Super User is a question and answer site for computer enthusiasts and power users. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Me and my brother purchased and built a custom budget computer for our little brother for Christmas. It has been running fine for about a week, but last night it would no longer start. When the power button is pressed, the PC's fans (including the CPU fan) start up for about half a second before the PC shuts down again. Unless the power cable is removed, the PC will keep trying to start and die every ~3 seconds.

I have tried removing and reattaching the power cords (both the 24 pin for the motherboard and the 8 pin for the CPU), removing and reinserting the RAM chips and disconnecting the SSD. The PC still won't start, so I've got to assume that one of the components have died, I'm guessing the PSU.

But how do I figure out for sure which component is faulty? As far as I know, the only candidates are the PSU, the mother board, the CPU and the RAM.

Update: I have left every PSU cable connected to the PC, but removed the 24 pin from the motherboard and used it to jump start the PSU. That works! All the fans are spinning and the PSU is showing no sign of the behavior I'm experiencing when the 24 pin is connected to the motherboard. When I reconnect the 24 pin to the motherboard, the behavior I described resumes. Does that rule out the PSU being a faulty component, or could it be that the PSU is suffering from severely reduced efficiency and chokes when I connect it to the motherboard? (It's a 650W power supply connected to nothing but a mini-ATX motherboard with an i5 CPU and 4 fans)

Update: Also, I forgot to mention, the green lamp on the motherboard is shining happily when the power is connected. I don't know if that's relevant to the problem in any way.

share|improve this question
might help to mention what the parts are, and if there's anything at all, lights for example that might give a clue. Lots of newer boards enchew the traditional speaker for other means, but still have a speaker header - which might help if you connected something to it – Journeyman Geek Dec 31 '12 at 9:11
The description to me looks like a short-circuit somewhere. PSU starts, detects overload and shuts down. After that it waits a while and tries to start again. Check if any solder joints are touching the case or if there are any loose components in the computer. – AndrejaKo Dec 31 '12 at 10:21
@AndrejaKo: I have completely dismantled the computer and put it back together, now only with the motherboard and the CPU. No RAM, no SSD, no fans, and the issue still remains. I'm going to try with a different PSU later. – Hubro Dec 31 '12 at 23:14
@AndrejaKo: Please see my question update. I can't rule out short-circuiting, but I can't figure out a way to rule it out either. No solder joints are touching the case and no lose screws are rolling around below the motherboard. Additionally, every component has been detached and reattached at least twice since the problem started happening. Thanks for your time – Hubro Jan 2 '13 at 18:00
@Codemonkey - Sounds like the motherboard should be replaced instead. – Ramhound Jan 2 '13 at 18:08

Well using your ideas of the faulty component...

If you can get to your BIOS this immediately rolls out the CPU, PSU, and the Motherboard, leaving the RAM.

You can try resetting the CMOS chip by looking for the yellow jumper on our board and switching it over. It could just be a faulty BIOS. (This has happened to me before)

If you have access to another desktop, try switching the PSU and see if it works on the alternate machine. If not, It's the PSU.

Look for any damage to the motherboard. Otherwise you can't differentiate between the CPU and the motherboard.

share|improve this answer

Honestly, my instinct is that your CPU is overheating. This can even be caused by the thermal paste being too thick. I have had this happen to myself when I was the one who attached the heatsink to the CPU. In my case, I just had to wipe off the paste and do the process again, but with a much much thinner layer of paste.

share|improve this answer
Can the CPU overheat that quickly? The power is not even on for half a second before it shuts back down. Remember that the PC was in use for a week before this problem started happening, and the temperatures were very low. – Hubro Jan 2 '13 at 18:43
There's no chance that the heatsink was detached or anything subsequent to the week of good operation? I'm not sure of all of the material properties involved, but I also kinda wonder if the paste could form bubbles or leak out over time to create the bad situation you now have. – Dane Jan 2 '13 at 20:34
Anyway, do you have a second known-good PC with a similar socket for the CPU? If you do, swap the CPU from the bad PC into the good one and see if it boots. If the CPU works in the good PC but not the bad one, I'd say you're down to the motherboard. Bad RAM should show up during the RAM check and shouldn't reboot you beforehand. – Dane Jan 2 '13 at 20:38
I'm assuming from your updates that the PSU is good. While you might have an esoteric problem with the PSU, the behavior you're describing from a newly purchased PSU makes that one of the less likely scenarios. Of course, if you have the known-good system around, you can always do a brain-transplant and test the PSU with the known-good system. – Dane Jan 2 '13 at 20:43
I'm sorry, I just can't possibly believe that the CPU is suddenly overheating in about 0.3 seconds the instance I power the PC on, out of nowhere. The CPU heat sink is screwed tightly to the motherboard, and even without a heat sink it would take at least a few seconds for the CPU to overheat. I don't have any other systems I can test an i5 processor in. I agree it can't possibly be the RAM either. As I've updated in my question, the PSU is working fine, so the only remaining hypotheses I have is that either the CPU or motherboard is faulty and needs to be replaced. – Hubro Jan 2 '13 at 22:14

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .