Help! Newly built PC powering off itself

Dingatron

Just built a new system for one of my friends. Specs are
MSIKT300 MB
Qtec 300 W PSU
RADEON 9000
ATA133 WD 60GB HDD
CDRW
Generic PC2700 DDR 128MB
AMD XP2000
Coolermaster fan.

I would not consider myself a novice at building PCs etc. Anyway built a new one, installed everything and left him to it. Next day I found out that the PC ran for 1-3 hrs and powered itself off. On further investigation checked bios settings, power cables etc. I did not see anything out of the ordinary. After about 3 hrs I witnessed it power off again. The only thing I noticed was through MSI PC ALERT, that the 12 volt meter was running at 12.18 volts. Mine runs normally at a constant 12 volts, but the new pc does have a lot more fluctuation? Could the problem be with the PSU? Also the pc is plugged into a power surge protector and that looks fine. As my mate intrusted a lot of hard earned cash I'd like to be able to resolve this quickly. Do you reckon its the power suppy?

elexes

do you know if its overheating ?


dont think it would be the psu

sutty

It sounds like the PSU may be the problem, you may want to do a search to see if the PSU as caused problems for other people with the same MoBo.

Thorbar

On some AMD mobos when the cpu starts to over heat it automatically shuts them down. Happened to me before, gave no warning or explanation, took me ages to figure it out. Try underclocking the cpu and see if that fixes it just to find out if heat is your problem. After that I suppose buying a more powerful fan or leaving it underclocked.

Saruman

Leave it in the bios for a few hours and see if it happens... if not then it could be software or driver related.. unlikely as it sounds its possible as i have seen it before.

More than likely though its a cooling issue or lack there of. The voltage thing could be it.. not sure..

Dingatron

Humm should have given CPU temp. CPU is at around the 41 c mark. CPU fan is 7200 rpm. Only one I had. A noisey f***er. The system temperature is at 24 c. I haven't had a chance to examine PSU fully yet but will try leaving pc in bios mode. Talking to a sparks and he didn't think the voltage of 12.18 was that serious. It has been tried in a totally different location and still reads this. will swap PSU out tonite and cross fingers :-(
Would faulty memory have this effect? I'll also underclock and see. Thanks for the quick responses.

Inspector Gadget

I don't think the 12v line is the issue - the ATX specs mean that component designers have to build their components to expect up to a 5% fluctuation on each line either way - in other words, 11.4v to 12.6v. Obviously, the closer your computer is to 12v in a "resting" state (where the initial current inrush at startup is over and you're not overtaxing the machine in terms of computation) the better, but it's not a problem.

Have you looked at the APM settings? It could be something as simple as that you computer goes into standby if left idle... otherwise, it could be something like a thermal cutout in your PSU... this will reset when it cools down, so it doesn't look like a hardware failure. It could also be a CPU overheat protection, as Thorbar mentioned earlier.

Personally, though, I'd be suspicious of the PSU. Once the machine's been running for a while, check the exhaust fan on the back to see if:

1. There's an airflow. If the rear fan isn't working, your PSU will overheat.
2. The air in question isn't very much hotter than the room is
3. That you don't get a strong smell of burning. Component burn-in (what happens when you run a newly-built electronic circuit for the first few weeks) will cause a very mild smoky smell for the first few weeks from the PSU exhaust, but it should be very mild, and certainly not visible, if you follow me.

Another thing that strikes me is that a "weak" 300w PSU could buckle under the load you're placing on it. Some PSUs are definitely better than others - and as 300w is a combined figure (giving a total amperage across the positive and negative 3.3, 5 and 12v rails) the output combination of your PSU may not be much in excess of the needs of your system.

Hope this helps,
Gadget.

Diabolus

If these suggestions have failed then let me suggest this one. It sounds to me like a shorting out problem. Take out the motherboard and check the pins under it that it rests on. It is possible that you placed the board down on a pin that is not used for a screw and is pushing into teh board causing that problem. This happened to me once in my PII days.

Dingatron

Ok. To keep you up to date. I left it in the bios for a few hours no problem. I also ran Memt86 and the tests all passed. Back into W2K and it still crashed. Took out all cards bar the Radeon 9000 AGP. CPU temperatures etc are all fine,plenty of airflow etc. The PSU unit has a combined 5v + 3.3v 190 wattage. As I have only HDD, CDRW,FDD, XP2000 CPU and Radeon card attached I am ruling out the PSU. Reinstall of W2K made no difference. Found an app from MSI called Test AGP. I ran that and within 5 seconds the machine powered off. The Radeon 9000 is a Sapphire one from Komplett. (It was the cheapest 9000). At this stage fairly convinced it is the Radeon. Anyone know any other tests I could try to confirm this is the case? Will put a 450 W psu just to verify the PSU is alright but fairly sure its ok. Any other suggestions would be appreciated. Thanks.

The_Conductor

Disable APM (Advanced Power Management) in the BIOS as Inspector Gadget suggests. I am 90% confident it will solve your problem.

The_Conductor

The Radeon is a power hog- but have been running a similiar card on a 400 Watt PSU for a good while without any probs. Disable all API/APM settings, and set PCI palette to "snoop".
The later Radeons have a seperate power lead taking a feed straight from the PSU along with the AGP slot, meaning the draw through the board itself is minimised.

You will get assistance from the ATI people if you ring them on 01-807 7826

Note- When you ring you are given a few questions as to wheter your card is "Powered by ATI" or "Built by ATI"- choose "Built by ATI" (all the powered by cards are OEM and ATI will terminate the call without connecting you to the techies).

If you download the latest ATI driver package (the big 16Mb one.....) it has a definitive suite of testing utilities. You will be asked by the techies to download this and to run the diagnostics- do it yourself first of all, it'll save you time.

Drazhar

If you've tried the 450w and the system is fine then problem solved. Check the combined output for the 3.3 and 5v rails, if its less than 200w then you're going to need a bigger PSU

Having the same problem with my GF4 if i have my two HD's in, basically the system doesnt get enough power, so restarts

Dingatron

Disable all API/APM settings, and set PCI palette to "snoop".

These are disabled but I don't have an option to set PCI palette to snoop unless its called something else in the bios??

This actually made the problem worse as W2K refused to boot after I changed these settings.

Can't seem to find the correct ATI file. Is it the catalyst suite or have you a quick link to the file?

Dingatron

Ok. I'm going around in circles. I uninstalled the radeon drivers and grapics card. Put in a MX440 card (less power etc.)and was starting to install the drivers when the thing powered down again? This takes me back to the fact that it could be the AGP slot on the mother board. had a look at the latest bios releases and there is nothing pointing to problems, fixes etc. Do you think this is enough investigation? I don't want to send it back to komplett without full proof that the MB is faulty.