Desktop: Shuts down and comes back on continuously

SgtBuck

Honorable
Nov 3, 2013
16
0
10,510
I've had this desktop for about a month now and I recently decided to switch the PSU from a Coolmax ZX Series 500W to a Thermaltake T2 600W PSU.

While using the T2, the computer overheated twice so I decided to switch back to the Coolmax. It's now been about two hours since the computer initially overheated and all I get now is my machine coming on for ten seconds, shutting down for three, on for 10, off for 3 over and over.

What should I do? Keep in mind: The computer ran flawlessly with the Coolmax PSU installed beforehand.
 
Did you remember to re-connect the fan?

Check all the connections, you may have inadvertently knocked something, and if you can get into BIOS, choose default settings.

Make sure that if there is a setting like 'Update ESCD', it is enabled.

And try again.

If push comes to shove, re-install the other PSU and lets find out why you are over-heating in the first place.

Does leaving the cover off your machine help?
 


I reconnected the plug for the heatsink just to be safe and re-checked the connecitons. Everything seems to be in place. The side cover on my machine is off. I still get the recurring on/off process and the machine doesn't stay on long enough to view the BIOS.
 
And does the machine behave in the same way with the other PSU installed?

Also, you might remove the CMOS battery and reset CMOS. This will eliminate BIOS from the suspect list.

Or not.

If you have a graphics card installed, I would remove it and temporarily use the on-board VGA if possible. If the machine starts, then I would re-install the graphics card.

Does your graphics card have a connector for an extra power supply? And is it connected? The motherboard may not be supplying enough power to the card.

Does the fan start at all? Try rotating the fins before you start it in case the fan has stuck.

I am sorry to say though, that if none of this causes any change at all, that this is starting to appear like a CPU problem.

If you can't get into BIOS, that usually means either you've 'bricked' your machine by incorrectly updating BIOS or that the processor isn't carrying out instructions as determined by BIOS/CMOS.

Which means, in my view, that either the BIOS chip has failed, which is highly unlikely and is something I've never come across, or it is the processor that has failed.
 


All I've tested is plugging the Thermaltake PSU back in and the same thing happens. The fan is loose and I can spin it with a finger. It doesn't spin up in the ten second window that the machine is on, though.

I still haven't tried resetting the CMOS or removing my graphics card. It's a GeForce 650Ti Boost. 500W is enough power to run it, so it shouldn't be a matter of power not being directed to the board.
 


What I meant was, is the six-pin power connector properly connected to the graphics card?

But do see if the machine starts with the card removed just to narrow the problem down further.
 


I mistakenly said "board" and not "card." Regardless, removing the six-pin and the card entirely doesn't fix the issue.


 
Can you say something about the two times your machine overheated?

I really am starting to think that your processor might have been saying, 'going, going, gone'.

You are using the on-board graphics, you have reset CMOS - I'm running out of ideas.

I'm thinking that perhaps there was a fan malfunction that caused the processor to overheat twice and on the third time, it gave in.

Is your system over-clocked?

Why do you think your machine overheated twice?

What happened to make you think that?
 


#1. I haven't reset the CMOS yet. I'll have to do that tomorrow after classes.

#2. I was running a game both times the machine overheated. First was Sniper Elite: Nazi Zombie Army, then Mount & Blade: Warband. The night before I was running Dark Souls, Saints Row III, and Payday 2 without a hitch.

#3. Nothing is being overclocked. Everything is, to my knowledge, running at stock speeds.

#4. This isn't the first time its overheated. Before it would overheat while running any game but turning on automatic fan control in my graphics card's settings seemed to fix it. Up until plugging in the 600W PSU, anyways.
 
Well, I have my fingers crossed for you dude.

All I can say for now is, I would configure my system to have as few components as possible connected to the mobo.

I'd reset CMOS as this will cause the defaults to be loaded which should rule out video card problems.

I would remove any cards that are installed, I would only have one stick of RAM inserted and I would swap the memory in and out to make sure that the memory is not causing the problem. (Try and restart the machine with one stick - no go? - remove the stick and put a different one in - any change?)

Even disconnect the disk drive for now to reduce the load on the mobo even further.

Also, I would try restarting without the system hard drive connected and then hope that I got an 'Operating system not found' message.

So basically, I would configure my system to be comprised of motherboard, CPU, one stick of RAM and a monitor that is connected to the on-board graphics port.

If there is no change after the system is 'pruned' to such an extent then you will have narrowed the problem down to either the motherboard or the processor.

The next thing I would do after that is to remove the CPU as well in order to see if the start-up cycle is changed at all.

If the system behaves differently then the CPU would be my main suspect. I would infer that it is the presence of the CPU that is responsible for the ten second - three second start up cycle.

If there is no change then the motherboard would be in the frame.

If I had come this far and still not got the system running then what I would do is:

1) Take the CPU to a computer repair shop.

2) Tell the repairman that you have this chip but you are not sure if it is working.

3) Tell him that you are putting a system together and would like to know if it is worth getting a motherboard that is compatible with that CPU and that if the CPU is working then you would like a motherboard for it.

4) If the engineer thinks he might sell you a motherboard, he might be willing to test the CPU and advise you on what to do without it costing you actual money.

5) If the CPU is faulty then you can discuss your options with the engineer and you can tell him 'you'll have to think about it since the cost of a motherboard and a CPU are a little more than you would like to spend today'.

6) Go home, go online and order either a new CPU or a new motherboard.

At least this way you don't end up buying a new CPU to install into a faulty motherboard or vice versa.
 


Tried removing the CPU. The machine starts up for maybe a second and immediately shuts down, probably because the CPU isn't recieving power and relaying data. Should I assume the CPU is the culprit by now?

Edit: Plugging the CPU back in gives me the same 10 on/3 off pattern. The processor itself doesn't look burned or damaged in any way. I'm starting to think it's the motherboard.

 
I'm thinking CPU unfortunately.

My logic is this:

With all the gear attached and with almost none of the gear attached, the system behaves the same way. The screen remains blank and after about ten seconds it restarts.

Is that right?

Nothing at all on the screen?

The only thing that changed the behaviour of the system was the removal of the CPU.

So I'm thinking that when the CPU is in, the motherboard 'allows' a period of time, ten seconds, for the CPU to present its 'ready' signals.

After ten seconds the motherboard decides there is no CPU present because the 'ready' signal didn't come and the machine restarts and repeats the cycle.

Remove the CPU, the board immediately detects that there is no CPU and shuts down.

It seems to me that the motherboard is doing what it is supposed to be doing.

The CPU appears not to be.

I'm sorry.
 
The screen remaining black suggests that the processor simply isn't running during those first ten seconds and it take the motherboard that long to notice.

Take the CPU out the boards says 'Oops! Nothing there' and shuts down.

Put the CPU in the boards says, 'Well, it's there but it won't wake up.

I think the motherboard stops because the CPU didn't start.
 


That makes sense to me. I'm still going to take my desktop to a local repair place to test the processor, just in case.

I guess this is what I get for buying a used processor, eh?

 


Yes, I would have the processor checked out first but don't beat yourself up over it.

And there is still and outside chance that it is the motherboard.

Hobson's choice right?

Sorry dude. I hope you fare better with the next one.
 


In the case that I do need a new processor, how much do you know about them in terms of performance? The one I had was an i5-2500k Quad Core with an LGA 1155 socket. Amazon has them for almost $300, which is too high for me.

The i5-3570k is a customer choice on Newegg. It's also a quad-core and has the same socket as the 2500k, plus it's only $219. That'll chew up most of the money I have left, but I really want a brand new processor so I can be sure that nothing's wrong with it.

Would the 3570k be a suitable replacement for the 2500k? I've got no intentions of overclocking my system.

 
Very nice but I am afraid I like a well stacked processor with a nice pair of double cores...

... and the 3770K has enough hyperthreads to weave sheets of silk.

Rrrrrr!

I'd save up the extra 100, 200 maybe, if I account for the cooling system that I would insist on.

But I'm a patient man. And the price might come down over the next month.