A power supply in one of my homelab boxes died last week, and instead of binning it I opened it up, found the fault, and fixed it for the price of two capacitors. This is not a heroic story, the fault was about as common as faults get, but I want to write it down because the instinct to throw electronics away the moment they stop working is one I think we should resist when we reasonably can.
First, the caveat that matters, because this is the one part of the post I am not being wry about. A power supply contains capacitors that can hold a lethal charge after it is unplugged. The primary-side bulk capacitor in particular can sit at hundreds of volts for a long time. If you do not know how to discharge it safely and you do not have a way to confirm it is discharged, do not open one. There is no measurement inside worth your life, and I mean that plainly.
The symptom
The box would not power on. No fans, no post, nothing. The first job is always to work out whether it is the PSU or the thing it is feeding, and the cheap way to do that is the paperclip test: with the supply disconnected from the motherboard, bridge the green PS_ON wire to a black ground on the 24-pin connector and see if the fan spins. Mine did not, which pointed firmly at the supply rather than the board.
I confirmed it with a multimeter on the standby rail. The purple wire on the 24-pin should sit at around +5V whenever the supply has mains, whether or not it is switched on. Mine read nearly nothing. A dead standby rail is a classic, and it usually means the small standby section has failed, which usually means capacitors.
Opening it up
After unplugging it and leaving it to sit, I discharged the primary capacitor through a suitable resistor and confirmed it was at zero volts before I touched anything. Then I looked at the board, and the fault announced itself the way these faults love to: two electrolytic capacitors on the secondary side with domed tops where they should be flat, one of them weeping a little crust around the vent.
Bulging or venting electrolytics are the single most common failure I see in consumer power supplies of a certain age. They dry out, their capacitance falls, their equivalent series resistance climbs, and eventually the supply cannot hold a rail steady or cannot start at all. The "capacitor plague" of the 2000s made this notorious, but it never really went away for cheaper kit; heat and time do for them regardless.
I noted the values and voltage ratings printed on the side of each, two 1000µF 16V parts here, and ordered replacements rated for 105°C from a reputable supplier. The temperature rating matters: a 105°C part lasts far longer in the hot innards of a PSU than an 85°C one, and the price difference is pennies. I did not skimp.
The repair
Desoldering the old parts was the fiddliest bit, because PSU boards are often single-sided with thick traces and a lot of thermal mass, so the joints take real heat to flow. A higher iron temperature, a bit of fresh solder added to the old joint to help it melt, and some desoldering braid got them out without lifting a pad.
The golden rule with electrolytics is polarity. They are polarised, the stripe on the can marks the negative leg, and the board marks the negative pad too. Fit one backwards and it will, at best, fail again immediately and, at worst, vent loudly. I matched the stripe to the marking, double-checked, soldered them in, and trimmed the legs.
checklist before reassembly:
- bulk cap confirmed discharged (0V)
- replacements rated 105C, correct uF and >= original voltage
- polarity: stripe to marked negative pad, both caps
- joints shiny and fully wetted, no bridges
- no stray solder balls on the board
Reassembled, the paperclip test spun the fan, the standby rail read a healthy +5V again, and the box posted first time. Total cost was a couple of capacitors and about an hour, most of which was waiting for the iron and being careful.
Was it worth it?
Strictly on money, a new supply was not expensive, so the maths is not dramatic. But that is not really the accounting I care about. I now know exactly what is inside that PSU, I know it has fresh capacitors that will outlast the old ones, and I did not send a perfectly repairable lump of electronics to landfill over two components worth less than a coffee.
There is a quieter return too. Every repair like this makes the next one less intimidating. The first time you open a power supply it feels like defusing a bomb, and after a few it becomes a known job with known risks and a known method. That confidence is worth more than the tenner I saved, and it compounds.
I am not going to pretend everything is worth fixing. Sometimes the fault is a failed switching transistor and a controller chip you cannot source, and then the bin is the honest answer. But bulged capacitors on the secondary side are common, cheap, and well within reach of anyone with an iron and a healthy respect for the primary capacitor. Look before you bin. You might only need two parts and an hour, and you will learn something either way.