PC keeps shutting itself off - eep!

ixoy · 10-10-2004 3:04am #1

Within the last month, I've noticed a horrible tendency for my PC to switch off suddenly, without warning. It's an Athlon Barton 2500XP chip, running at 2GHz (I can't recall if that means it was overclocked when it was assembled for me) on an ASus A798X motherboard.

The only common thread I can see as to when it's turning itself off is when I'm in the middle of playing a game which leads to wonder if its something to do with the CPU heating up due to being over-taxed. This is further evidenced by the casing feeling disturbingly hot. It particularly doesn't seem to like Doom III which would be the most graphically intensive game I have but I still thought all that heavy GFX work was offloaded to the graphics card (GeForce 4 Ti4200 for what it's worth).

This problem has only recently manifested itself and I'm wondering why is this happening all of a sudden? The only recent upgrade to the PC was an additional HD and, if it is the CPU I can't see why it would be affected. The two fans at the rear seem to be working judging by the air venting out of them (and the blades aren't stationary) so it doesn't appear to be the fans either. What could be causing it? I don't want to see my PC die but I'm at a loss as to why it's suddenly switching off and the system logs don't seem to say anything useful (unless I'm reading them wrong).

Anyone got any suggestions/advice on what might be wrong, how to track it down, and how to save it?

pickarooney · 10-10-2004 7:48am

It most likley is overheating, it has all the symptoms. An extra hard drive will always increase the overall system temperature. Next time it shuts dow, restart it immediately and go into the BIOS to check the system and CPU temperature. Alternatively, there are software thermometers which can show you if you're getting close to critical temp. A bigger or extra fan is probably all you need.

KdjaCL · 10-10-2004 8:09am

Ram set at higher FSB than it is?

Run Sandra http://www.sisoftware.demon.co.uk/sandra/ and check the mainboard and CPU tabs down the bottom will be warnings if something is not right.

kdjac

Ciaran500 · 10-10-2004 8:32am

Could be a PSU problem, what wattage/make is it?

Mutant_Fruit · 10-10-2004 9:23am

Sounds like a Qtec PSU that is aging gently.

tba · 10-10-2004 10:47am

I have almost the exact same problem. ie new hard drive, random crashes.

But I have noticed that I only crash when I use programs of the new hard drive, have you tried running said games off your original HDD?

ixoy · 10-10-2004 2:39pm

Thanks for that info so far.

Well I installed Sandra as suggested and, unsurprisingly, it's giving out about the mainboard temperature.

Warning W2518 - Mainboard temperature is too high (i.e. over 50ºC). This should appear only if the mainboard has a temperature-monitoring sensor installed.
Fix: Power down the system immediately. Open the case and check that all the fans are working and check whether components are unusually hot inside. Make sure the components have good all-round ventilation. Use more fans if required.

I examined the system and all three fans (CPU fan and 2 external fans) are spinning. The heatsink, beneath the CPU fan, seemed very hot but I'm not sure if that's normal. Yet I also got this warning:

Warning W2519 - One of the mainboard fans has failed or is spinning too slowly. This should appear only if the mainboard has a temperature-monitoring sensor installed and the fans have tachometers. If the fan is speed controlled, this may be normal.
Fix: If the fan is not speed controller, power down the system immediately. Open the case and check that all the fans are working and check whether components are unusually hot inside. Make sure the components have good all-round ventilation.

Again the fan blades are spinning around, so I can't see how a fan is not working. There are fan options in the BIOS settings but it looks, from what I can tell, that they're set at maximum speed although it's generally unclear I find with these things.

Could it be a PSU problem? I actually thought I had a 400W PSU, but it's only 350W. In addition to the GFX card and Barton 2500XP mentioned previously, I've also got a floppy drive and 3 HDs (120GB/160GB/160GB) attached but I'm not sure offhand how to monitor their power consumption. Would this possibly heat up the mainboard (the main heat I feel from the PC is coming from the PSU area)? As a temporary measure is it worth leaving the side panel of the PC exposed or is the fact it's on a carpeted surface and exposed to possible dust mean it's not worth the risk?

Optikus · 10-10-2004 2:44pm

Could be the overclocking on the CPU check to see if the CPU voltage has been upped if so then this is what is causing the heat.. if it is reduce it, you will also have to reduce the FSb for it to work on the lower voltage.

ixoy · 10-10-2004 2:53pm

Optikus wrote:

Could be the overclocking on the CPU check to see if the CPU voltage has been upped if so then this is what is causing the heat.. if it is reduce it, you will also have to reduce the FSb for it to work on the lower voltage.

Right I suspect this could be it but eh... how do I know? Alas a friend did all this over-clocking for me when I set up the PC so I'm wondering how to switch it off (assuming it's on!) What speed should the chip run at? I've an odd feeling it's something like 1.833Ghz and currently it's at 2.0GHz but I might be wrong on this.
The Multipler says 10/1x and the core voltage rating is set to 1.65V and the CPU voltage, it says to 2.47v with the CPU fan at 898RPM (which produces a little warning icon). Now from these... is there any particular way of working out what they should be?

Optikus · 10-10-2004 3:06pm

That voltage is fine.. althought at stock the mutli is 11 and FSB is 166x2 (333), so that shouldn't really be a problem as the temps dont really raise untill the ther voltages is upped .. if you want to test you could set everything to stock. what Heatsink fan combo are you using this could be the problem?

Carpo · 10-10-2004 3:07pm

Three easy suggestions for you,

First thing to do is, as you say, leave the side panels off and trying running something intensive like Doom 3 for a while. Dust shouldnt be a problem, at least not in the short term. If it was crashing regularly whilst playing before this should give you a fairly good indication wether its over heating or not.

Alternativly (assuming you havent tried it already) take out the new hard drive and run it for a while. Heating sounds more likely but it seems like you have a lot of devices for a 350w psu so a power problem isnt out of the question either.

Last easy suggestion is to check the cpu fan and heatsink to make sure theyre not clogged up with dust.

MrVestek · 10-10-2004 3:10pm

It doesnt sound like an overclocking issue per say, however it seems to me that a 350w PSU wouldn't be able to handle 3 hdds, 1 cd rom and a good enough gfx cart like that, not to mention a motherboard that's specifically designed for overclocking. Get a new 400w PSU, 450w preferably.

Optikus · 10-10-2004 3:10pm

the voltage should be 1.65v for CPU ..im not to sure what you mean by 2.47v.. a easy way to set everything to stock is to Load fail safe settings, and then if you want to overclock it safly from there we can help.

the_syco · 10-10-2004 3:37pm

Firstly; Is any of your hard-drives / CPU / graphic cards set on "silent mode"? It runs silent by not having the fans going to fast. It makes it silent, but also doesn't get rif of the heat. This may be one of your problems. Since it seeme to be caused by the new hard-drive, this is most likely it. Maxtor, and a few other drives have this setting.

Sedcondly; Where is your PC? Can it suck in lots of cool air, and expel it out the back easily? I ask this, because if your room is very hot all the time, your PC wil only be sucking in hot air, which can't be good.

Thirdly; I'm no expert, but someone may be able to answer me here; if you overclock your CPU, aren't you meant to get a better fan?

Finally; Check the insides of your case. Is the new hard-drive blocking a "wind tunnel" through your PC? The wind tunnel basiclly goes from the front of your PC, and out the back. This sucks most of the hot air out. If there is not much of a wind tunnel, the heat may just be trapped inside your machine.

Optikus · 10-10-2004 3:42pm

Achilles wrote:

It doesnt sound like an overclocking issue per say, however it seems to me that a 350w PSU wouldn't be able to handle 3 hdds, 1 cd rom and a good enough gfx cart like that, not to mention a motherboard that's specifically designed for overclocking. Get a new 400w PSU, 450w preferably.

I agree a 2500+ @ 2.0ghz is hardly a stressfull overclock. so have you tried Doom3 without the new HD, you should do this to elimanate/pinpoint the problem.

Ciaran500 · 10-10-2004 3:55pm

It depends on what make of PSU it is.

I'm no expert, but someone may be able to answer me here; if you overclock your CPU, aren't you meant to get a better fan?

Unless you increase the voltage the tempreature does not change when you overclock.

Optikus · 10-10-2004 4:10pm

If its a Q-Tec then this is more than likely the problem.. get a motherboard monitoring application to see if the 12v and 3.3v are what they are supposed to be at, as some Q-Tecs give out overvoltages and if so would deffinitly be causing the mentioned over heating.. if not try doom3 without the new HD.

ixoy · 10-10-2004 4:24pm

Optikus wrote:

what Heatsink fan combo are you using this could be the problem?

I'm using the default standard ones that I got with the chip when I bought it. Does anyone know what the RPM on this fan should be? It says it's running at 898RPM - is this too slow? Oddly though ASUS probe reports a value of 3792 and it lists the CPU temperate at 56C (and won't report the status).

As a temporary measure check, I've disabled the latest HD. I forgot to add that, in addition to the 3HDs, I've got a DVD-ROM and a DVD-R/W drive so I could be pushing the 350W PSU, which is an Altec Lansang brand supplied built into the case when I bought it.

the_syco wrote:

Is any of your hard-drives / CPU / graphic cards set on "silent mode"? It runs silent by not having the fans going to fast. It makes it silent, but also doesn't get rif of the heat. This may be one of your problems. Since it seeme to be caused by the new hard-drive, this is most likely it. Maxtor, and a few other drives have this setting.

Not that I know. I can't, again, see anywhere to set something like silent. The new HD is a Western Digital Caviar.

the_syco wrote:

Where is your PC? Can it suck in lots of cool air, and expel it out the back easily? I ask this, because if your room is very hot all the time, your PC wil only be sucking in hot air, which can't be good.

It's in the main living room in the appartment. The room isn't particularly warm - average room temperature. I've got the front panel, that holds the drives open and have now opened the side panel to try and get maximum air coverage to it.

The 350W PSU is an Altec Lansang brand, supplied with the Altec case that I bought.

optikus wrote:

a easy way to set everything to stock is to Load fail safe settings, and then if you want to overclock it safly from there we can help.

I assume that's the next last resort? I've gone to the Chipmanagement settings in the BIOS and set it to "Optimal" - is there something else that would need to be done? All other settings seem to be at their "lowest" (VCore, etc) now.

Optikus · 10-10-2004 4:31pm

Ok now you are at default settings and have disabled the new HD can you run D3 ok, has it changed the temps?

Optikus · 10-10-2004 4:35pm

ixoy wrote:

I'm using the default standard ones that I got with the chip when I bought it. Does anyone know what the RPM on this fan should be? It says it's running at 898RPM - is this too slow? Oddly though ASUS probe reports a value of 3792 and it lists the CPU temperate at 56C (and won't report the status).

My fan speed is nearly ten times faster than that. the ASUS reading sounds a bit more realistic, but you cant be to sure on application readings.. if its not the new HD generating the extra heat then a a new fan/heatsink might be the solution.

The Muppet · 10-10-2004 5:09pm

ixoy wrote:

Again the fan blades are spinning around, so I can't see how a fan is not working. There are fan options in the BIOS settings but it looks, from what I can tell, that they're set at maximum speed although it's generally unclear I find with these things.

You should replace the fan that is giving that error as the board is detecting that it is spinning too slowly and so it is not doing its job properly.

Mutant_Fruit · 10-10-2004 5:37pm

The Muppet wrote:

You should replace the fan that is giving that error as the board is detecting that it is spinning too slowly and so it is not doing its job properly.

There are some ultra low RPM fans (500rpm) that show up as defective, but they are actually operating correctly. But i doubt a fan that comes with a mobo would be one of those types.

BloodBath · 10-10-2004 5:47pm

My Northbridge fan spins anywhere from 4000rpm to 8000rpm depending on temp if yours is spinning at 800 then that might be a problem.

To be honest a system just pwering off without warning like that sounds like the psu. I'd be almost definite on this. Athlon xp's do not have thermal throttling so they will not shut off if they get too hot. They will just fry. The only way it can power off is if you have a shutdown temp set in the bios. Check your bios and see if there is a setting that powers off the pc at a certain temp and see what it is set to.

If it isn't that then it is almost certainly the psu. I had the same problem while overclocking my prescott. The 12v rail which powers the cpu and mainboard dipped as low as 11.5 and caused the system to lose power. I modded it and now it doesn;t dip as much but i still get the same problem when i try a higher overclock and only a new psu will fix it.

Download prime 95 and run the second torture test. Use mother board monitor 5 to monitor your systems rails especially the 12v rail while it's running. It should not drop below 12v. If it is then this is the problem.

BloodBath

Optikus · 10-10-2004 6:14pm

Xp's dont have built in thermal throttling but most socket A boards do, and have preset and definable settings for shutdown at certain temps, so it probably the CPU overheating under stress that is causing the shut down. it is strange that it worked fine untill you installed a new harddrive. As i said before get a motherboard monitoring app to check the 12v and 3.3v to see if it is the CPU if its not then try and borrow a better heatsink/fan to check if thats the problem.

ixoy · 10-10-2004 6:31pm

bloodbath wrote:

To be honest a system just pwering off without warning like that sounds like the psu. I'd be almost definite on this.

Is there any appreciable way to find the power consumption of various devices? Offhand there's the GFX card, 3Hds, 1 floppy drive, 2 DVD drives, and a couple of USB devices that would - I assume - draw power.

After disconnecting my old HD and leaving the side panel open, I've noticed none of the temperature heating that I felt previously (by touching the casing). I played Doom III for a while which, previously, caused the shutdown and had no bother.

optikus wrote:

As i said before get a motherboard monitoring app to check the 12v and 3.3v to see if it is the CPU if its not then try and borrow a better heatsink/fan to check if thats the problem.

Currently, according to Asus Probe, the +12v rating is at 12.48. Is there a particular range of acceptable values? I'm pretty sure it dipped to 11.48v when I had the new HD installed.
The +3.3v rating is, again according to the same software, at 3.28V right now.

Optikus · 10-10-2004 6:42pm

Tbh thats sounds fine and by the look of the dip you got when you used the new harddrive id say that it was the prob so if you want to run all that equipment without it turning off get a new PSU.

Giblet · 10-10-2004 6:42pm

You should be running at 1.83 ghz, don't mind what anyone says about 2 ghz not being much of an overclock. It CAN make a big difference as I've found out.

BloodBath · 10-10-2004 6:44pm

Is there any appreciable way to find the power consumption of various devices? Offhand there's the GFX card, 3Hds, 1 floppy drive, 2 DVD drives, and a couple of USB devices that would - I assume - draw power.

You can try this but there is a lot more to it than simply the stated wattage. My psu is a 480w and my system uses a max of 330w yet I get these shut downs at times because of the low amps on my 12v rail. The amps on each rail have a lot to do with how it performs. If your system is sucking too much power from a rail that doesn't have enough amps then that rail will droop. It is usually the 12v that does this.

The rails are supposed to be within 5% of their stated output but 5% under really is a lot.

There is no point monitoring the rails while the system is idle. They will only droop a lot when under load. If it dropped to 11.48 with the new hard-drive then that is definitely around the area where the system will power off like that.

Check your BIOS for the shutdown temperture to see what it is set to to see if this is the problem. If it's not then it is your psu.

BloodBath

Optikus · 10-10-2004 6:47pm

Giblet wrote:

You should be running at 1.83 ghz, don't mind what anyone says about 2 ghz not being much of an overclock. It CAN make a big difference as I've found out.

1.83 to 2.00 is nothing unless you Don't know how to overclock and didn't lock the AGP/PCI bus and or overclocked your ram to much.. as i said the OC is fine your problem is just the PSU.. and my advice still stands leave the new HD out untill to get a higher rated PSU.. Hope this of help to you Ixoy.

Giblet · 10-10-2004 6:50pm

Optikus wrote:

1.83 to 2.00 is nothing unless you Don't know how to overclock and didn't lock the AGP/PCI bus and or overclocked your ram to much.. as i said the OC is fine your problem is just the PSU.. and my advice still stands leave the new HD out untill to get a higher rated PSU.. Hope this of help to you Ixoy.

It's people like you causing people like ixoy to buy new PSU's all the time.

He took the side off his case, problem disappeared: HEAT

Running a barton with a stock heatsink is bad enough, running an overclocked one is even worse. Could be tipping it over the edge.

StRiKeR · 10-10-2004 6:52pm

Giblet wrote:

It's people like you causing people like ixoy to buy new PSU's all the time.

He took the side off his case, problem disappeared: HEAT

Running a barton with a stock heatsink is bad enough, running an overclocked one is even worse. Could be tipping it over the edge.

didn't he mention that his voltage was dropping?

theres no harm having a good decent PSU at the end the day, but dont just going by the wattage, higher wattage doesn't really means its better!

BloodBath · 10-10-2004 6:55pm

Giblet the heat of a processor only increases if you raise the voltage of it. Raises the mhz alone will not do this and his small overclock is more than safe. If it was heat then his BIOS shutdown temp is set too low and simply raising it 10c would fix his problem and he could keep his overclock.

Just because it ddn't crash after a few mins of doom3 doesn't mean the problem is fixed.

BloodBath

Optikus · 10-10-2004 6:57pm

Giblet wrote:

It's people like you causing people like ixoy to buy new PSU's all the time.

He took the side off his case, problem disappeared: HEAT

Running a barton with a stock heatsink is bad enough, running an overclocked one is even worse. Could be tipping it over the edge.

From what i hear he has had the system for quite a while, pre-overclocked when he bought it, now if there was a problem with the OC why has it only surfaced now.. wanna know why because there isnt.. his voltage was dropping due to a new harddrive sucking up the power and his PSU not being rated high enough to handel it..hence the unexplained shut down. please do not sterio type me with out even knowing me, im just trying to help him, i work in computers and know what i am talking about .. HEAT is not the problem here, as he removed the HD and the problem was resolved if heat was the problem he would have know about it a long time ago.

StRiKeR · 10-10-2004 6:58pm

overclocking without rasing the voltage will still rasies the temperature but not much, very very little compare to rasie of voltage, not enough to make it overheat!

StRiKeR · 10-10-2004 7:00pm

Optikus wrote:

From what i hear he has had the system for quite a while, pre-overclocked when he brought it, now if there was a problem with the OC why has it only surfaced now.. wanna know why because there isnt.. his voltage was dropping due to a new harddrive sucking up the power and his PSU not being rated high enough to handel it..hence the unexplained shut down. please do not sterio type me with out even knowing me, im just trying to help him, i work in computers and know what i am talking about .. HEAT is not the problem here, as he removed the HD and the problem was resolved if heat was the problem he would have know about it a long time ago.

that true, removing the HDD doesnt reduce the heat of the CPU, but it does reduce the stress of the PSU!

Optikus · 10-10-2004 7:00pm

BloodBath wrote:

Giblet the heat of a processor only increases if you raise the voltage of it. Raises the mhz alone will not do this and his small overclock is more than safe. If it was heat then his BIOS shutdown temp is set too low and simply raising it 10c would fix his problem and he could keep his overclock.

Just because it ddn't crash after a few mins of doom3 doesn't mean the problem is fixed.

BloodBath

Agreed this decision cannot be made just on D3 running .. may i suggest that some bench markss be run to prove my point.

BloodBath · 10-10-2004 7:08pm

overclocking without rasing the voltage will still rasies the temperature but not much, very very little compare to rasie of voltage, not enough to make it overheat!

No it doesn't. I've tested running my machine at 2.8ghz and 3.4ghz at the same voltage and the temps were the same. In his case going from 1.8ghz to 2ghz without a voltage increase definitely wouldn't make a difference.

It's a simple case of running prime 95 while monitoring his temps to see if it's heat. There is point debating until he does that.

BloodBath

StRiKeR · 10-10-2004 7:11pm

i dunno about intel chip but if you wanna call over and see my AMD chip overclocked without increasing the voltage and see the temp goes up by just 1c???

but thats not enough to make the CPU over heat and shut down!

P.S. thats clock from 2gig to 2.2gig

ixoy · 10-10-2004 7:36pm

Right then, the computer isn't as warm now but is it because the side panel is open or because the HD is disconnected, is the CPU or the PSU.

Stress tests have been suggested and I've got the latest edition of Prime95 - what sort of test should I work with? Keep stressing it until it shuts down and note the temperature (I assume it logs such things)? Should I try it both with and without the HD connected then to spot the difference (if any)? Thanks for all the help so far of course...

Ciaran500 · 10-10-2004 7:36pm

Giblet wrote:

He took the side off his case, problem disappeared: HEAT

He took the side off his case and disconnected the new hdd and the problem dissappeared.

ixoy: I definatley think its a psu problem. You need a new psu with a strong 12v line of at least 20a. You should download motherboard monitor. THis will log your temps and voltages to give you a definite cause.

Optikus · 10-10-2004 7:44pm

When running the stress test Prime95, just try it as normal with the door closed the HD out ..if it passes all, which it should as the problem only arose when you got the new HD .. then try again with the HD in and i guarantee it will not pass, thus indicating that you need a higher rated PSU, to support the new drive.

StRiKeR · 10-10-2004 11:11pm

ixoy wrote:

Right then, the computer isn't as warm now but is it because the side panel is open or because the HD is disconnected, is the CPU or the PSU.

Stress tests have been suggested and I've got the latest edition of Prime95 - what sort of test should I work with? Keep stressing it until it shuts down and note the temperature (I assume it logs such things)? Should I try it both with and without the HD connected then to spot the difference (if any)? Thanks for all the help so far of course...

sure give a ago it wont do you any harm! but I think Optikus was right!

BloodBath · 11-10-2004 12:59am

Stress tests have been suggested and I've got the latest edition of Prime95 - what sort of test should I work with? Keep stressing it until it shuts down and note the temperature (I assume it logs such things)? Should I try it both with and without the HD connected then to spot the difference (if any)? Thanks for all the help so far of course...

You have been told several times what to do. Run the second torture test on prime 95 while running motherboard monitor 5 and monitor both the cpu temp and the psu rails. You will spot either problem here.

Also check the bios to see what the cpu shutdown temp is set to.

BloodBath

ixoy · 11-10-2004 8:21am

BloodBath wrote:

You have been told several times what to do. Run the second torture test on prime 95 while running motherboard monitor 5 and monitor both the cpu temp and the psu rails. You will spot either problem here.

Also check the bios to see what the cpu shutdown temp is set to.

Uh huh, thanks I'd have tried that last night but for various reasons I didn't want my PC dying on me. Instead I just left it on, unstressed and, unsurpsingly, nothing happened. I can assume that the occasional brief spike (at one point, for a second only, CPU temperate shot up) is probably a software glitch than anything else (given how suddenly it would drop afterwards).

I'll properly test the bugger tonight. What I meant to more ask was how long you should run a stress test for to be sure of a decent result. I'm thinking overnight should do it?

Optikus · 11-10-2004 3:01pm

If your worried about stressing it to much try 3DMark first (any one) as this is more stressul as D3 and should highlight a problem, try it the way i suggested without the new drive, then again with it. I dont think you have a temp problem bartons are ok up to about 65c anyway. Then try the Stress Test, after a few mins of that you will know if there is problems, so say 15min is a good enough time to run the test.

PC keeps shutting itself off - eep!

Comments