Sign in with
Sign up | Sign in
Your question

New system - running WAY hot and can't find a problem

Tags:
  • CPUs
  • Overclocking
Last response: in Overclocking
Share
July 1, 2007 4:01:09 AM

This is about the 10th system I've built. Specs are:

Gigabyte GA-965P-DS3 mobo
Intel Core 2 Duo E6420
Antec 550W True Power
2GB Crucial PC2-6400
SATA drives and a GF8600 VGA, not that I think that matters.

Anyway, everything was going pretty much as normal, but then I noticed the system getting really sluggish. At first I thought there was something wrong with the network because accessing the internet was just painful (getting new drivers, win update, etc.) Finally all but hung the system. So I power-off, reboot into BIOS, and it failed to even make it to the BIOS before it shut itself off. Reseated everything, fiddled around and found nothing wrong, removed all extraneous hardware to eliminate possible points of failure, then tried it again.

This time I made it to the BIOS, and that's when I noticed the temps were already into the 80s*C off a cold boot straight to BIOS. I pulled the CPU heat sink, even though it looked fine, cleaned the OEM thermal paste, wiped with alcohol, and replaced it with some silver based paste I had on hand. Made no difference.

I’m stumped. This is a major system upgrade and I have no spare parts to test with. Any ideas? I don’t really feel like ordering an entire second set of core system components from NewEgg (first time ordering from them, not going well).

One other data point, the chip set heat sink was loose when it arrived. Didn’t seem to have hurt anything, so I snapped it into place. I figured I would reseat (with new paste) it too while doing the CPU, but those pins won’t come back out without getting rough, so I left it alone.

BTW, this is no over clocking or any fancy stuff. In fact, I even tried cranking all the clocks DOWN to cool it off and it made no difference!

More about : system running hot find problem

July 1, 2007 5:17:50 AM

Sounds like bad heatsink contact with the CPU. I've heard that the LGA 775 stock HS have problems seating correctly on the CPU and improper seating can lead to high temps. Does the HS get hot to the touch when your system is on? If not, then definitely poor contact between the HS and CPU as no heat is being transfered.

Suggestions:

- Lap the heatsink
- Buy a new HS/F
a c 88 à CPUs
a b K Overclocking
July 1, 2007 5:41:45 AM

If its hitting 80c, then the heatsink isn't properly attached. They can be a PITA to get it attached right. I suggest you take the motherboard out of the case, and then attach the HSF. I find it a lot easier to do this task if you can look at all four sides and visually varify that its on correctly. I'm not sure I'd bother lapping it yet, unless the HSF came to you warped.
Related resources
July 1, 2007 6:26:45 AM

Thanks. And you're right, the heat sink does not get hot to the touch, that was one of the first things I checked. But the paste gets flattened/smeared. But then again, I was reaching under to touch the ZIF lever and it wasn't hot either. No direct connect, but at that heat I figured it would be pretty hot.

I’ll give it another try tomorrow, right now I’m fed up with it and disgusted.

About that heat sink. Are all these designed that cheesy? That’s the most half-assed mount I’ve ever seen. Expanding twist lock snap plugs (or whatever they are called)? Most of the others I’ve done have one or another different ways of strapping them on tight with spring loaded hold downs. This doesn’t look like a very good design from what I can tell. It flexes fairly easily even when locked down. Maybe I’m just not getting it all the way down. I’ll post with any results, can’t get replacements till next week anyway. And just my luck, a holiday week to boot. <sigh>
July 1, 2007 6:46:05 AM

Chances are you just need to flash your bios. I'm a PC Technician, and I have seen this problem plenty of times. Especially considering it's hitting 80 degrees on a cold boot. The slow down is probably caused by some sort of heat protection, since the temperature is so high, the cpu is scaling down to lower temps.

However, I could be wrong, looking at Gigabyte's website and all the previous fixes for that motherboard, no where do they list an incorrect cpu temp fix, however, it doesn't hurt to try.
July 1, 2007 6:49:48 AM

CPU thermal protection is internal to CPU: no related to any bios upgrade
If it was a bios thermal sensor reporting error, system won't hang up
July 1, 2007 7:06:51 AM

Well, I just couldn't let it go. So I pulled it out yet again, and sure enough, one of those silly little pins had popped loose. I fiddled around and finally got it to hang under the ledge well enough, I hope. Got it back together, and it’s running about even with the system temp around 30ish, even after a series of cycled reboots, still holding steady.

That’s the good news. Bad news is that vista is still flaky. I’m not sure if it got corrupted somehow due to the earlier system instability during it’s install, or if I have other problems (possibly heat damage?). It couldn’t seem to hold a network connection for browsing the web, or coping files across my LAN. Explorer itself kept locking up too.

I tried Vista months ago and hated it bad enough to go back to XP. Now I have no choice due to work, so I figured I would build a new system that could keep Vista from being such a dog. So here I am… Just to see if it’s Vista in general having trouble with some hardware aspect, bad hardware, or whatever; I am currently installing XP to see how that runs on the new hardware. If that works, I’ll use that for a week or two until I’m confident in the hardware, then give Vista another whirl (after previously downloading BIO patches and such so I don’t need the network right away!).

I’ll post how it turns out, though it is getting late here in dusty AZ…
July 1, 2007 8:34:34 AM

Well, XP runs MUCH better than Vista, but still have flaky network, though the behavior has changed. Now it finds the network, tries to “acquire an IP”, then says network cable is unplugged. That series cycles about once every 2-3 seconds. And that’s after the chipset driver install, can’t get to the net or another machine to get an update if there was one. So now I’m figuring that’s a mobo issue. I’ll see if I can find a plugin net card to try tomorrow.

Oh, and on the temps, the CPU has been staying a bit cooler than the reported system temp, no higher than about 43 as I recall. Thanks MUCH for prompting me to look at that heat sink again.

That Ethernet port has been trouble since the beginning. It won’t turn loose of the plug because the face plate (not the punch out in the case, but the plug itself) is too small and has a small edge sticking up on the bottom that locks the bottom (opposite the latch) so that it can’t be removed. Add to that the chipset sink flopping around loose when it came in, and I’m not so sure I even want a replacement Gigabit mobo. In the past I’ve used mostly ASUS with a few Abit/Soyo here and there. Is Gigabit a good board? Seems that what I’ve read indicates this particular board is very good on all points. Not quite sure what to think now.

If it must be replaced (as I now suspect), should I just RMA this one and get a comparable ASUS? Is it likely that this chip is damaged by the over heat? XP seems ok with everything now except for the network hardware…
July 1, 2007 9:31:54 AM

Test stability with Orthos, than use some heavier benches like PC Mark series
If system is stable on 3 loops of PC Mark and on Orthos, then it is less probable heat damaged something. Maybe your LAN is bad
July 1, 2007 7:09:01 PM

Well, looks like it's the integrated NIC that's bad. But I can't find my old plugin card, should be around here somewhere...

This post is being made using the cable (and router port) that was reported as unpluged last night. The new system fired up this morning (after being fully cooled all night) and seemed to have decided to work. But very quickly things went down hill. Throughput started slowing rapidly, then just stopped.

I haven't yet seen the "no cable" message again, but the mobo seems to be defective, and perhaps heat related? Recall that the main chipset heat sink was loose when the mobo was first opened? It is distinctly hot to the touch, so it's making contact, but maybe not fully? But if that were the case I would expect to see other issues rather than just network.

Correction, it JUST decided to start repeatedly popping up the “unplugged” balloon just like last night. So it must be heat related…
July 1, 2007 7:11:21 PM

BIOS reports 42* system and 38* cpu, for what it's worth...

And I'll have to burn a CD to get a stress program onto that machine with no network, which is problematic since the burner is on that machine... <sigh>
July 1, 2007 7:28:13 PM

Well, good news.... of sorts.

I found the NIC and plugged it in. Works just like it should, no other problems found... yet.

So, I've downloaded Orhos and will give it a run. Any other utilities I should use? What is best for monitoring temp while in the OS? What is that nice utility that I see on lots of this sites images that displays temps and all in a vertical column?

So at a minimum, looks like I'll have to RMA the mobo. Any suggestions on a different brand? Or is Gigabyte good and I just got a bad one?
July 1, 2007 8:50:52 PM

BIOS and driver updates (from web) make no difference.

One more data point. Even though my meter reports +12.08V, the BIOs (and windows monitors) report ~0.7V-3.3V on 12V. -12V on my meter is about -12.6V and reported by BIO/Monitor is -16.9V. I think it's pretty clear I have a DOA motherboard. Any other possibilities I might be missing?

I'm running Orthos now, default settings. CPU temp is up to 55 and holding. No errors so far. Any options I should try?
July 1, 2007 9:43:56 PM

Quote:
CPU thermal protection is internal to CPU: no related to any bios upgrade
If it was a bios thermal sensor reporting error, system won't hang up


Ah good point. Now I feel dumb :( 
July 1, 2007 9:51:38 PM

If you reseated the heatsink, you should have scraped off the old thermal compound and applied fresh, something like arctic silver 5 or equivalent. Make sure you follow the instruction to the letter. Any thermal compound is a one shot deal. Unless you are overclocking, your idle CPU temp should be in the low to mid 20's C.

Stock CPU voltage should be 1.187 - 1.325v.
July 1, 2007 9:57:55 PM

Best thing I can say to do is definitely RMA that board, use an older system for a week or two, however long it takes them to get you the new board. Gigabyte boards have been used for years, and are traditionally good boards. I have heard many people recommend that board, so it would be worth RMA'ing it. Unfortunately that is just the nature of electronics. They may work at the factory if they test them (don't know if they do or not), but once shipped damage can happen and the board goes bad. Thank God for warranties :D 
July 2, 2007 12:07:08 AM

As stated, I did wipe the old compound off, cleaned with alcohol and applied new “Extreme Cooler Compound”. Instructions are long gone, but I recall admonishments to not use “too much”. So I started with a small dollop on the smaller contact surface (since it therefore defines the total contact area) and smeared it somewhat evenly with a smooth clean spoon handle. Did I miss it badly?

It should be 20*C??? The ambient in my house is 27, and in the case (located in my office, which is the hottest room in the house due to equipment) is something like 30+. Before you say something about the temps, I live in Phoenix and creep the house temp up slowly through the spring so that going outside, 115 right now, is not so painful. Without active cooling there is no way it could be in the low to mid 20s, and assuming it generates any heat at all, it would have to be into the low 30s even at idle. Am I missing something?

I decided to give Vista another go and so far so good. I finally got Vista to accept my old 3COM card and it’s working very well now. But all I have to do to lock it up is turn on the embedded NIC. Hard power off and disable in BIOS, no more problems.

I hope NewEgg is open tomorrow, should be. If so, back it goes. I’ll just buy a new identical board and have it sent ASAP. I’ll be running this board till the new board gets here, swap them, then I’ll return the first on RMA for refund. That way I can get on with my work in the mean time. That work is what prompted the whole Vista, and therefore the upgrade, in the first place; so just going back for a week or two would have been a problem. I really wanted the long holiday weekend to get the new system up and stable. Little did I know it was going to take almost all of it! <grin>

The only remaining unknown problem is the misreported voltages, which probably point to other board issues I suppose? Anybody got any info on this? I don’t know where those sensors are, but as long as they are not in the CPU, getting a new mobo should fix it, right?

Thanks much to all of you for taking time to help. Getting me to look more closely at the heat sink (shoulda known that myself… sigh…) was the key to the whole thing. Once that was fixed, things got back into the realm of something I could work with.
July 2, 2007 2:10:28 AM

over how much time are we talking? I am considering upgrading to a C2D proc, but with a very limited budget constraint. But I also don't want to burn my proc. Depending on the amount of time needed, I will get a after market HSF later on...
!