Log in

No account? Create an account
Previous Entry Share Next Entry
(no subject)

The fan on our main PC has been misbehaving for the past couple of months (buzzing intermittently), which is a fairly good indication that the bearing has been going. The bearing seized completely within the past few days, but I'm not exactly sure when it happened.

So, time to replace things. The last time this happened (over a decade ago, on my machine of that time), I replaced the fan, which seemed to do the trick. This time around it looked a little more tricky; I'd have to unsolder the old fan from a very crowded circuitboard in the PSU. Discretion being the better part of valour, I decided that the easiest option would be to just buy a new power supply, and get one that was a bit quieter than that on the old machine.

So, hippety-hop to Maplins, and back again with a new PSU. Removed the old one and fitted the new one, being careful to ground myself before doing so. New PSU is indeed quiet, and pulls air through in a way that the old PSU no longer did...and the machine as a whole is dead. Doesn't boot, doesn't even get as far as the BIOS prompt. Looks very much as though, despite my precautions, I've just killed it.

And this is why I'm a computer scientist, and not a computer engineer.


  • 1
Did you sit the garklet down in front of you to watch ? From the photos I've seen of him, I can imagine him folllowing you very intently. ( I know, woman, thinking of babies.....I just see you as being the cool kind of dad )

Here's a fun one for you. A box dual-boots Windows XP and GNOME/Linux.
  • Boot it into Windows. Receive random bluescreens, often with a bias towards I/O.
  • Boot it into Linux. Be told that / is corrupt. fsck it, and watch the kernel "oops" itself.
  • </ol>Let it warm up, and everything is fine for both, and all apparent I/O problems have vanished.

    If you diagnose this as "bad memory when booted cold", congratulations. I'd love it to be that, but memtest86+ tells me that it really, really isn't. Voltages, temperatures, etc. are fine, too. I guess it's something more irritating, like a motherboard fault that has appeared out of nowhere.

    Don't even get me started on the high-end graphics card that fell into mediocrity in the time it took me to get it working (over six months). Bloody PC hardware.

    * Exceptions are made for robots. Also, Bromptons, and cars not made by Americans or the French.

Re: I hate technology*.

Yes, when it comes to that irritating combination of complexity and unreliability, motherboards come top of the list (followed closely by graphics cards).

More irritatingly, the march of new processors means that motherboards get deprecated very quickly. Want to replace a motherboard in a four year old machine? You might as well replace the processor as well while you're at it. And the memory.

Hi Nick,
OK a couple of things to check through.
1. Does the Fan on the new PSU start up ?
2. Any odd smells near it ?
3. What is the power rating rating of the new PSU compared to the old PSU ?
4. Is the PSU switched to operate at 220V instead of 110 V ? (yes I know it would have gone bang but its a check to see if it soft failed).
5. Unplug the power supply from the mother board and see if it powers up (ie the fan). (it may not as some PSU's need a minimum load.)
6. Check the power cable from the mains for damage.
7. Check the fuse in the power is either not blown and is the right rating for the PSU. There may be a large inrush current so you might have to put a higher rated fuse in to cope.
8. Check the connectors for damage, these days the pins are 'formed' pins and are easily damaged. The PSU side should be 'sockets' check that they haven't become deformed.
9. Not tooo sure of the pin out but see if you can get a DVM onto the pins and see if there are voltages available but point 5 may come into effect.
10. Plug the PSU into a different mains socket.
11. Check the plug to see if the wires are actually secured.

Hope that helps.

Best Regards


In order:

1. Yes, fan on new PSU starts happily.
2. No odd smells.
3. Old PSU had a rating of 300W, new has a rating of 400W
4. It's 220V only (no switch, and sticker where the 110V/220V switch would have been)
5. Yes, fan still starts.
6. Tried with two different power cables.
7. Fuse isn't blown.
8. Connectors all looked okay.
9. Haven't tried this yet - will this evening.
10. No change.
11. No change.

The behaviour I see is this. When booting from cold (after power has been switched off at mains), the machine begins to start up (PSU fan starts and I see the bios screen briefly), then shuts down (before getting past the bios screen, PSU fan stops). Machine then restarts (PSU fan comes on again), but it doesn't get to the bios screen again (no output from graphics card, but some peripherals are powered - tray eject on the DVD drive works, for example).

When warm booting, all I see is the second state.

Switching back to the old PSU (which worked, bar the duff fan) makes no difference.

So, all I can think is that I've managed to kill the motherboard.


It's still possible you have killed something less vital (or that it coincidentally died by being moved about rather than getting static zapped).

I'll assume you have already checked that all expansion cards are properly seated and that both ends of all data cables are still firmly in etc. So the next step is to take them all out! Remove any expansion cards apart from the graphics one, remove all data cables from the MoBo end and unplug the power from all drives, and remove any external connections except keyboard and then see what happens. Dead CDs etc are surprisingly good at making machines look dead until you unplug them, been there, spent days diagnosing that.

If that doesn't work, you could even try taking all the RAM and the graphics card (assuming it is a card) out and see if you at least get a beep code.


I'd already checked that the cards were correctly seated and the cables were connected.

I've now removed all but the graphics card and all the data cables (etc), and I've got the same result as before.

Haven't yet removed the RAM and graphics card, but am not hopeful.

What makes this more awkward is that I've not been able to get hold of the manual for the motherboard; it's an OEM motherboard made by Gigabyte for Fujitsu-Siemens. FS give an incredibly short datasheet for it that doesn't include any useful information about diagnostics such as beep codes, and Gigabyte denies ever having made such a motherboard.

Is the BIOS by Gigabyte? If not you might be able to find the beep codes at somewhere like BIOS-Central if you can see the screen long enough to check whose it is.

But yes, I am purveying straws here.

I think that the BIOS manufacturer is written on the top of the BIOS ROM (off the top of my head, I think that it might be Phoenix).

Is there an additional 4 pin power connector on the MB that you missed connecting up ?


No, that's connected as well.

It's worth pointing out that the system no longer works with the *old* PSU connected (which was previously working fine, bar the issue with the fan).

The most common component that's failed for me has been hard drive controllers. Which usually means getting a new motherboard, (+memory & cpu).

Or, if you've gotten fed up of this happening every year, you bought 2 identical motherboards last time, and stuck one in a cupboard for when this happened again.

I've not had this happen on any ATX computers, but I do recall getting wierd interlock problems on AT machines. I usually had to unplug all the power connectors from everything, power the machine up and down, and add each connector back in turn (mobo, hd's, fd's, cd).

  • 1