HP e-pc rejuiced

Brutal revival of a PC suffering from dried up capacitors

20:th of April 2007

A common problem with electronics exposed to high operating temperatures is that the electrolyte in its electrolytic capacitors eventually dry out. Another problem, that was especially common around 2002, was that electrolytic capacitors leaked. In my former job as a system administrator I ended up spending lots of time replacing motherboards with leaking caps (several times over for some PCs). This link has the full story about those. Both the dried up and leaking problem will change the capacitance and thus the properties of the capacitor, causing erratic behaviour. Leakage may also be harmful for the rest of the electronics on the board since the electrolyte may act as an electric conductor.

Anyhow, I got my hands on some old HP e-PCs of which some had typical dried-up-capacitor-symptoms. Since the e-pc is a really usable piece of equipment, due to its low noise and integrated soundcard/usb/ethernet/etc in a very small format, I set out on a mission to try to whip at least one of them back into shape by doing some serious recapping. This is my story, which I decided to contribute back to the internet community as a token of gratitude for the countless number of advice I've myself got over the years by nice people sharing information. The methods used here are actually quite brute-force, and I really don't recommend anyone else doing this, but hey, there are more fun things to spend time doing than repairing broken stuff the correct but time-consuming way.

The HP e-pc doing a full Monty

I connected everything needed to make up a complete PC on the desktop and observed how the thing behaved. It was instable with spontaneous resets, especially when doing something that required more power (such as having the CD-ROM motor spinning). The computer was also a bit hard to get started, on many power-up attempts it died again within a second. I checked the memory for errors but it was ok, so was the CD-ROM and hard drive. All these symptoms hinted that the caps were no longer able to cope with the power surges.

By inspecting the caps on the e-PCs motherboard I noticed that they weren't bulging so I pretty much could rule out leakage. The box being very compact would probably mean that the relative high temperature inside would have dried the caps up. But to make absolutely sure I didn't have to redo the job, I decided to also take care of any possible leaking that might have taken place.

before de-soldering anything I reverse engineered the circuit board's every electrolytic capacitor. I came up with the following values (NOTE! the schematic may contain errors).

I decided to try to replace only capacitances over 1000uF, these was also the ones closest to the warm CPU and would have been the ones taken the most punishment. In those cases where I couldn't buy the exact right component values I went for the next capacitance or voltage above the original one.

Unfortunately I don't have any proper equipment for doing de-soldering (such as a vacuum-soldering iron, and enough patience), so I never tried to get the holes cleared from the solder clogging them after the old caps was de-soldered. Trying to do this with my crude tools would probably had done more harm than good.

All sensitive engineers look away now! To eliminate any possible electrolyte lingering around, the board went into the sink for a rinse. I know, it's insane, but it actually removes any electrolytic residue.

The board was then hung to dry for a day or so above a radiator. In order to get the caps on in spite of the clogged holes I left the leads on the new caps quite long and soldered them on from the top of the motherboard. I did say I used brute-force, right?

The schematic below shows what was eventually replaced and soldered onto the board. I may as well say it out loud; I screwed up here. The green caps in the schematic have too low power rating, they should be 25V, not 10. In my infinite laziness I never got around to fixing this, so I'm still using them (fingers crossed).

I reassembled the whole shebang and said a quiet prayer that it would boot.

It DID boot and it DID sustain 3 days of continuous burn-in without any malfunction whatsoever. I've now been using it for a couple of weeks and it's working perfect !

Thanks for reading !

Peter Sandström

[Back] [Home]