Hi there.
The problems started when I decided to install an Arctic-Cooling NV silencer 5 on my MSi 6800GT AGP. The screws used for the Voltage Regulator heatsink interfered with the cooler (They were taller than the heatsink itself), so I had to replace them. I accidentally knocked off the Voltage Regulator heatsink in the process, and put it back on without doing anything. I heard that you should never re-use thermal paste / pads, but this heatsink had a very thick thermal pad underneath it, and replacing it with paste wouldn't be very smart since it would probably start leaking everywhere.
Everything was fine after this, and I even managed to OC to 425/1225 without any problems. However, about 2 months later my games started to crash to desktop very often, and I would get occiasional nv4-disp.dll BSODs. I thought it was a driver issue, since my core temps never exceeded 60c. However, installing different drivers didn't help, and one time when I was playing C&C Generals I got a "Serious" error, stating that it was probably caused by malfunctioning / overheating hardware. This gave me an idea. I decided to underclock my graphics card to 320/900, and.. it worked! No more crashes to desktop!
Ofcourse, I was not satisfied yet. Underclocking my graphics card is not my idea of fun, so I decided to try and fix it. I took off the Silencer and the Voltage regulator heatsink. I then removed the thermal pad, and dremeled a bit of copper off the screwholes to decrease the distance between the voltage regulator chips (The heatsink was cooling 3 tiny chips) and the heatsink. Note that I didn't touch the part of the heatsink that makes contact with the chips, only the screwholes were too thick. I then put a very small amount of thermal paste (Cooler Master PTK-blabla) on the chips, put the heatsink on top of them and used screws to tighten it. I cleaned the memory chips / core / Silencer, and put new thermal paste on the core and silencer. I then put the silencer back on, and put the card in my system. So far so good.
However.. When I turned on my system, the monitor didn't get a signal! So I took out the graphics card, removed the heatsinks and made sure there was no thermal paste shorting the card. This was not the case, and I don't even think the Coolermaster stuff conducts electricity, so I put the card back in to try again. This time the monitor did get a signal, but the screen was messed up. All I saw was miscoloured pixels. No BIOS screen or anything resembling it. I turned off the computer and slapped a 80mm fan on the side of the Silencer to blow air on the VRM heatsink using strong double sided tape.
I let it rest for 5 minutes, and turned it back on. Again, the screen was completely messed up. At this point I was afraid that I had wrecked my card, since I don't have the money to buy another one. I removed the graphics card again, and made sure there was no dust on the card or in the AGP slot. I also made sure the Silencer made contact with the GPU core. I put it back in and turned on my computer. Surprisingly, the screen was fine, and it booted into Windows. However, as soon as I accesed the Nvidia Control panel to check my temperatures, the screen got all messed up again. I rebooted again, and decided to use Rivatuner to check my temps. 47c core and 30c ambient. At this point I reinstalled drivers, knowing that it probably wouldn't help, but I had to do something. After reinstalling the drivers (84.21) I checked my temperatures in the Nvidia control panel again, and it showed 47c. The card did not crash. I decided to test the card by playing the Dark Messiah demo. I launched the game, and watched the intro movies.. so far so good.. However, as soon as it reached the main menu, where it was supposed to render a 3d character, the screen got all messed up. I rebooted and tried it again, with the same result. I tried downclocking the card with RivaTuner, but as soon as I pressed the test button, the screen got messed up.
I messed something up, but I don't know what. Was it a bad idea to put thermal paste on the VRM heatsink? I am certain that the heatsink makes contact with the chips, even without paste. Should I get a new VRM heatsink, or does it sound like something else is the problem?
I'm going to bed now, but I will test the card again tomorrow. It appears to improve with every reboot, so who knows..
EDIT: If it's important.. I've had this card for about a year now.
The problems started when I decided to install an Arctic-Cooling NV silencer 5 on my MSi 6800GT AGP. The screws used for the Voltage Regulator heatsink interfered with the cooler (They were taller than the heatsink itself), so I had to replace them. I accidentally knocked off the Voltage Regulator heatsink in the process, and put it back on without doing anything. I heard that you should never re-use thermal paste / pads, but this heatsink had a very thick thermal pad underneath it, and replacing it with paste wouldn't be very smart since it would probably start leaking everywhere.
Everything was fine after this, and I even managed to OC to 425/1225 without any problems. However, about 2 months later my games started to crash to desktop very often, and I would get occiasional nv4-disp.dll BSODs. I thought it was a driver issue, since my core temps never exceeded 60c. However, installing different drivers didn't help, and one time when I was playing C&C Generals I got a "Serious" error, stating that it was probably caused by malfunctioning / overheating hardware. This gave me an idea. I decided to underclock my graphics card to 320/900, and.. it worked! No more crashes to desktop!
Ofcourse, I was not satisfied yet. Underclocking my graphics card is not my idea of fun, so I decided to try and fix it. I took off the Silencer and the Voltage regulator heatsink. I then removed the thermal pad, and dremeled a bit of copper off the screwholes to decrease the distance between the voltage regulator chips (The heatsink was cooling 3 tiny chips) and the heatsink. Note that I didn't touch the part of the heatsink that makes contact with the chips, only the screwholes were too thick. I then put a very small amount of thermal paste (Cooler Master PTK-blabla) on the chips, put the heatsink on top of them and used screws to tighten it. I cleaned the memory chips / core / Silencer, and put new thermal paste on the core and silencer. I then put the silencer back on, and put the card in my system. So far so good.
However.. When I turned on my system, the monitor didn't get a signal! So I took out the graphics card, removed the heatsinks and made sure there was no thermal paste shorting the card. This was not the case, and I don't even think the Coolermaster stuff conducts electricity, so I put the card back in to try again. This time the monitor did get a signal, but the screen was messed up. All I saw was miscoloured pixels. No BIOS screen or anything resembling it. I turned off the computer and slapped a 80mm fan on the side of the Silencer to blow air on the VRM heatsink using strong double sided tape.
I let it rest for 5 minutes, and turned it back on. Again, the screen was completely messed up. At this point I was afraid that I had wrecked my card, since I don't have the money to buy another one. I removed the graphics card again, and made sure there was no dust on the card or in the AGP slot. I also made sure the Silencer made contact with the GPU core. I put it back in and turned on my computer. Surprisingly, the screen was fine, and it booted into Windows. However, as soon as I accesed the Nvidia Control panel to check my temperatures, the screen got all messed up again. I rebooted again, and decided to use Rivatuner to check my temps. 47c core and 30c ambient. At this point I reinstalled drivers, knowing that it probably wouldn't help, but I had to do something. After reinstalling the drivers (84.21) I checked my temperatures in the Nvidia control panel again, and it showed 47c. The card did not crash. I decided to test the card by playing the Dark Messiah demo. I launched the game, and watched the intro movies.. so far so good.. However, as soon as it reached the main menu, where it was supposed to render a 3d character, the screen got all messed up. I rebooted and tried it again, with the same result. I tried downclocking the card with RivaTuner, but as soon as I pressed the test button, the screen got messed up.
I messed something up, but I don't know what. Was it a bad idea to put thermal paste on the VRM heatsink? I am certain that the heatsink makes contact with the chips, even without paste. Should I get a new VRM heatsink, or does it sound like something else is the problem?
I'm going to bed now, but I will test the card again tomorrow. It appears to improve with every reboot, so who knows..
EDIT: If it's important.. I've had this card for about a year now.