OT: Help needed,my system becomes unstable

[Skybird]

Symptoms:

during the last two weeks, the system shut down completely (black death, power cut) two times while running idle, or running Words (so, no stress for the CPU). Two or three other incidents when it did not shut down, but started to reboot suddenly.

Three or four times, the booting sequence took up to 30 seconds to even start with showing the first BIOS screen.

Today, after having left a game (Oblivion), the desktop was heavily blurred and washed-out that I needed to reboot. But even the BIOS screens showed the same blurness and washing-out. I cut power and waited three minutes, started again then: blurred and lots of white noise. Again completely shut down, and waited some more minutes. Booting went fine again.


Setup:

I run my tower with an almost open back and a room ventilator behind it, it is more silent, and very effective - I feel a gentle flow of air at the front's openings with my hand.

I have inserted a new graphics card two weeks ago, and equipped it with a new cooler.

System was stable since 1.5 years.

It is very hot over here, room temperatur often above 27°C. That'S why i open the wide side of the tower, too.


My thoughts so far:

The new graphic card (nVidia 6800) needs more power than the AC can provide. It is a 350W device. However, one would assume that the problems would appear when the board is under maximum stress, what is the case when playing Oblivion, and I play it a lot, currently. However, the problems never rose when playing Oblivion. Sometimes, they even started when the system was in idle.

Overheating. The CPU cooler is an Arctic silent 4 (P4, 3.0). The CPU has a temperatur of 37-38°C in idle. It was no problem in the last 18 months, or last summer. The graphics board is equipped with a Zalman VX900 at around 2200 rpm. when touching the copper fins, they are handwarm. Again, one would asusme the card overheating when under stress (Oblivion), but it runs Oblivion well.

Hard Drive.

RAM (2x512).

Any other thoughts? I am clueless and don't know what to do. I checked cables and connections. Cables inside are nicely bound together.

Could it be a power problem, or a colling problem? Can anyone make sense of the symptoms described?

[mog]

I'm far from an expert, but:

1) It certainly does sound like overheating
2) Removing the side panel of your case could actually be detrimental by disrupting airflow from the front to the back.

[jumpy]

Originally Posted by Skybird

The new graphic card (nVidia 6800) needs more power than the AC can provide. It is a 350W device.

I have a 6800gt card which is essentially the same as yours. Before I bought it a couple of years ago I had a look on somewhere like toms hardware gude for a review. I seem to recall it being mentioned that the card required at least a 400w psu to run correctly without risk of crashing.
If the motherboard cannot get enough power due to the graphics card hogging all the juice, could that cause your machine to crash?

Worth a shot maybe? Besides, you can never have too mahoosive a power supply :p

I wouldn't have thought you could get a serious temperature issue with the side of the case off... but having said that what's the manufacturers operating temp for the cpu?
My case has 2 120mm case fans at ~ 3000rpm and an 80mm cpu cooler at 5500rpm. Recently it has been pretty hot here, hotest so far this year, and current temperatures are: case-36c cpu-56c which is a little above average for my machine due to it being gernerally hot.

Maybe not relevent, but how old is the 'backup' battery on your motherboard?

[Skybird]

Graphics boards consume the more power and produce the more heat the more busy they are. I am irritated that the problems do not appear when the system is performing heavy duty, but is at idle. I have had no problems so far while playing Oblivion at max details and high resolution. If the card is undersupplied, one would expect it to fail when doing this or Far Cry or Il-2. but it does not. at least not until now. I had the crashed when the system was at idle, or running words.

The side of the tower is open, and the back is very much open. From the corner of these two open sides a 25 cm ventilator blows diagonally into the tower. The ventilator and the wall are arranged in such a way that the airstream flows from back to front. It reaches all areas inside the tower. Ypu can feel the air easily coming out at the front of the chassis when holding your hand there. I do like that since years, and with two earlier systems as well. Never had problems with cooling.


I will need to wait for a period of colder days then, and see what changes then. Maybe when I am done with Oblivion I reinstall my older FX5900 card which served all other sims and pruposes well. If the new card, despite the contradiction of working well under stress, but failing when working nothing, is the porblem, the symptoms should disappear with the old card then.

No way I am willing to buy computer systems with AC beyond the 500W limit. Changing from 200W to 350W already marked an impact on my electricity bill that was to be felt.

A p4 of 3 GHZ is said to trigger the emergency shutdown at around 70°C, I recall from some Intel table. I am sure, and have checked that with a temp scanner as well ,that the CPU does not reach such levels. It is at 37°C in idle, and around 45-50°C under full stress.

Oh, and wanting to say that dust is not the problem, I keep good care of that, since the tower is open so often.

The battery is the one the sywstem came with: 1.5 years, a Lithium CRxxxx, often used type. In watches, this battery usually lives up to five years. but I will measure it the next time I start working in the inside.

[scandium]

The P4 will throttle if it overheats, so the shutdowns/reboots you are experiencing are not the result of your CPU overheating. I don't think its the GPU (your video card) overheating either. My money is on your power supply having gone flaky after you replaced your video card with one that would (especially in conjunction with the notoriously power hungry P4) push even a good 350W PSU to its limit, and perhaps beyond it.


By the way Skybird, the electricity used by a computer's PSU is not determined by the rating of the PSU; it is determined by the load placed on the PSU by the computer's components and the PSU's efficiency, and the way to save on that is to use computer components that use less power, or are at least more efficient (and thus any newer AMD CPU is to be preferred to its Intel equivalent), in conjunction with a high efficiency PSU (good efficient PSUs advertise their efficiency).

My recommendation is that you buy a high efficiency replacement PSU in the 450W range.

[Hellcat]

Sounds like a power supply issue to me, what brand are you currently using? I've had a generic brand psu's fail on me after 4 years of steady use.

ideally do you have any componets kicking around you can swap and test? like the old vid card. (edit i see you still have it!)

[kiwi_2005]

Sounds like symptoms of a overclocked CPU a overclocked cpu can make your computer crash evenwhen its idle. But you never mentioned you overclocked it, so check your PSU see if you can borrow another with the same watts and connect that up then see if the problems stop.

[TteFAboB]

That's a very clean CPU case for someone who runs a tower fan in the rear.

I have to clean my entire case at least once a month or every two months, I always wanted to switch to liquid cooling just to be able to weld shut my case and stop any dust from getting in but the entire liquid system is too much of a hassle and the risk far too great.

Do you put air-filters in your tower fan or do you live in a negative-pressure vacuum sealed enviroment?

[Skybird]

No, i just clean it regularly and take care when doing so. When you run an open tower and put a vent near to it, you really need to have an eye on the dust level

[NEON DEON]

Hi Sky,

I was wondering if you could post a dxdiag file and your complete systems specs?

It might help to see what the OS is saying about your computer.

Sounds like you should be good on the PSU capacity. Whether or not its operating w/in specs would only be guessable if your motherboard has SM bus and monitoring utlility can function.

I don't think cooling is suspect either, considering the open case. However, open case does present the potential for ESD. Do you have an animal in the house? Electro static discharge (ESD) can wack components so schell wie einnen Blitz und Tod Schlag. Most technicians work with wrist straps. However, this is not absolutely necessary if one grounds themselves by touching the chassis (with PSU still plugged into mains) prior to disconnecting unit from mains and touching any components withing system (and being very carefull while working on the system no to suffle feet on carpet). Touching running water stream from faucet is also good way to ground oneself (do this if you have to walk away from workaread and system is apart). Only handle board components by holding board at PCB edges (do not touch surface mounted components). Its o.k. to touch anything inside box while its running provided one drains any accumulated charge before so doing by touching the chassis first (PSU must be plugged into mains for this to be effective though). Never remove/insert anything from system board while its plugged in (or running).

All identical shaped plugs coming from PSU should have identical rated power specs. Although a bad contact within some connectors could be at fault.

Do you have a HDD utility, such as Norton Disk Doctor? I'd run it (or the Windows equivalent scandisk) and see if it finds any major problems.

Your problem strikes me as electrical, in that morning sickness seems to be the main problem. Also, at idle computer components may cool sufficiently to break contact.

1. Disconnect each and every component from motherboard and reconnect/reinsert (component connections can get loose over time).
2. Run disk scan utility
3. disconnect unit from mains, remove CMOS battery, replace CMOS battery 5 minutes later, reconnect PSU to mains
4. run memory test utility
5. reflash BIOS (after initial flash, be sure to load default configuration, and then configure with system specific configuration). You could check to see if a more recent BIOS version is available and flash that, but I've seen flakey system problems disappear by merely reflashing the very same original BIOS. Why would a BIOS flash go bad? Uh, ESD? Dunno. Happens. I recently had my system unplugged from mains for three days. When I plugged it back in all sorts of strange issue cropped up during boot. After much head scrathing I figured the system was 6 years old, so I ran out and bought a new CMOS battery. That helped enourmously, but ultimately also had to reflash the BIOS before all the bugs got shaken out.
6. If all that fails to pinpoint/resolve the problem, I'd strongly suspect failure of a controller on the motherboard.