System instability after GPU coolers install

[Soup]

I recently installed an Accelero Xtreme 7970 VGA cooler on each of my Reference Sapphire HD 7970's.

The cards stay far cooler and are much quieter than the reference coolers. The problem is whenever they are put under load, after a few minutes, my monitors will go black(as if put to sleep) and the system will become unresponsive. I cannot wake it and must hard reset it, in order to bring it back up.

I was initially afraid that maybe the thermal glue holding on the various small heat sinks was shorting something out on the board. I just learned however that the "G-1" thermal glue is non-conductive.

The only thing I can think of that I changed is the crossfire cable. I am now using the back connectors instead of the front.

Help?

 

[erocker]

The only thing I can think of that I changed is the crossfire cable. I am now using the back connectors instead of the front.

Try swapping them. Also, you should be able to view the VRM temperatures using this program: http://www.hwinfo.com/download64.html
Make sure those aren't overheating.

 

[Soup]

I've attached the HWiNFO64 log.
It appears that it is certainly the VRM overheating. The problem now is that all of the glue-on heat sinks appear to be properly affixed to their corresponding board parts. Can you think of a way I could narrow down which VRM is the culprit?

 

[Bo$$]

seems like a few others have this issue too!
http://hardforum.com/showthread.php?t=1674481&page=6

get some better thermal tape, or use a copper shim

 

[Soup]

I suppose I'll re install those annoying little sinks. One more thing, would over heating VRMs cause the symptoms I listed? It seems reasonable to me... Opinions?

 

[erocker]

I suppose I'll re install those annoying little sinks. One more thing, would over heating VRMs cause the symptoms I listed? It seems reasonable to me... Opinions?

Yes, absolutely. Are you positive you have everything heatsinked that needs to be heatsinked? If possible could you provide a picture depicting what you have heatsinked?

Thanks!

 

[Bo$$]

this is how they should look

Try reseating it, it would definately help.

 

[Soup]

Yeah, I'll pull the cards tonight when I get home... Still at work.

 

[Bo$$]

actually, i took a look at your XLS document, your VRMs temps are perfectly fine!

Something else is at fault here....

reseating might help still. one looks a tad high...
Try slightly unscrewing the card. You may have overtightened the card

 

[erocker]

actually, i took a look at your XLS document, your VRMs temps are perfectly fine!

Something else is at fault here....

reseating might help still. one looks a tad high...
Try slightly unscrewing the card. You may have overtightened the card

98c isn't really fine. Granted this is the temp while the cards are still working.

 

[Soup]

That XLS only shows a few seconds of load. The temps of the VRM's on card#0 jump considerably while the VRM's on card#1 stay at a constant 33°c. When I get home I will stress the system again to the fault point while logging the results.

 

[Bo$$]

That XLS only shows a few seconds of load. The temps of the VRM's on card#0 jump considerably while the VRM's on card#1 stay at a constant 33°c. When I get home I will stress the system again to the fault point while logging the results.

thats crossfire, disable it and see if it crashes

 

[Soup]

I removed the card in slot one, this is the card i suspected of being my problem child, and ran load on the remaining card. No faults. Runs great.
Pulled the cooler off the die and two sinks immediately fell off, one from the VRM's. That thermal glue is a joke. Hopefully I can find something better.

Thanks for the help guys.

 

[Soup]

I have an update on this issue.
I recently purchased some thermal tape from a member here at TPU. I cleaned off the thermal glue leftover from my initial install of the VRM heat sinks, applied the tape to the now clean sinks and set them in place on the VRM's.

No change.
At all!

I'm still getting immediate temperature climbs that will exceed 100c if left to run. I think my last resort is to replace the reference cooler which sucks.
Any ideas?