Replacing burnt FET's to a Power Inverter

[tribat]

I had a brand new 12V->220V 1000W Power Inverter that started smoking after only 10 minutes of usage. Opened it up but nothing looked burned, noticed however that one of the 8 mosfets it had was poorly mounted to the heatsink. I desoldered all the mosfets and did some basic resistance tests between their pins, they numbers were all over the place for each part. So I'm guessing what happen is after one of the fets failed under thermal stress, the rest followed.

The device has 2 rows of 4 mosfets on the each side of the device, 4x IRF840 and 4x SUP85N03. Now the IRF840's seem to be really widely available part and you can get them really cheap off ebay. However the same cannot be said for SUP85N03, hardly anyone is selling them and getting them would be really expensive. How do I go about finding a suitable replacement part that is as cheap and as widely available as possible?

[Kiriakos-GR]

I bet that every reader of this story , will ask just this : ( if this is an new device) why you did not RMA it .. on the spot .

[Zero999]

It's likely that other components have failed and just replacing the MOSFETs won't help.

My advice is solder the old MOSFETs to the board as well as you can, put it back together and make it look like you've never touched it. Return it to the supplier and ask for a refund.

In future, if it's brand new always peruse the seller for a refund rather than attempting to repair it yourself as doing so will invalidate any warranty.

[tribat]

It cost only $40 and there was no warranty with it hehe. I heard that often the case infact is that the driving circuitry gets destroyed when the mosfets fail, however I can't see anything that was damaged or burned. I already bought a new one with a proper warranty but I would still like to try and repair this one if I can do so cheaply.

[Zero999]

In most jurisdictions, it doesn't matter whether it it has warranty or not, the seller still has an obligation to sell you a product which works as intended for a reasonable length of time.

[cjk2]

I bet the other FETs are good, or at least could be. FETs usually fail with all pins shorted together. Sometimes they fail high resistance if you damage the gate oxide which you did not do in this case (that takes overvoltage or possibly too much dv/dt). I would guess that you measured "all over the place" because the FETs had various amounts of gate charge when you measured them. Short the gate to source on each FET. The pinout will be Gate, Drain, Source if you are looking at the front of the package.

While the gate is shorted to the source, stick the positive lead of the meter on the drain and the negative lead on the source. Measure resistance, you should see and open. If you measure a short or less than thousands of ohms then the FET is dead. If you switch the leads you will see a diode voltage drop because the intrinsic diode will conduct. Also measure from gate to source, you should always see an open.

Hope this helps.

[tribat]

I was looking for SUP85N03 replacements from taydaelectronics.com (http://www.taydaelectronics.com/servlet/the-Transistors-cln-FETs-%26-Mosfets/Categories) and it looks like STP80NF55 comes pretty close with the specs. Could it work?

SUP85N03 Datasheet:
http://datasheet.octopart.com/SUP85N03-04P-E3-Vishay-datasheet-12129.pdf

STP80NF55 Datasheet:
http://www.st.com/stonline/books/pdf/docs/8260.pdf