 8-bit scope set to 20V/div I could discriminate ~1V superimposed on a 100V common mode-signal. Will I be able to tell if something is wrong with the driver? Any other ideas?

[ComfyChair]

I was wondering wether i could power the board with 24V (i.e. via TP2) instead ov mains voltages so that I can attach the oscilloscope directly. What do you think?

[ComfyChair]

Continuing my journey through the repair I decided to apply power to the 24V rail from an external supply, instead of mucking around a device directly conencted to mains voltage.

So I connected 4 wires near Q6: 2 for power and 2 to probe the voltage before (purple trace in osciloscope photo) and after the capacitor C5 (yellow trace).

When I turned on the supply I saw that Q6 was indeed switching but after the capacitor there was no voltage. When I added an external capacitor, also the yellow trace started switching.

A capacitance meter on C5 reads nothing so it's completely gone!

Even with C5 open, the IGBT is still turning on somehow. This is rather strange 

I do not have a 1205 50V 1uF cap, can I replace it with a leaded capacitor of similar rating? I have a 1uF 400V leaded polypropilene cap. Will it work?

Am I missing something else?

[ComfyChair]

Here we are for the latest episode in this installment. I managed to solder in two 2.2uF 25V 0603 capacitors in series. Everything is working now!

[Swake]

Nice repair Thanks for the documentary