Tests -- at the bottom !
We have tested several types of thermal paste and we used our Peltier testrig for this purpose you can see a pic off the rig in action here .. We attached a 58W peltier during the experiment to really torture the paste. Well we discovered one thing pretty fast - the variations are within 3'c for all off the tested pastes thermal pastes containing silicone performed worst.... - not much to write home about .. in other words a paste is a paste.. ?? Yes and no .. the most difficult paste to apply - 3dfxcools "goop" was actually the best ..when applied correct. One tip on "howto" apply this paste : heat it under the water tap letting hot water run over the syringe for at least a couple of minutes - or even better - boil a kettle with water - ensure that the cap is tight on before dipping the syringe into the water.. This goop when applied correct lowered the heatsinks temperature 2'C more than the best of other tested heatsink compounds
The other "thinner" types off thermal compound is best applied using a soft brush. Apply only a thin evenly layer - pic - off course - I assume you have gritted/polished both the CPU slug and the heatsink to ensure they are 100% leveled ...
BUT - a while ago I was refurbishing a turbocharger and I bought a tube off Bostik never seez at the local gas-station.. I tried it - just for fun - and found it to be better than any Thermal Compound I ever have tried ! ... :) It can withstand heat up to 1093'C (2000'F) and is called a "extreme pressure lubricant" and the stuff is thin and easy to apply - pic off the dark silvercoloured compound - Pic off the tube - here .. I have seen "better than Thermal grease" compounds advertised on a website .. I actually think it's about the same stuff as the never seez I got ! Anyway this stuff was able to lower the heatsinks temperature 3'C more than the 3dfxcools goop..and a whooping 7'C more than the worst heatsink compound we tested..
Some people are also using ordinary grease as "Thermal Compound" .. I tested it a year ago and found it to be far worse than "normal" pastes...But if you dont have anything "normal" it is better than nothing..
And what about copperpaste -- not too bad just be sure you get a fine compound and you can expect better heattransfer than ordinary grease and even silicone based paste with the "right type" copperpaste..
Thermal pads ... remove them --FAST ! And don't listen to the sites telling you to use a knife (!) to remove the remainings off the heatpad after you have removed the factory heatsink - use electric cleaner - pic - or lighter fluid (gasoline) and a soft cloth to remove it...
This is for you VapoChill users : stay away from silicone based and types containing too much water - I actually found a way off dealing with the water problem : put it in the oven ! But watch out so you don't end up with a dustpile :)
NOTE : its off extreme importance that your heatsink and CPU slug is perfectly flat and that you don't apply to much compound...how to lap in your CPU here and the same procedure can be used for your bx-chip heatsink - graphic cards heatsinks : in other words all heatsinks !
Danfoss metaloxide-paste took our C366 to 660MHz with OcShoot's peltier .This paste is only sold by GlobalCool - this paste beats everything we have tested previously. This type is a new and improved silver based paste. An older version with the same number is still available- but it's not as good as this one. Guess what -- Asetek supplies this one with their new "socket VapoChill" !
Here is this heat conduction compound under test :
-35' Celsius and not frozen - highly recommended to you VapoChillers and peltierusers !
Related articles :