One of my test power supplies in a rather large switching unit from Power One in the SPM5 series. It's basically a mainframe of sorts with power factor correction, timing and a fan on the backplane and 5 slots for various factory installed output modules. Unit is rated 1500 watts. Date code on the outside is 1990.
unit is pretty much this thing with the small slot having a different card
Mine has two double wide 24 volts 32 amp modules in parallel and the fifth slot filled with a 12 volt 20 amp module. Very handy thing to have around.
The fan started to go crazy as in slowing down and speeding up, stopping and stuff like that, then I noticed the outputs all lost regulation, 12 volts output dropped to about 9 under load and other weird stuff.
About two dozen screws later, the thing opened up and I got the modules out. Fuses seemed ok, and the DC bus had 320 volts on it (sales sheets say300 volts, but whatever). There's some sort of SCR based active power factor correction trapped under heatsinks, thermal grease and dust. At this point the 24 volt cooling fan was still erratic then basically made sad noises maybe twice a second as it tried to start but didn't have enough voltage. It could not bootstrap or regulate and was restarting. Amazingly the main board only has some 1980s type 8 pin SMPS regulator, not a crazy module thats impossible to work with. Swapped that, no dice. No burned parts, but lots of discolored under the solder mask traces.
To complicate things the control board where the fan attaches was covered in something greasy and apparently corrosive as seen from the blue colors on some leads. Parts cleaner and a brush got that off, mostly.
Probing around didn't show much other than the 24 volts for the fan was pulsating at under 2 volts. It also looked like each module slot has a dedicated winding off the same ferrite transformer to supply isolated power. Those were also way too low. Voltage to one 7805 regulator was zero. It appears that each module gets 320 volts, something around 24 volts and a timing signal and from that they all do their own thing.
Finally found a shorted diode, but the markings were worn off from the cleaning process and whatever. The other similar parts were MUR something and from motorola. I just went ahead and stole one from a backplane slot what would never have a module card edge attached. All the caps on these outputs were rated the same 35V, so it seemed safe enough. I've also noticed the American power supply companies like to keep the part type count low so many parts appear to be grossly overrated, but it keeps assembly simple. Turns out the diode was probably just a MUR115, 150V 1A fast recovery rectifier, judging from the others.
Powered up the chassis and it worked fine. Closer inspection showed one contact for a module slot was broken off from corrosion. Now I have to transplant a heavily glued down card edge connector from the slot I disabled to the slots that does get a card.
Lazy, but it should be OK in the end. Replacement power supplies of this type range from about $150 to $2000 on ebay. Not sure who's paying list prices from decades ago for this stuff.