Thermal Probes on CPU

[bigge]

I am currently trying to test the effectiveness of various thermal compounds (TIMs) on cpu processors. I am currently trying to find the thinnest thermal probe/thermocouple I can find for this project.
My current idea is that I could place one prob ontop of the CPU and then apply the thermal compound. I would then place another probe and then attach the heatsink.

From there I could measure the heatchange and calculate the thermal conductivity of the product.

Any advice or input is appreciated!
Thanks

 

[scorpinock2]

Just use the CPU thermometer built in, or measure the intake air then the exhaust air OR measure and compare idle temperatures and full load temperatures. If you put a probe on the CPU and then apply the compound and heatsink, that probe will just give you the CPU temperature as it is touching it, it will always be warmer than what It is because its directly making contact with the CPU as well as taking away surface area for the heatsinks base to make contact with the CPU to transfer heat away.

 

[getochkn]

The other problem is many thermal pastes need hundreds of hours of cure time, so if you're going to test, you need to cure according to the makers instructions.

I agree, use the internal cpu temp rather than trying to sandwich a TC in between there, you're not going to get a good reading as the TC is going to be covered in thermal paste and unless you can get it in the EXACT same spot every time, your test is going to be flawed because it will give different temps on different areas of the die, depending on whats underneath that part of the die.

 

[bigge]

thanks for the responses.

I originally planed to use the cpu probes inside the processor as well until i did some reading. The probes inside the cpu are vary in-accurate and are not ment to be used for polling data. The sole purpose of the thermal probes inside the processor is to determine if the current temperatures are potentially damaging and warrants a shutdown.

In addition, the heat dissipation of processors are not constant, depending on what the processor is doing there is a different heat dissipation. The use of stress tests help in all, but the operating system still has control over the computer which means there could still be fluctuations of power consumption. So i was planning on creating something that would have a constant heat dissipation anyways.

what intel does as of 2005
http://www.intel.com/assets/pdf/designguide/302553.pdf: