Computer/Internet Issues & TroubleshootingAnyone with questions about the terribly complex world of computers or the internet should try here. We will also try and help with troubleshooting any technical problems you may have with the forums.
About six months ago I upgraded one of my PCs with a new motherboard, processor and CPU cooling fan (ASUS A7V233, AMD XP 1800 and GlobalWin respectively, if memory serves). Everything worked perfectly, but the cooling fan was incredibly noisy - the system also has a PSU fan, chipset fan on the motherboard, and two SCSI disks, but this CPU fan drowned them all out by far!
So, I recently changed the fan for a (more expensive) GlobalWin Copper Superquiet (or something). Sure enough, it's much quieter, but the machine started locking up occasionally. Suspecting overheating, I loaded up the ASUS 'Probe' software to check the temps, and also removed a coupl of the blanking plates from the 5.25" bays to let some of the hot air out of the case (it's a full tower).
The machine no longer locks up at random intervals, but the Probe software reports (and here comes the point of this rather long message) that the CPU runs at 71C idle, and 81C when under full load (crunching SETI work units).
This strikes me as being far too hot, although everything seems to work fine. What sort of CPU temp should I be looking for? Unfortunately, because I never checked the temps before changing the fan, I can't quantify how much difference it has made.
The PSU also runs *very* hot - almost too hot to touch. Is this a sign that it's being overworked?? I'm not sure what its power rating is - 300W, I think...
Tom's hardware site suggest the optimum idle to be around 50C. Certainly my machine (P4, 1.8) standard fan installed reports a whisker under that from the BIOS (the system in general is reported as running at 45C or thereabouts).
I don't think it's a good idea to remove blanking plates as it tends to destabilise the airflow. You might want to get an exhaust fan to force the hot air out of the box. I found it to be very effective.
Last edited by amanoffewwords; 4th Oct 2002 at 21:35.
I can assure you that a watercooled P4 2GHz on an ASUS board with plenty of fans and massive airflow, including the mentioned 450W PSU runs in idle at around 58 degrees, also under extremely heavy load never ever went above 68.... At around 89 degrees you approaching the temp which the tiny little bits and pieces inside the CPU destroy.... Everything above 80 degrees should be considered as dangerous and you should re-consider the cabinet and/or devices providing a better airflow...
Thanks for all the replies, everyone. I checked out the Tom's Hardware and Overclockers sites - very handy. I've just had the case off the machine again and re-routed a couple of ribbon cables to help with airflow etc.
After powering up, I noticed that the PSU fan was blowing and not sucking! Which is always bad, as we know - possible damage to hard drives etc, ooo-eerrr matron, etc.
So anyway, I flipped the fan over and have now got some actual airflow through the case, and CPU temp down to 61C idle / 66C loaded. I've also fitted, but not connected, an intake fan to the front bottom (ooo-err etc, see above), so I'm now going to plug that in and see what happens. I'll let you know...
It turns out to be only a 250W PSU, so I think I'll upgrade that, in the next couple of months or so. Cheers,
I have the same motherboard, and for different reasons I have been fiddling with it a lot, trying to fix a problem. I can tell you that the CPU temps on them mean little. The reason is that my motherboard started off with BIOD version 1.05. With that, I got CPU 100% loaded temps of about 49°. I then upgraded to BIOS version 1.11, and with no other changes the temp then read 62°. I'm now on BIOS version 1.13, and the CPU temp is 54°. The ambient temps are close enough to being the same, so the BIOS has given me a good 13° odd temperature variation. The rule of thumb, if you can, is that the CPU temp should be no more than ~30° more than room temperature, and no more than 90° at any point. The Asus motherboard has overheat protection, but they are sometimes unreliable in shutting down in time, from what I hear.
Thanks for the tip, 18-Wheeler. I was wondering how accurate the temperature sensors were... Tom's Hardware seemed to imply that some of them were very accurate, but it sounds like it depends a bit too much on the BIOS (calibration issues??).
Anyway, the intake fan (see above) didn't make a lot of difference - the 100% load CPU temp now down to 64C or so, although the motherboard temperature is now only around 28C, with ambient temp around 22C. The PSU runs *much* cooler, which is good - I wasn't happy about the amount of heat that was building up in there.
18-Wheeler - is the ASUS overheat protection meant to trigger a controlled shutdown of Windows, followed by a power-off? If so, it certainly didn't work properly in my case, for whatever reason - the thing just locked solid. Hmmm...
I have an Athlon XP system in an aluminium Coolermaster case (pretty much all components from overclockers - excellent site). The chip is the 1.8Ghz version sitting on an Abit board, under a Zalman Flower copper cooler, which not only cools extremely well but can be silent if the fan attachment is unused (depending on the layout of your case and power supply, the additional fan may not be necessary).
The chip is not overclocked and currently idles at about 38 - 39 degrees C, and even if pushed hard for a half hour or so won't exceed about 45.
Also, note that the cooling compound you use to mate chip to heatsink can have a large effect on cooling potential. I have used Arctic Silver, which seems to perform very well. Compound is probably the cheapest component in your system and returns considerable performance benefits, worth looking into!
FWIW the computer magazine I read tried a stack of different heat-sink compounds and there was naff-all difference between the most expensive and the cheapest. The application and smoothness of the surfaces was far more important.
Yes, the COP overheat protection is supposed to shut down the computer, but I havent't read how it does that. Your lock-ups may or may not be heat related - to the CPU - If the inside of the case is getting too hot it might even be the memory locking up ... ? I leave the side of my case off and have found it runs a good 6° cooler by doing that. If you do the same and the lock-ups disappear then that's a pretty good indication of it being a heat problem.
One point which seems to have been missed here is the possibility that the lockups may be being caused by "brownout" (one or more of the PSU output rails dropping momentarily). From what CBLong has said so far it seems clear that the PSU is heavily loaded (& possibly overloaded) so it is conceivable that under certain circumstances it drops out for a fraction of a sec. In fairness though I would think you would notice this from the sound of the fan changing but I think the point is worth raising anyway.
Cbl, is the PSU fan now extracting air (ie does it blow towards you when you are behind the PC)? This is how it normally should be.
Out of interest, as the point was discussed, the thermal drop between the heatsink and processor topside is normally negligible (less than 0.1 deg C) compared to that between the it and the silicon die within, not to mention the difference between heatsink and local ambient air temperature. Hence even a 10 fold improvement in heatsink compound effectiveness does not really count for much if all else is correct. (The soft "spongy" heat transfer pads used by some systems are somewhat less effective though).
I would always be concerned over parts within a computer which were too hot to touch (ie much over 60 deg C). Whilst semiconductors can generally operate up to 150 deg C (actual silicon temperature) reliability suffers badly, also thermal "runaway" can affect certain components - this is when the device heat dissipation actually increases with temp' (due to physical phenomena) to the point where the cooling arrangements are inadequate and the device is destroyed.
Well there are quite a few independant tests that would verify that the type of thermal compound you use can have a bearing of up to 5 degrees on your chip's operating temperature (especially on hot Athlons) but I'm not going to labour the point. It's something that costs about £4 a tube at the very top end, so it's hardly a major outlay!
I had an associate of mine lose three 2Ghz XP chips on the trot to heat damage, eventually determined by the supplier to be insufficient contact twixt chip and heatsink.
FujiFlyer - re: possible PSU 'brownouts', I guess that's a definite possibility, the PSU was running very hot. I can't figure out if the PSU fan was deliberately installed to suck through the PSU, or if it was a mistake at the factory. The fan is on the bottom of the PSU and was sucking air through the back of the case, via the PSU, and blowing it onto the CPU area of the motherboard - I guess this might have been deliberate??
In any case, I've now turned the fan over so that it blows through the PSU and it's running a lot cooler.
The thing that makes me suspect that the lockups were due to CPU overheating rather than PSU overheating was that the only change I made was to swap out the CPU fan - this may well have had an indirect effect on the PSU temperature, but it must surely have had a much larger effect on the CPU's temperature?
I'm happy now though - it runs without problems, so that's pretty much that. I might try some 'proper' thermal compound next time I've got the case off, but other than that I'll just keep an eye on the temps... :o