Idea to get rid of thermal compound/pads?

[corrado33]

So thermal compound and thermal pads are there to fill in the "gaps" in the metal.

 

So why doesn't LTT do a video where they take the IHS off a CPU, and take the block of an air cooler, and get them machined flat and level, then polished to a mirror shine. Basically machine the two surfaces so they mate together perfectly. Then test without any thermal paste. 

 

Alternatively, just get a piece of glass with a piece of high grit sandpaper and do the lapping yourself. It's super simple. Literally place glass on surface, place sand paper on glass, rub surface on sand paper. (Glass is perfectly flat.) Then go up through the grits until both surfaces are polished. (Like literally ALL the grits, if it's not polished, it's not worth doing it as sanding introduces scratches. )

 

If there are no "little holes" to fill, then we shouldn't need thermal paste. 

[Vengeance1020]

2 minutes ago, corrado33 said:

So thermal compound and thermal pads are there to fill in the "gaps" in the metal.

 

So why doesn't LTT do a video where they take the IHS off a CPU, and take the block of an air cooler, and get them machined flat and level, then polished to a mirror shine. Basically machine the two surfaces so they mate together perfectly. Then test without any thermal paste. 

 

Alternatively, just get a piece of glass with a piece of high grit sandpaper and do the lapping yourself. It's super simple. Literally place glass on surface, place sand paper on glass, rub surface on sand paper. (Glass is perfectly flat.) Then go up through the grits until both surfaces are polished. 

 

If there are no "little holes" to fill, then we shouldn't need thermal paste. 

as far as I am aware, microscopic craters will always be present, therefore this test may prove inconclusive

[Sakkura]

You'd have to make each surface atomically flat, which is incredibly difficult. And it would still take very little to ruin the flat surface.

[Cereal5]

I think this is about as good as it gets. It's not as simple as "make it flat", as stated above.

[corrado33]

9 minutes ago, Cereal5 said:

I think this is about as good as it gets. It's not as simple as "make it flat", as stated above.

Na that's different. I don't want to apply the cooler to the die directly, I want to try to mate the IHS with a cooler much better. 

[Marcus Torre DeProspero]

Just now, corrado33 said:

Na that's different. I don't want to apply the cooler to the die directly, I want to try to mate the IHS with a cooler much better. 

well you are talking about not needing paste or pads but there is paste meeting the ihs and the die itself so getting rid of the ihs contact point makes it much better in the first place

[Cereal5]

8 minutes ago, corrado33 said:

Na that's different. I don't want to apply the cooler to the die directly, I want to try to mate the IHS with a cooler much better. 

Yeah I know, the guys above told you why it's not possible. The closest you can come to that is by soldering the cooler directly to the die.

[Amazonsucks]

Even if you could get it that flat, which would be super expensive, you might run into issues with CTE mismatch between the IHS and heatsink.

 

Also glass is not perfectly flat and has microscopic surface porosity.

[Marcus Torre DeProspero]

2 minutes ago, Cereal5 said:

Yeah I know, the guys above told you why it's not possible. The closest you can come to that is by soldering the cooler directly to the die.

I like the idea of a bigger and better ihs soldered onto the die

[LogicalDrm]

1. Glass, as in generic window pane, is not perfectly flat. You need highly specified machinery for smoothing it perfectly flat. These are done for high level scientific purposes. Like space and ground telescopes.
2. Metals normally used as materials aren't easy to machine so flat that they wouldn't have imperfections. Copper and aluminum both are rather soft materials. So any touch from something harder will cause micro or nano scale damage. Material would need to be something harder so that mounting process itself wouldn't cause issues.
3. This is far from easy or cheap.

[Dylanc1500]

43 minutes ago, corrado33 said:

So thermal compound and thermal pads are there to fill in the "gaps" in the metal.

 

So why doesn't LTT do a video where they take the IHS off a CPU, and take the block of an air cooler, and get them machined flat and level, then polished to a mirror shine. Basically machine the two surfaces so they mate together perfectly. Then test without any thermal paste. 

 

Alternatively, just get a piece of glass with a piece of high grit sandpaper and do the lapping yourself. It's super simple. Literally place glass on surface, place sand paper on glass, rub surface on sand paper. (Glass is perfectly flat.) Then go up through the grits until both surfaces are polished. (Like literally ALL the grits, if it's not polished, it's not worth doing it as sanding introduces scratches. )

 

If there are no "little holes" to fill, then we shouldn't need thermal paste. 

You'd be better off with submersion cooling with something like florinert FC-72. Heck you could run it through a chiller and have everything in your system at 0C.

[wONKEyeYEs]

2 hours ago, corrado33 said:

Basically machine the two surfaces so they mate together perfectly.

Can be done, not by hand.

Parallel blocks might be accurate enough.

And you know, if you have to ask, you can't afford it...

http://www.crystalarmatech.com/crystal-gauges-blocks.html

[Amazonsucks]

This also ignores the actual state of the art in cooling, which are technologies like:

 

Microfluidic direct cooling which is in development.

 

Silver or copper diamond matrix composite heat spreaders and single crystal diamond heat spreaders, phase change immersion in 3M Novec all of which are currently in use or available.