Computer keeps turning off randomly.

[Dantoshou]

Hey all, gonna try to be concise here, basically i believed this problem to be my power supply, the computer randomly turns off while in use, sometimes a few times a night, sometimes once a day, but i tried new plugs that hasn't changed it, also getting a Belkin surge protector just in case anything bad happens.

Now i took the PC away from the supply after one of these power downs, looked inside, felt around and found the graphics card to be red hot whilst the other components seemed fairly cool comparatively, while i know graphics cards to run quite hot, I'm just wondering could this be the cause of these power downs? I'm playing cod4 currently and there's shutdowns, when i play guildwars its generally fine, but has shut down a few times, so this is quite sporadic a problem.

what do you guys suggest as its really bugging me that it keeps powering off randomly, I've taken the side metal plate off just to make sure there's more airflow, but if you guys have any insight for me I'd appreciate it, the card is a 470? from Zotac, apparently these have 5 year warranty so i guess i can send it back if it is faulty?

Also could I change the thermal paste on this card and not void the warranty, as I'm thinking this may be the key to the problem?

Many thanks in advance,

Dantoshou

 

[Dantoshou]

Hey guys in reply to you all, i have a 1000w psu, Ive just been monitoring the temps with speed fan and the results from last night shocked me a bit, ill link you a screen from last nights idle, last night under load, and this morning idle, this morning load.

so idle last night was

Load was

After letting it cool down overnight
idle this morning (started at 30 but has gone up slowly)

and load today is

what concerns me is the high temp my cpu is reaching, i thought cpus weren't meant to reach that high, it is an overclocked bundle at 4.2ghz but its been fine for months.

i can try onboard graphics, not sure if i have a onboard slot but i will have a look.

 

[Dantoshou]

You say your psu is 1000W, but what make - this makes a huge difference.

Temps honestly don't look too bad. Not great, but nothing I would be overly concerned with personally.

The psu is an OCZ 1000w proxstream, they no longer make these and its past the 3 year warranty point, although when i bought it the box said LIFETIME warranty, kinda defeats the point.

 

[Dantoshou]

so has anyone got any idea what could cause this, i played for hours on guildwars today, and its just had this problem again twice :S after hours of solid use without any problems

 

[Dantoshou]

okay i went into the clocking took a while to figure out but i managed to get it to 3.00ghz, it seemed more stable but still had the problem, I've just set it to 3.3ghz as thats apparently what this core is as standard.

Yea i hope its just the PSU and nothing else, although everything BUT the psu is still in warranty, so sods law states it should be that! (although the fiancee will be pleased if i do change psu as she says this one is noisy)

I'm testing this under clocking and standard clocking by playing black ops, managed to get through one match before it crashed this time.

 

[Dantoshou]

75c for the CPU while playing guild wars is NOT fine imo. I wouldn't be surprised if it hit 100c (the thermal shutdown value iirc) during excessive load.

exactly what i thought sad, i thought that cpus were sposed to hit 70 on something like crysis, this cpu used to idle very low compared, i think ill get some more thermal paste soon and reapply that, but first i need to establish whats wrong with it, currently Ive returned it to stock speed, and it hasn't yet crashed, but ill have to leave it on to see what happens.

 

[Dantoshou]

Okay it just crashed at stock speed, what do I do now!

 

[Dantoshou]

When i said return the CPU to stock speeds I meant do a CMOS reset on the mobo (return to factory settings). Sounds to me like you have been randomly changing settings......not a good idea really

cmos reset? is there a cmos button for me to press, i think i know where that is and ill give it a go

 

[Dantoshou]

Your screenshot showed your 12v rail at 10.xx volts. That is not within the 5% acceptable tolerance for voltage rails in PCs and could very well be the cause of your problems.

Acceptable is 11.4v - 12.6v for the 12v rail.

P.S. Your home telephone number is visible in the Windows Sticky Notes in your 2nd screenshot.

okay ill try a cmos reset and change that voltage to 12! as for the number its alright nobody ever phones us (and i realized this too )

 

[Dantoshou]

75 degrees in guild wars is pretty bad actually very bad... I hit 62 with 5.0 Ghz in battlefield 3 on ultra... What is your CPU cooler?? My AMD used to freeze and shut down, I went to check temps and they were around 90 when I touched CPU radiator I got pretty bad burn on my finger... Cleaned the heatsink from dust and it was fine again.

My money is on CPU shutting down due to high temps... I would not be surprised that if you open your case you will find a lot of dust which is causing cpu and gpu to work pretty hot.

yea i have the coolermaster thing that came with the raptor bundle from last year? so i assume its a good cooler, ill just buy some diamond thermal paste or something to sort out the temps, but first i need to establish the whether these other things are causing it.

 

[Dantoshou]

Okay so i pressed the cmos button at the back, which made the computer shut down, so that's done ( i hope that's correct!)

now I'm wondering where the voltage control is for that 12v thing? all i can find is cpu and ram voltage.

do you know where that would be listed?

 

[Dantoshou]

Don't mess with the voltages for the love of god. Just do the CMOS reset. This removes the overclock and the CPU will run cooler at stock and should be stable at these settings.

The temps at load look high. Did you fit the heat sink yourself? How much paste did you use?

i used a thin scrape of paste, yea it didn't come prebuilt just the overclocked bundle, ill have to buy some more paste, is the 24 carat diamond one actually good at keeping temps low?

and okay, ill make sure it fully resets and just leave the voltages as they were when it was overclocked!

 

[Dantoshou]

"Try this or try that" without any reasons to do so? Shotgunning. Long before trying to fix anything, first learn what is wrong.

Numbers from "Speedfan" said a power system (which is more than just a power supply) is defective. Unfortunately, "Speedfan" can only read numbers from a multimeter built on the motherboard. These are badly calibrated. Therefore the voltages could be OK. But the on-board meter is reading defective.

Your problems could have existed months ago. Typical is for a defective supply to still power up a computer. But the defect could have been seen on day one if using a multimeter. The same tool necessary to calibrate the on-board meter.

Those posted numbers identified a defect. Unknown is whether those numbers are properly calibrated. Anyone doing computer work has a multimeter - even £7 in Maplin. So that even a new supply can be verified long before its warranty expires.

BTW, a 1000 watt supply for hardware that typically consumes 200 watts? Or is your computer as hot as a four slice toaster?

so what you're saying is i need to get a multimeter to get readings from the power supply to make sure this is still functioning correctly?

how do you mean hardware that consumes 200watts?

I wouldn't know how much power it consumes,
its "Radon Utahraptor" Intel Core i5 2500K 3.30Ghz @ 4.40Ghz Overclocked Bundle - MSI
with ZOTAC GeForce GTX 470 1280MB GDDR5 PCI-Express Graphics Card
4 hard drives
8gb ram (after cmos reset is running at 1333mhz)

Are you saying i don't need that much power. i think i originally bought this as a replacement for a 600watt supply that blew up.

 

[Dantoshou]

It's very hard to troubleshoot a PC "virtually". You should know full well a CMOS reset is the first port of call. His overclock is drawing more power from the PSU for a start and it's increasing the temps. It's good practice to troubleshoot from stock settings.

I've already said the PSU is a candidate for the problem (post #11), trouble is his temps are quite high.

2500K intel spec sheet

the Tcase is 72.6 degrees (that's not the max safe temp just the highest the Tcase sensor should ever read) so you are looking 80 degrees before heat is causing issues.......he is showing temps close to that. If he tried realtemp and a few other programs he'll see that the recorded temps can vary from program to program.

I'll make some assumptions and go out on a limb and guess the PSU is faulty. Infact maybe even the fan has failed in the unit. At 50 degree temps it will struggle to maintain it's output. Seeing as it's an old PSU it's not going to putting out was it was new anyway.

If the fan was sucking warm air in from the case it would run warmer but at the same time was helping to cool the CPU. The hotter your components are the shorter they live is the rule of thumb.

If the machine is still failing at stock speeds. Investigate the PSU, easiest way is to use a known working PSU. Can you borrow one from a friend (a good 500W PSU is all you need for that system)?

I would bet that will solve your problem, If you have had the PSU 3 years thats good going to be fair but I am suspecting you need to look at your case cooling too. What case and fans do you have installed and where is the computer kept?

Okay, well im currently on the stock settings, still at 80 degrees, very high temp even on the stock settings with two copies of gw open

instead of trying to explain the setup ive made a video of it!

this shows all the components, all the installed fans and obviously that i keep the side off, previously i DID have the side on but decided to take that off thinking it may be causing overheating issues, also I couldn't put a side fan on as the cooler is so big it fills that area of the case.

I can try to reconfigure the power cables aswell, but as you can see the airflow of the system shouldn't be too bad, although its a compact space in the case. http://s272.photobucket.com/albums/jj179/dantoshou/?action=view&current=CompVid.mp4

there you go, that hopefully shows the system, i also tried to show the speedfan status, but that was blurry.

currently seems to be 83 degrees gpu and 80 cpu.


*edit*

just for clarification, heres the speedfan on a screenie

*edit again!*

Everyone is pointing at your PSU, but I still maintain its a CPU heat problem. Your cooler is not dissipating heat effectively for whatever reason. Heck, you may have left the plastic sticker on the CPU when attaching cooler. I know some people have done that.

30c boot
35c idle after a few minutes
+5mins of GuildWars
58c load

PC on for long period I assume
53c idle
+3mins of GuildWars
75c load

Have you logged temps while gaming for more than a few minutes? It's highly likely you're hitting the thermal shutdown temperature during heavy load (large fight scenes for example).

yeah, ive been on the pc about 2 hours now, ill keep logging them as i progress but it seems to stay at 80+ during gaming/video editing.

 

[Dantoshou]

Buy a new case mate.......cable management is non existant

Why is the case so bad? Your PSU is mounted at the top of the case. Hot air rises, the PSU is sucking all that hot air into it (direct from the CPU heatsink too i might add) and exhausting it out of the case. Modern cases have the PSU at the bottom of the case where the air is cooler and quite often there are vents at the bottom so the PSU fan can suck in cold air from outside.

You still need to test the PSU, I think you'll find some of the vault lies here, it been just slowly cooking itself to death in that case over the last 3 years. A new case and a PSU (modular would be nice) would be a wise investment.

YOUR BASKET
1 x Zalman Z9 Plus Tower Case with Fan Controller - Black £49.99
1 x OCZ ZS Series 550W '80 Plus Bronze' Power Supply £46.98
Total : £109.27 (includes shipping : £10.25).



I don't know what budget you would have for parts so this is just to give you an idea.

I understand what you mean, I should really have got some cable ties and sorted out the cables, but the case is actually fairly new, its the coolermaster elite 300? ( i think)
and i have only had it since last year some time, i understand what you mean about hot air rising, but when i touch the psu, its feeling cool, even while running the pc under load, doesn't feel like the air is rising to it,


I do understand where you're coming from though and if the PSU is the faulty part
I will be looking at modular ( this one has FOUR GPU pin cables, which are the thick red wires, about 10 black wrapped wires aswell, so you can imagine the MESS of wiring ive hidden in the floppy drive bay area)

I just completed a full burn in test for 15 minutes aswell the results made me laugh.

the warnings seem to be about my ssd, which only has a few things on, so I wasn't bothered about the whole BIT and errors

Interestingly, the CPU and GPU performed a lot cooler than in an actual game, even though they allegedly used 100% of their processing.

anyway, it looks like I'm still without solid proof as to whether its the CPU needing new thermal paste, or the PSU being faulty, so am I to believe i should be getting one of these multimeters to actually test the psu, and also getting some more thermal paste to bring temps down?

 

[Dantoshou]

I use CPUID Hardware monitor a lot of the time. Download that and try it. See what voltages and temps it reports.

If it is still showing a low voltage (10 rather than 12), then the PSU is looking very guilty. Yes you could buy a multimeter, if it shows the same as the software then you still need to cough up for a PSU.....you will wish you used that £10 or whatever to go towards the PSU then.

YOUR BASKET
1 x BeQuiet Pure Power L8 530W '80 Plus Bronze' Modular Power Supply - With 120mm Silent Wing Fan Built in £66.98
1 x Cooler Master Silent Pro Modular 600W '80 Plus Bronze' Power Supply £65.99
1 x OCZ ZT 550W '80 Plus Bronze' Modular Power Supply £63.98
Total : £196.96 (includes shipping : FREE).



That's some suggestions for a modular PSU for a single GPU setup. The OCZ has a five year warranty too. If it doesn't cure the problem you end up with a new warranty and a spare PSU for testing in the future....not a complete fail is it really?

Your temps look better now. Intelburntest isn't great really, running for 15mins doesn't really tell you much at all. I do use it to quickly test overclocks, then move onto prime95 for fine-tuning.

Your temps whilst gaming were in the high 70s what are they now the overclock is gone?

hmm i hope i got the right program, its

Yea maybe ill admit defeat and get a new psu and get some heat paste aswell, probably that 24 carat diamond stuff.

That temp is at 51 now which is weird, but earlier it was 80 while on guild wars, very weird as its dropped dramatically.

Overall the cmos reset SEEMS to have worked, It hasn't crashed since, im just gonna play black ops now (done my daily war in kryta quests) so hopefully that will force any shutdowns out as its a higher demanding game.

 

[Dantoshou]

okay guys all i can find in the bios is these screens, there are others but they are for general options

i guess ill have to get one of those multimeters and test for myself, i know you guys say that it may take a few components with it, but aren't they covered under warranty if they do die out? as i stated before i bought this system last august so the components are all covered via the manufacturer for at least 3 years.

but anyway, ill get a multimeter from amazon or something, ill have to learn to use it as it will be the first time Ive ever used one!

and test the voltage aswell, hope you get some insight into what the problem is via the screenshots also

 

[Dantoshou]

I didn't say the multimeter damages it, I said that if the psu blew up and damaged components would those components not be covered by warrant for such things ( its just the psu that isn't still in warranty.) yea i am sure its easy to use a multimeter, ill have to look where i can buy one ( I recently moved house and I don't know the shops round here.)

yea i see how to actually change the multiplier, i saw the drop from the ocuk setup.

 

[Dantoshou]

okay i've just bought a multimeter and some IC Diamond 24-Carat Thermal Compound, so ill test the psu and reapply heat paste when they arrive, once I've tested it and find that the rail is probably still only running at 8v i guess ill buy a new one (and probably a new case too) and put the other one on the mantlepiece.

And heres a picture of a short OCCT test, if you need me to test for longer I can do later today.

 

[Dantoshou]

hmm, i put on that IC diamond 24 carat stuff, but i was told that its better not to try to spread it, however in trying to reattach the cooler i ended up spreading it everywhere, probably wont be very evenly spread and i may have to reapply, its a weird cooler to get back on, very awkward.