Questions about Lapping CPU IHS & Heatsink

dreamcat4

dreamcat4

dreamcat4

Associate
Joined
3 Sep 2015
Posts
369
Location
UK
Hi there,
Got some questions about lapping...

Not in any hurry to do it right away. I've read plenty of articles about lapping but they don't specifically mention certain considerations.

My plan: I plan to remove the IHS from the die before lapping. And after lapping then apply CLU to both sides of the IHS inside and out. So that CLU will be the only TIM used throughout. I have already checked my CPU heatsink base is a copper one and not aluminium.

Questions:

* My sandpaper has arrived in the post with a few creases in it. None are full sharp creases. It is more like 7-8 of mild crinklies spread around all over.

So will such a surface un-even ness really matter once it's been taped onto a pane of glass? (to get the flatness). And once the object is pressed down against the surface? I ask because most of the lapping tutorials say 'just use hand weight', and 'don't press down too hard'.

* is there any good trick / solution how to sand the inside of the IHS, given that it is recessed and difficult to access? The same method cannot be used as for the outside.

* My CPU heatsink is supposed to weigh 420g. Can I use the same technique as the IHS topside? Meaning is it too heavy for that? Moving the object around on stationary / fixed sandpaper surface as the base. No lapping tutorials really seem to show of discuss the matter at any great length.

* Can leaving a slightly rough surface on the copper help increase the total surface contact area vs mirror finish. Thus enhancing the thermal performance / heat transfer? Is there any articles somewhere about that? What is the correct grit to stop at to achieve the correct scale peaks / valleys? In particular for CLU TIM.


My sandpapers are 600,800,1000,1500,2000,2500 grit. One of each, size 28cm x 23cm.
 
I did the IHS on my 4670K when I delidded it as it was very concave. I went all the way to 2500 grit so it's nice and shiny. I then applied CLU to the die (after putting nail varnish over the transistors) and glued the IHS back on with gasket sealant. Between the IHS and my block I have Gelid GC4. I got a drop of around 15 degrees C which made it worthwhile although I was never temp limited in the first place.
 
If your using Liquid pro on the die then you dont really need to lap the inside of the IHS just stick sum liquid pro on that too, i wouldnt know how to go about that tbh either.

As for the grit size just keep going smaller till it gets nice and shiny like pasty says, ive never done a big air cooler only water coolers so ive had no experience with that, but i imagine youve just gotta get a grip as close to the bottom as possible so you can keep it flat to the surface, and yeah dont push hard or youll just ruin the paper.

If the creases arent sharp then just put them between 2 flat surfaces and put summat heavy on them over night to flatten them out.
 
CLU = Liquid Ultra (not pro). It's pretty similar stuff though.

OK thanks. Yeah the only way I could imagine someone doing the inside (without re-engineering the sides). Would be with a dremel or something like that. Although I am doubtful about how to keep it a flat surface. As such it would wise leave the inside as-is and not to push my luck out to far.

And being a Skylake chip i can't really get a spare IHS (from old CPU) to use instead. Since the vertical heights have changed to acommodate the different thicknesses. It needs to be the same IHS.
 
Yeah the only difference i could see between the 2 was with one you got a brush and the other a bud so i just call them both liquid pro :)

If your worried about the underside of the IHS you could always just make a shim and cool the die directly or use a lapped copper peliter plate. Ive tried both as i didnt like the idea using the IHS but i found in the end just using some lquid pro under there gave the best results.
 
BluD said:
but i found in the end just using some lquid pro under there gave the best results.
Click to expand...

Ok then. I shall follow protocol and not waste my time when you've been kind enough to already try out such things. Thank you. It solves my biggest problem / hassles.
 
be wary of the liquid stuff I hate it. I put some on my i5 haswell when I delidded it was a nightmare to apply. I didnt get the results I liked either. So when i tried to clean it off it was really difficult to get rid of it.

Used AC MX5 instead and got much better results.

Dont take my word for it a lot of people like the liquid stuff but in my opinion its horrible stuff.
 
OK thanks for that. I'm not rushing into it and will be using AS 5 for time being. My current understanding (just purely based on other people saying things on the net). Is that:

* CLU *is* better than Pro in terms of application / ease of use. And you didn't mention whether you were talking about CLP or CLU. Which matters i think.

* CLU should be 'ok' to apply / remove (isn't certainly nothing like paste). Ahm. But for a liquid metal CLU shouldn't give too much / excessive trouble *provided* that you didn't over-apply for the specfiic surface area cm^2 (which is somewhat case-by-case depending what you are doing with it) and use too much of the stuff in the first place. i.e. it takes patience / time to get the spread out.

=====

Anyway I have another question now about Lapping which is not related to liquid metals:

I received my cooler today. So was able to assemble things for the first time to test all thee bits are working (mobo, CPU, ram, etc).

The thing is: it doesn't upon casual inspection seem there's a lot of downwards adjustment for the cooler height. Like the clamps are already pretty close together. So I'm just wondering now after seeing that if it's anything to worry about. Because obviously lapping the IHS will reduce the z-height somewhat. Would that risk lessening the pressure on the socket / pins resulting in a poor connection?

I know there's always shims but i'd like to avoid that if possible (as it's extra layers). And not you know, get the IHS out-of-spec and become too thin. Is there any kind of physical test / known method I can make use of to ensure I don't sand away too much of the IHS thickness?

I know it's probably me just being overly-paranoid about these things again. Just thought I'd ask again in here. If anyone else happened to have such similar concerns / experiences with their own heatsink.

Mine's a Raijintek Pallas. They are pretty rare ones.
 
You must log in or register to reply here.
Back
Top Bottom