Hi all. Having a very frustrating time with my home desktop, particularly annoying since I use it to work from home. The problem is this…

.

Recently the computer has not been starting correctly. I press the power switch, and one of three things will happen.

1. The fans and lights come on for a few seconds, then everything turns off again. After a few secons, the fan and lights come on again etc. The cycle repeats until either 2) or 3). happens.

2. The fans and lights turn on and stay on. The fans stay on at max power, whereas normally they would power down after a few seconds, and nothing is displayed on screen. At this point I have to shut the computer down again.

3. The computer starts up succesfully. This is the rarest scenario, but when it does happen the computer appears to operate faultlessly.

When the computer does eventually run, the opening screen suggests that some kind of overclocking fault has occured and offers to let me change my BIOS settings. This happens despite me never overclocking the machine.

.

I have already tried the following:

a) Removed all dust from the insides of the computer (this was causing problems previously, and I wonder if this may have caused longer term damage in the past).

b) Removed memory sticks alternately to see if one of them is faulty.

c) Sprayed contact cleaner into the off/on switches. Not sure how much contact cleaner actually reached the contacts.

.

My impressions so far:

I don't think it's the graphics card, because I would expect the computer to start but simply not to be able to see anything, whereas the computer does not start at all.

I suspect it is either power related, dust damage(?) or a dodgy connection…

.

So… sorry for the long winded post but if anyone has any diagnostic help I would be EXTREMELY grateful!

I would go for the power supply. You could get a multimeter and see if it's actually kicking out the right voltages on startup – the added current draw of starting every motor in the PC can drag it down, which results in reluctant starting.

Useful info. Luckily my father is visiting tomorrow, so I can get him to bring his.

A bit naive, but I have never got to grips with a PC PSU. Do they output a single voltage, and if so what is it supposed to be? 12V?

Also, it was a cheap PC brand so the components are unlikely to be the highest quality…

Sounds to me like the processor is overheating or possibly reporting a higher temperature than the safe limit is set to.

You could try cleaning and re-applying a new layer of thermal paste to see if that improves things.

Remove the heatsink and fan assembly, clean off any residue from the thermal paste or pad using a lint free cloth and isopropyl alcohol (nail varnish remover will do if you have any 😉 )

Remove the processor and clean the die as above.

Apply a new layer of thermal paste and put it all back together.

edit : you could also try setting any clock speed related stuff to "auto" in your bios.

A bit naive, but I have never got to grips with a PC PSU. Do they output a single voltage, and if so what is it supposed to be? 12V?

There are a variety of 12V and 5V supplies, I believe. Google for the pinout of the powersupply connector – you'll need to test it under load.

It's a bugger of an issue, this. It could be a fault on the power supply, the motherboard, or a CPU / heat issue as Haze suggests. You could try disconnecting everything non-essential (ie, all the drives) and see if it's happier – if it boots reliably, that would point towards the PSU. In honesty though, the only way you're going to diagnose it reliably is to start swapping bits out to see if it stablises. Maybe borrow a PSU from a doner PC if possible?

There are a variety of 12V and 5V supplies, I believe. Google for the pinout of the powersupply connector – you'll need to test it under load.

+1 to this; the PSU supplies -12v, +12v and various others inbetween. A modern (ATX 2.x) system will primarily use the 12v rails and step-down accordingly; on older (ATX 1.x) systems, the 5v and 3.3v rails are used a lot more instead. The cutoff is generally sorta mid-Pentium 4 days.

Donor, not doner. A PSU out of a kebab probably won't work.

Oh,

It's also worth disconnecting / reconnecting the power connectors to the motherboard (including secondary connectors to the CPU and graphics card where appropriate).

I had a PC on the bench the other day with the same symptoms, rigged up a new PSU and it worked, refitted the old one and it carried on working and has been fine since. I can only assume that it was a slightly loose connection somewhere.

This is all very useful stuff, thank you very much! I think I will try disconnecting/reconnecting all power connections first, then try disconnecting non-essential power connections. This might isolate whether it's the PSU. If it doesn't seem like the PSU I might have a look at the processor…

Thanks once again.

PSU or memory is my guess but you've tried memory it would seem unless it's all faulty but that's exactly what mine does with dodgy memory.

The Overclocking message is because it thinks there's a fault with that because it switching off immediately it just needs told to carry on.

Doubt it's overheating given you describe it as being off and it's when you power it up add the fact it can run all day with no worries once started.

The Overclocking message is because it thinks there's a fault with that because it switching off immediately it just needs told to carry on.

Without the exact message this is a guess, but on some OC-friendly motherboards, you can end up rendering the system inoperable by overclocking it too hard. To reset it (again, on some boards) you can repeatedly power it on and off in succession several times, which tells it to reset to 'safe' values. It's possible that this is what's happening in this case, it's interpreting the failed boots as an attempt to reset OC'ed settings.

Aye that's exactly it cougar but mine will stay at the message unless I go into bios. I don't change anything just save the settings and restart.

*nods* Not saying that's the cause of the problem, just the cause of the OC message.

I understand. That would point the finger more strongly at the PSU then.

What brand / age is it by the way? Before you go swapping bits, in the days of yore Dell and a few others used PSUs and Motherboards with exactly the same connector as the ATX standard, but with all the pins jumbled round.

I suggest a quick google may be in order to check it's not non-standard.

Bought about 18 months ago, from Eclipse. Not sure what brand the PSU/motherboard are, but I suspect something cheap. Dual core Intel processor (can't remember which one).

Probably fine then. It's obviously a big faff to make a non-standard part, and even Dell have given up with it in the last few years.

Hmm, sounds like your PC gone mad

Doubt it's overheating given you describe it as being off and it's when you power it up add the fact it can run all day with no worries once started

Very true, could always clean the heatsink and die anyway to rule it out.

It'll only cost the price of a tube of paste.

If the motherboard emits a set of "beeps" on start up that means it's detected a fault and the sequence of beeps is a code which identifies the problem. Google for a list of codes.

Ok, after a Sunday afternoon spent ignoring my wife, with my head buried in my PC case, I am about ready to admit defeat.

I have now tried:

1. Adding/removing memory cards. Fault seems to occur with any combination.
2. Unplugging disk drives to reduce power load. Has no effect/fault still occurs.
3. Unplugged every power connection (apart from some very small fiddly cables, the purpose i’m not sure of) and reconnected firmly. No effect.
4. Removed the CPU heat sink and cleaned thoroughly of dust (may need to apply more thermal paste?). No effect.
5. Removed the graphics card and cleaned fan/heatsink thoroughly of dust. No effect.

I will apply some more thermal paste to the CPU, although I am skeptical that this is the issue.

If this doesn’t work I think I might have to take the PC to a commercial computer repair centre/shop (recommendations?).

Otherwise I could buy a new PSU on the off chance, but it seems a bit risky?

Anyway, thank you all for your help it was very useful in bringing me this far!

Oh, and one final piece of bizarre behaviour that may help with diagnosis is that up until the computer failed, it would ocassionaly freeze if you knocked the case (still outputting to the screen), requiring a reboot.

Also, no beeps on startup.

Thanks again!

If you’re anywhere near me geographically (East Lancs), I’d be happy to take a look at it for you.

Did you manage to measure the PSU voltages?

At the last place I worked, our tame electronics guru took a bog standard PC PSU to pieces. He was trying to figure out how they could build them with such cheap filtering capacitors.

Well it turns out that the capacitors are massively under-rated, with far too much current going through them, and should only be expected to last a few months of continuous use.

OK, long shot but I’ve seen it before – check the plastic clip that the heatsink attaches to is undamaged.

I once saw one where the heatsink was only fastened down properly on one side due to a broken clip, cue random reboots etc.

Might explain the “freeze if you knocked the case” symptom?

and should only be expected to last a few months of continuous use.

PCs generally aren’t designed for continuous use, that’s what servers are for. There’s an expectation that they get switched off (or at least, enter standby / power-saving) occasionally.

Might explain the “freeze if you knocked the case” symptom?

To be honest, pretty much any loose connection could explain this, along with simply a particularly sensitive hard disk.

Yeah like I said though, long shot!

Thanks for the offer Cougar, but i’m down South!

Unfortunately I didn’t manage to measure the PSU voltages; I couldn’t get hold of a multimeter at the weekend. It’s a bit annoying as it would certainly help determine whether the PSU is a problem – i’m not sure that buying a new one is sensible without confirming it as a problem first.

No mates with PCs who’d loan you one?

Well it turns out that the capacitors are massively under-rated, with far too much current going through them, and should only be expected to last a few months of continuous use

PCs can and do work for years on all the time, or almost. That must’ve been a crappo PSU. My in-laws typically run their PC on all the time as do many offices and other folk. Less so nowadays of course as people start to realise how much it can cost.

PCs can and do work for years on all the time, but that’s not the model they’re designed to run with.

+1 to the ‘crappy PSU’ comment though.

Ok… sorry to dredge this up again!

I bought a PSU from Maplin, on the basis that I could (hopefully) return it if it turned out that it didn’t fix the problem.

I tried the new PSU (780W new vs 500W old) and got the same problems. I also tried recoating the processor with thermal paste. Again, same problems.

Not sure what to do now… I am tempted to try an old graphics card (borrowed from an old PC at work). If that doesn’t change anything, then I guess it is probably the motherboard, processor, or else I damaged something else with all my faffing around. In my paranoia I worry that I could have damaged the processor or something – after all I am not an expert at handling with electronics equipment.

Oh well, thanks again and more advice welcome if anyone is feeling generous…!

Sounds like you need a new motherboard 🙁

Very long shot: have you taken the motherboard out completely? Perhaps there’s something shorting it out underneath?

Ah yes, I forgot, that was on my list of ‘things to try’ – remove everything from motherboard, remove motherboard and check, reassemble.

Also, does anyone think that either of these could have killed the processor (cack-handedness alert…):

1. To remove the original thermal paste I used a cotton bud and some solvent. I earthed myself first (radiator), but I suppose I was worried that using a cotton bud could be a no-no…?

2. I am slightly paranoid that I might not have had the CPU heatsink properly clamped down one time when I powered up the machine. Argh. I am not sure how dangerous this is to the CPU health – presumably during start up it is not doing much and won’t generate much heat, but I just don’t know. It’s difficult to tell because it didn’t work before and doesn’t work now…

Anyway, new plan of action…

1. Dissasemble/reassemble motherboard
2. If no success buy new motherboard
3. If still no success could be processor at fault(?), not sure how to test, or what the symptoms would be.

Thanks again all.

CPU death is comparatively rare compared to motherboard failure.

With the previous generation of CPUs, the Intels had fairly effective temperature management, in comparison to AMD which had none and would fail in catastrophic (and spectacular) fashion if overheated. One of the geeky websites (Tom’s Hardware? Maybe) did a video demonstrating it.

With current CPUs, I’m not sure if that’s still the case. I’d hope that AMD have caught up by now, but I’m a little out of date on CPUs these days so I don’t know.

Ah, found the video.

Cougar.. it’s lovely to have someone on STW so knowledgeable about PCs 🙂

I’m soon to upgrade my desktop I think, so might be posting on here…

Well, its an Intel CPU and looking at those videos i’m hazarding a guess that I didn’t kill it! Quite a difference there…

And yes, quite a few knowledgable people here. It’s been such a great help, as it would be a right pain/faff to get a professional to look at it. IME people that have an interest in computers tend to have an interest in problem solving too, which has been very handy for me!

Longshot but have you tried changing the battery on you MB?