MX-2 Thermal Compound

[jtleon]

Small margins of improvement do not indicate a better compound...

After visiting all these sites that review MX2 against the competition, I can only say that none of them have convinced me that AS5 does not reign supreme.

Frankly, a compound's true performance cannot be measured without precisely identical applications of each competing compound. I have seen no site that actually recognizes this fact and puts such into practice.

Even AS5 can yield crappy results if it is not applied correctly. I know from my first try in mid-winter months with office temp of 64°F. I did not know I needed to preheat anything at that time (preheat surface of processor, heatsink, tube of AS5, etc).

My first application of AS5 gave me around 65°F load temps and with the same HSF my last AS5 application achieved around 58°F. Both cases having all other factors being the same. AS5 has such high viscosity, that it must be spread with something that will yield a film residue, rather than a puddle, or smeared imprint.

In fact, if you can just begin to see the cpu markings through the AS5 film, then you have achieved an optimum AS5 application. Only a very, very tiny amount of AS5 is needed (pinhead bead) for most cpus. I use a ballpoint pen cartridge to "roll" out the bead into a film that covers the entire mating surface. This takes time, but the result is worth it.

Regards,
jtleon

 

[Mussels]

spread? most of us follow the instructions, put a blob in the middle and let the pressure do the spreading.

 

[jtleon]

Indeed - the blob method is not optimum

spread? most of us follow the instructions, put a blob in the middle and let the pressure do the spreading.

The blob method is great if you are clamping to pieces of 200 grit sandpaper together, but in most cases, the cpu and heatsink mating surfaces are of much finer quality, and only a thin film of compound is necessary to fill the very small voids.

Note that the blob approach, thanks to high compound viscosity, will never allow the entire mating surfaces to be covered (i.e. transfer heat).

Regards,
jtleon

 

[Mussels]

The blob method is great if you are clamping to pieces of 200 grit sandpaper together, but in most cases, the cpu and heatsink mating surfaces are of much finer quality, and only a thin film of compound is necessary to fill the very small voids.

Note that the blob approach, thanks to high compound viscosity, will never allow the entire mating surfaces to be covered (i.e. transfer heat).

Regards,
jtleon

covering the entire surface wasn't a good idea on 939, as you'd get suction and the CPU would get torn out of the socket when removing the cooler.

Also, is it needed? heatspreaders negate it a bit, since you only need grease above the cores, not across the entire thing.

If you've got some links or anything to back you up, fire away - but in my experience the blob method when done properly, yields temps as good as painstakingly spreading it.

 

[jtleon]

My experience is similar...

covering the entire surface wasn't a good idea on 939, as you'd get suction and the CPU would get torn out of the socket when removing the cooler.
Also, is it needed? heatspreaders negate it a bit, since you only need grease above the cores, not across the entire thing.
If you've got some links or anything to back you up, fire away - but in my experience the blob method when done properly, yields temps as good as painstakingly spreading it.

My experience is similar - when removing a cold heatsink. However, the suction problem disappears when only a film of compound is spread on the IHS (the film is almost translucent). Typically, suction is caused by using too much compound and trying to remove a HSF when the system is cold.

Understand that the compound is the least thermal conductive substance in the joint, and thus the less that is used, the better the heat transfer. As I mentioned before, I only apply a small pin head volume of compound and carefully spread that out using a ball point pen refill cartridge (like "rolling" out the dough).

I'll look for some links for you, but let me share my engineering experience with you. Consider the thermal conductivity of copper (the typical IHS material and better HSF material in use today) to be around 400 W/m-K. The thermal conductivity of Arctic Silver 5 is 8.9 W/m-K. As you can see, heat flows much more easily (4500% faster) in solid copper than in AS5. This means that the core heat flows to the outer edges of the IHS almost instantly, so that the total IHS surface is a uniform temperature - no hot spots. The heat flow from the IHS to the HSF is directly proportional to the area of coverage of the AS5, such that if only 50% is covered, then only 50% of core heat may pass into the HSF. This fact alone is justification for spreading the compound for full coverage, regardless of the blob mentality.

I cannot find the W/m-K rating for MX-2 anywhere online, however I suspect it is much lower than AS5. If you find such a rating, please post it. For example, event Tuniq's TX-2 only has a 4.5 W/m-K rating as is reported by TPU.

Regards,
jtleon