Adventures in lapping

unklerosco

Read through most of this thread....

Lapping makes sense if ur HSF has a bad surface on it... The tuniq tower seemed to have a bad surface so i can understand lapping it. What i dont get is why u seem to think having a mirror reflection means its perfect, it could be convcave n reflective, could be at an angle n still reflective. Remember reading up about it before that a dull surface transfered heat better than a shiny one.. be it only 1-2c but considering thats what ur doing this for...

Also, is it not the point(aswell as transfering heat) of the thermal past to fill any imperfections on the surfaces.

An KRAZY 8... Do not lap the G5, Silver is alot softer than copper n its very very easy to make a mess of it.. Cathar machine the block himself for the best thermal results.

Cuddlesworth

Will do, seems like a waste anyway. I will be doing my E6600's Ihs though as it is terrible.

gline

unklerosco wrote:

Read through most of this thread....

Lapping makes sense if ur HSF has a bad surface on it... The tuniq tower seemed to have a bad surface so i can understand lapping it. What i dont get is why u seem to think having a mirror reflection means its perfect, it could be convcave n reflective, could be at an angle n still reflective. Remember reading up about it before that a dull surface transfered heat better than a shiny one.. be it only 1-2c but considering thats what ur doing this for...

Also, is it not the point(aswell as transfering heat) of the thermal past to fill any imperfections on the surfaces.

An KRAZY 8... Do not lap the G5, Silver is alot softer than copper n its very very easy to make a mess of it.. Cathar machine the block himself for the best thermal results.

totally agree with this, only really to be used on bad blocks, paste will fix the rest

You Suck!

You can argue about the effiency of it, but none the less it's great to see here.

L31mr0d

unklerosco wrote:

What i dont get is why u seem to think having a mirror reflection means its perfect, it could be convcave n reflective, could be at an angle n still reflective.

Having it flat is a given, making it reflective is secondary, plus it would be impossible to get a concave HS completely reflective, as there would always be a part of it that is not getting sanded evenly, if it was convex you'd have to roll it to get it reflective, which would be noticeable if you are sanding on a flat surface. Flaccus already made this point, but the more reflective it is the less imperfections there are in the surface, so there is less need for TIM, which will give you more metal to metal contact. If it was possible, there would be no need for TIM at all if you could get 2 perfectly flat, flaw free surfaces to butt up against each other.

unklerosco wrote:

Remember reading up about it before that a dull surface transfered heat better than a shiny one.. be it only 1-2c but considering thats what ur doing this for...

Proof? Not being smart but i'd like to read that, as I was led to believe from numerous tutorials and guides that the shinier the better for thermal conductance.

unklerosco wrote:

Also, is it not the point(aswell as transfering heat) of the thermal past to fill any imperfections on the surfaces.

It is, but TIM has a much lower thermal conductance than copper, so if possible its better to have copper to copper contact rather than copper to TIM to copper.

Either way, if you don't believe it, don't do it. Nobody's twisting your arm. I'm not going phase or water cooled anytime soon so if doing an hour of sanding can help, then so be it, you'd spend longer setting up a decent watercooled system.

As regards temps, well i'm using AS5 for now as i wanted to test to see if the cpu was still working, before I use the liquid pro. I was idling at 50'c at stock settings before, i'm at 35'c now. The only reason I considered lapping in the first place was because of my ridiculously high idling temps, and it turned out my 165 was very concave. Shaving 15'c was worth the $6 for sandpaper and an hour of my time imo.

gline

L31mr0d wrote:

As regards temps, well i'm using AS5 for now as i wanted to test to see if the cpu was still working, before I use the liquid pro. I was idling at 50'c at stock settings before, i'm at 35'c now. The only reason I considered lapping in the first place was because of my ridiculously high idling temps, and it turned out my 165 was very concave. Shaving 15'c was worth the $6 for sandpaper and an hour of my time imo.

do you not think the temp drop could be more down to you using a different tim and reseating the heatsink??

also instead of lapping the heatspreader, could you not just have removed like a lot of people do??known to give a few degrees drop and takes less time, which means you are getting cpu-tim-heatsink and not cpu-heatspreader-tim-heatsink

L31mr0d

gline wrote:

do you not think the temp drop could be more down to you using a different tim and reseating the heatsink??

I am using the same TIM as before? I was using a zalman 9500 before and had reseated it numerous times, using various amounts and methods of applyling the AS5, i'd then give it a few days of thermal cycles to cure, the temperature would always end up settling at 50'c+ idle or above, and when i'd remove the HS the TIM on the IHS barely looked like it had been touched. Admittedly i've changed the heatsink now to the Tuniq which has also dropped the temps, but it required little effort to sand them and was a helluvalot cheaper than phase or water. If it only drops my temps by 3-4'c then it was worth it for what little effort went into getting it there.

gline wrote:

also instead of lapping the heatspreader, could you not just have removed like a lot of people do??known to give a few degrees drop and takes less time, which means you are getting cpu-tim-heatsink and not cpu-heatspreader-tim-heatsink

This is also a lot more dangerous, you'll find most of these "people" are also not using air cooling as the weight of the HS on the die can cause the corners to be chipped when moving the case around. People will usually remove the IHS if they are going water as the weight of a cpu water block is a lot lighter than full blown HS. Plus I wasn't going to take that chance with the Tuniq, as its one of the, if not THE tallest air cooler on the market. It's already known to stress the motherboard when a case is moved around.

gline

L31mr0d wrote:

This is also a lot more dangerous, .

1 broken pin later lol

L31mr0d

gline wrote:

1 broken pin later lol

Yeah but if i'd found a guide like this one before i'd started there wouldn't be any broken pins. So hopefully others that consider lapping will find this and take the needed precautions when lapping a cpu that has pins. Most of the guides out there atm are for c2d which has pads instead of pins so the same precautions aren't needed.

unklerosco

L31mr0d wrote:

its better to have copper to copper contact rather than copper to TIM to copper.

Why isn't there a copper TIM avail????

I've lapped many HSF's before, as a need rather than for a performance gain.

Either way, u do what u enjoy.. I've spent over 6 hours polishing the door of my case... N it aint gonna make my PC go any faster or run cooler but i still did it(and broke a bench grinder in the process!!)

L31mr0d

there is afair. But its not liquid copper, its basically micronized copper instead of the silver you get in AS5... and as silver is a better conductor of heat then copper, its not nearly as good as AS5. But still if you look at the comparison chart on the first page of this thread you'll see even the likes of liquid pro and AS5 are far down the scale of thermal conductance compared to copper itself, due to the impurities and non thermally conductive chemicals added to the TIM.

But like you said, I lapped, like you, because there was a need. Under normal circumstances if my temperatures where acceptable I wouldn't of even considered it. I wasn't going to have my Opty 165 running at 1.8Ghz when I know it can do 2.7Ghz or above, so I had to sort out the temperature problem, and from reading around lapping was the simplest and cheapest solution... and it seems to be working. Most people will never even attempt to lap their CPU, but if you're unlucky enough to have one horribly concave or convex then apart from removing the IHS, lapping is the easiest solution.

L31mr0d

I bought a HR05 the other day, and guess what... yep I lapped it It was actually very flat I have to say, even from the first stroke across the sandpaper I could see that it was sanding evenly. Still I wanted to get rid of that dull nickel plate and show off some of that naked copper :thumb:

Does anyone know the reasons behind coating the copper in nickel? Is it to stop it tarnishing or something?

Here as the pics, same process as before:

Fresh out of the box:



Euro test before sanding with the nickel finish:



600 Grit, using the SC stuff this time not the micron stuff



Voila, after 10 minutes on 2500 grit and a few rub downs with some brasso :cool:

DanGerMus

Just one or two things. someone mentioned that a rough surface is better at conducting than a smooth. well that all depends doesnt it. If your talking about a metal(alluminium/copper) to fluid(air/water) transfer like in your common forced convection (fan) heatsink,or water block, then that would be true because the more surface area you have the more easily heat will be conducted ergo rougher surface will mean more material in contact with the air. the same goes for fluid/water. but when your talking metal to metal transfer the **smoother** and flatter the surface the better because in that case this gives you more material in contact with material and so better transfer.

About the nickel coating your on the right track with the tarnish because tarnish as we know is a kind of oxidation or other build up depending on how greasey your fingers are or whatever. tarnish would mean that you basically have a whole new chemical in the mix sandwhiched between you HS and CPU surfaces and that said mix could possibly have negative effects on your heat transfer...hence they use nickel which doesnt tarnish or oxidise as easily or quickly as copper would.... but seeing as your useing pastes and hoping you keep your surfaces clean you should be safe from anything happening to your bare copper surfaces as long as you put it in place very soon after polishing it cos pure copper tarnishes quickly but i doubt they use pure copper.

L31mr0d

Yeah I read up on it and the reasoning is that it is better to have a flat rough surface than a shiny concave surface for heat transfer. The best is to achieve both together if you can, but the majority of the benefit to lapping will come by just getting the surface flat, making it shiny might get you an extra 1'c drop in temps.

Anti

Stickied. as this is a great informative thread.

DanGerMus

BECAUSE YOU CAN GOD DAMN IT! IF you stand to gain at all from an hours effort, an hour that might other wise be spent scratching your arse in front of the telly, then why not do it. Time spent constructivly in any form is time spent well. Also it's just soooo shiny and purdy. Why have 95% of your potential when you can easily gain 100%. If i'm being phylosophical then call me Homer.

bushy...

Maybe getting the finned aircooled bit of the cooler gritblasted would help too ?

L31mr0d

bushy... wrote:

Maybe getting the finned aircooled bit of the cooler gritblasted would help too ?

I think I speak for everyone when I say "what???"

Anti

He means sand blast it, As some reports say a "roughed" up surface can work well too.

L31mr0d

Anti wrote:

He means sand blast it, As some reports say a "roughed" up surface can work well too.

Yeah but sandblast what? He's talking about the fins of the HS. Maybe it's supposed to be sarcastic or something.

Anyway. I need to make a revision to my original method of lapping.

I've since changed my method to one which I personally have found to be more efficient and leads to less errors and scratches which will cause you to go back to a lower grit to remove.

I work up through grits 200, 400, 600, 800, 1000, 1200, 1500, 2000, 2500 and then 10 microns.

The majority of the sanding will be done at 200 grit as this will get the surface flat as soon as possible (this is a good thread which illustrates the difference between a shiny surface and a flat one), if you can't get it flat at 200 grit, you aren't going to get it flat at 400 and so on. You'll need to use something to check that you are getting a flat surface (note: it won't be anywhere near a mirror finish at this grit so using graph paper is out of the question) I personally use a machinists ruler (i.e. those metal ones for tech drawing) which usually have a perfectly straight edge. Line it up perpindicular to your lapped base with a light behind it. If you can see light through between the 2 surfaces then the base isn't flat and needs to be sanded more

Now to begin, I sanded at 200 grit backwards and fowards on the surface for a few minutes, checked if it was flat for the direction I was sanding then rotated the HS by 180" and continued. Once it was flat in that direction I rotated the HS by 90" and repeated. Until both directions of the HS where flat (i.e. N to S and E to W).

Now the hard part is over, you have a flat surface. If you are so inclined you could leave it there and use some TIM to fill in the scratchs. If you want to go that extra mile and get it shiny AND flat then read on.

Moving up through the grits, I found it took as long to get up through the grits as it did to get it flat, so getting it flat should be half of the job. The benefit of sanding in one direction per grit level then rotating 90" for the next is that you can be sure that the surface is evenly lapped because the scatches from the previous level should be completely removed, it's because of this that I recommend NOT to sand in figure 8's or circles, as you can't be sure that you are evenly sanding.

At each grit level clean the flat surface you are working on (i.e. flecks of metal on the glass under the sandpaper could cause deep scratches in the surface) also everytime you lift up the HS to check if it is sanding evenly and if it is still flat you should clean off the metal dust that is on the sandpaper you can do this by blowing on it, using a can-o-air or wiping it with a cloth.

Also once you get to a point where it is reflective start using graph paper as well as the machinist ruler to check if it is flat. Also don't worry if there is still small scratchs in the surface when you are finished. It doesn't have to be perfect, and these scratchs will be filled by the TIM anyway, the main thing is to have it flat and have the majority of the surface smooth for better contact. Some slight scratches are not going to affect your temps that detrementally. Also I found that using brasso left the surface speckled with tiny pock marks (if you look close after using it you'll see this) which is why i'd recommend NOT using brasso if possible. IMO I feel it is better to have a few tiny scratchs on an otherwise flat surface then to have a surface covered in tiny pock marks. Anyway, the choice is up to you. After this give the surface a clean with some alcohol or some Articlean and your done. Be sure to install teh HS asap after lapping as the bare copper might oxidise if left out to the air over time.

Any comments or ideas are welcome.

DirkVoodoo

mathias wrote:

Im also a mech/Elec engineer , when I first saw this I thought , no way could this be just to improve thermal conductivity , but lo and behold it was , working in heat systems for years tells me this is way too much effort for very little return

As an engineer, you shouldn't be asking these questions. It is a fairly simple solution, and I for one would call a 1-5 degree temperature drop significant for a modest amount of lapping.

To the OP, informative post, proved a good read.

sebastianlieken

hey guys, great innovative thread btw.

i was completely ignorant to lapping till now, seems like a solution to my high temps with my e6600. ive ordered some as5, gonna get rid of those stupid puch pins on my scythe ninja plus,and replace with bolts. tracking down some lapping gear for my scythe aswell. but it would make alot more sense if i was going to lap e6600 aswell then. question is, is lapping a c2d high risk?, anyone done it yet?

can i expect a good temp drop after doin all this?....as in is it worth the trouble?

oops, double posted!

L31mr0d

sebastianlieken wrote:

hey guys, great innovative thread btw.

i was completely ignorant to lapping till now, seems like a solution to my high temps with my e6600. ive ordered some as5, gonna get rid of those stupid puch pins on my scythe ninja plus,and replace with bolts. tracking down some lapping gear for my scythe aswell. but it would make alot more sense if i was going to lap e6600 aswell then. question is, is lapping a c2d high risk?, anyone done it yet?

can i expect a good temp drop after doin all this?....as in is it worth the trouble?

oops, double posted!

check out my build log (link in sig) i recently lapped my E6600. You will only see temp drops really if the surface is either convex or concave, if it's already flat, lapping probably won't do anything for you. In fact if you do it wrong and leave scratchs on the surface you could actually make your temps worse.

I'd practive on the stock heatsink you got with your E6600 first to work out what method works best for you. I'd also check to see if the surface is flat or not.

A simple way to check is to spread out a thin layer of TIM on the IHS, then attach the flat HSF, when you remove it, have a look at the TIM on the IHS and see if there are areas that haven't been touched by the HSF. If there are then one of your surfaces isn't flat.

Get your kits from easypckits.com, get the premium kit and ask the seller to reduce the price and not ship you the glass. Hes fairly easy going and will do this for you.