lapping thermalright heatsink

6

64dragon

Weaksauce
Joined
Jul 18, 2006
Messages
115
i'm thinkin about lapping my heatsink (thermalright si-97), obviouslly the best thing would be to get a better heatsink but i dont have $50. while doin some searchin on lapping i came across this article that suggests going over 600 grit is a waste. what do you guys think? have you heard anything similar
http://www.3dgameman.com/forums/showthread.php?t=42746&highlight=lapping

cant wait to see the gains due to the copper under the layer of nickel, i'm gona check out the gains after the 600 and take it from there. maybe i'll see more than the 2-3 from this site http://www.phoronix.com/scan.php?page=article&item=212&num=2
 
there are fews things more indefinite than whether or not one should lap their heatsink (not to mention the ICS of their proc)


it's really hard to tell WHO to trust. the desired "finish" of the surface is more debateable than the flatness - the heatsink probably should be flat, but MAYBE not so flat its mirror-like
 
has anyone tried lapping board heatpipes (the bases on the north and southbridge)
 
lapping is usually a bad idea because just being flat does not mean you are level. being human mean you will likely put more pressure in some spots even on glass which will create a bias. spring type mounts will compensate but any other mounting hardware will not and you could effectively make everything worse. because of this most precision machining is not manual.

mirror finish is a whole different monster. if you look at a wall with texture that is also glossy the shine does not indicant the flatness of that wall. shine or reflection only shows that light i less refracted by peaks and valleys on the microscopic level. in reality this would only count if your die was microscopic. that is where thermal transfer media comes into play, those peaks and valleys.

many swear by lapping but good quality products are machined well and usually are not mirror finished.

when it come to things like the chipset, they are concave unfortunately and not much can be done other then use more goo whch is not ideal but more akin to a bandaid.
 
not lookin for the mirror finish, i couldnt find 600 grit but did find 800 so i'm gona stop at that. i dont have time to do it till thursday or friday but i'll take pics along the way and post them. to test for flatness i'll try the water drop and scribble with a sharpie on the bottom and watch how it wares during the lapping

link to water drop check for flatness
http://www.techpowerup.com/articles/cooling/air/39
 
Tangent inc. said:
being human mean you will likely put more pressure in some spots even on glass which will create a bias.[...]because of this most precision machining is not manual.
Click to expand...

I know optics experts who make 1/20th wave optical flats by hand. Amateur telescope makers can grind glass to a much, much finer finish than 600 grit routinely. It isn't the material or the equipment that counts, it's the technique.

If you rub two surfaces together with a fine abrasive between them, you will create either a sphere or a plane. To ensure you get a plane you use two surfaces and keep moving the work between them.

It's true that a fool with an attitude can destroy a heatsink. With a soft material like copper, mistakes can happen fast. But if someone knows what they're doing and take the time and care, they can make a reasonably flat surface in good time.

You do bring up a very good point - if the surface is flat but not perpendicular to the vertical line through the heatsink, the thing will sit tight and crooked - the Leaning Tower of Pisa heat sink. If the mount does not compensate, the heatsink might as well be lumpy rather than flat.
 
Unless you're using 100 grit sand paper, it's going to be pretty darn hard to make a heatsink not flat by lapping. When you get into the 1000grit+ range, it starts to move to more polishing than grinding. Anyway, use the weight of the heatsink rather than any force to lap it. Just go slowly and be careful as the SI-97 is oddly shaped for lapping.
 
i couldnt take waiting any longer so after my test in engineering i decided to do it today, i'm disappointed i didnt get any improvement, i'll play around with the amount of AS5 i use(d). also considering using 1000 grit or 1500 but dont want/cant afford for my temps to get higher.
 
heres a few pics for the hell of it, my cameras not that great and had issues focusing on the base, it looks better in person

DSCF2060.jpg


DSCF2064.jpg
 
Ive always said anything over 600 is a waste ( and this is true ) but after making my first DICE pot, I had to make her shiny ...... My cpu was previousely lapped to 600grit, but since I have some 1000 and 1500 grit leaftover, I might as well make my cpu purty...... :D

lapped.jpg


mplap2.jpg
 
luihed said:
Ive always said anything over 600 is a waste ( and this is true )
Click to expand...

yea, i read that a few places while lookin into lapping, i dont think there would be a huge difference between 600 and 800 but i'm disappointed that i got 0 gains from it. on a good note my base is now flat, i was actually surprise to find out how much it was off, the nickel coating wore off from the outside in (concave), the one portion took quite a while

as i mentioned earlier, i was/ didnt care to get a mirror finish...considering that 600 grit is supposed to be best
 
You must log in or register to reply here.
Back
Top