Guide: lapping your HS/F

[WeStSiDePLaYa]

LOL!!!

dude, this is the WORST lapping guide ever. i wouldnt be suprised to see temperatures increase if someone followed your guide. thick glass is the only way to go.

unless you can show me how to make your hand flatter than glass. noob.

[cantankerous]

damn, you guys are real, real harsh. 2gigs wasn't even going to do a guide but did so cause myself, biohead etc asked him if he would be so kind to take a few mins and do one. If you guys don't like it, don't follow it but don't push him down into the mud over and over again. I appreciate what he has done and I hope others do too. I really like this site but sometimes the stuck up elitest attitude of most members is totally uncalled for. There are a million things people say on here that I don't agree with but I respect their opinions as people and let it go, maybe you guys should start doing the same.

[[XC] DragonOrta]

It's not so much that the people are stuck up, as it is that they are trying to give him advise on how to correct things in his guide, and he refuses to take that advise.

People here don't have much patience for people that do things wrong but are bullheaded on "doing it their way"

[mad mikee]

Nothing except swifty stuff touches my HW unlapped

If you can find a piece of leftover granite countertop - GOOD stuff.
I found a couple of pieces in an alley when I was walking home
Find somewhere that makes them and ask for their 'leftovers'.

Get some 220 and 400 grit also since you never know how bad the surface is to start, had some REAL bad thermalrights over the years.

My way is to start w/ 400, then if I find spots are NOT getting sanded and they are LARGE, 220 to the RESCUE . Once you are flat, you can move up rather quickly since copper or AL is relatively soft.

Soap - No you don't NEED it, but it helps IMMENSELY, by keeping stuff in suspension when sanding so that it doesn't clog up the paper.

1 tiny drop is more than you need

I use ceramique in a solvent (alcohol, etc) and a paper towel for a final finish (smooooottthhhhh)

another thing im not sure u need the dish washing soap.

[Uncle_Mojo]

Jeez, fellas! Take it easy on the poor guy. While I don't necessarily agree with everything in his guide, 2gigs took the time to do this because he was asked. Give him a little respect.

On the other hand, 2gigs should expect to see some critical responses and be objective about suggestions to make the guide better.

JM2C

[WeStSiDePLaYa]

sorry, but when its a guide that in all reality will probably cause worse perfomance, im not to gratefull. especially when he wont take advice from consesus, and experience.

[Overconfidence]

LOL!!!

dude, this is the WORST lapping guide ever. i wouldnt be suprised to see temperatures increase if someone followed your guide. thick glass is the only way to go.

unless you can show me how to make your hand flatter than glass. noob.

Lap

5 a : to dress, smooth, or polish (as a metal surface) to a high degree of refinement or accuracy b : to shape or fit by working two surfaces together with or without abrasives until a very close fit is produced

Obviously there is some misconception in what we are aiming for in lapping.

Most people take 5a as the general definition of lapping.

However, the better way is 5b. In fact, if you could, lapping your CPU's IHS against the heatsink would produce the best result.

The method with glass is a little bit of both, and probably the best way. In fact, anything that has a flat surface (And that you can work back and forth fairly easily, such as a piece of relatively flat wood or plastic) will work fine. Granite and glass are great suggestions.

[dinos22]

LOL!!!

dude, this is the WORST lapping guide ever. i wouldnt be suprised to see temperatures increase if someone followed your guide. thick glass is the only way to go.

unless you can show me how to make your hand flatter than glass. noob.

there's no need for this dude

[[XC] moddolicous]

This is my personal opinion, but I will trust cathar when it comes to lapping:

Well in the hunt for the best die-baseplate thermal "connection" with my waterblock research, I've found the "best" method that I've been able to achieve, and it falls in line with what's accepted and been tried elsewhere. However, me knowing me, I don't believe stuff until I try it for myself.

Basically I've lapped bases with:

150 grit
220 grit
320 grit
600 grit
800 grit
1000 grit
1200 grit
1500 grit
brasso + strong cloth
toothpaste + cloth
fine-cut cleanser
jeweller's rouge

All in incremental steps, to varying degrees of finish from rough, to a copper finish you can shave in.

So what's worked out the best?

Go to Bunnings. Grab a small board of 12mm thick MDF (Medium Density Fibreboard). It's flat as glass and a whole lot cheaper. Should cost about $5. Pick up 1 sheet each 150grit, 320grit, and 600 grit wet'n'dry sandpaper (black colored).

Take your heatsink, and lightly scribble over the base with a black permanent marker. Color the whole base in if you want.

Stick your 150 grit paper onto the MDF.

Take your heatsink/waterblock and scrub lengthways (if the block has a longer length in one direction - otherwise just pick). Move the block up and down the sandpaper while pushing down firmly. Do not move in a circular motion, just straight up and down. You'll need 3 hands. One to hold the block, and two to hold the edges of the sandpaper down so it doesn't "grab" and fold and rip. If nature didn't supply you with 3 hands, sitting on the floor and using a foot works just as well. Keep scrubbing until you can't see ANY of the ink/black. I periodically look at the base to see how it's going, this is how I can tell if a block is initially convex, concave, or whatever. This can take anything from 1 minute to 30 minutes (or more) depending on how flat the base is to start with.

Once the ink is gone, use the 320grit and scrub cross-ways. This time keep doing it until all the cross-lines left by the 150 grit sand-paper are ground away. Just move the block straight up an down - no round motion. This should take about 3-5 minutes.

Then use the 600 grit sandpaper and moving straight up and down lengthways sand the base until the 320-grit cross-ways lines are gone (another 3-5 minutes).

Finally follow up with a circular grind of the base on the 600-grit sandpaper for about 30 seconds. Wipe clean. This should leave the base with a dull reflective surface.

You're done!

Like I said, I've tried anything from 150-grit up to a mirror polish, but the dull 600-grit finish works out the best. Surprisingly, a 220-grit finish, and a mirror finish give about the same results (at least for me), while the dull 600-grit finish gave me about 0.5C better than either of those two for a ~100W CPU load.

I know that the above has been said here before a few times, so consider this as a re-affirmation of the procedure.

[nomad3700]

i think the glass thing is prolly the biggest oddity on the guide to use hands instead of glass... i think the glass thing is quite improtant, because flatness is teh KEY...

Another thing THAT you can do is to lap your IHS and HS to the same level of flatness by using the same procedure on each and thus maximize flatness/contact points.

BUT alll this really goes to waste if you dont get the perfect amount of thermal paste onto it, because huge amounts of thermal paste will basically kill this contact effectiveness. A lot of things are at play here.

LAPPING is not for people who dont understand the concept behind it..... (not aimed at 2gigs, just in general for other ppl). IF you really really want a good lap do some research on how other ppl do it, whats the best method and onwards from there.


Nomad.

[Delirious]

Now that is what i would call an excellent guide.

This is my personal opinion, but I will trust cathar when it comes to lapping:

People look at the reflectiveness of a heatsink and think that because its like a mirror its flat. All that means is that the bottom was polished so smooth its basically a mirror. Thats not to say a heatsink cant be really flat and a mirror.

[WeStSiDePLaYa]

i think the glass thing is prolly the biggest oddity on the guide to use hands instead of glass... i think the glass thing is quite improtant, because flatness is teh KEY...

that is the whole reason he is recieving critisism.

as using a perfectly flat surface is the whole reason it reduces temps.

if your gonna use your hands, you might as well lap your block with a slice of bread.

[dinos22]

This is my personal opinion, but I will trust cathar when it comes to lapping:

yeap that's what i was referring to

[[XC] moddolicous]

Yea, I had a tough time finding it. I knew I had it quoted somewhere, just couldnt find it, but eventually with a little work I found it.

[dinos22]

i did mine in a similar way....3 hands and all hehehehe.......i continued with 1200-grit as the heatsink was also going on an IHS-less Opteron.....

I lapped the base with 180, 360, 600, 1200 grit paper on granite kitchen bench...

i wasn't after perfection as you can see but something that will get rid of the major grooves that's all.....looks better now

[WeStSiDePLaYa]

dino, that finish looks a bit concave from the last 3 pics.

did you lap enough from begining with low grit?

[Uncle_Mojo]

dino, that finish looks a bit concave from the last 3 pics.

did you lap enough from begining with low grit?

LOL!

[Delirious]

Its kinda hard to tell from the picture, its probably just the lens on camera distorting things slightly.

dino, that finish looks a bit concave from the last 3 pics.

did you lap enough from begining with low grit?

[dinos22]

could be concave don't know.....it's flippin' hard doing it properly as some of you probably know


anyways the whole idea for me was to have a flat IHS in the middle to cover the core properly and make sure temps are good..................check out the load temps on this sucker with Opteron 146 overclocked to 3.1GHz @ 1.63v

great load temps.........i was priming at 3GHz 1.42v and temps were at 31-32C

2.9GHz SuperPI load temps



3GHz prime...load temps around 32-33C




this is the best CPU i've owned

12 Hour Prime Stable at 3.1GHz


i was wondering whether the temps were out but having just used a thermal probe it seems they are OK :drool:

the HSF is not even warm when i touch it .................unlike my dual core.....when i prime my dual core at 2.6Ghz 1.44v it's hot as

just pulled out my temp probe...........
Coldest Ambient temp measured: 23.3C
Temp probe on side of heatsink: 30.5-31.2C
Temp probe on top of heatsink where the heatpipes going in: 32.1-32.7C
Temp probe on the base of heatsink...slid it towards the CPU: 33.7-33.9C

[2gigs]

never use brasso, good way to stuff up, like your ihs

[WeStSiDePLaYa]

i personally dont use any form of polishing compound, i only use wet/dry auto body finish sandpaper, bottled water, and a sheet of glass.

heres results i have gotten, used upto 1500grit paper.

notice how the edges of the block are still crisp and sharp, and not rounded off. and the scratch in the top left from dropping it

i also prefer the circular motion, and switch up between clockwise and counterclockwise, my thing against a straight back and forth motion is, that if there is any little air bubbles under the sand paper, constant back and forth on the same area of that raised paper, will eat a valley into the block.

kinda like if you drip water all over a brick, it will eat away at it more uniformly, but if you kept dripping water in same spot, it will eventually eat a hole into it.

[Sam666]

This is a nice lapping guide I come across the other dayhttp://www.bigbruin.com/reviews/lapp...dex.php?file=1

[omga14]

is the flatness and not the mirror finish the most important thing? Cathar mentions that his best results after multiple trials was with 600 grit as the end paper used. i can only get 600 here where i live as far as i've looked and used that to good effect on my zalman(don't laugh) and got decent results from that sad little cooler.

[WeStSiDePLaYa]

is the flatness and not the mirror finish the most important thing? Cathar mentions that his best results after multiple trials was with 600 grit as the end paper used. i can only get 600 here where i live as far as i've looked and used that to good effect on my zalman(don't laugh) and got decent results from that sad little cooler.

yes, flatness is KEY.

[-aDaM^]

nice