After build troubleshooting assistance needed

[eds2103]

I just built a nice computer. It is described here:

http://arstechnica.com/hardware/guid...-edition.ars/2

Except I have a Western Digital Caviar Blue 640 GB and I am at the moment using my friend's nVidia nForce or geforce 6 series or something -- not sure -- but its not state-of-the-art or anything.

I got it to boot, and I configured the BIOS. The computer can sit in the BIOS setup utility forever without issue. Then I installed windows 7 off of a DVD. The installation had basically completed and the system needed to restart. After the restart, when it was loading windows 7, it shut down instantly as though the power supply gave out or it short-circuited, or something. This continued to happen every time I booted up the computer, while it was loading the installed windows 7. It happened at variable times, though, not at the same time. Somtimes really early, maybe even still in BIOS. However I could take the thing to the BIOS set up page and as I said it could still sit there forever, which makes me think I can eliminate that the mobo is being short circuited with the case (can I make that assumption?).

Then I tried installing Ubuntu Jaunty. Firstly, I was able to run it off of the Live CD fine (anyway, i did so for 10-15 mins). Then, I installed Ubuntu off the CD. The installation finished, and after reboot I was able to play around with ubuntu for about 10-15 mins. Then the system restarted. And then it started restarting every 1-2 mins, before I even got past the splash screen. However, sometimes it didnt restart so fast. Sometimes I would get to use Ubuntu for like 15 minutes to over an hour before it would restart; or I would turn it off to install my wireless adapter, or to restart for a driver installation. Usually, after a restart it would go back to dying within the first 1-2 minutes -- around the time of log-in screen, or earlier, or sometimes just after logging in. But after a while I would usually be able to get it to start up and run for long periods of time, like over an hour. This was true even after I installed my driver for my graphics card and was playing Nexuiz at highest graphics settings for 20 minutes.

What could be the problem? I was thinking perhaps the power supply is giving out b/c my friends graphics card takes power directly from the motherboard, whereas the graphics card that was recommended with the system by that guide that I linked to has its own power sockets. Perhaps the motherboard doesn't have enough power for all the stuff thats on it?

Another possibility: my CPU heatsink came with thermal pads on the bottom. I had installed the heat sink and the CPU had been running for a while. Then after the CPU had run for at most like 30 minutes (aggregate over many different runs -- i was trying to figure out why i got no visual output at all -- a problem I already solved -- hadnt plugged in the single 2x4 12v cord) -- after the cpu had run for like 30 mins aggregate, I took the heat sink off to see what the bottom looked like. Then I put the heatsink back on. Maybe the heatsink needs new thermal compound since I took it off?

Perhaps the power supply isn't enough? The guide I used (linked above) recommended it. And a PSU calculator suggested 340 or something. I suppose it should be enough?

Could it be that the operating systems don't work with some piece of my hardware, or they need the drivers or something?


If its none of these things, i guess some piece of equipment must be faulty? Any other ideas? If some piece of equipment is faulty, can you please help me narrow it down. For example, this couldn't be happening because the hard drive is faulty, correct? Nor the cd-rom drive, correct? I know the video card works. So it would then have to be either the PSU, the mobo, the cpu, or the RAM? Or is there one of these I can eliminate? Could it be one of the fans or the heatsink?

Thank you for your help.

[LeftyAce]

If you remove the heat sink, you have to clean off the compound and apply new thermal compound. Clean off the CPU and heatsink with isopropyl alcohol, and then apply a bb-sized dot of Arctic Silver to the center of the cpu, and reapply the heat sink. Your BIOS should have a temperature monitor. Let us know what temps it reports.

To start diagnosing a new build, follow these directions: http://forum.pcmech.com/showthread.php?t=168456

You must take the motherboard out of the case to eliminate possible shorting issues.

Did you use the powersupply in that guide?

[eds2103]

Hi, thanks for getting back to me.

Yes, I'm using the powersupply in the guide.

The CPU temperature is ~42 all the time, even just after it shuts down -- i checked it a lot actually b/c i thought maybe that was the problem.

Are you sure there could be a shorting issue if it works in BIOS and it works for hour+ in linux? I don't want to take it out of the case for nothing, as it was hard to get in there. I saw the sticky on troubleshooting but it seems like it applies to when you can't even get visual output of the BIOS.

Do I really need to scrape off old thermal compound and add new if the CPU had only been running for like 5 minutes and it's not overheating?

[LeftyAce]

You _can_ continue testing without taking everything out of the case, but if it doesn't become perfectly clear what's wrong, that is the next step.

First, let's check the cpu temperatures. Again, it's possible you're ok without taking it apart and re-applying thermal grease. Download coreTemp for windows, and set that up to start when the computer starts. See what temps you get just sitting at the desktop; see if they're climbing steadily or if they're stable.

Also make sure you have disabled "Automatic restart on error" I'm not sure where to do this in windows 7, sorry. It will prevent windows from automatically restarting after a blue-screen error. If the computer is giving you such an error (although it doesn't sound like it) you'll be able to read what it says.

Can you tell us exactly what video card you put in there? I am inclined to think that the 380 watt powersupply is too weak for the build, but someone else who's more knowledgable on that subject should interject.

So, get back to us with temperatures....my next target to test is the RAM. Memtest86+ is a free download (bootable cd).

[eds2103]

Well, I meant to say that I Windows was never able to start after I installed it. It always died during "windows is starting up" screen, or earlier. So then I installed linux on the whole drive, getting rid of Windows. Linux I was able to get running several times for over an hour, but today and yesterday my computer dies before/simultaneously as GRUB is loaded, so it seems like Linux has stopped runnnig, perhaps because I installed drivers for vid card? I dunno. Could be a short circuit that jiggled back into play. I can't imagine how there could be a short circuit. I mean the mobo has obvious screwholes where I put standoffs and screwed the mobo into the standoffs like all the guides say. Can someone explain how I would even look for a short circuit even if it worked out of the case?

[flanzig1]

There could be an extra standoff touching someplace on the motherboard. Also, the I/O shield could be installed incorrectly; a grounding tab could be stuck into one of the output ports on the motherboard.
As for the PSU, 380w is pushing the limits even if it is a quailty unit. When I build PCs, I like an about a 20 to 25% higher wattage rating than what the build specs out at. This will cover any load spikes and added hardware. Your shutdown issues do sound like classic weak or underpowered PSU issues.

[eds2103]

So I can't get coreTemp and I cant seem to boot linux as i have it now (although I could reinstall it off of boot disk and then get some temperature program for it.) BUt like I said even after I was running this computer game at highest possible graphics settings for like an hour, I checked the CPU temperature immediately afterwords via the BIOS and it was 42 degrees, which is about what it always is. one time is was 44 degrees, but pretty much every time i check it its about 42 degrees. one time it was 41 degrees. I checked it about 40 times. And I made sure to check it when it crashed after having used the OS for a while. but at the moment its shutting down just after GRUB every time, right when its starting to load jaunty, and CPU temp is 42. So i dont think its a CPU temperature problem.

I just ran Memtest86+ via GRUB. It past 100% w/ zero errors.

The video card is an nvidia geforce 8600 GT.

[LeftyAce]

Ok, follow the link I gave above, and take the computer completely out of the case. It does sound like a PSU issue, but there's only one way to be certain, and that is to do out of the case troubleshooting.

[eds2103]

Okay, I went to Best Buy and got an Antec EarthWatts 500 Watt PSU and now the computer is running fine.

But I have a few questions that maybe someone could help me with:

1. Is this a good power supply and an appropriate amount of power? Should i just keep it? Or would you recommend getting something else? --a link to something that explains how to select a psu for a new build would suffice

2. This new PSU, in addition to its 24-pin ATX 12v cable, also has a 4-pin ATX 12v (I guess this is 2x2) and an 8-pin EPS 12v (i guess this a 2x4?). My motherboard has a socket for the 24-pin and it also has an 8-pin socket which can accept either the 4-pin ATX or the 8-pin EPS. Right now I have the 4-pin ATX plugged in. Is that right? Or should I plug in the EPS? Or neither? The motherboard manual just says it "supports 2x2 12v or 2x4 12v" and it says that using a 2x4 is recommended for "intel extreme edition cpu," but says nothing more. I dont have extreme edition so I don't know. How can one determine this, for future reference?

3. I have just one 120mm TriCool case fan in the back of my antec solo. Is this enough for my setup w/ the 500 W PSU? It has three settings: High, medium, and low. Which should I use -- i would like to use the lowest one that's safe since that makes the least noise. I can install a 92mm fan in the front to blow on my 640 gb caviar blue. Necessary? How can one determine these things, for future reference?


If you can just provide me with links to good guides that explain these things that would be great.

Thanks again for your help.

[eds2103]

A hunch -- I bet when my radeon 4890 comes I will be able to use the seasonic S12 II 380 W, because the 4890 (as well as the gtx 275) have their own sockets for power -- in fact they have 2 psi-e 6-pin sockets. I bet the 380 W PSU wasnt able to power my corei7 AND my friend's geforce 8600 gt via the 24pin atx and 2x2 pin atx power connectors going into the motherboard. Because the arstechnica guide said this power supply should work with this rig. hard to imagine they would make a big mistake like that. thoughts?

[glc]

I would recommend you use the Earthwatts, you made a good choice there. Use the 8 pin EPS if the motherboard has a EPS socket. You should be fine running the fan at low speed and you shouldn't need any more fans.

[eds2103]

Thanks!