SEARCH
ASUS Member ID
Password
Forgot password?
Join member

ASUS Support Home Location : Forum > Motherboard > M2N32 WS Professional  
  Forum  
  Topic : [Problem]Chipset overheating problems New Topic
Rumpknull
Member
Posts : 17
Level :
Tech Points : 0
From: Sweden
Posted:2/22/2008 6:33:00 PM # 1

After 3 months of use, my motherboard started to develop problems - initially freezing, then increasing to spontaneuos reboots. I started to get a lot of "GPU reset" messages from my ATI driver, I initially thought my GFX card was faulty, but further observation eventually narrowed the problem down to my northbridge, which was literally running red hot.
 
I had installed speedfan early when I built my new system, and there were two entries there from the ADT7475 sensor chip that were very high - around 100 C - and I thought it was just either the VRM modules or just a misread. Now, when I put a 60 mm fan blowing on the north bridge, one readout dropped from 100 C to 92 C. Further experimentation showed that the first readout is from the northbridge, and the 2nd from the southbridge. My system had been running hot all the time, despite being in an open chassis, with the motherboard lying down so all heat convected upwards freely. I thought the problem might have been lack of thermal compound under the heatsinks, I disassembled my system and checked the heatpipe system - it was outfitted with thermal compund all right. But it was so incredibly light! The entire heatsink and heatpipe assembly cant be more than 100 grams in mass! I reassembled my system and downclocked the chipset as much as possible by reducing the HT multiplier from 5x to 1x. My system is now stable with 2x 60 mm fans and 1x 40 mm fans blowing on the chipset coolers in addition to the 120 mm fan on the big tower heatsink.

What am I going to do? RMA and get another motherboard which may give up after 3 months? My motherboard is stable again downclocked and with heavy cooling, but chipset temperatures are still 95-100 C during light load. Are there more than 1 variant of the chipset cooler?

Here are the total readouts from the temp chips onboard using Speedfan:

Chip:             Temperature
IT8712F-J     34, 28, 36C  (CPU, motherboard, unknown)
ADT7475      99,98,0 C  (Northbridge, southbridge, probably not installed)

System:

K8 65 nm 2x2.1 GHz dualcore
2x2 gig 667 mhz Kingston ECC modules
ATI X1800XT 
A smattering of hard drives

         

  Reply Reply Quote & Reply Quote & Reply
GASUS
VIP Member
Posts : 220
Level :
Tech Points : 42
From: Hong Kong
Posted:2/23/2008 6:38:00 AM # 2
No thermal greese when assembling in factory!!!  ;)

  Reply Reply Quote & Reply Quote & Reply
ASUS SABERTOOTH 990FX R2.0 mainboard; BIOS 1708; CPU Athlon II x4 640; ADATA XPG™ Gaming v2.0 DDR3 1600MHz 2GB non-ECC RAM x 4; WD Caviar Black 1TB harddisk; Liteon DVD Writer; Huntkey Jumper 400W 80Plus power supply; CoolerMaster Case; ATI HD6770 Display Card.
Rumpknull
Member
Posts : 17
Level :
Tech Points : 0
From: Sweden
Posted:2/23/2008 7:13:00 AM # 3
It did have thermal grease or thermal pad under all contact points. The point is, that the entire heatsink + heatpipe assembly is very much lacking in mass. Heatpipes do not help with dissipating heat, they only transport it - and the heatsink mass is very low when the heatpipes are subtracted.. This has to cool the northbridge, the southbridge, a nec chip and the power chips. Has anyone else installed speedfan and checked out the temperature readouts?

  Reply Reply Quote & Reply Quote & Reply
Jolt
VIP Member
Posts : 6
Level :
Tech Points : 0
From: Romania
Posted:2/24/2008 2:01:00 AM # 4
Hi Rumpknull ,

   I have also installed Speedfan and it shows some strange temperatures regarding Nvidia chipset, in my case : [ 93,103,0 ] / C grades - in idle !. But my system is stable so I guess those values are wrong. Use instead Nvidia Ntune, to get NVidia Chipset and  your Geforce temperature (if apply ) , and guess what : nForce temperature : 55C,  GeForce temperature : 47C . Both are in idle. So my conclusion was that speedfan is not reading right temperature level.
 
Ps : Don't leave stock termal paste for your chipset radiator ,it's a cheap crap ,replace it with a brand  thermal paste ( In my case this helpt to reduce chipset temperature with almost  15 C ).

  Reply Reply Quote & Reply Quote & Reply
Sapper
VIP Member
Posts : 2
Level :
Tech Points : 0
From: United Kingdom
Posted:2/26/2008 6:36:00 PM # 5

Quote:
Author: Rumpknull Posted: 2/22/2008 6:33:00 PM

After 3 months of use, my motherboard started to develop problems - initially freezing, then increasing to spontaneuos reboots. I started to get a lot of "GPU reset" messages from my ATI driver, I initially thought my GFX card was faulty, but further observation eventually narrowed the problem down to my northbridge, which was literally running red hot.
 
I had installed speedfan early when I built my new system, and there were two entries there from the ADT7475 sensor chip that were very high - around 100 C - and I thought it was just either the VRM modules or just a misread. Now, when I put a 60 mm fan blowing on the north bridge, one readout dropped from 100 C to 92 C. Further experimentation showed that the first readout is from the northbridge, and the 2nd from the southbridge. My system had been running hot all the time, despite being in an open chassis, with the motherboard lying down so all heat convected upwards freely. I thought the problem might have been lack of thermal compound under the heatsinks, I disassembled my system and checked the heatpipe system - it was outfitted with thermal compund all right. But it was so incredibly light! The entire heatsink and heatpipe assembly cant be more than 100 grams in mass! I reassembled my system and downclocked the chipset as much as possible by reducing the HT multiplier from 5x to 1x. My system is now stable with 2x 60 mm fans and 1x 40 mm fans blowing on the chipset coolers in addition to the 120 mm fan on the big tower heatsink.

What am I going to do? RMA and get another motherboard which may give up after 3 months? My motherboard is stable again downclocked and with heavy cooling, but chipset temperatures are still 95-100 C during light load. Are there more than 1 variant of the chipset cooler?

Here are the total readouts from the temp chips onboard using Speedfan:

Chip:             Temperature
IT8712F-J     34, 28, 36C  (CPU, motherboard, unknown)
ADT7475      99,98,0 C  (Northbridge, southbridge, probably not installed)

System:

K8 65 nm 2x2.1 GHz dualcore
2x2 gig 667 mhz Kingston ECC modules
ATI X1800XT 
A smattering of hard drives

         



As anyone contacted ASUS regardig this issue. 

My Windows Hoem server is built around this motherboard and until this weekend has been stable as anyhting.  It then started to crash, or freeze regularly.  It became apparent that after a period of cooling, the system was stable for longer.

Having read this thread yesterday, I have placed a 80mm fan close to what Ithink is the component in question which seems to give soem stability.

But it is still at the moment ineffective as a PC due to the instability issue - Having just re-installed the OS, I still cant get it long enough to install Speedfan!

How did you down clock the chipset to impove stability?

A quick glance at soem online retailers seems to suggest that this motherboard is no longer available.  What other motherboard has 10 SATA connections and accepts an AMD Am2 CPU?

Adrian

  Reply Reply Quote & Reply Quote & Reply
Rumpknull
Member
Posts : 17
Level :
Tech Points : 0
From: Sweden
Posted:2/28/2008 4:10:00 AM # 6

Further investigation indicated that my GPU crashes were really GPU crashes. It seems that the default ATI cooler spinup profile was too soft and manually adjusting the fan profile with ATItool solved this. Unfortunately that also meant a lot of noise of the most annoying kind, since the standard fan on an X1800XT sounds like a cordless electric hand drill when running at higher (> 60% speeds). I replaced it with a coolermaster coolviva pro se, which is very quiet, but also takes up another slot because of its greater width.

WRT the motherboard temperature measurement problems, installation of Nvidia Monitor showed crazy temperature values and no readout at all on voltages etc, it appears that it doesnt work for me.

The 4.34 beta 38 release of speedfan can correctly interpret the output of the ADT7475 temp sensor - it seems that the author used temperature readings offset from 0 C when they should be offset from -64 C since that is the bottom of the temperature range the ADT7475 can measure. So my readings were 64 C too large. The new ones from 4.34 are < 40 C in idle. I now noticed that the ACPI temperature is 40 C and that as soon as "internal temperature" or ADT7475 reading #2 reached above 40.5 C the system would freeze or crash. I started to suspect that the "crashes" were the system trying to shut it down itself via ACPI. Upgrading the bios to 1703 from 1501 seems to have fixed my crashes, I just ran a series of tests and managed to raise the readout of internal temperature (and other sensors) above 41 C without crash.

Adrian, if you want to clock down the chipset you have to reduce HT speed to minimum (200 mhz) and HT multipler to 1x on all connections. However, it seemed to be a bios issue as i wrote above. Which bios version do you use? Which type of CPU cooler do you use? A tower type that blows downward or a wind tunnel type that has the fan blowing sideways?

  Reply Reply Quote & Reply Quote & Reply
Rumpknull
Member
Posts : 17
Level :
Tech Points : 0
From: Sweden
Posted:2/28/2008 8:08:00 AM # 7
FWIW, I managed to retrigger the shutdowns. I was playing total annihilation when my system suddenly froze, then a few seconds later it shut down. I decided to find the new limits, which appear to be some degrees higher than before. I first tried a slow heat build up while running orthos for cpu stress and atitools find artifact scanning for additional stress on the chipset. I had slowly reduced the fan speeds while watching the temperature creep up and eventually my system autoshutdown with ADT7475 sensors at 51 and 49 degrees. Next step was sitting with the system in idle with the fans turned off. It eventually shut down after several minutes of slowly rising temperatures which were not as high as during the stress test but still much higher than before this bios. With automatic fan management I can now have a reasonably silent system.  
The ACPI temperature limit is still 40 C so I guess it doesnt have anything directly to do with this.

  Reply Reply Quote & Reply Quote & Reply
MF
Member
Posts : 2
Level :
Tech Points : 0
From: Germany
Posted:3/17/2008 11:30:00 PM # 8
I have solved the "Overheating/Freeze - problem" for me this way:
The Nec chip (pci-x bridge) covered by the heatpipe has some kind of silicon pad. This pad is about 3mm height - too much for the northbridge chip "hanging in the air" with a lot of heat-conductive paste. So I changed the silicon pad (to some with about 1mm) and fitted the other components with good heat-conductive paste (AS5). You have to take care that the northbridge chip makes a clear rectangular mark on the copper side of the cooler and having good thermal contact. This was everything I did. The freezes went away, even running Prime for several hours (previously freezed after some minutes).
Hope this works for all having the "freeze" problem.

greetings

  Reply Reply Quote & Reply Quote & Reply
Rumpknull
Member
Posts : 17
Level :
Tech Points : 0
From: Sweden
Posted:3/26/2008 10:36:00 AM # 9
MF, after I read your post I realised that there was one more thing to try: I increased the voltage for the southbridge from 1.5 to 1.6 volt, shutdown temperature now increased by 10 C which means that the computer is now stable within reasonable temperature ranges. Furthermore, more experimenting has shown that the ADT 7475 sensor #1 appears to be the voltage regulator - whatever it is, its temperature is highly dependant on CPU load, and once the CPU load is gone, it drops very fast, within seconds. The internal temperature sensor has to be the southbridge.

  Reply Reply Quote & Reply Quote & Reply
MF
Member
Posts : 2
Level :
Tech Points : 0
From: Germany
Posted:4/16/2008 6:32:00 PM # 10
Interesting...
In my bios (1501) it seems that I'm not able to change the SB voltage from 1,5 to 1,6 volts... the bios always put it back to 1,5???
The ADT7475 temarature thing is the same for me in the speedfan readings. But how did you find out that this temp must be the voltage regulator temperature? If you're really sure I could try to do some more voltage regulator cooling enhancements.

  Reply Reply Quote & Reply Quote & Reply
1 2  Next  Last
Previous Topic:: Next Topic
Feedback for our Website ModeratorModerator: New Topic Related FAQ Related FAQ Forum Home Forum Home
Forum Model :
Legal Information | Privacy Policy | Sitemap | Contact Us | ASUS Global
  ©ASUSTeK Computer Inc. All rights reserved.