Has anybody here done this?
Is so what did you do? :banana:

Pretty simple actually. Get some 400grit, 800grit and 1000grit sandpaper. You will use 4-5 sheets of each. Some like to continue with 1500grit and 2000grit - you can if you like - but it shouldn't make any difference in temps.

Get a nice flat surface, if you have some surfaced aluminum or you can use a good counter-top or marble if you have it. Basically looking for as flat of a surface as you can get. Attach your sandpaper sheet to the flat surface (tape?) and with light pressure (and keeping the heatsink flat - don't tip it!) work in slow circles. When the surface of the heatsink looks uniform, move to the next grit (400 -> 800 -> 1000 and so on). DO NOT USE POLISHING COMPOUND - it will affect the performance of the heatsink.

All there is to it! Post some pictures when you get it done! :cheers:

EDIT: Forgot to mention, a lot of people like the wet-sand too.

Pretty simple actually. Get some 400grit, 800grit and 1000grit sandpaper. You will use 4-5 sheets of each. Some like to continue with 1500grit and 2000grit - you can if you like - but it shouldn't make any difference in temps.

Get a nice flat surface, if you have some surfaced aluminum or you can use a good counter-top or marble if you have it. Basically looking for as flat of a surface as you can get. Attach your sandpaper sheet to the flat surface (tape?) and with light pressure (and keeping the heatsink flat - don't tip it!) work in slow circles. When the surface of the heatsink looks uniform, move to the next grit (400 -> 800 -> 1000 and so on). DO NOT USE POLISHING COMPOUND - it will affect the performance of the heatsink.

All there is to it! Post some pictures when you get it done! :cheers:

EDIT: Forgot to mention, a lot of people like the wet-sand too.
thanks. :)

thanks. :)

No problem, hope it comes out nice!

circles arent good, linear motion is what you want, circles will create a convex surface.

start at 220, make sure to wet the paper, soak it preferably in soapy solution, this solution will cause the paper to stick to the flat surface(I use a ground granite plate)
lap the heatsink with its own pressure(maybe a little more) forward and backward in a linear motion, you can rotate it a few times while lapping.

you can(if you want to check it) mark the bottom with a sharpie marker then lap it some more and see where its not so flat, then move up to 600grit and stop there.

Jon

Does this really make a difference??

Does this really make a difference??

Maybe on a sub atomic level, but I doubt you will see it, unless you ran temp test at the core. :) It's like running high octane fuel in your car instead of regular, you feel like your engine runs better, so yes you will feel like it's making a difference.

I am thinking this is something "ger" would really get into the analysis of the difference! Did you see his analysis on the "surface finish" problem....pretty amazing...it's something that NASA at one time was capable of doing!

I am thinking this is something "ger" would really get into the analysis of the difference! Did you see his analysis on the "surface finish" problem....pretty amazing...it's something that NASA at one time was capable of doing!


Surface finish is very import on a microscopic level, but grease and water cooling can easily IMO give you just as good result, again unless you have the equipment to measure you would never know if you had a cooler CPU even by than you maybe got it 2% cooler.

Viper, depending on how much lapping it really takes, it can. or if the heatsink is concave as is, circular motion would actually speed up lapping a little then to go to linear. The flatness does help a ton because it will cause the heatsink and core of the cpu to mate better and increase heat transfer. You can actually carefully lap a cpu and heatsink together(not recommended) to get the best transfer, but its likely to screw something up.

Jon

Sorry, but I (and most other modders out there) disagree about linear versus small circles when lapping. I have always used small circles and I can get a nice mirror finish in my blocks. Again, I also recommend starting with 400 grit and not anything coarser...

Lapping a heatsink is all about getting a "smoother" finish than the way they ship. Some heatsinks have a very poor finish, other are lapped at the facotry - depends on who you buy. You want a good surface for your paste, whether you are using Arctic Silver or just basic "white goo". I have seen a properly lapped HSF drop temp by 2*-3* after lapping. Might not be much, but if you are overclocking it might mean the difference between stable and not.

Javadog, I have been in watercooling for over 3 years now, have made around 50 waterblocks, lapped many more. Check around ocforums,ocau,procooling etc.

A mirror finish means NOTHING, I can get a mirror finish by hand sanding the base of the thing.

Ron, what heatsink do you have that you want to lap? if its a swiftech, thermalright, or any of those better ones, DO NOT lap it, you wont gain anything.

Jon

A mirror finish means NOTHING, I can get a mirror finish by hand sanding the base of the thing.

Never said it did...

You lap blocks the way you see fit, and I will continue doing it the way I always have.

Come one, be honest - most of the time it doesn't make a damn bit of difference - so no point in getting our panties in a bunch over technique. :rolleyes:

Javadog, I have been in watercooling for over 3 years now, have made around 50 waterblocks, lapped many more. Check around ocforums,ocau,procooling etc.

A mirror finish means NOTHING, I can get a mirror finish by hand sanding the base of the thing.

Ron, what heatsink do you have that you want to lap? if its a swiftech, thermalright, or any of those better ones, DO NOT lap it, you wont gain anything.

Jon

Really?

You guys all machine things, right? Ever heard of a fly cutter? You'll get a great surface in seconds, and you can save your sandpaper for something else, not to mention having more time to do something other than give your arm a workout. All you're trying to do is get the surface as smooth as possible, so your thermal compound will get good heat transfer. In my experience, most people use too much compound, which in a thick layer becomes an insulator.

Flycut the heatsink, removing only a few thou, put a thin layer of compound on, assemble the heatsink and cpu, take them apart, and look to see that there is a uniform coat of compound on each part with no bare spots. Wipe off any excess that squeezes out and keep assembling and taking apart until you have the uniform coat with no excess. That will give you your best thermal transfer. I used to mount huge Triacs for high power (10KW and up) furnace supplies and I never had one fail, so I know this works.

Dave

The secret here is to get the contact as uniform as possible, the flatter the better. The more intimate contact the better. Then the proper application of a high quality transfer grease like the arctic silver series does make a difference. As much as 3-5 C.

Hello. This sounds like an interesting topic. So is the problem that you're not getting enough heat transfered to the heatsink? I'd be more concerned with adding area and better ventilation to the heatsink otherwise. The heatsink might get saturated after the improved contact. I've never seen a CPU heatsink that wasn't smooth enough for proper heat transfer but maybe the ones aren't? Do you also polish the CPU to get the two sides to mate (please say no)?

JR

so where are the thermal transfer experts? if the goop put between the chip and sink has the right thermal conductivity, what's it matter what the finish is? either way you still have "chip |goop| heatsink" barrier in 100% contact? presumably the goop is a good conductor of heat? maybe its the reduction of the thickness of the oxidization layer that is the trick?

even if someone really knew what they were doing with lapping and you got it flat and evenly finished to a millionth of an inch, guage block quality stuff, who says the chip surface is flat? or are thes heat sinks that poorly made in the first place? I'm not say it's an invalid thing to do, just that it would be nice to hear an engineering/scientific reason for the improved heat transfer.

now liquid cooled i can see - sure wouldn't want to spring a leak though. what do you use for the 'rad'?

you still want to lap it after fly cutting typically, some deflection could mess the surface up real good.
Theres no goop that can get that good of a transfer. reducing the thickness of the goop increases heat transfer really well.
The surface needs to be extremely flat for good heat transfer, very smooth too.
the CPU surface is typically very good, some do try to lap it but its usually foolishness because its such a small peice.

Watercooling you too need a good surface. They make some specialty radiators for the computer watercooling, you can also use tranny coolers or heater cores.

I have more info but no time, take a look around proforums and ocforums


Jon