Lapping....

[SamDaMan]

There seems to be some confusion about lapping the copper heat sinks.

Unlike what others have suggested you don't use sandpaper - if your heatsink comes from the manufacturer needed 600 grit sandpaper purhaps you might consider tossing it in the trash...

You use tripoli compound for cutting and then White (maybe red) rouge to finish...

http://www.gomog.com/allmorgan/polishingmetal.html

METAL POLISHING TERMS
by Lorne Goldman
M21 SMOOTH SPECULAR – initial cutting using aluminum oxide compounds.Beginning coarse and finishing with 320 grit, followed by buffing with aluminum oxide buffing compounds.
M22 SPECULAR – cutting with compounds the same an M21finish, followed by a final light buff.
M31 FINE SATIN – wheel or belt polishing with aluminum oxide or silicon carbide abrasives of 180 – 240 grit.
M32 MEDIUM SATIN – same operation as M31, except 120 –180 grit is used.
M33 COARSE SATIN – same operation as M31, except 80 –120 grit is used.
#3 FINISH – a polish obtained by finishing with approx. 100 grit abrasive.More commonly used as a precursor to the final finish.
#4 FINISH – general purpose, bright, clean finish, 150 –240 mesh abrasive used following prior grinding with coarser grit abrasives.
#4 NON-DIRECTIONAL SATIN – 80 – 150 grit with no grain direction.
#4 LONG GRAIN – not the rice but a long continuous grain.
#6 FINISH – a soft satin finish, with lower reflectivity than the #4 finish.Use a medium abrasive oil
#6 FINISH MIRROR BRIGHT – a highly finished product by buffing a surface which hasbeen finely ground but still has grit lines heavier than a #7 finish.
#7 FINISH – a highly reflective finish produced by buffing a surface that has been finely ground, the grit lines are not removed during the final buffing process.
#8 FINISH – standard polished finish. It is quite reflective and is achieved by polishing with successively finer abrasives, then buffing with a fine compound.The surface is basically free of grit lines, although not completely invisible.

Finishing Compounds
Green Rouge

Primarily used in final finish buff operations on stainless steel, steel, brass, aluminum, nickel, and chrome. The green rouge is a chrome oxide, and is considered the best all around luster compound.

White Rouge

The white rouge is the softer than green rouge. Primarily used in the final finish of steel, stainless steel, and zinc. This compound is also a favorite in coloring aluminum and brass.

Red Rouge

Primarily used in the final finish of gold and silver, it is the finest of all rouges. It also produces an excellent finish on brass.



This is what the tripoli and rouge looks like:


http://www.wmhtoolgroup.com/JETWood/Accessories/575850.html

http://www.wmhtoolgroup.com/JETWood/Accessories/575851.html

This stuff might be sold at yur local hardware store in the section with the buffing wheels ...

 

[Randyman...]

I'm sure that is the "proper" way, but I'm sure I saw benefits using some 600/1000 wet/dry I had laying around in the garage (I KNOW I saw and felt the physical difference once done - I never tried it w/o lapping, so I can't say conclusively).

My SLK-947U DID INDEED need lapping, but I was FAR from throwing it in the trash! Dropped my temps 11*C.

Later

 

[CrystalMethod]

I think Sam is a little confused as to what we're trying to achive by lapping our heatsinks. When you lap a heatsink you're not trying to achive a polished finish. You're trying to get the contact surface flat. Like I've stated many times before, the polished shiny (sometimes mirror like) look is just a by-product of the lapping process. "polishing" and "lapping" are two different things.

 

[SamDaMan]

Well it would seem that flatness is achived by polishing, although a polished spoon is shiny but far from flat.

Basically you want to use a super fine abrasive regardless. Clearly the finer the abrasive the flatter the surface. And the flatter the surface the more uniform light reflects off it..... thus a mirror has the "flatest" serface....

I am sure you would agree that you get a flater serface with a higher grit sandpaper. Thus the smaller the partical the flater the serface.

Also given that the artic five is supposed to "full in the voids" I have to wonder how much is actually achived with all this...

Here is a product with even smaller particals:

http://www.gotgrit.com/product_info.php/products_id/103

Its .30 microns

Here is the final word on all this:

http://www.stellafane.com/atm/atm_mirror_ref/atm_grit.htm

The smallest abrasive you will find is the .3 micron stuff listed above. The rouge is like .5 microns... 80grit is 165 microns...


This guy suggests that .7 microns is fine for glass...

http://homepages.ihug.co.nz/~ddowning/feb02/03.html

 

[Randyman...]

I thought the "common" belief was a descent lapping job (with regular wet/dry) will get POSSIBLY 1-2*C drop, and a HIGHLY POLISHED lapping might get 3*C. 5x the work for 1*C? I'm happy with my 45 minute Lap job...

Latr

 

[SamDaMan]

Well...

Ya I only spent 4 mintues with tripoli on mine... I really can't see what is gained by all this but if you want a super flat surface then may as well do it right. Spending an hour will no doubt make yur heatsink surface even flatter then the cpu die.... Like how flat is that with their name stamp on it????? Well 2degree C for 1 hour of work... worth it to some waste of time to others...

And eveyone would agree that the swifties are really super flat from the company whereas other companies are a joke (including my thermaletake tower112) by comparison. Clearly you have to do some lapping/polishing

 

[nosilverman]

The reason that you use sandpaper and not polish is that when you are lapping with sandpaper, it is on a flat surface (many people use glass). With polish, you may be able to remove scratches, but the surface may still be slightly warped. Some people will lapp up to 2,000 grit, but most don't see any real improvement over 600. It is true that thermal paste like AS5 fills the gaps, but copper conducts heat many times better than any thermal paste. You use it because it is much better than air, but the less paste you need to fill gaps the better. Do a search if you want to find out more.

-Noah

 

[Randyman...]

After the first minute of lapping, it is painfully clear where the surface is not truly "flat", and the macnine ridges are very apparent. Polishing would NOT be able to knock down these ridges - only polish the existing ridges to a "pretty shine" - this would not help heat transfer at all IMO.

4 Minutes - That is no watse of time IMO. I thought this was a 4-5 hour project - as it WOULD be if starting out with a fine compund - and expecting this fine compound to knock down the ridges. Either way, you need to START with 600Grit to knock down the ridges, then polish all you want to (not much for me). 1000Grit is my stopping point - some will go to the micro polishing steps you recommend - just not me!

Later

 

[nosilverman]

A good way to tell where the ridges and vallies are is to cover the surface of the heatsink with marker. As you lapp it, some parts will get the marker rubbed off first. When it is all gone, it is fairly flat. I just did a stock amd64 heatsink today that looked like someone lapped it with 100 grit or somthing, lol. I had to stay at 400 for almost an hour until i could go to 600. If it starts out smooth, it should go a lot faster.