An interesting problem

May 02, 2013 23:30

Or at least to a nerd...

I've been having troubles with my main PC lately that I have really struggled to pin down. It started off with IO timeouts and generally poor performance to one of my hard drives, a Western Digital Green. I replaced the drive with another WD (A blue), but the new drive rapidly started having the same problem. I tried updating drivers, using old drivers, changing all manner of settings, disabling AHCI, etc... the problem only got worse, and for the last few days the BIOS had been reporting errors at boot time and didn't want to come online for long periods of time. I sent in for an RMA for that drive, and the BIOS wouldn't even see the new drive half the time. One of my other drives in the box always seemed to be thrashing as well (A Samsung EcoGreen), but it at least worked reliably, and I had no problems with my SSD boot drive.

So here I am, thinking that I have an issue with my mobo, and I start ordering a new mobo, CPU and RAM since mine are all 3+ years old. I wasn't happy about it, because this system otherwise works just fine and more than meets my needs. But something is nagging at my mind that this is not right. I also have a USB smart card reader that has started dropping out and coming back in, and I notice that the RMA drive is spinning up and then spinning down rapidly. The system randomly shut down at one point today, and I'm even having performance issues just moving between screens in the BIOS.

On a hunch, I dug out the old trusty multi-meter and hooked up to a 12v wire on a Molex plug, and what do I find? 10.5 volts! Yikes! The 5v wire was only getting about 4.5v. I took the whole PSU out, and jumpered it on just in case I had a really nasty device sucking all the power, and was rewarded with the hiss of a bad capacitor, and a 12v rail that still only puts out 11v, even completely unladen.

Suddenly it clicks! The drives that were unaffected or least impacted were low power drives (Green, EcoGreen and a solid state drive).

Time will tell if this really is the problem (new PSU will be here tomorrow), but this is probably the trickiest problem I've ever tried to nail down with a PC. It started in October or November or something, but wasn't really that bad until recently. Still, it's rare that a PC problem takes me more than a day to figure out.
Previous post
Up