Is this the right way to put on thermal paste?

[mariushm]

A pea sized blob, a thin line, an x , it really doesn't matter, as long as you don't go overboard with the paste.

 

The paste is meant to be a thin layer between the heatsink and the metal surface of the cpu... so often less is better. 

[KarathKasun]

6 hours ago, mariushm said:

A pea sized blob, a thin line, an x , it really doesn't matter, as long as you don't go overboard with the paste.

 

The paste is meant to be a thin layer between the heatsink and the metal surface of the cpu... so often less is better. 

With good mounting pressure (looking at screw down mounting here) you can just slap a glob in the middle, it spreads itself.  Just dont do a ring, that will create a central void.

[AngryBeaver]

For most intel chips the pea sized dot in the middle seems to be best... for something like the threadripper though you need either a line or a X to get proper coverage.

 

When it comes to the paste.. it is only there to fill in the micro-spaces. So in the way of things you have your IHS/block which is very efficient at moving heat, then you have your TIM (Thermal Interface Material) which is worse at moving heat, but pliable so it can fill the small gaps between your two metal contact points, then you have air which is terrible for transferring of heat.

 

So if you just put two blocks together you will have air crapped in the small imperfections (gaps)  between the two surfaces.. as air is a terrible at transferring heat temps would be much higher. Now if you put too much paste then you are now increasing the distance between the two surfaces meaning the heat has to travel farther through the tim which is less efficient at transferring the heat. Then you have a very thin layer of TIM which is just enough to fill in the micro-gaps, while also not increasing the distance between the blocks by a noticeable amount.

 

The last scenario will provide you with the best temps. Now this is all assuming normal TIM and not liquid metal, which is similar but different. The key to getting the best transfer of heat is to using the correct amount of TIM. Too much is better than none, but having the right amount is always going to be superior.. which is why the rice and pea sized method are quoted as they are much closer to providing the right balance of TIM for the majority of chips.

[HTWingNut]

I just realized that gif says "Hellmans" on the tube... Mayonnaise.

[KarathKasun]

17 hours ago, AngryBeaver said:

For most intel chips the pea sized dot in the middle seems to be best... for something like the threadripper though you need either a line or a X to get proper coverage.

 

When it comes to the paste.. it is only there to fill in the micro-spaces. So in the way of things you have your IHS/block which is very efficient at moving heat, then you have your TIM (Thermal Interface Material) which is worse at moving heat, but pliable so it can fill the small gaps between your two metal contact points, then you have air which is terrible for transferring of heat.

 

So if you just put two blocks together you will have air crapped in the small imperfections (gaps)  between the two surfaces.. as air is a terrible at transferring heat temps would be much higher. Now if you put too much paste then you are now increasing the distance between the two surfaces meaning the heat has to travel farther through the tim which is less efficient at transferring the heat. Then you have a very thin layer of TIM which is just enough to fill in the micro-gaps, while also not increasing the distance between the blocks by a noticeable amount.

 

The last scenario will provide you with the best temps. Now this is all assuming normal TIM and not liquid metal, which is similar but different. The key to getting the best transfer of heat is to using the correct amount of TIM. Too much is better than none, but having the right amount is always going to be superior.. which is why the rice and pea sized method are quoted as they are much closer to providing the right balance of TIM for the majority of chips.

You dont need to get "the right balance" for most coolers, you just need to not have too little.  The mounting force is sufficient to eject any extra paste, its just messy to clean up whenever you disassemble.

 

Ive never seen a negative impact on temps from over-applying thermal paste with screw down mounting.

[JoostinOnline]

On 2/15/2018 at 6:58 PM, Tsuki said:

actually your better off using less.  its only there to fill the microgaps, less really is more.

"If less is more, think how much more would be!"

- Peter Griffin

[AngryBeaver]

5 hours ago, KarathKasun said:

You dont need to get "the right balance" for most coolers, you just need to not have too little.  The mounting force is sufficient to eject any extra paste, its just messy to clean up whenever you disassemble.

 

Ive never seen a negative impact on temps from over-applying thermal paste with screw down mounting.

Actually the mounting pressure does not eject all of the paste to make it right... and depending on the paste some are conductive which will result in a fried board.

 

The mounting pressure will help to make sure the thickness of the paste is uniform and will eject some, but for the most part once the paste is uniform across the IHS then the down force is quality distributed so the paste will not be ejected... it will just increase the thickness of the paste between IHS and Block.

 

It doesn't take much extra paste to add another 4-6C to your cpu temps. If you don't believe me feel free to test it. put a massive X of paste on your IHS, then do the rice sized dot. Record temps of both. The results WILL surprise you.

[Highflow]

2 hours ago, AngryBeaver said:

It doesn't take much extra paste to add another 4-6C to your cpu temps. If you don't believe me feel free to test it. put a massive X of paste on your IHS, then do the rice sized dot. Record temps of both. The results WILL surprise you.

ofcourse, use common cense, the paste is there just to fill the errors, theres nothing better than direct metal contact if both surface are very straight on microscopic base.

In old days adding paste wasnt enough in some cases as the cpu`s surfaces were everything but flat like the P4`s....

so the get better cooling performance people did lapping the cpu, but very time consuming and not really something

for basic noobs who just build a system for first time.

[KarathKasun]

3 hours ago, AngryBeaver said:

Actually the mounting pressure does not eject all of the paste to make it right... and depending on the paste some are conductive which will result in a fried board.

 

The mounting pressure will help to make sure the thickness of the paste is uniform and will eject some, but for the most part once the paste is uniform across the IHS then the down force is quality distributed so the paste will not be ejected... it will just increase the thickness of the paste between IHS and Block.

 

It doesn't take much extra paste to add another 4-6C to your cpu temps. If you don't believe me feel free to test it. put a massive X of paste on your IHS, then do the rice sized dot. Record temps of both. The results WILL surprise you.

The viscosity of most pastes is not high enough for a thick layer to form under pressure.

 

Ive found the difference with modern coolers to be less than 3c when comparing what many call too much and painstakingly spreading the compound out to a paper thin layer.  Ive done this testing every 3-5 years for the past 20 or so years.  The increase in heatsink clamping force has drastically reduced the importance of TIM application method.  (Remember those horrid spring/lever clips from Socket 7 & 370?  They are one of the reasons for painstaking TIM application.)

 

The only time that the thickness matters enough to do the whole paste spreading thing is when going sub zero.  And then its not because of transfer characteristics, its because of mechanical characteristics.

[dcgreen2k]

I use as much as a cooked grain of rice or pea sized amount. Kinda hard to mess up.