Sunday, 13 June 2010

SG45H7 power supply failure

A few weeks ago my shuttle box stopped booting. For the record, it's a one year old core2quad box with 8 gigabytes of memory and 1.5 Terabytes of hard disk. All packed in a little box that is no wider than an A4 sheet of paper.

When I hit the power button the fans would start rotating a little bit, maybe 25° and then nothing. The LEDs of the motherboards stay lit but nothing happens.

I opened the box, wiped the dust from the fans, measured the voltage and current coming out from the pins of the power supply. Both current and voltage were a bit lower than expected, but to my surprise, the box did boot again. I made sure my hourly incremental backup (rsync powered!) was doing OK, and I went back to hacking G++ as usual.

Then last Tuesday the box stopped working again. This time I removed most of the parts from the motherboard: Hard disk, DVD burner, memory sticks one by one. Each time I remove of a part, I'd try to power the box on. Still nothing.

At that point, I thought it was either the mother board or the power supply that was at fault. But how would I know? I bet it was the power supply as it was the less expensive part to replace :) So I did the "power supply paper clip test" and it appeared that the odds of the power supply unit being broken were quite high.

I figured I should probably order a more powerful power supply as the 300 Watts of the current one were probably a bit too tight for my usage.
I ordered a 500 Watts PSU (power supply unit) and got it two days later. And yes, the PSU was indeed the culprit!

Let's see how long is this new PSU going to serve.