Computer keeps restarting

Status
Not open for further replies.

davids

Posts: 87   +0
Hi have been working on a computer, here is the approximate spec:

AMD athlon 2800+
1GB ram
DVD rom and CD writer
80gb HDD
Creative sound card ( a fairly decent one)
Nvidia Gforce FX 5950 Ultra
New 500Watt power supply

I have a problem in that when any sort of intence graphics take place, it switches off, and will not turn on until it is unplugged and re-plugged into the mains.
For testing purposes I have been using 3d Benchmark, and When I run this it inveriably shuts down before the end of the tests (normally on CPU Test - is this Necessarily relevent?).

I have tryed the Video card on another machine and that works fine.

I have changed the power supply, no joy.

I have just re-pasted the heatsink to the processor as I thought it was loosing contct (the old paste was very dry).

The problem still persists, and whats more, nothing seems to be getting overly hot. I have felt the cpu heatsink and the graphics card after shut down and they seem ok, Also when I reboot and look at the temops, the CPU is at about 120 degrees whic doesnt seem exccessive does it (concidering my P4 runs at about 175 - and by the way, is that ok?)?


If Anyone has any assumptions on whtat the most likely cause is, I would very much appreciate it..

Thanks in advance for any help.
 
if you're getting BSODs, zip up the last 2 or 3 minidumps for analysis and submit them as an attachment
 
Is your cpu and ram at stock speeds ? Somtimes this can happen if a certain piece doesn't recieve enough voltage .. so if you play with your clocks .. you might have to ajust your voltage.
 
CPU overheating. Check that your fan speed is normal, 120F = 50C and that is what it is after you get it to reboot and then run whatever program you do to read the temp. What do you think it was just as it shut down? a lot higher is my guess. Sound like your CPU fan is dieing or the heatsink is dirty, but I'd think you cleaned that when replacing the paste. Double check the heatsink/fan is reinstalled correctly and sitting flush with the CPU die.

Cheers.
Edit: What amps are on the 12V rail of your PSU. The 12V is what runs all your fans and motors. Most PC's with a few case fans a couple of HDD, DVD CD need more then 22A.
 
just about to heck that out

Thanks for the replys.

Tedster, What are BSODS? are they like stat logs?

Radwen, I think the CPU is running at the standard speed it doesnt appear to have been overclocked, the memory Im not sure about, how would I tell?

Kirock, I am just about to test out the heatsink as I have borrowed another heatsink an fan off a friend so I will clean that up and try that on there. I will have a look at the amps on the PSU also.

PS I talked to my friends in my local computer shop and they think it may be a heat damaged cpu or possibly damaged board, is this possible?

Thanks or all the help guys
 
I talked to my friends in my local computer shop and they think it may be a heat damaged cpu or possibly damaged board, is this possible?

I don't think so, in fact your mobo and CPU are doing what they are suppose to do. Shut down when critical temp is reached! BUT..what might be happening is the mobo temp sense circuitry or the thermistor on the cpu die has aged and they are "tripping" at a lower temp. They think the temp is critical, but in real life it might still only be 50-55C.

See if you have an adjust for this in your BIOS, should be under PC Health.
Also remove the heatsink and cooling fan, clean off the old thermal paste and apply new stuff. Clean fan and heatsink too. Reattach. Use Everest to check your Temps and fan rmps.

BSOD=blue screen of death. OCing questions: Set BIOS to fail safe defualts, save and exit.
To enter BIOS, hold DEL or F2 key (in most cases) at staet up/power on.

Let us know what happens.

Cheers.
 
Thanks Kirock

Blue screen of death; I see! No I havnt even had this (which is one good sign I suppose). It just shuts off, although once so far, it did crash and the whole screen went pixelated and just froze, but I think that might of been the temperature monitor program I was running in conjunction with the 3d benchmark.

Anyway, since my last post, I have tryed a different fan and heatsink (andcleaned all surfaces and applyed new paste). The replacement was not brand new, but the heatsink seemed better, and the fan was much more efficient, so I think the problem of overheating as such, is probably ruled out.

Kirock, it makes sense what you say about maybe the sensors being wrong, I didnt think of that. What I will do now is reset the bios to failsafe, then see if I can disable or lower the shutdown temp in the bios. Then I will post the results back here. If that doesnt help I will really e at a loss!

Thanks again, all your help is much appreciated
 
Back again...

Hi, I have just tryed loading the optimim values in the bios, and I also disabled the CPU shutdown temperature.

After doing so, the 3d benchmark test competed the first time that I ran it, but the second time, rather than shutting down, the screen totally froze and went all multicoloured pixelated if you know what I mena - like when you put a graphics card in a machine that totally does not work (but as I mentioned earlier, I have tested the graphics card in anoterhr machine and it worked fine).

So now, once again I am at a loss.
Any ideas?
 
Well I sort of came in half way through and it immediately struck me as an overheating issue. Sounds like we've covered the CPU, now time for the GPU! :) I know you just want to smack me with a giant heatsink, but you really need to check this possibility out too.

Also btw, I really don't recommend you diable the overtemp in BIOS, but just maybe move it up a little. Have you tried Everest yet and see what it says your temps are? Sometimes a second software optinion helps.

OK back to the video card:
Under normal circumstances, the heatsink and fan combo on your graphics card does its job relatively well. However in a closed case with other components also pumping out heat, the temperatures rise to a point where any overclocking can tip the balance from stable to unstable, particularly in warmer months. As soon as you start experiencing freezes and crashes, and strange blocks or lines of colors appearing on your screen, or even small white dots, then you know that additional cooling is quite likely. Even if you don't overclock the card, when ambient temperatures are very high, you will need to provide additional cooling. To supply plenty of fresh cool air to the graphics card, try doing one or more of the following:
That was taken from this link:
http://www.tweakguides.com/ATICAT_10.html

When you are statisfied you solved this cooling/overheating issue and you still have probs, there is still a few things we can try in the BIOS.

Cheers.
 
Back again

Kirock, Thanks for the link I have read the tips on there. It does say in there that the screen going snowy etc can be a simptom of over heat - and that does tie in with some of the symptoms on this system. However, I did try the graphics card and ran it for 4 hours and it worked fine - running 3d benchmark over and over, also this system here is reasdonably cool - as I said Ive got a powerfull CPU fan, a case fan and the GPU fan which also seems reasonable - and it has a little plastic cover to direct the flow onto the heatsink, I am just running it now to have a good feel for any part getting hot.

As Tedster said though, I tihink it may be the motherboard, and having said this, I have just noticed a capasitor (the cylindrical things, do I have the right word) near the CPU has a convexed top ie it is bulging out, in my computer shop they said this would be a sing something had gone on the motherboard, is this right?
 
Downloaded everest

Ok I have now downloaded everest, its a smart program, but Im not sure how to interperet everything in it, I didnt see any temperature readings, but on the temperature front I am beginning to think everything is ok.

In the graphics card settings you can read the gpu temperature, it is running idle at about 30 degrees and I havnt seen it pass 40 degrees at any point - and the core slowdown threashhold is 127 degrees!

I am really lost here guys, the only solid sign Ive seen of any problem is that bulging component on the board, so can anyone give me a rundown on what I am looking for in that respect...
 
Extra cooling

Just some more info: The case is open and lying on its side, and I have also managed to fashion an extra fan directing cold air onto the GPU heatsink, so I am now pretty sure I have a cool environment...

And it is still crashing!

But strangely, it now seems to be doing the "snowey effect more" rather than actually turning off, any ideas why this would be?
 
Power consumption

Hi, I dont know if anyone is still following this link or if I would be better starting a new topic, but here goes:

I ended up replacing the motherboard and processor to try and solve this problem, as it seemed like the only thing left that could be wrong (also as I mentioned, there was a dodgy looking transistor on the motherbaord).

Unfortunatly, when I installed the new hardware, the problem STILL persisted!

SO I went right back to the beginning, and it has now transpired that the power supply was causing the problem - even though it was the first thing i replaced, by some stroke of bad luck I managed to get a similar power supply, and it seems that it is not a very well made one.

To cut a long story short, I have now put another power supply in the origional system and everything seems to work fine - I have been testing it for about 1 and a half hours non-stop and it runs like a dream.

The problem that I now have, or rather the question, is this:

The power supply that I have put in the system is only 230watts - and I have been told that the graphics card that I am using (Geforce FX 5950 Ultra) needs a minimum of 350 Watts.

My qualm is that it is working fine on the 230Watt power supply, why should I change it when it may just present me with the same problem all over again?

I have also been reading up a little bit on power consumption, and it appears that amps are the thing to measure when you want to find out how much power your system uses all together, and when looking at the 230Watt power supply compaired to the 450watt power supply, there is little difference in amp output.

If anyone has any pointers on what to do here, I would be very glad to here them./


Thanks
 
Hello davids,

Sorry I didn't mean to drop this thread, just missed it.

Power= VoltsXcurrent (amps). Power supplies have three main outputs, 3.3, 5 and 12Volts. The most important is the 12Volt rail and should have at least 18Amps (this is for a 2 HDD and NOT top of the line video card.) The 12Volts runs the fans and the motors (Hdd and optical drives). The 3.3V is used by the mobo for the CPU and RAM and the 5Volts is used by the rest of the electronics in the whole computer, HDD, optical drives, Gfx card etc.

A typical GOOD PSU would read like this:
12V @ 22Amps (264 Watts)
3.3V@30Amps(99 Watts)
5V@20Amps(160Watts)
This would then be a total of 463W and the unit would probably be rated at 450W. That rating is peak and it is not recommended to operate the PC such that you are drawing near the max amps/power of PSU continuously.

Here is a good site to calculate your power requirements:
http://www.extreme.outervision.com/psucalculator.jsp

I suggest you play with adding a higher vidoe card or a 2nd hdd etc and see how the power draw increases so you can get an idea of how as you add extras and upgrades you must consider upgrading your PSU too. This is why we always recommend buying a very good powerful PSU from the start then you don't have to worry.

Cheers and let me know if you have any questions or any more probs.
 
to follow your question of the bulging component on the mobo - if it is a capacitor (cylindrical thing wraped in plastic, two pins to the board, silver top with usualy a + sign) that has bulged up, then it may have as well caused the problem at the first place, some wont give much damage, others produce some very weird errors. Its just due to a chemical imbalance that causes this effect, and can be sometimes common with some electronics (or a bad batch of em) - google leaky/popped capacitors

the psu problem: do as kirock says, check your system with the calculator to get your minimum recomended watts.. a dodgy psu will only give you a 70% rating, good quality ones abouts 80-90% efficency rating. most common way is to look at the brand on the box - nonames/looks cheap vs known names/looks well built eg Coolermaster, enermax etc.. just google on them brands if you dont know whats which..
 
thanks guys

Thanks for all the help guys, I really appreciate it.

Right, I looked on the power calculation site (very good link thanks!).

For the system I have, they recommend 279watts power (although the graphics card maker said we needed 350watts).

If you dont mind Ill list my powersupply specs here, because Im still not sure I totally get it:


COLORsit 230watt power supply (the one that seems to work fine)
3.3v - 14a = 46.2
5v - 28a = 140
12v - 18a = 216
total = 402 - which doesnt add up with the 230watt spec, but does agree with the fact that the psu works fine on the system.

And the other power supply (that didnt work on the system):


Q-Tec 450Watt Dual fan switching power supply
3.3v - 28a = 92.4
5v - 35a = 175
12v - 16 = 192
total = 459

So by these calculations, the 450 watt power supply is not much different to the 230 watt is it??


N3051M - thanks for the info about the capacitor , Yes your discription is what I saw, and I thought it looked a bit dodgy.
However, I am testing the machine out this morning and if it all works fine, then I may leave things as they are for now - but I will tell the owner about the capacitor so that he is aware of it.
And Kirock, if the power output was insuficient, could this cause major problems, or just a lack of power??


Lastly, with the parts that I have bought and now dont need (motherboard etc) I am thinking about building linux system - but I will probably start a new post on that in the right section.

Thanks alot guys
 
they recommend 279watts power (although the graphics card maker said we needed 350watts)
279w = absolute minimum you should be using (actualy should be rounding it up to 300w - headroom) if the vid card says give it 350 then thats probably because some of us mere mortals dont have a clue what brands are good and just giving that operational headroom (vid cards eats power for breakfast), or just 350 with all devices running at full throttle....

if the power output was insuficient, could this cause major problems, or just a lack of power??

goto the top of this thread, there's a search (next to "Today's Posts", not the one at the top top..) and search for psu's.. that will probably answer your question..

but if you wont bother, then simple answer is:

lack of power = devices not running at optimum/recomended levels = psu have to run harder to keep up (and fans spin more noisier at some cases) = more wear to psu and chance of psu killing the mobo/other components

the Colorsit psu seems to be correct, and remember that those maths kirock mentioned was for a "GOOD PSU", so for that one to even it out would x by the efficency rating, which for that one is abouts 60-72%.

the Q-tec.. havent heard of that one, but then again i dont usualy shop for psu's that much... maybe mid range psus? the only theory i can come up with that psu not working properly with the pc is maybe due to the amps on the 12v rail not enough (18 vs 16)

but thats from what i understand from reading through various posts and tech sites.. and so just hang on for kirock seeming he's got the experience/expertise in this feild of pc hardware :)
 
Thanks n3051m

Thanks for the help, but I have to confess Ive made ANOTHER mistake!:

I have just opened up the case again, looked at the power supply (the good one) and it actually says 350 watts!! Which ties in with the amp value.
I was sure that it was a 230 watt power supply, I know I saw that printed somewhere; Im gonna take the psu out later and look on all its surfaces, but Im sorry about that mistake.

So now, as the computer has been running all morning without any errors, I am beginning to think all is solved.

Thanks again for all your help, I appreciate it. I have got out of a lot of sticky situations thanks to this site.

Now Im off to play badminton:)
 
Well things are starting to make more sense now.

The 350W PSU is being rated by the manuf as 85% and the Q-tec guys are saying theirs is a wopping 99% effecient unit! Obviously the Q-T is nowhere near this and the 12Volt@16Amps rail was not enough power and it was probably pulling down the 3.3V and 5V rails underload. The inter-rail dependency is a matter of quality design, a cheaper unit will have more rail interplay. As one rail/output draws more current, the others rails loose current and thus their voltage drops. For high speed electronics (like GHz PC circuits) the voltage levels are critical.

This was from my 1st post on this thread, after I mention the prob was most likely CPU overheating.
Cheers.
Edit: What amps are on the 12V rail of your PSU. The 12V is what runs all your fans and motors. Most PC's with a few case fans a couple of HDD, DVD CD need more then 22A.
Somehow I missed coming back to this point and it got lost in the details.

Take care and let us know if there's anything else we can help with. Cheers.
 
I believe I was stating the fact, but was also being sarcastic:
...the Q-tec guys are saying theirs is a wopping 99% effecient unit! Obviously the Q-T is nowhere near this and...

I'm just mad at myself for letting this thread move on without really sticking to a point/question I made right at the start to clarify the PSU. Oh well, type and type and type and learn!
 
Status
Not open for further replies.
Back