"Polishing" guide wanted :D

[Dustswirl]

Does anyone have a guide to"polishing" a heatsink base plz?
I bet many of you have some experience and theories on how to obtain the best surface for an improved heat transfer!

Just got no namereallycheap HSF (but with a TTsamartfanII on for 5$ ) that is quite useful for shredding cheese over a nice hot pasta plate since it's base is almost like a nail-file and worse!

 

[Peter D]

like.. Lapping a heatsink basically? try searchin on Google for like 'heatsink lapping'

 

[DivideBYZero]

Don't search google for 'Clam Lapping', however.

 

[onza]

Originally posted by: DivideBYZero
Don't search google for 'Clam Lapping', however.

not even sure how you even thought of that one LOL

 

[Dustswirl]

If i wanted i would have searched for it in google... but in fact i want to interact with ppl not a search engine coz usually ppl are smarter...usually

 

[Eric1285]

There's a guy who sells heatsink lapping kits, which is a really good deal, since if you went to get all those grits of sandpaper by yourself, you'd spend at least $20. His kits are like $4...try googling for lapping kit.

 

[Dustswirl]

Normally i use car polish hardware (different heads for different softness), the results are gr8 but not "mirror like" like some friend of mine who uses his knowledge in chemistry and makes a "perfect" surface.
Am not at all new into lapping/polishing etc etc in fact i even make my own heatsinks (even molded one out of silver...yes Ag, the best "normal" metal for heat and electricity transfer...it sucked btw coz no normal fan can dissipate heat fast enough and actually am working on a hybrid version of this "expensive" but highly efficient crazyness)

I just wanted to see if anyone in here pushed the limits of air cooling (not too exentric stuff though) and finding a way of polishing the surface to a mirror finish not just lapping it to a "normal" surface.

Btw the surface quality is so important in heat transfer but few ever venture in this. AMD just realized that and just look at the new cpus (opteron and A64)...anything "special"?

 

[dbwillis]

Go to an auto paint store and get some sandpaper from them, 320, 400, 600 grits

go home and cut the sheets into strips a bit wider than your heatsink is..tape the paper to a nice flat surface (mirror, glass, etc)
then wet the paper and maybe even put some soap on it to keep it nice and slippery

work the heatsink back and forth until it gets smooth and the 'grain' is all going one way..then switch the heatsink 90 degrees and do it again till all the grain is going the new direction.
then switch the grits until you are at 600 grit..at this point, the heatsink should feel real smooth..

after its done, you can use some polish like Mothers on it to shine it up if you like.

takes about 20-40 minutes for me to do an Alpha 8942 and P4 chip

Alpha
P4
Polishing

 

[WarCon]

I don't know, but is lapping to a high mirror finish really that beneficial? I know lapping it flat is important in order to remove any concave or convex imperfections and that usually gets fairly mirror like just using 1000 grit. If you were matching it to an equal mirror like surface and could manage to apply the exact amount of thermal compound so that you weren't floating the smooth surfaces completely apart on a layer of thermal compound, then maybe it might make a bit of difference. (One member uses his Arctic Silver as the liquid for wet sanding so that the final grooves that are left have a better chance of being filled with thermal compound versus the silica/aluminum oxide that sandpaper is made from).

Always wondered if a laser would work well. It should make a fairly flat surface, but will it turn the surface into an oxide of the material from heat?

If you do this please let us know if it did matter though, I am curious.

And what kind of silver heatsink are you making? Pins or fins? You might try taking the shine off of the radiating fins/pins or even run them over a sander to make deeper grooves in them. This simple action can increase surface area and improve turbulence (which aids in transferring the heat into the moving air). Also soldered is better than pressure fit for pins or fins, but if you are using pins at least remember to drill them smaller than the pin size and then freeze your pins and heat your base so that the base will shrink around the pins and the pins will expand to make an extremely tight pressure fit.

 

[Dustswirl]

WarCon >>> am testing an electrolysis procedure to ge the perfect surface for the base that'll be made of high grade silver. The fins are ondulated (~), not straight for more surface/per unit ratio and are soldered to the base. Each fin is molded in high grade extra thin modling paste for perfect mirror like finish after electrolysis (for each of the 45 fins).
One important thing. the side that'll be facing the memory modules is designed in such a way that the hot air won't hit them like in some normal HSF units.

Another design that is still on paper is a ducted HSF with the air blowing outside the case (i guess am not the one who invented that) but it's designed in a way that the whole thing is but a big heatsink (ouf course the air will be sucked out not blown in)

btw want to know how delicate the surface problem is? well believe it or not, since all system's mobos are vertical and that cpu/HSF contact plane is perpendiculare to the horizontal, if your final surface has grooves(no matter how small they are) that are also perpendicular to the horizontal your ASIII or whatever fluid heat transferring stuff will just "shift" and follow gravity... i didn't believe it myself until i saw results in some PC lab... still can't see what's important in that but i got told it has an impact on the cpu. (!?)

(in case you're wondering... my sister is a jewel designer i get to play with gr8 weird contraptions lol)

 

[screw3d]

Sounds really cool.. any pics of the silver heatsink? I'll imagine it'll cost u a bomb to get that much Ag to build a heatsink ??

 

[Dustswirl]

marvie>>> when i finish i'll get a shot @ and post some pics and even a test and comparison! as for the price well it sure is costy but since my sis leaves "trash" behind while working (expensive trash if i may say so lol) i collect them and mold them!
Right now we're testing (using the good old calorimeter) different metal fins, looks like copper is gr8 but i can't use machines made to handle precious metal (very delicate) for plain copper stuff

cheers

 

[jdurg]

Dustswirl, what do you plan on using to prevent the silver metal itself from reacting with the air? While I know that silver is a "noble" metal, it isn't really all that noble. If you leave silver lying around in the air for too long, it will react with the oxygen and any sulfur gasses in the air. (Sulfer containing gasses + Silver = black/brown tarnish). With the silver metal getting warmer from the contact with the CPU, its reactivity with sulfur gasses may become even greater. Are you planning on treating the metal somehow to prevent that type of reaction, or are you just banking on the fact that even a tarnished silver metal would be a better heat conductor than copper/aluminum? (Which I believe it may indeed be.)

 

[Dustswirl]

jdurg >>> very intresting note you're making here!!! actually i dunno what i expect out of the tarnished silver but afaik heatsinks like the thermalright SLK800/900U family and all those units tarnish after some time but their effectivness doesn't seem to be affected. now sulfur gas is less likely to be around cpus what am really afraid of is normal air with high humidity whereas oxydation isn't the problem at first but the N+O+H2O (and heat) action that will defenetly blacken the metal! as long as there's no corrosion action then it's ok. Hey i might chrome or nickel the whole base (like the new HSF units by that are packed with new AMD cpus)... but dunno how physically!!! (i can use electrolysis...but $...)