Power Failure...

A few weeks ago I was experiencing major graphical glitching/artifacting. I went through and physically cleaned out the tower, removing and fitting the parts back in. I put a clean windows XP install on and everything has been keen for those few weeks, no more problems playing games, etc...
Suddenly I started getting an issue where games would just lock up, and in some instances the entire screen would go black with the machine still running but no video coming out. That has progressed into an entire power failure, so far only while running games. System and CPU temps are within safe parameters, but I CAN'T find a way to check the temp on my ATI radeon 9800Pro 256mb video.
Following the power failure the only way I can get the machine to boot again is by pulling the power from the wall -> powersupply AND the powersupply -> motherboard. once they are placed back in the machine will boot, no problem.
Again, I can run the machine as long as I like until I throw open a game, then I've got a 15-20 minute window before it simply shuts down. I've had no problems with temp for the past year and a half, and have even upped the # of fans: 2 x 120mm, 2 x 80mm, a pretty heavy cpu fan/heat sync, and finally a small copper Vantec fan (2500rpm) to replace the shoddy one that came on the video card.

My question is what is the most likely component that is failing out on me?

System: 1.5 gigs kingston ddr 3200
nvidia nforce 2 an35n 400 shuttle mb
allied 400W power
ATI radeon 9800Pro 256mb video
AMD barton 2800

 

My first hunch is that your psu is overheating and tripping a safety switch,thats a really cheap psu for a 2800 barton and radeon 9800 just did a quick search I get rough price of around around $25 Its usually best not to get a psu below $50 unless its a really low end web browsing system-not a gaming machine,I'd change that psu out just for peice of mind

If you havnt got a temp readout in your video card properties your card probably hasnt got a temp sensor theres no way to check it,ATM it does look like your video card or psu is overheating,does it do this with all games? Try running 3dmark on loop see how long your computer lasts,pay specific attention to the graphics,if you start to notice strange pixels,rainbow colours,loss of colour ect its a sure sign of your video card overheating,also feel how warm your psu is getting during the test,it should feel just slightly warm but not hot,have a sniff see if it smells like its cooking

http://majorgeeks.com/download.php?det=1712

Also go to control panel,system,advanced,startup and recovery-settings,uncheck auto restart see if you get a blue screen,Also go to control panel admin tools,event viewer check system,application for any errors around the time of the crashes,I doubt this will be of any help though a consistent crash every 20mins sounds like overheating

Double check you have fitted the the fan to the video card right and the design of the new fan will work with the heatsink arrangement and the fan is spinning in the right direction,if you have your old fan try replacing it see if the symptoms subside,have you removed the heatsink at all?It might be a good idea to reinstall the heatsink with some fresh thermal paste,the contact with the core could have jarred when you replaced the fan,could you post a pic of the new fan you have installed or is it a new heatsink\fan combo