[Pentium]

It was only a matter of time before this became a problem and now I'm here confirming it is indeed time to start considering performing preventative maintenance on your Cherokee power supplies. In fact, it looks like it should of started ten years ago.

Not long after the fall of Nekochan I was running my O2K 24/7 and one day I came into the shop and it was silent with a power fault on the MSC. When I cycled the circuit breaker it went completely dead and when I pulled the PSU out I had the face of an IRFPG40 MOSFET fall out. What caused you to blow up?
Well, I was so buried in other problems I ignored it and have not come back to it until now because these power supplies look like hell to service. As it turns out they aren't and I spent today figuring out what happened.

So if you take out all the flush screws and the two handle screws the shell of the PSU comes apart in three panels. You can remove the DC lugs by unscrewing them from the inside, then it's a few more screws to separate the DC secondary board. All the screws up to this point are grouped into two identical sizes. When unplugging the harnesses to the blukhead mark them so that you know where they all went during reassembly. Also note the black plastic baffle.






The lower board that is the primary AC/DC board comes out with similar screws and three very tall standoffs and a few more cables.




From here I could see that Q113 had grenaded and the 7A 250v fast blow fuse had vaporized. Because this is leaded solder a chisel tip iron and a soldapult makes easy work of pulling the MOSFET pair and heatsink out.





As of this post Vishay still sells the IRFPG40 and my suggestion is to replace them as a pair and do not forget to transfer the ferrite bead over when doing so. Keep in mind as well that these are $10cdn each so they are very expensive. This one went pop for a reason so we need to figure out why. The first thing we see is a trace that don't look too great and the legs of both MOSFETs look corroded. I got bad news for you.




The nearby Nichicon capacitors have all started to leak. The cap spooge made its way over to the lone trace and ate through it while the machine was running. This in effect caused the gating signal to the MOSFET to go open and fail and that caused the MOSFET to lock itself up in a state and go boom. In the process the avalanche failure was able to arc over the break and backfeed into the gate driver circuit, blowing open a 10 ohm 1/2w resistor before the rest of the short could be dumped to ground and blow the fuse. I've pulled the surrounding circuitry apart and it seems the damage ended here with a zener diode and the sacrificial resistor doing its job. So the first course of action will be a partial recap and component replacement before putting it back together, putting it back in the system and seeing if I can at least get to the PROM, then it's back apart again because I got more bad news.



Following with >>an older thread<< this would explain weirdness with phantom PSU faults. Power supplies will require a complete recap down to the control boards. There are over 70 capacitors in these power supplies. I do not have a capacitor BoM yet and no, I'm not looking forward to this.

Edited: Pictures of the trace curing cleanup. you can also see where ther was carbon tracking between Gate and Drain which I'm sure also would not of helped. The break happened right at the pad and could only be seen under a microscope. the repair laid a new piece of wire over the break before it was insulated with conformal coating.

[Raion]

I got three of these PSUs. Seems like I gotta crack one open and give it a once-over.