Maker Pro
Maker Pro

Replacing fets for a DC-AC inverter?

fastline

Aug 9, 2010
13
Joined
Aug 9, 2010
Messages
13
I guess worth a shot here. Posted this elsewhere but no help on this. I have a "1000w" inverter. I use quotes before I know and knew the 1000w is a joke on this inverter, but I never even pushed it! Probably a max of 200w ever, and only got about 10hrs use out of it. China is just so good at making things last a few hours.

Anyway, I found two shorted output fets. gate drivers are still good. The marking on the fets is 85gt33sw. From what VERY little I can find online, these are 85A, 330V BUT that is ALL I can find. I even asked one chinaman on ebay that is selling them and he has no specs. I have no interest in replacing with the same type of fets. I highly doubt the above specs are qualified or vetted. There is some indication the fets might be rip off AP85GT33SW which are IGBTs but not sure why a cheap inverter would have IGBTs? Does not seem likely to me.

In any case, help would be appreciated!!
 

Bluejets

Oct 5, 2014
6,901
Joined
Oct 5, 2014
Messages
6,901
Show photos especially clear close up of the suspect devices.
Also model/brand of inverter.
There may be a perfectly good reason why it is branded as 1000W.
 

fastline

Aug 9, 2010
13
Joined
Aug 9, 2010
Messages
13
This is about the best I can do. This site won't take larger files. Someone should fix that. The rack of 4 fets is where the issue is. 2 of those are shorted out between all pins. Failure was not very eventful. Flipped the switch and noticed the watt meter on it was showing 15w with no load, and there was a noise that sounded like fan noise. Rebooted and it started its soft start routine and would pull 150A against the battery bank, then go into a fault. No smoke or bang!

All fets transfer heat to the case but are electrically insulated with pads.

The inverter is just an off brand. I have looked for specs and info. Just smoke and mirrors.

Honestly, the electrics don't look all that bad. Just baffled what it failed. I knew not to push a cheap inverter.
 

Attachments

  • 20200515_001741 mod1.jpg
    20200515_001741 mod1.jpg
    278.7 KB · Views: 6

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700
not sure why a cheap inverter would have IGBTs? Does not seem likely to me.
Why not? IGBTs can have a much lower Vds at high currents than MOSFETs. I can't locate the exact datasheet, but one for a very similar component AP85G33W.

I have no interest in replacing with the same type of fets.
Assuming your inverter works from a 12 V input (why am I assuming this?), 1000 W is equivalent to 84 A input current. The two IGBTs are probably on/off 50 % of the time, which means the average current per IGBT is on the order of 42 A. Add losses for an efficiency of 90 % results in an average current of 47 A per IGBT. But the max. current is close to the limit of these transistors. Replacing them with ones with a higher rated current seems a good idea.
However, to minimize the risk of the replacements burning up equally soon, you'd have to find the root cause why these IGBTs shorted in the first place, especially as you didn't use the full power capacity of the inverter. These are my thought:
  • Check the attachment to the heat sink (if there is one at all). Are the IGBTs firmly mounted and thermal paste used to improve the ransfer of thermal energy to the heat sink?
  • Check the load circuit connected to the IGBTs. There should be a transformer, make sure its windings are ok, not short circuited.
  • Last not least, but hard to verify, if possibl echeck the timing of the gate drive. It should be non-overlapping to avoid simultaneous conduction of the IGBTs.
 

Martaine2005

May 12, 2015
4,932
Joined
May 12, 2015
Messages
4,932
There may be a perfectly good reason why it is branded as 1000W.
Yep, Cheap, typical chinese specs.
Seems you found a replacement. They are not expensive, they average at £1 each.
Between £8 and £12 for 10.

Martin
 

fastline

Aug 9, 2010
13
Joined
Aug 9, 2010
Messages
13
Martin, if I replace with the same devices, I can probably expect the same results? Run it at 10% capacity and expect it to fail in a few hours? I am still puzzled why it failed!

Harold, no, paste was not used, but they used a type of insulating pad as all fets are electrically isolated from the case. I do have a FLIR camera which may prove helpful if I replace and run it again.

I can only confirm there are no shorts between primary and secondary on the transformers. I do not have a way to check for winding shorts as they will always meter test as low ohms.

I hear what you are saying about the fet timing but I just wasn't aware of concerns beforehand. At some point, I can only waste so much time on it but hoping the FLIR camera will spot any big hotspot concerns if we fix it.

Any way to verify what we have as the components? IGBTs? Any way to know for sure or verify? IIRC, gate voltages get a bit higher, but IGBTs "should be more robust and reliable?
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
13,700
Joined
Nov 17, 2011
Messages
13,700

Martaine2005

May 12, 2015
4,932
Joined
May 12, 2015
Messages
4,932
if I replace with the same devices, I can probably expect the same results?
Maybe. But there is a chance that a defective device was an accident waiting to happen. It may prove a reasonable unit once repaired.

Martin
 
Top