Customer has a computer that will frequently restart for no reason.

At first thought, it sounds like a typical power supply problem.

But the problem happens after I install a new power supply.

OK, lets try pulling out the modem card, unplugging the HDD, and booting from the CD: problem happens less often… but it keeps happening…

So I unplug the CD drive (and reconnect the HDD), and after a few minutes, the PC freezes… hmmm, so far, its been sudden restarts, but no lockups.

I reboot again, and everything seems normal… but the problem seems be happening even less often now…

I try removing the RAM, and then installing it into a different socket: now the PC makes a loud beeeep.

Ah! this is familiar: faulty RAM, or faulty RAM socket, or just some “dust” preventing a good electrical contact.

I re-seat the RAM, and the PC starts.

Good, I insert my Mepis live CD, and run memtest.

In the past, memtest has found memory faults within 5 minutes (even though it doesn’t complete all its tests).

This time, memtest says everything is fine.

I’m starting to run out of options here.

I’m starting to think CPU or motherboard.

So I pull out the CPU, clean off the thick, hardened layer of thermal paste, and apply a new thin layer.

But that doesn’t help either 🙁

So I take the computer away for more thorough testing.

I give it to Andrew (and tell him I suspect a MOBO fault), and he swaps CPU and RAM, and eventually diagnoses a RAM fault (after a few hours of running memtest!).

Its the first time I’ve seen a RAM fault that has been particularly difficult to pinpoint.