Computer overheating

[velis]

Sorry for the long post:
Background:
I recently had my NB fan fail on me on my ASUS A8N-SLI DeLuxe. Since the computer itself is stil under warranty, I had it replaced with a new one (not much better IMHO). It came together with a new heat sink.
At the same time I added a new HD (WD 3200KS) and rearranged the cables in my Antec Sonata I so that far fewer are visible. I tucked the data ones under the MB and most of the power cables are now fastened in the 5.25" drive bay area. Overall I think my case clutter is now much less than it was before I replaced the fan.
Other than NB fan I also have Zalman CNPS 7000 AlCu for CPU, Papst 12cm for case exhaust (just behind CPU socket) and an Antec TruePower II 380W PSU's two fans. GFX card is passive 6600GT.

The problem:
Since the NB fan failed, my NB tends to overheat. Before I replaced it with a new one, I had a 12cm fan blow on the bare heat sink.
I don't know the temps with the original heat sink + fan, but I had that fan turned down to 4000 RPM and it never overheated. I never knew my NB fan failed until I restarted the computer due to some SW installation. I could've beed dead for days without me knowing it. I didn't play much at the time though 😉
While I had the 12cm "replacement", the NB would (according to SpeedFan) start at about 48°C and reach 58°C before it started overheating (there were some interesting glitches on the display when it overheated)
Now I have the replacement and the NB temp never drops below 58°C 😕 The temp is quite stable though, but the NB will start overheating at 62°C.
Due to high overall temperatures I turned my Papst all the way to the max speed allowed by FanMate. CPU fan is still controlled by MB BIOS depending on CPU temp.

The question:
Why is my NB now hotter than it was before the heat sink was replaced? Note that the ambient temps didn't change much. It was hot while I had 12cm and it's just as hot now. The CPU and case temps haven't changed.
How can it be that with the new heatsink the overheat temp is 4°C higher than before? This would suggest a much better heat transfer between the NB and it's HS, but is it really that? And why the heck isn't the temp lower then?
Any suggestions how to fix this problem?
I'm sorry to say this, but I'm beginning to think there was some permanent damage done to the NB while I was waiting for the replacement HS + fan.

I don't want to bug my supplier with this until I've exhausted all other options. The computer has been in heavy use since feb. 2005 and already serviced 4 times due to other components failure 🙁 (note the PSU - it's a replaced one....).

 

[nullpointerus]

1. Where is the NB in relation to the passively cooled 6600GT?
2. What is the temperature of the GPU?
3. Did you overclock the GPU?
4. Can you try underclocking the GPU for a bit?

 

[velis]

NB is in terms of airflow "before" the GPU which is at 66°C idle, 100°C loaded. This did not change from original fan to replacement fan and can not be the reason.
There is a slight chance that the third disk recently added would somehow change the airflow through the case so that NB would not receive the fresh air it once did, but I believe this option to be a remote one at best.

 

[dawza]

The A8N-SLI Deluxe does not have a temperature sensor for the NB. In fact, there are only a handful of NF4-based boards that have a built-in sensor (some of the DFI mobos come to mind).

That said, I am not sure what the temp in question is in reference to. The fact that you have noticed display glitches suggests a GPU issue. If I understand correctly, you originally had a 120mm fan pointed at the chipset, and now that you have replaced the chipset HSF, you no longer have the 120mm fan blowing on it? If so, it very well may be that the 120mm fan was providing enough airflow to your passive 6600GT, and now that it is gone, GPU temps occasionally reach the point where artifacting becomes visible (although if that 62C reading is indeed your GPU, it is way too low to be the cause of artifacting).

Here is what I would suggest:

1. Put the 120mm fan back on, but this time direct the airflow directly at the GPU's heatsink.

2. Monitor GPU temperatures with either Everest Home edition or NVtemplogger (google for links). See if GPU temps drop with the fan ON vs. OFF (should drop a few degrees C during non-3D activity).

3. Run a graphics-intensive program- RTHDRIBL (google it, will be the first hit) is useful, as you can run it windowed, allowing you to monitor temperatures concurrently. Do this with the fan OFF. See if you notice the artifacts as GPU temperatures increase. If so, shut RTHDRIBL down and let the GPU temps drop.

4. Turn the 120mm fan on, and let GPU temps drop back down to idle levels. Fire up RTHDRIBL again, and see if the additional airflow from the 120mm fan prevents artifacting.

If you see artifacting at temps below 85-90C for a GPU, then something is wrong with the card and/or the card's temperature sensor. GPUs are rated for upwards of 115C, although it is generally not recommended that one run them in the triple digits for the sake of the card's health. Also, "safe" does not mean artifact-free, although as I just stated, even 85C is not unusual for a GPU (especially a passively cooled one), and should not be the reason for artifacting.

As far as the NB is concerned, it seems that you are concerned with having a quiet system, so I would strongly suggest that you look into replacing the stock NB HSF with a passive solution. The Zalman NB-47J is popular, although not recommended for highly overclocked systems with inadequate airflow in the NB region. The new Zalman NB heatsink, the NBF-47, looks promising. For $20 total, you can also order the heatpipe NB cooler as seen on the A8N-SLI-premium from the Asus estore. I have the A8N-SLI-deluxe, and replaced the whining stock NB HSF with the heatpipe, and could not be happier.

It is difficult to say if your NB was damaged during the period between when the original NB fan failed and when it was replaced. There are scattered reports of people who have run their A8N mobos (unknowingly) with dead NB fans with no apparent ill effect. I imagine that these same people were not heavily overclocking and had some air passing over the NB area as well, though. Your problem really does sound graphics card-related, though, and that is where I would begin troubleshooting.

 

[velis]

No, you got that wrong. I only put the 120mm AFTER the original fan failed and also AFTER the computer exhibited the first overheat. I installed it back today, but in the designated position just after the drive bays. Previously I had to pint it directly to the NB heat sink to achieve some degree of cooling. Installing it helped somewhat dropping the temps about 3°C for the NB and 10°C for the GFX card. Will test some more today to see what happens.
It also isn't the GFX card. The glitches will appear when it's at 50°C as well as at 100°C. As for the three digit temperature, I was worried about that at first too, but made my peace with it when I was assured by Gigabyte tech support that the temp is normal for this card. Having no problems with that temp also helped 😉 Oh, I also replaced it with a spare one to see and the glitches appeared with the replacement as well.
What I'm suspecting to be wrong here is the PCI express interface on the NB. That would explain why the glitches always exhibit themselves on the video and no other problems occur. Perhaps it might also be the hypertransport interface, but then I would get some filesystem corruption, wouldn't I? Will try reducing the multiplier.
Thanks for sensor heads up. I assumed the particular temperature was the NB temp since it is actually the only temp reported by SpeedFan that exhibits the correlation with NB overheating. The other temps remain stable, only this one raises through the roof 😉