thermal paste application

bigjimmyauk

bigjimmyauk

bigjimmyauk

Soldato
Joined
23 Nov 2009
Posts
6,038
Location
North Leicestershire
well just throwing in some info i found a couple of days back applying thermal paste in the centre using the rice size amount is no good neither was using a pea sized amout this guy tried about 10 different ways of applying it and the blob method never fully coverd the cpu when pressed down into place.

this test had clear photo evidence supported by a massive write up. so why does everyone insist on using a blob

you can see the results here
 
Last edited:
we used to spread a thin layer accross the surface of the cpu only to be told that a blob in the middle was better, is there a diffinitive answer as to how it should be applies as intel seem to use 3 small strips (the ones i have used do anyway) and amd have a big square on theirs.
 
i think the 2 lines method looked the best for a square based cooler obviously too much and it's gonna drip all over the socket and mobo too little and it won't do it's job properly think theres a fine line between perfect and messing it up
 
An excellent review there, thank you for the link.

However I fear I must answer your question. The reason most people use a blob is that it's quick, easy, and results in damned near exactly the same temperatures as any other approach. About the only way to get thermal paste wrong is to use far too much, and by telling people to use a very small blob you greatly cut down on how many people use an entire tube. The page specified shows many images, the flaws in your deduction are the assumptions that the entire surface has to be covered (untrue) and that thickness of cover is irrelevant in order to get the best results.

The best approach remains lapping the heatsink and ihs down to virgin copper and soldering the two together, but for obvious reasons people aren't keen on this.
 
only problem i seen with the solder method was actually joining the 2 sufaces perfectly without frying the cpu. something i want to google later
 
Soldering them? I've never seen it done, but the reasoning seems solid enough. The processor die can be considered to be a piece of silicon with some copper and stray doping atoms in. It then has the ihs soldered to the top of the die. This strongly suggests that the die itself can take high enough temperatures for soldering. Add to this that graphics cards are reballed by putting them in an oven (which melts the solder but does no particular damage to the cards) and I'm sure it'll work without harming the processor.

The biggest issues are making sure none of the solder runs off the surface of the cpu onto the socket and actually taking the thing off later. The first isn't so bad, the second I haven't thought of a solution for yet. There's various gallium based liquids (liquid pro for one) which achieve a similar end result, even down to a mechanical join between the two surfaces, but can be broken off by torsion. So that's slightly less extreme than solder, shouldn't perform much worse, and a lot easier.
 
JonJ678 said:
Soldering them? I've never seen it done, but the reasoning seems solid enough. The processor die can be considered to be a piece of silicon with some copper and stray doping atoms in. It then has the ihs soldered to the top of the die. This strongly suggests that the die itself can take high enough temperatures for soldering. Add to this that graphics cards are reballed by putting them in an oven (which melts the solder but does no particular damage to the cards) and I'm sure it'll work without harming the processor.

The biggest issues are making sure none of the solder runs off the surface of the cpu onto the socket and actually taking the thing off later. The first isn't so bad, the second I haven't thought of a solution for yet. There's various gallium based liquids (liquid pro for one) which achieve a similar end result, even down to a mechanical join between the two surfaces, but can be broken off by torsion. So that's slightly less extreme than solder, shouldn't perform much worse, and a lot easier.
Click to expand...

i might try it with my 939 stuff when i get the new i7 might first tho start of trying to bond the heatsink to a piece of copper for practice. think the hardest obstacle will be stopping the heat dissapation of the heatsink and pre heating the proccy suface so it dosen't bond on contact and create a air pocket
 
What you also have to remember, is that the core of the CPU doesnt actually cover the whole IHS. It is a heatspreader.

Plus, It doesnt state what compound they are using. Usually it takes a few thermal cycles to spread the compound around the chip. It isnt a thorough test and is lacking somewhat in the 'real world'.

I have always used the grain of rice approach, and every time my heatsink has been removed it covers the full die. Not just on m y PC, but many others I have built. All using AS5.
 
You must log in or register to reply here.
Back
Top Bottom