Lapping?

[Kiwi]

What exactly is lapping? I've seen it mentioned a few times in some guides and posts, but I've yet to find a guide that actually explains what it is or how you do it.

 

[Sony]

It is when you sand your HSF it steps normally starting with 600 Grit the 800 the 1200 the it's your choice how high grit you go depending how smooth you want it. The reason this is don't is it makes the HSF more effective and it works better I went all the way up to 4000 Grit I got a super smooth HSF.

 

[taz1004]

I've seen people lapping the heck out of their heatsinks. Poor heatsinks. Lapping with higher than 300 grit sandpaper isn't recommended because if it's too smooth, there's no place for thermal compounds to go in. As long as the surface is flat, (not necessarily smooth) thermal compound will do the rest.

I've seen guys lapping for hours trying to get it to mirror like finish. Ends up shaving up to 1mm off the heatsink. This is insane. If you sand the bottom of the heatsink too much, it will likely decrease the thickness of the heatsink and therefore decrease the pressure on the core which the clamp is designed for. And might cause ineffective cooling.

--BrianC

 

[Warlord2]

I lapped my fop38 with 2000grit and got it 8c cooler!!!!

 

[JaY_III]

taz1004 (Jun 18, 2001 02:21 p.m.):
I've seen people lapping the heck out of their heatsinks. Poor heatsinks. Lapping with higher than 300 grit sandpaper isn't recommended because if it's too smooth, there's no place for thermal compounds to go in. As long as the surface is flat, (not necessarily smooth) thermal compound will do the rest.

I've seen guys lapping for hours trying to get it to mirror like finish. Ends up shaving up to 1mm off the heatsink. This is insane. If you sand the bottom of the heatsink too much, it will likely decrease the thickness of the heatsink and therefore decrease the pressure on the core which the clamp is designed for. And might cause ineffective cooling.

--BrianC

Thermal paste is bad (sort of)
the entire reason of lapping is to make the surface falt, thermal paste fills in the small holes that you can not see so that contact is made with what ever you are trying to coo.

Now a lapped surface, doesnt have as many holes, and much more metal will touch the surface of what you are trying to cool ,better contact, better cooling. Now less thermal paste is needed, and that is good because thermal paste is not anyware close to efficient at transferring heat as heatsink to core(providing 100% contact)

 

[Carousel]

what are you guys using to make sure you keep the surface flat? standard tables aren't machined to any sort of tolerances and lapping blocks are expensive. I also thought Thermal grease was to fill in imperfections in your contact area, so if you had perfect mating you wouldn't (in theory) need it.

 

[taz1004]

"the entire reason of lapping is to make the surface falt, thermal paste fills in the small holes that you can not see so that contact is made with what ever you are trying to coo."

These thermal paste molecules are not that fine as you'd expect. Silver is even larger if you use AS2. You need some degree of roughness so the holes can be filled with thermal compounds. Mirror like finish wont let the compounds in. Generally 300~400 grit sandpaper is right size. Try it. Apply some thermal compounds on a mirror and scrape it off with a razor. How much of it is staying on the mirror?

"Now a lapped surface, doesnt have as many holes, and much more metal will touch the surface of what you are trying to cool ,better contact, better cooling"

I didn't say lapping is bad. I said excessive lapping is bad. Yes there wont be as many holes if you lap with say 4000 grit. Problem is, holes are too small for thermal compound molecules.

"Now less thermal paste is needed, and that is good because thermal paste is not anyware close to efficient at transferring heat as heatsink to core(providing 100% contact)"

You can never achieve 100% contact even with a mirror.

Few web sites did testing on this. I can't seem to dig it up.

--BrianC

 

[Phil]

Your right that you can never achieve 100% contact, but the arguement over how smooth you need it is not one thats can just be settled by saying 'this is how it is' I would expect the temp differances between a surface lapped with say up to 2000 grit and then polished with brasso to be almost the same as those of one only lapped to about 1200 grit. I have mine lapped with 1200 grit not to a mirror polish but it actually came out quite dull and milky which is odd as I have lapped the same sort of heatsink but only gone to 800 grit and got a silvery smooth but not shiny surface. I would recomend going up to around 1000 grit though, it would be an interesting comparrison to take a load of heatsinks and compare the differances each grade of lapping makes to them.

 

[CrystalMethod]

Phil (Jun 18, 2001 04:45 p.m.):
Your right that you can never achieve 100% contact, but the arguement over how smooth you need it is not one thats can just be settled by saying 'this is how it is' I would expect the temp differances between a surface lapped with say up to 2000 grit and then polished with brasso to be almost the same as those of one only lapped to about 1200 grit. I have mine lapped with 1200 grit not to a mirror polish but it actually came out quite dull and milky which is odd as I have lapped the same sort of heatsink but only gone to 800 grit and got a silvery smooth but not shiny surface. I would recomend going up to around 1000 grit though, it would be an interesting comparrison to take a load of heatsinks and compare the differances each grade of lapping makes to them.

Which sort of brings up the next thing that should come to mind. Lapped your heat sink? Great! Even got a perfectly flat surface to do it on? Even better!
...how flat and level is the die on your CPU? I don't know about the rest of you, and i've read some posts about some people that managed to get away with it without damaging their CPU's, but lapping the CPU? Are they nuts?!? Personally, I'm not that ready to make a perfectly good procsessor into a keychain. I'd rather go on top of my roof, and light 100 dollar bills on fire, for all the world to see the money I have to waste...
Speed, and lower temps are what we're all looking for, but at least use some common sense before you do things. Think twice or even three times before you do something.

 

[Hoot]

If the finish on both mating surfaces is so fine that they won't let AS particles into them, then you don't need AS. Nothing to be gained by it. You don't prep the mating surfaces to suit implementing AS. You use AS to make up for shortcomings in those surfaces. Lets get who is serving who in the proper perspective here. ;D

Hoot

 

[JaY_III]

taz1004 (Jun 18, 2001 04:31 p.m.):
"the entire reason of lapping is to make the surface falt, thermal paste fills in the small holes that you can not see so that contact is made with what ever you are trying to coo."

These thermal paste molecules are not that fine as you'd expect. Silver is even larger if you use AS2. You need some degree of roughness so the holes can be filled with thermal compounds. Mirror like finish wont let the compounds in. Generally 300~400 grit sandpaper is right size. Try it. Apply some thermal compounds on a mirror and scrape it off with a razor. How much of it is staying on the mirror?

"Now a lapped surface, doesnt have as many holes, and much more metal will touch the surface of what you are trying to cool ,better contact, better cooling"

I didn't say lapping is bad. I said excessive lapping is bad. Yes there wont be as many holes if you lap with say 4000 grit. Problem is, holes are too small for thermal compound molecules.

"Now less thermal paste is needed, and that is good because thermal paste is not anyware close to efficient at transferring heat as heatsink to core(providing 100% contact)"

You can never achieve 100% contact even with a mirror.

Few web sites did testing on this. I can't seem to dig it up.

--BrianC

i get ya now