It's that time again when another peice of my IT equipment has died. When I say died it's more akin to requiring the replacement of several major organs at the same time. Thurday night I was going about my usual activities of checking email and surfing the web with my windows desktop and 24/7 linux server.
I have been using the linux box alot recently due it collecting my rss feeds for me, the windows box being used for emails and games. The screen for the windows box had gone dark, not a big deal the screen saver kicks in if it's not being used. The discovery of a problem came when it didn't wake up when I moved the cursor to that screen (I use synergy to share kb&mouse across 2 machines). Ok not a huge problem, sometimes an error occurs with the kb/mouse sharing so I used my Belkin Switch2 to connect directly to that machine, still no response.
By now it was getting a little worrying, it was going to be a more significant problem. Next on the list was the TFT backlight, I had 2 identical monitors and I had experience backlight problems on one in the past. Backlight was fine, cables connected ok. A reboot indicated a worrying development, I didnt even get a post screen from the bios. Time to get inside the case.
With the side off the dusty interior was exposed; not overly dusty, the intakes are filtered, but enough dust that once the problem was sorted should be clean out. The first thing was to check the gfx card, if nothing is appearing on screen its the logical step. It was seated correctly in the motherboard, powerconnector in place and fan spinning. Swapping the monitor cables didn't improve matters.
Then I spotted the problem, a small cheap looking fan on the motherboard wasn't rotating, maybe it was clogged with dust and system was protecting itself as it could detect it wasn't spinning. A prod with a screwdriver immediately let me know dust wasn't the issue. The fan had totally seized, it wasn't going to move again. How do you get around a problem like this?
Chipset fans are not always essential things, they are there to improve the heat dissipation of some chips. In my case the chip underneath was the primary bit of the nvidia chipset rather than someting like a voltage regulator. The chip had gone to silicon heaven all because of a cheap fan. I select high quality components for my home built systems, I even make sure that my motherboards have 100% high quality japanese capacitors to guarentee a good lifetime. What you can't check in advance is the quality of the fans.
I removed the motherboard from the case (getting covered in dust in the process) and examined the fan. It must have been tilting about 5 degrees of its axis, it was gone alright. I removed it from the motherboard completely and disassembled it as much as possible. The failure seems to have been caused by the central post around which the fan spun having melted. Now given that this was a chipset fan designed to be working in a computer case I would expect the plastic to be stable at around 50-60 degrees Celcius. Therefore it must have been hotter than that to melt enough to cause the fan to fail. Over 60 degrees for an extended period would have killed off the chip before too long.
So thats all there was to it, a crap fan causing a chip to fry. You can bet the replacement motherboard is fanless.