How to test an inverter without blowing up the output transistor?

[Pentium100]

My big UPS blew an output transistor module (with some collateral damage). I think I have repaired what I thought was the source of the problem, put everything back together turned it on and the transistor blew up again. I found one more problem, fixed it.

However I would like to test it this time without blowing up an expensive transistor module if I missed something. I am thinking of adding a couple of 4.7Ohm resistors in series with the fuses that go to the transistor so that if something is shorted the resistors would blow while limiting the current and protecting the transistor.

Is that a good idea or a really stupid one?

[cnqhdszq]

I had not  repaired UPS ,but I had  repaired  induction heater  .
After I replaced the fuse、bridge rectifiers、IGBT、and the drive transister .I usually put a filament lamb is series between the heater and the power strip . if there is something wrong with the heater and the IGBT is stated  “on ”.the filament lamb will lighted . I usually use 220V   100W lamp .
In the end .We should select the suitable power of the lamp


I think adding a couple of 4.7Ohm resistors in series with the fuses that go to the transistor is not a appropriate solution.

[filssavi]

What I do when bringing up a new converter for the first time or after a transistor swap these are the steps I take

1) test functionality at low voltage open load and with heavy current limit
2) gradually increase dc-link voltage up to full value again with heavy current limit
3) start testing with a heavy inductive load that keeps the current constant, while increasing currents

[Pentium100]

I cannot really reduce the DC voltage since it is controlled by the firmware and the MCU tests it before starting the inverter. As for the lamp, the DC voltage is +- 400V and the idle current is about 10A (because of the output capacitors). That's why I thought about resistors in series with the DC voltage to the transistor. At least I would be able  to see the output waveform befdore the resistors overheat and fail. If the waveform looks OK, then try it without the resistors and hope for the best.

[aqibi2000]

Remove the output transistors and repair the 400V boost circuit and make sure it works

Then put the output stage back in

[Pentium100]

The rectifier/booster works, what blew up the first time was the output and the drivers for the output transistors. I think I know why it happened the first time and the second time and I think I fixed both problems.

I think I figured out how to run the outputs on low voltage - I do not need to use the DC rails, I can instead just connect the output stage  to 12V and see how the output waveform looks like before the MCU decides there is a problem and shuts everything down.

[TheMG]

This may or may not work depending on the circuit topology of the UPS, but what I have done repairing a UPS before is actually to power it up will all the inverter output transistors out of circuit and checking for proper switching waveforms at the gate connections of the transistors. Usually you do get a few cycles of switching before the UPS fails its self diagnostics and shuts down the inverter drive.

The reason I do this is that more often than not, output transistors blowing is due to a problem in the drive circuits rather than the output circuits themselves. If one or more transistors aren't getting proper switching waveform, you get a nearly instant bang, usually due to one or more transistors in the H-bridge coming on when it shouldn't be.

Once I've confirmed proper gate switching waveforms then I put all the transistors back in circuit and hope for the best. I've had rather good luck using this repair strategy.

Important note: don't just lift the gates of the transistors, remove them completely from circuit. Floating gates can cause the transistors to inadvertently switch on resulting in a big bang - don't ask me how I know.

[Pentium100]

Thanks for your suggestions. I ended up connecting a pair of weak 12V batteries to the output transistors and seeing how the output waveform looks, it looked OK, so I installed the fuses and tried it and very quickly found out that I forgot to glue part of the output filter inductor. Didn't blow the transistor though.