Sudden PSU Failure

Associate
Joined
3 Feb 2009
Posts
942
So like the title says, my PSU failed, fairly suddenly. The problem is that it's not a complete failure. The PC still boots, gets into the OS, runs a CPU stress test (Prime95 8 threads) and runs various other tests with no problem, but it looks like the 12V rail is running at around 3-3.5V.

I have run an OCCT PSU test which resulted in a complete lockup of the system which obviously points to a PSU failure, my question is; why has this happened?

I have been running the same hardware setup for a year now and nothing has changed. The PSU i'm using is a Corsair TX750W and I got it on recommendation of the members of this forum (obviously I don't blame them but usually when people trust a brand there's a reason for it) and I thought 750 would be plenty for what i'm running. I have an i7 920 runnin at 3.6GHz (has been that way since I got it), a BFG 295GTX and 3 HDD's.

If anyone has any suggestions as to why the PSU would have failed like this so suddenly I would greatly appreciate it. At this point I want to try and figure out if it's worth RMA'ing or just replacing with something even more beefy.
 
If the PC boots, and plays fine then there is most likely nothing wrong with your PSU. Just a reading error somewhere.

If your 12v rail was putting out 3.5v you would know about it.
 
If that's the case then can you offer an explanation as to why the pc locks up completely on the OCCT PSU test, and why online games have very odd connection issues, but nothing apparently wrong with the game itself?

Also, I spoke to someone at OCUK who, when I told them about the OCCT test results, immediately agreed that the PSU was kaput.
 
I think your PSU is likely to be fine, otherwise the computer wouldn't have booted up at all. The Corsair TX 750W is a very good PSU and is more than capable of powering your components. I would take out the PSU, connect it to a case fan to put a little load on it, and test it using a paper clip. I would measure the voltages with a multimeter. If you don't have one, I recommend buying one as you can get one cheap. After getting some advice from people here, I bought one a couple of weeks ago to test my Corsair HX 520W, and the correct readings I got made me put any doubts I had aside :)
 
I'm going to get hold of a multimeter tomorrow to check for certain but I haven't actually seen any reasoning as to why it wouldn't be the PSU other than 'I don't think it is because your pc boots'.

As I said before an OCUK employee seemed to think that after the tests I ran it was a PSU issue and said RMA it straight away.
 
I suspect either the cooling of the CPU isn't as effective as it once was, or the voltage regulation isn't as good as it was, or the CPU itself has degraded a little over time.
If you want to check with a multimeter, do it with the PC on and running prime95 so you can see what it's doing under proper load. Don't bother disconnecting it and turning it on with a paperclip.
 
Not even sure I'd buy another corsair PSU.

Mine went pop a week ago and I didn't realise when you RMA their stuff you have to send it to the netherlands. Well that is if you have had it over a year.

Luckily I had an old tagan psu so at least I can still use my pc.
 
Voltages will be same the under load as they are idle.

Not so much. Voltage will go down with load.

Your 12V rail isn't at 3V, the computer wouldn't boot if it was. However the psu may well be faulty and the software making things up.

Put your processor back to stock settings and see if it still crashes. If it does, then try with a different graphics card (295gtx draws a lot of current).
 
No, they won't be the same under load, especially if it is a failing PSU.

ok maybe i mis-worded what i meant..

They key is to have stable voltages within an allowable tolerance regardless of the load but allow for some deviation of around 5% either side of the expect rails. So for 5volts you need to allow between +4.8volts and +5.3volts and for 12volts between 11.4volts and +12.6volts.

Anything outside this range is certain to be a problem with the power supply.
 
Back
Top Bottom