My cheap crappy 1KW grid tie inverter (GTI) seems to keep blowing IGBTs. When I first got it, it had two failed IGBTs. It used four 60N100's (TO264). I bought 5 from a dubious source (fleabay) and replaced just the two that were dead. The GTI worked for almost two years but recently one of the two that I replaced failed. Since I had three spares, I replaced the dead one. Then a few weeks later it failed again. Weird. Well since I bought dubious China-sourced IGBTs from fleabay it could very well just be a mixture of fakes. I replaced it, but then a day later it failed again! I put in the last spare I had and the inverter works again... but then two weeks later it failed again. I'm all out.
The weird thing is that there are 4 IGBTs in the H-tree and one would imagine that it should be a fairly even distribution of which one fails but it was always the same one, implying perhaps I got fakes, but that high side seems to be okay...
Well, since I was out of spares I went and fleabayed more IGBTs, this time 80N60's (TO247) as at 120VAC I don't need 1KV isolation and apparently the inverter was likely populated with 60N100s because that's what they had on hand as the same FR4 PCB was used for both 220V and 120V variants. Once again this is fleabay and chances of fakes are high. (BTW, it seems that China once again has a new scheme dealing with the tariffs...)
Anyway, as a proud ghetto mod poster, I was even cheaper than just getting possible fakes from China. Instead of replacing all the IGBTs I replaced just that one failed 60N100 IGBT with the new 80N60.
Now who wants to make bets or at least some theoretical discussion:
- When will the GTI fail again. Not only am I using dubious transistors, I have a mismatched H-tree now. Heck since that one transistor failed in a day, it could also fail in a day!
- which IGBT fail: options are
1. The (smaller TO-247) 80N60 I just put in which could be fake.
2. The other (TO-264 giant cheap fleabay possibly fake) 60N100 I replaced two years ago. AFAICT the two failed transistors were of the same half bridge, so the remaining possibly fake transistor is the high side driver and the one that keeps failing is the low side.
3. One of the other two 60N100s on the other half bridge that were in there the whole time and were never replaced.
I was wondering how good TO-264 was compared to TO-247 and TO-3P for that matter. Seems manufacturer and bonding differences may affect ultimate thermal resistance and the smaller TO-247 looked better on paper... Alas there's more to it than just this. Who knows.
BTW the 80N60 is *significantly* faster than the 60N100, and slightly more efficient. As of posting this the GTI is working even with the mismatch...
The weird thing is that there are 4 IGBTs in the H-tree and one would imagine that it should be a fairly even distribution of which one fails but it was always the same one, implying perhaps I got fakes, but that high side seems to be okay...
Well, since I was out of spares I went and fleabayed more IGBTs, this time 80N60's (TO247) as at 120VAC I don't need 1KV isolation and apparently the inverter was likely populated with 60N100s because that's what they had on hand as the same FR4 PCB was used for both 220V and 120V variants. Once again this is fleabay and chances of fakes are high. (BTW, it seems that China once again has a new scheme dealing with the tariffs...)
Anyway, as a proud ghetto mod poster, I was even cheaper than just getting possible fakes from China. Instead of replacing all the IGBTs I replaced just that one failed 60N100 IGBT with the new 80N60.
Now who wants to make bets or at least some theoretical discussion:
- When will the GTI fail again. Not only am I using dubious transistors, I have a mismatched H-tree now. Heck since that one transistor failed in a day, it could also fail in a day!
- which IGBT fail: options are
1. The (smaller TO-247) 80N60 I just put in which could be fake.
2. The other (TO-264 giant cheap fleabay possibly fake) 60N100 I replaced two years ago. AFAICT the two failed transistors were of the same half bridge, so the remaining possibly fake transistor is the high side driver and the one that keeps failing is the low side.
3. One of the other two 60N100s on the other half bridge that were in there the whole time and were never replaced.
I was wondering how good TO-264 was compared to TO-247 and TO-3P for that matter. Seems manufacturer and bonding differences may affect ultimate thermal resistance and the smaller TO-247 looked better on paper... Alas there's more to it than just this. Who knows.
BTW the 80N60 is *significantly* faster than the 60N100, and slightly more efficient. As of posting this the GTI is working even with the mismatch...