How to Apply Thermal Compound optimally?

[Sachit]

Firstly that the thermal paste is used to ensure full contact between the surface of the micro core and the surface of the heat. Why? it because if there are gaps (and between two surfaces that are not entirely smooth, as is the case, there) the thermal conductivity of air (which is what would have within that gap) is lower than the thermal paste, ie , for thermal conductivity is better than the hole are filled with air thermal paste. But we must also say that no matter how good the thermal paste (I mean the ones in the market) will never have the conductivity index is aluminum (and in no case copper), for Thus, any amount that exceeds what is necessary is to act as a thermal resistance (and heat dissipation hurt). One comment, it is true that if you catch a chipset heatsink and take away (or the micro for a graphics card) you usually find a good dollop of thermal paste (usually white). I suppose the reasons are twofold. First, the surface of the chip is different to that of a micro AMD XP. There is no well defined core in these components (no overhang) so what makes contact with the cooler is the entire chip, so you have to apply paste on all of it (like for example if you ride a bus with intel heatspreader or a new AMD64).

It is therefore a sum check to ensure that through the pressure of the heatsink itself extends over the entire surface (as if you hit a jam sandwich.) The ideal would be to apply a film on the whole chip but in an assembly (I guess it will be for this) is easier and faster to have a machine that will put a dollop in the center and stop trifles.

Recently I had an idea I am about to see, to deduce what is the ideal case where the application of thermal paste. I want to measure temperatures in the following situations :

1. No thermal paste (or holes filled with air)
2. With thermal paste in the holes (Core).
3. With thermal paste in the holes (Core + sink).
4. With excess of thermal paste (using more pasta)
5. Strictly thermal paste in the holes.

In which I think is closest to the ideal situation. Be obtained by applying a film, after placing the sink for the pressure to extend the pasta .. and then take off the heatsink cleaned with a tissue surface of both the micro and sink .. But this does not bring you back to the 1st case? because I think no, because if we imagine the core surface and sink high power view with something like we saw teeth .. on the tips of which has passed the toilet paper but not on the gaps between teeth (which are holes to fill with pasta) pasta would therefore ONLY in the gaps.

This will maximize the contact surface (the principle that applies to the pasta) but instead minimizes aquaplaning between pulp and core (that gaps do not absorb all of the paste is applied and between micro and sink area, putting thermal resistance) .. I think this would be the best way to apply thermal paste.

Finally, it occurred to me the other day re-reading forum posts and not if one leg of pot or if you have a real basis (must wait to take measurements and see what happens).

[Sachit]

The Experiment :

The case is not about assessing how good or bad the whole: Heatsink + fan + thermal paste. What I see is the temperature differences in relation to the application of thermal paste.

I plan to do so as follows :

• Idle = 15 minutes looking at the desktop without even moving the mouse (only program monitoring temperatures and voltages on my Abit NF7-S 2.0) .. After this period the temperature record.
  
• Full = 15 minutes using a program to check the stability of the micro (the Burn Wizard or something like that).

1st No thermal paste: obviously nothing to mount it without pasta .. First Idle temperature record, then the full ..

2nd and 3rd with thermal paste in the gaps: ie using a film of thermal paste as thick as a cigarette paper. Idle and Full. This section I have considered splitting into two parts to add more cases to the experiment. On the one hand will apply a film thickness of a cigarette paper on the core of the micro. Later repeat measurements with an equivalent amount on the sink also.

4 With excess thermal paste: using more amount of toothpaste on the q already there .. Idle and Full

I am interested in measuring the temperature not to take the microphone itself, if not changes that occur to see what's the most efficient way to apply thermal paste.

[Sachit]

Participants :

• Heatsink: Arctic Cooling ND20L (Aluminium with copper base)
• Microprocessor: AMD AthlonXP 2500 + Barton
• Thermal Pasta: pasta Titan (the typical giving their heatsinks). The Artic Cooling, also brought, but I preferred to use the paste Titan because I included the Artic Cooling thermal silicon instead of pasta (a compound that has silicon in its composition and adhesive, yet less effective as a driver. Besides being adhesive difficult to remove and put the sink)

Simply place the microphone and the sink without any thermal paste.

• Idle: 45º C
• Full: 51.50º C
• System: 22º - 26º C

If you look closely you see that there are traces of thermal paste on the surface of the mike. This is because once the experiment I noticed that some pictures were blurry, so I had to clean the components (there is always something of pasta) and repeat. In the latter case, I apply a film of thermal paste on the surface of the core the bus, the width of a cigarette paper. This paste applied on the core, and extend it with your finger.

• Idle: 36º C
• Full: 44º C
• System: 21º - 23º C

In this case, removing the heatsink, it looks like the effect of the pressure surface has been impregnated by the thermal paste that was in the core. Then apply a little paste on the surface of the sink and extend it with your finger, to give a thick layer of cigarette paper.