about polishing heatsinks...

Solution
You don't use sandpaper.
You use wet and dry and a very hard perfectly flat surface.
You don't use anything that you can't make perfectly flat.

However modern Heat Sinks are fine. 10yrs ago it was a problem. Not any more really.
Why do you want to polish it? If it is dirty, you should be able to clean it with 99% rubbing alcohol and a soft cloth (microfiber). Polishing can introduce an uneven surface if you end up removing material on accident.

A heatsink surface that is not 'polished' is likely just fine and like that by design.
 


In order to maximize surface area therfore heat dissipation, the bottom of a heatsink is scratched with something like sandpaper when its made. That being said, I agree this is not something you want to do if the heatsink already has this. But if this heatsink isnt specifically for a cpu or just doesnt have any abrasion on thee bottom of it, its not a bad idea to do so.
 
When you say bottom do you mean the surface in contact with the heat source? If yes that sounds like a bad idea, yes it technically increases surface area but that those scratches will be filled with air or thermal compound so will negatively effect heat transfer.
 


Yes the part that touches the heatspreader. it would be a bad idea if there was no thermal paste, but due to the thermally conductive properties of thermal paste which is mostly metallic, it can transfer heat almost as well as the copper itself so its not a problem to have to dissipate the heat slightly further due to a scratch being deeper.
 
Are you sure about that, I've always read that thermal compound is better than air at conducting heat but not as good as metal to metal contact. I have alway read that too much thermal compound has a negative impact on a coolers efficiency which also suggests its not as good. I thought its main purpose was to remove air and replace it with something better at conducting heat.

Seems counterintuitive to me but I may very well be wrong.
 
Wet 2,000 grit sandpaper on a flat surface making a perfect copper mirror. Been there, done that. It does indeed make a difference. If you do not have the means, don't worry about it. Clean the surface & CPU well, apply some quality thermal paste in the prescribed manner, and enjoy.
 
As 13thmonkey said, it's not really the problem it once was. Years ago the base of most heatsinks were machined roughly and may not have been flat or the machining left grooves in the surface. Newer heatsinks are generally milled pretty nice and flat. Keep in mind some heatsinks are purposely machined with a very slight convex surface rather than perfectly flat to help increase contact pressure to the middle of the cpu's ihs. Sanding it flat will remove that.

It's generally done with very fine grit sandpaper, around 800 grit to start and working through several grits until reaching 2,000 or higher along with a bit of water. People usually use a piece of glass or mirror for a perfectly flat base. It's called 'lapping' the heatsink. I did it years ago and it improved temps, haven't done it with any of the coolers I've used the past 6yrs or better. No need to. Extra care wouldn't be a bad idea if it's a heatsink with exposed direct touch copper pipes since you don't want to sand through them and compromise the heatpipes.
 
Lapping and water sanding are two way different processes. First of all I'm not at all certain why you would want to introduce water to a CPU. There's no guarantee it's even water-resistant, let alone waterproof. Lapping, like you would do to the intake and exhaust valves in an engine head uses lapping paste. Something more akin to jeweller's rouge would be more appropriate for a CPU. You simply put a dollop or two of the rouge on a flat block of steel, preferably stainless and a couple inches thick, and swirl it under light pressure until you get the desired effect (affect?). You would use the same process with a heat sink. When it's all done you simply clean off the lapped piece(s) with 99% alcohol.
 
First off. We are talking about lapping the heatsink, not the CPU.

As a tool & die machinist, allow me to clear up a few misconceptions there. Lapping and sanding are two different names for a similar process. There are basically 2 kinds of lapping, fixed abrasive, as with a piece of sandpaper, and loose abrasive, as with your valve lapping compound example. Water is used as a light lubricant and to help prevent the sandpaper loading up. It is also critical for washing/cleaning parts when changing grits. For instance, you will get nowhere with 1,500 if your work piece is contaminated with 800 grit sized particles. A common mistake with beginners. Furthermore, shape follows form. This is why a flat piece of glass is used under the sandpaper to lap the heatsink surface flat.

 
Rule of thumb: simplify when you can.
Guaranteed sand-paper will roughen the surface. Now, if you have some uber-tech goop that loves that sort of surface? (Some adhesives love it totally! You wouldn't believe properties of some of the stuff we required in avionics.) then hey, go for it.

But //polish//?!! Nope ... back up ... think again.


 
Good on yuh mate.
Polishing? we've been doing that since before we built the pyramids.

Want the best historical documentation on this? Reflectors for telescopes.
Lapping the surface of uber-pure glass ... glorious.

 


Heat sink, not cpu...
 
Sandpaper does not equate to a rough surface. "Polish" is merely a matter of grit size. As an example, here is one of my AthlonXP heatsinks from the old days. Mirror finish on copper with nothing but 2,000 grit paper and water on a piece of glass for flat surface.

aaUY8Rq.jpg


Vast improvement over the heavy milling machine marks on the original surface.

UwJxb3i.jpg
 
I've worked with guys like you.
Score points by sophistry instead of actually getting the work done.
"2,000 grit paper" is what most folk think of as "sandpaper"?

Sandpaper <> rouge.

"Polishing compound is either cerium oxide or rouge, the former being much easier to work with ..."
http://bbastrodesigns.com/JoyOfMirrorMaking/Polishing.html

Folk who put others down as though for sport ... really loath them.
==I'm OUTTA here==


 
As a machinist and metalsmith, polishing & lapping everything from high carbon steel hydraulic diverter plates, to chrome, to gold, I have learned a few things. Simply trying to have a discussion & educate people. To bad it came across to you as deceit & subterfuge. Any beguilement is purely your own perception.

 


Why polish it? It only makes sense that a rougher surface translates into more surface area enabling better heat reduction.
 


Incorrect, there will be metal to metal contact only at the peak to peak interfaces. This might account for 1% of the surface area, the thermal compound (a worse conductor than bare metal) has to fill in the rest. A perfectly flat surface with no peaks would have close to 100% metal to metal contact.
If you could pattern it so that they fitted together perfectly then you would be right, but as a microscopic level you'd still be looking at peak to peak contact.