How do you know a PSU is (or is going) bad

[Illyest]

How do you know a PSU is (or is going) bad?

I have a Silverstone ST85F. It's been running for months even though this thread http://www.ocforums.com/showthread.php?t=496655 had me a little concerned. My rig has never experienced any low power message or anything though.

What's wrong is occasionally while just browsing My Computer my rig will just shutoff... No BSOD or anything. Initially I thought I was getting "dirty power" so I bought a Backup UPS VA1500LCD (awesome investment). Unfortunately that didn't fix the problem.

At first my rig would shut off and all I had to do was press the power button to turn it back on. Nowadays though, "most" of the time when I hit the power button the blue lights on my 680i will just "flicker" and the board wont POST; I have to toggle the power switch on the back of the PSU to fix it.

I dont think temps are a problem, I don't use a thermal diode or anything but my rig and room are never "hot". My CPU is usually 34c idle, 60c Prime95 load. Load in games are around 47-55c depending on the game. The 8800GTX is 50c idle 83c "fur" load and around 64-76c in games. Whats strange though is the random shutoff doesnt happen in games... Like I elluded earlier. I will open a folder on my NAS or another folder then the computer will "freeze" a couple secs then shutoff... Completely random...

It is completely obvious my PSU is going bad or what? I plan on doing a mass component swap in the next couple weeks. However I'm starting to think the $1500 I've already "invested" should have included a new PSU...

Thanks in advance.

Edit: Forgot to mention I checked to make sure my mobo wasnt shorting out on the case

 

[Super Nade]

Well, if you don't have dedicated test equipment, component swap would be the fastest way to check. In your case, I'd take a hard look at the Mobo and RAM.

 

[Illyest]

Well, if you don't have dedicated test equipment, component swap would be the fastest way to check. In your case, I'd take a hard look at the Mobo and RAM.

Well what do you know. As I alluded earlier I had already invested in some new components; one investment being 2x2GB OCZ Reapers. I had ran Memtest86+ on my Crucial PC28500 Tracers a few weeks ago, 2 sticks were bad the other 2 were good, so I thought.

Without running Memtest86+ again I just popped the 2 remaining sticks of Crucial out and threw in the OCZ...BAMM...works like a charm, It seems anyways.

As a bonus, in addition to the random shutoffs I was getting the dreaded nvlddmkm.sys error again (thought it was SP1 and new driver related), the new ram fixed that too...

That makes 6 sticks of Crucial Ballistix Tracers that have died on me, and I was running them all STOCK!!!! First time and last time I've ever bought or ever will buy Crucial


Thanks Nade

 

[Super Nade]

No problem. I'm running the old Ballistix PC6400 with no problems. Crucial seems to be a hit and miss. Try upping the Voltage to 2.2V and see what happens.

 

[PorPorMe]

This might be off base-I'm no tech and what I know about memory is diddly.
That being said, I had the same happen to me in almost the same situation. I had 1gb pc2100 and needed more for Vista. I ordered a 1gb pc2100 from 1st memory.
I originally had 3 sticks:
512mb pc2100 1st bank
256mb pc2100 2nd bank
256mb pc2100 3rd bank
I received the new stick and installed it in 1st bank-then the 512mb-then a 256mb. It booted fine and ran for a couple hrs and died. Like you said-just dead, but it booted rt back up. then later did it again. and again. Sometimes it wouldn't power up at all for a few, or maybe boot just to loading CMOS and halt. Got me a bit nervous. Ran memtest and passed all tests. Problem didn't go away. I started to do different mem configurations thinking there might be a conflict. I even pulled the new stick and went back to the original 3. That didn't even work 100%. During all this, I noticed that in some situations, the problem would be either be better or it could be worse. As I said, even the original wasn't good and there was the clue for me. If the original was doing it too-it could not be the memory itself. But it had to have something to do with the memory.
I finally figured out it was the seating of the chips. If they didn't get seated just right-somewhere along the line-they would miss a signal.
I made a VERYconcerned effort to make sure all 3 seated solid w/out a bit of rocking and I haven't had a problem since.
This is a bit long winded but i hope it helps.