Power Supply Resurrection

[An_Onion]

For the last month or so I've been working on resurrecting an older APC power supply for use with some older computers. I started by replacing all of the electrolytic capacitors, because they were quite old and several of them were swollen. With those replaced I figured I couldn't fry too much by powering it on. It would try to start, but one of the relays wouldn't stay on. I figured out which one it was, and there were two identical ones on the board. I switched them and the problem moved with the relay. I figured just replace them both, since they were both old. Now the thing would start up, but there was a lot of noise from the (lonely, unheatsinked) 317T voltage regulator. So I replaced that and added a heatsink to the new one.

Hopefully that all made sense, and now you're caught up. Now the darn thing will start up, but will cut out and restart after it's been running for 15-30 minutes. The onboard software doesn't record the crash or whatever it is as a power failure, and does record if I pull the plug. I checked all of the film capacitors (out of circuit), since those are often iffy on old stuff, but they all test perfect. Next on my hit list is all of the MOSFETs. The main ones are all ST, marked "P80NF 55@ / GK03X V6 / CHN 828" I haven't been able to find the exact data sheet for that one, but I think this one is probably close enough. I've tested the forward and reverse voltage drop across gate and drain with the FET on and off and resistance across gate and drain with it in and off with my multimeter. On one of them I get a lot higher voltage drop across when the FET is on (2x the average of the others) and it has higher resistance when it's on (again, 2x). I know V=IR, so that makes sense.

My intuition tells me that the outlier is probably bad. Since the thing runs for a while, maybe it's getting hot. I know Joule heating, our ancient enemy, is P=I²R, and with higher resistance it's getting hot. But it's got a MASSIVE block of aluminum to sink the heat away. So maybe not. The Al block never gets very warm.

I know the test conditions on the datasheet for the MOSFET are not what my pos multimeter can put on it, but it's the best I can do with what I have. What do you guys think? Is that MOSFET probably the culprit? Or is there another more failure-prone part I'm overlooking? I know I could easily replace the whole thing, but I'm using this to learn. Plus, it's a LOT more alive than it was when I started, so I feel like I'm making progress. :V


I'll include a pic of the board below.

Test equipment I have:

• Klein MM400 multimeter
• an old analog 20MHz scope I picked up for a song
• A few DC power supplies
• Basic soldering tools and skills
• one lonely brain cell rolling around in my head like a ball bearing in an empty paint can