Thermal Paste Application

[philjohn]

I quite enjoyed this video which shows how different application methods spread - he uses a clear piece of perspex:

There's also a first video which shows some more methods (and why spreading it out before application just leads to air bubbles):

 

[Cookeh]

-SNIP-.

Thanks for that, was a decent read!

So a pea sized blob (around 5-5.5mm) spread by the weight of the cooler is generally the best method for all types of compound?

Also, if the HSF has "channels" on the base, is it wise to fill then with compound as well, or leave them be?

 

[Plec]

Also, if the HSF has "channels" on the base, is it wise to fill then with compound as well, or leave them be?

Below is advice for exposed heatpipes (channels/grooves etc) on the base only - using AS5 (but i've used this method with other gunk if i felt it was required.):

If you have exposed heatpipes (ridged surface) you could try tinting the surface of the heatsink first to fill in the voids made by the exposed pipes on the base (artic silver 5 advises this but i've used this method with other gunk if i felt the heatsink required it.). You're basically filling the gaps, created by the exposed pipes, but scrape of all the excess so the base of the heatsink only has a very slight tint of gunk (credit card is great for this). The theory is that it fills, the sometimes pretty large, gaps that exposed heatpipes can cause on the base of the heatsink.

After tiniting the heatink you then only need to apply either a rice size blob or a single line (less than a mm thick) down the middle of the core (vertically) for quad cores - but don't go from edge to edge leave a 5mm gap at each end.

This method should ensure that the pipe gaps don't suck up all of the gunk on the cpu and allow it to spread as it would on a shiny flat heatsink base.

I will stress that you only need to use this method on exposed heatpipe heatsinks (and even then not all of the time) - there is no need to use this method with perfectly flat based heatsinks.

 

[IC Diamond]

Thanks for that, was a decent read!

So a pea sized blob (around 5-5.5mm) spread by the weight of the cooler is generally the best method for all types of compound?

Also, if the HSF has "channels" on the base, is it wise to fill then with compound as well, or leave them be?

5.5 mm bead is best

For DTH

http://www.youtube.com/embed/rlO6Gx7Uf6g

Fast Simulation of Laplacian Growth

http://www.youtube.com/embed/AHmgPtOJsYk
Laplacian Growth in thermal compounds

Laplacian Growth - a better visual explantion

http://www.physics.ubc.ca/pitp/archives/theory/2004talks/wiegmann.pdf

 

[MatteH Oxford]

They must have money to burn, thermal compound per ml is more expensive than ground diamonds!

 

[EdGey]

They must have money to burn, thermal compound per ml is more expensive than ground diamonds!

Heh yea it is a rip off.

 

[IC Diamond]

Originally Posted by MatteH Oxford View Post
They must have money to burn, thermal compound per ml is more expensive than ground diamonds!

Heh yea it is a rip off.

Not that simple

By that metric you could buy the raw materials for MX4 and most other compounds for about $0.05 USD per gram and contrast that against AS5 with about 3 grams silver approx @ 4-$5 USD per 0.5ml

AS5 and ICD are bargains and MX4 would be the biggest rip off ever.

There are fixed costs to bring this stuff to market and material costs are generally a minor component. Shipping costs, packaging, labeling, overhead, taxes, dealer markup etc. . Everybody would like to buy at cost but companies that bring product to market deserve a profit.

You can relax, competitive pricing all the way around,

 

[mikeyholt]

I do grain method, I find it to work pretty well.

 

[Phil2008]

Rice grain size is better. Spread it very thin across the whold cpu. (use your finger in a plastic bag of some kind)

I use that method too (even on gpus) and its never let me down. AS5 is prob the best when using that method as it spreads quite well.

 

[chrisarm]

I usually put a 3 to 4 small rice grain splodges on the cpu and spread a tiny amount on the cooler so when they push together it gives a full coating. Hadn't noticed if mentioned but you sometimes get slightly higher temps the first week while the compound settles in (well it always has with me anyway)

 

[Phil2008]

I usually put a 3 to 4 small rice grain splodges on the cpu and spread a tiny amount on the cooler so when they push together it gives a full coating. Hadn't noticed if mentioned but you sometimes get slightly higher temps the first week while the compound settles in (well it always has with me anyway)

Thats way too much, less is better as you just want a thinnest layer as poss on the cpu(not heatsink aswel) cos at the end of the day, your only filling the microscopic gaps in on cpu and heatsink. Plus applying a real thin layer on cpu, will avoid the paste dripping or spreading further a field from cpu, as if your using AS5 you might have a dead pc and a fireworks display, as AS5 is conductive

 

[Phil2008]

Plus thats prob why you get high temps at the start, because the heat cannot pass though all that paste.

 

[WingZero30]

Not that simple

By that metric you could buy the raw materials for MX4 and most other compounds for about $0.05 USD per gram and contrast that against AS5 with about 3 grams silver approx @ 4-$5 USD per 0.5ml

AS5 and ICD are bargains and MX4 would be the biggest rip off ever.

There are fixed costs to bring this stuff to market and material costs are generally a minor component. Shipping costs, packaging, labeling, overhead, taxes, dealer markup etc. . Everybody would like to buy at cost but companies that bring product to market deserve a profit.

You can relax, competitive pricing all the way around,

That's the cost accounting

 

[chrisarm]

Thats way too much, less is better as you just want a thinnest layer as poss on the cpu(not heatsink aswel) cos at the end of the day, your only filling the microscopic gaps in on cpu and heatsink. Plus applying a real thin layer on cpu, will avoid the paste dripping or spreading further a field from cpu, as if your using AS5 you might have a dead pc and a fireworks display, as AS5 is conductive

Never spreads past the cpu, it's not like I put the whole syringe on. The 3 small rice grain size pieces I use are less than a pea sized piece

 

[PR.]

That's interesting as I've always spread the paste all over the CPU now it seems it's better to leave it as a grain of rice to stop air bubbles building up

 

[Skunkworks]

I've always used the small X method, with the arms of the X pointing to the corners of the CPU.

 

[DJ_Hughesy]

Pea size is too big.

Rice grain size is better. Spread it very thin across the whold cpu. (use your finger in a plastic bag of some kind)

Not so thin that you can see the colour of the cpu under it, but you should not have to apply anymore than a rice grain size.

Its personal pref really, but this is the method I have used over the years, and I have had great results every time.

actualy depends on the paste...

http://www.youtube.com/watch?v=EyXLu1Ms-q4

pea sized n letting the heatsink spread it with pressure seems to be the best for most

 

[Phil2008]

actualy depends on the paste...

http://www.youtube.com/watch?v=EyXLu1Ms-q4

pea sized n letting the heatsink spread it with pressure seems to be the best for most

AS5 spreads much better then all the others, so Ive read.

If you spread it thinly enough you wont get air bubbles surely?

 

[DJ_Hughesy]

AS5 spreads much better then all the others, so Ive read.

If you spread it thinly enough you wont get air bubbles surely?

not sure tbh, although u can never be sure...

least with pea sized dot n heatsink spreading it you can bet the pressure forces the air bubbles out..

many diferent techniques all good with diferent paste/cpu.. theres isnt really a "best" way

 

[RizlaKing]

Have always used the pea sized method and my temps have always been good it really comes down to what you prefer i don't think there is a huge difference either way unless you put too much on .