48VDC to 230VAC inverter challenge

[Aurgelme]

I am trying to repair a 4kw / 8kw peek 48dc to 230V ac that I got from someone who gave up on repairing it.
First of all, it's to learn a little more about electronics and switch mode, which I'm quite new to, so maybe this should be posted in the beginner's forum?

Anyway, the problem is that module 1 (the one on the far left in the drawing) burns off the mosfet transistors as well as the 10ohm resistor and 10nF capacitor and the 4 20ohm gate resistors, even at low load load (20w).

If I then solder out the mosfets in module 1, the inverter works, but with a somewhat low voltage (218V ac)

I have checked all surrounding components, as well as soldered out all transformers and measured and megged. Changed transformer from the one in module 1 to the one in module 3, but same problem with module 1.

The signals from the driver circuit look ok to me, so any advice on where to look next is welcome.

The first time I use Kicad to create a sketch, so there are probably a lot of mistakes in the names and characters used, but the values of the components are correct.

[johansen]

do you really only have two gate drivers supplying 5 separate sets of mosfets and transformers? and only one set blows up?

something wrong with the circuit board.

[Aurgelme]

Yes, it is only module 1 that fails
I have also thought that there might be something wrong with the pcb layout, but no mechanical fault that I can find.
The only thing is that what I call module 1 is the one at the bottom right of the picture that has the furthest way from the driver board.

[Jwillis]

That spot for a diode between 4 and 2( I think it's 2)  transformers . Has that always been missing?
My thinking is if one bank isn't switching then that puts the others out causing a voltage drop and the current isn't shared properly. That creates more stress causing the weakest link to fail.

[Aurgelme]

Yes, that diode has never been there and there is old tin in the holes, so I think it is like that from manufacturing.

[Jwillis]

Seems odd that since all the others have a fly back diode except that one.
I cant see very well  if the two right ones have diodes under the ground wires.

[Aurgelme]

It is a 10A10, 10A high current standard diode Peak Forward Surge Current 8.3ms 600A.
I counted it as a short-circuit protection for incorrect reverse polarity since it is close to the fuses, but I may be wrong here.

[Jwillis]

Could be for reverse polarity. Hard to see what relation they have. Just saw one missing and thought it to be strange.

[Jwillis]

Just trying to get my head around a possible cause . Each transformer is 48V on primary side. I'll call each one a single bank . All 48V primary side banks are in parallel for high current drawn from battery. The secondary sides are all in series to create 48V x 5 for 240V low current. If one of the 48V banks are not working, that puts the system to 192V. The circuit tries to compensate by driving the 4 remain banks harder to raise the output back up to 240V but can only achieve the 218V you measure. Since each individual 48V bank will never draw exactly the same amount of current, one will always draw slightly more current than all the others. That is the weakest link. So my thinking is the one that keeps blowing isn't exactly the problem but a victim of the problem being the weakest link. So one of the other 48V transformer mosfet banks is the problem.

[Aurgelme]

I thought the same, so I soldered out all the transformers and measured them.

[Jwillis]

The transformers may not be the where problem but the whats driving the transformer. Did you check all the Mosfets on the low side. Each transformer has 4 Mosfets.Are all transformers drawing more or less the same current. The one that draws the least will most likely be the one mosfet transformer set that isn't working properly.

[Aurgelme]

Finally I found the fault, it was bad/wobbly in a through-plating in the pcb from the gate driver, which occurred a few minutes after the inverter was switched on.
Used an effect resistor instead of a fuse on the failed module, so that I had time to measure without burning the mosfet. After a couple of minutes, one mosfet pair was constantly left open.

Thanks for all the input!