A cautionary tale


— 10:04 AM on October 7, 2010

Boy, Tuesday was a rough day. Walked into my office in the morning expecting to get some testing done, only to discover that my long-tenured file server was producing lots of fan noise. I pulled it open and found what I thought was the culprit: loads of dust and cobwebs accumulated over time. I shut down the system and spent the next 15-20 minutes cleaning it up, creating an awful mess as my compressed air can propelled dust into the atmosphere around me. Man, it was nasty.

Once I was finished, I reapplied thermal paste to the CPU and went to boot up the system, but it wasn't meant to be: the box came on, but wouldn't POST.

It was so unresponsive, even holding in the power button wouldn't cause it to turn off.

In fact, I think the mobo had been on the way out all along, which is why the fan speed control had gone haywire in the first place. Quick CPU and PSU swaps confirmed that the mobo was the likely culprit.

So my day of testing became something else entirely: a day of IT support work, swapping in a new motherboard and recovering all of the data and config files on this system. I tried installing a new mobo with a similar, newer AMD south bridge, hoping the WinXP installation on the system would still boot properly. (The storage picture was complicated by the fact that this box had dual 250GB RAID 1 mirrors in it.) I took extreme care to get all of the individual pins of the front panel and USB headers connected properly from this older case to the new motherboard, so the installation involved quite a bit of time and tedium.

At the end of it, though, WinXP wouldn't boot off of the new south bridge. I had to reinstall the OS on a fresh RAID mirror and then recover the data from the original boot array. Then came the work of restoring as much of the original software config as possible, including the FTP server that accepts nightly backups from TR's web servers.

At about 3:30PM, I had finished most of the hard parts and wanted to make one final tweak to the hardware build: the installation of a fan speed controller on the CPU cooler, which was a little louder than I liked.

Now, I work on PC hardware constantly, mostly on an open test bench, and doing so has caused me to develop some bad habits over time. I didn't even think twice about leaving the system running while installing the fan controller between the mobo and the CPU cooler. I just popped the three-pin connector off of the header while the thing was powered on. But then I realized I'd removed the system fan connector and went to reinstall it.

It was at this point that my bad habit caught up with me, as I missed the alignment by a pin and pushed the connector down over two of the three pins improperly.

Immediately, the mobo buzzed, there was a quick "pop," and an acrid smell filled the air. I had released the magic smoke. On subsequent boots, the mobo would POST, but not a single fan on any header would spin. I'd fried it.

Right about then, I had to leave the office for a prior commitment, and my work day was supposed to be done. Instead, I canceled my plans for later in the evening and came back to the task after dinner, swapping in another new mobo—whatever I could hunt down in the back room, though it was less ideal than the prior one—and reinstalling the OS yet again. At about 10PM, I realized I'd reached the same point in the process that I'd reached at 3:30PM with my first attempt.

My reward at the end of it all: a new system that pretty much plays the same role in the same way as the old system had for years before that, although this one may be a little louder, since the mobo's power draw is higher. I really like building my own PCs, but sometimes, the support of systems on which one relies for business is just, well, overhead. That overhead is increased if you decide to plug in fan headers while the box is running, folks.

   
Register
Tip: You can use the A/Z keys to walk threads.
View options

This discussion is now closed.