PSU Repair: A Case Study

Background Story On Today's Patient

I originally purchased this power supply in 2005 when my Pentium III's supply blew up from power getting restored after an outage. I re-used the SL300 when I built my Core 2 Duo-based PC in late 2008. I had to re-cap its main outputs due to the the famous “Asus Anti-Surge” warnings shutting down the system in early 2010. After that, all seemed good again until mid-2014.

What follows may sound familiar to many of you: in May 2014, my PC would no longer turn on and I could not identify anything obviously wrong with the supply at the time. All voltages looked fine during the paperclip test, but the PSU would not stay on when connected to my machine. LEDs momentarily flashed on, the fans lurched forward and I heard an irregular hissing sound similar to water running through a faucet's strainer from the supply, then nothing. I had not changed anything in that PC in over a year, ruling out hardware changes and shorts from moving things around as suspects. Disconnecting everything except the motherboard also made no difference, ruling out dead peripherals beyond the motherboard. I opened the supply up for a quick look, and all of its caps still looked good with no apparent damage. Since I had no time to mess around with it and no suitable parts for the caps I didn't change the first time around, I shelved the SL300, put in an Antec VP450 I had bought on sale for $30 and my Core 2-based machine sprang back to life, confirming that I had a power supply issue.

Parts swapping is a crude troubleshooting method, but I cannot argue with results. For most people, including myself at the time, swapping in a presumably known-good component is often the quickest and most cost-effective solution after the paperclip and multimeter tests have turned out inconclusive.

A few months later, I popped the SL300's cover again and replaced the remaining original output caps; this did not help at all. In fact, the 5VSB supply ended up rising higher after I replaced the worst-looking one, and the main outputs no longer stayed on. None of the 5VSB components looked fried and everything seemed otherwise fine as far as my multimeter could tell. I tried touching up solder joints in case there may have been a cold joint somewhere, but that made no difference either. So, I dumped it in my pile of projects for later.

At the tail end of the year, I got the idea of loading the 5VSB rail using a crude current sink circuit and cranking current up until voltage dropped to 5.2V. This did allow the power supply to power up predictably, albeit at the expense of 10W worth of dummy loading. I would not be comfortable with this as a permanent fix, assuming the PSU would survive it, but I decided to call it good enough for occasional workbench use until I could go back in for a more permanent solution. That's exactly what I am aiming to do here.

On a related note, as absurd as it may sound, someone still lists the SL300 as new on Amazon for $125 at the time of writing and others are listing used units starting from $70. From what I discovered along the way, I would not buy a used (or new) SL300, even for $40. Buy the VP450 instead.