Lapping Theory

bigben2k · 07-09-2002, 09:36 AM

Here's my take on the lapping issue:

If the block is poorly lapped, then, as we apply the paste, air pockets may remain on the surface of the block, deep into those scratches or crevices.

If the block is polished to a mirror, then the effective surface of the block, within the contact area, is reduced/minimized.

In a properly lapped block, these crevices are not so deep that they cannot be completely filled by the paste, but they are also not so flat so that the effective surface of the block is as small as it can be.

For example, assuming that the surface is very finely scarred to form perfect rows of triangular pyramids, the effective surface area of the block would be tripled, in comparison to a perfectly flat finish.

I believe that a perfect row of pyramids with 45 deg angle sides, or rows of 45 deg "ramps", would be optimal. This would translate in an effective surface area 1.4 times that of a perfectly flat surface.

On the other end, the surface could be formed of very long microscopic points, but the effectiveness there would be reduced, as the tip of such points would not actually improve the thermal transfer much, if at all.

The thermal paste is also a big factor. It needs to be low in density, as least during the application, so that it effectively penetrates the crevices of the surface. It should probably be applied with great care, the same way that plaster is used to patch a wall: use a spatula (razor blade in our case), spread the compound around, from all directions, while applying some pressure, to ensure maximum coverage.

Realisticaly, an imperfect surface (that is relatively level) would have a surface that has been created by the cutting mechanism (saw, mill, etc...) and that may have been smoothed by a variety of processes (sanding, buffing), possibly using different substances (polishing compound, oil, mild acid, etc...).

Also realistically, we do not lap our blocks with electron microscopes. The tools that we have are simple, and limited:
-sandpaper, usually found in 600, 800 and 1000 grit.
-water (or any other lubricant used while sanding, but water should be best)
-buffing wheels, and sometimes buffing compound.

The process is very simple, as long as one understands that the block must remain perfectly flat while being sanded, otherwise, a curved surface (with irregular sanding) will be the result.

This is easily achieved with the use of a flat surface, like a pane of glass (because its flatness can be seen in the reflection). We can lay the sandpaper on the glass (which is in turn laid on a table), add a little water (optional), and run the block over it several times, in circular motions. The result can be "measured" simply by passing one's hand over the sanded surface. The result should also be visually inspected: an earlier attempt at lapping may have created an unusually large or deep scratch that a finer sanding may not be able to get to. All in all though, we have to keep in mind that the only area of interest here is the contact area with the core.

I believe that using buffing wheels, especially with buffing compounds, is a bad idea, for lapping a block. The buffing wheel creates too flat a surface, and some of the buffing compound would surely remain on the surface.

In any case, the resulting surface will be highly irregular, and will not have any form of perfect rows of perfect pyramids.

When I get my block (and the rest of my computer), I will test with 600, 800 and 1000 grit sandpapers, and report my results (unless someone beats me to it!).

redleader · 07-09-2002, 12:46 PM

Quote:

For example, assuming that the surface is very finely scarred to form perfect rows of triangular pyramids, the effective surface area of the block would be tripled, in comparison to a perfectly flat finish.

Assuming you want any compound at all in there. Pure copper is almost 100 times as conductive as silver paste. With a perfect finish, you could remove the paste altogether and have a much, much more effcient junction.

bigben2k · 07-09-2002, 01:44 PM

I see your point, but then, it's not possible to get a perfect junction, so we have to have something, otherwise, all we have are microscopic edges touching the core.

Maybe the paste's only purpose is to get rid of the air!

airspirit · 07-09-2002, 01:54 PM

I've heard just that: conductive paste is exclusively used to fill in the air gaps. If you were going for surface area, use a heat sink: that's what they are is increased surface area for air cooling. I reckon the idea with a water block is to absorb and xfer heat to the water itself, not to the surrounding air. You'd get better heat conductivity from perfectly smooth surfaces on the die and block. Since that is impossible, a tiny amount of thermal paste fills the air gaps and you're gold. In the meantime, lapping the block and die (I refuse to touch a processor die to sandpaper ... I've burnt one too many processors as it is and I don't want to tempt fate any more, which is why I'm migrating to water cooling on all my rigs) as smooth as possible will give you the best results with your silver goop.

What you suggest would make a kickass heat sink surface, but not a good water block, methinks.

bigben2k · 07-09-2002, 01:59 PM

Well, I didn't want to get to talk about the other side of the block. Let's keep that for later, shall we?

I agree, sanding the core is a big no-no, but sanding the surface of the block that touches the core is what this is all about.

MeltMan · 07-09-2002, 02:09 PM

heh chipping the core is another big no no.

airspirit · 07-09-2002, 02:10 PM

Actually that brings up a good point: if the surfaces are rougher on the sides that aren't touching the cpu, it might serve as a mini-heat sink in and of itself. But then, we're talking about such small amounts of space, would it even make a difference?

The guy above nailed it on the head, though: Copper conducts so much better than the silver goop it is obscene. You want as much copper as possible touching the cpu, and that means as flat of a surface as possible.

bigben2k · 07-09-2002, 02:22 PM

Ok, so why do some people report better results lapping to 600-800, rather than polishing to a mirror? Random error?

airspirit · 07-09-2002, 02:32 PM

That's easy. They're using too much silver compound. What happens is with a rough surface and large amount of compound, there are the "points" of the rough surface touching the CPU. With the smooth surface and large amounts of compound, it forms a film over the entire surface. When you have a mirror polished block or heat sink surface, you should apply the compound as thin as humanly possible, almost to the point that you can't tell it's there save for the sparkle on the metal. What many people are doing in error is following the directions on the box that says "apply an amount 1/4 the size of a BB and spread it evenly over the area that will contact the CPU." On a mirror polished surface that is about 10x too much compound. On surfaces that are that clean you need to use an extremely small quantity, since the air pockets are minimized. I guarantee that people that are reporting better results with a rougher surface are using way too much silver compound.

*Whew!* That was just a little long winded of me ....

bigben2k · 07-09-2002, 02:50 PM

I understand.

Then it all comes down to the viscosity of the thermal paste. If it's too thick, then it won't fill an unpolished surface, it will leave air pockets.

If it's too thin, then it still works very well, because it does still get rid of the air pockets.

If the surface is sanded to 600-800, the paste will fill the gap, and if it's polished, then the gap is still filled.

The problem may come from an unusual process by which the surface is polished, where the surface is like a mirror, but microscopically, it's a weird kind of flat, and the paste doesn't work very well.

Does that make sense?

airspirit · 07-09-2002, 03:17 PM

EXACTLY! A thinner, more fluid compound would work better because there are fewer and smaller air gaps to fill. If you had a liquid compound you'd use almost nothing at all because it would practically soak into the cracks leaving the excess to run away. Same thing, though, you're basically using a smaller amount of compound due to more metal on die contact (which is ideal).

BTW, if anybody knows of any liquid thermal compound that sets over time (like super glue), let me know, cos I'll be first in line to buy it.

bigben2k · 07-09-2002, 03:25 PM

Well, it seems that stuff like Arctic Silver is fairly liquid, but settles over a few days, so it's liquid for the purposes of the application, then solid for long term use.

So... obviously, once it has settled, one would not want to move it at all.

Also, as I opened this thread, I pointed out the importance of sticking that paste in there, with some pressure, from all directions.

Some people have resorted to bypass the whole thermal paste issue, by using a waterblock that shoots water directly on the core. What's wrong with that?

airspirit · 07-09-2002, 05:16 PM

If you could get away with shooting water onto the core without giving all of the guts a bath, you have the perfect system. Really, all the block is there for is to transfer heat to the water. By bypassing the "cold plate" of the block, you are getting maximum heat xfer and you will get better cooling than any other pure watercooling system you can put together. Direct water contact on the die will give you maximum efficiency. I just don't think too many people have the balls to do it (I sure the hell don't).

bigben2k · 07-09-2002, 05:26 PM

Yeah... a block like that, ideally, would be clamped to the CPU, not just the mobo.

gmat · 07-09-2002, 05:29 PM

Hmm you should search on this very forum. Actually there are quite a few ppl who tried a 'custom' direct core cooling.
What they learned was, one can't fight fluid dynamics laws.
Any system 'shooting' water on the core wont be efficient because of two main reasons:
1 - the exchange surface area is *very* small. Way smaller than in a WB. And you want the water to stay "longer" in the wb ie you need a longer path to optimize transfer
2 - the core is square and flat. The flow just over the core will be quite laminar, ie not efficient, whatever the way you 'shoot' water at it.
Overall it wasnt *so* bad (because of direct heat transfer), but it didnt outperform a Maze2.

Now about lapping. The finer the finish, the finer the past should be.
Rough finish -> arctic silver. Thick, with big particles
Mirror finish -> no-name quasi-fluid white 'goopy' paste. So fluid it will squirt on all sides. Apply only a film under the wb where the core will sit.
Of course mirror finish is better for obvious reasons...

airspirit · 07-09-2002, 05:33 PM

You know, I've heard of people doing that too ... I've just never seen a picture or diagram of a block that allows flow on the die. I can't see how you could do that without a ton of silicone around the die ... and even then, how do you prevent the water from soaking through the surrounding card? I think that somebody that would go to that extreme has to be:

1) absolutely gonzo crazy
2) rich, and doesn't care about frying one of their 17 gaming machines, or
3) st00pid, and doesn't know that electronics and water don't mix

I think it would be easier to peltier cool the thing ... it would be more reliable and ... uh, less panic-inducing to do it that way. It took me over eight hours to fully pressure and seal check my rig, and I still pop open the cover before I turn it on to make sure there aren't any leaks. Direct cooling like that would make me neurotic.

gmat · 07-09-2002, 05:37 PM

Quote:

Originally posted by airspirit
I think that somebody that would go to that extreme has to be:

1) absolutely gonzo crazy
2) rich, and doesn't care about frying one of their 17 gaming machines, or
3) st00pid, and doesn't know that electronics and water don't mix

Like i said do a search here. They used an O-ring in a square path on the base of a 'shell' that sits on the ceramic. Pretty high-tech aint it ?
They just has to insulate AMD L-bridges.

BTW i fit in category 3 i guess. I'm watercooled. 1 is quite ok for me as well

bigben2k · 07-09-2002, 05:50 PM

Quote:

Originally posted by gmat
Now about lapping. The finer the finish, the finer the past should be.
Rough finish -> arctic silver. Thick, with big particles
Mirror finish -> no-name quasi-fluid white 'goopy' paste. So fluid it will squirt on all sides. Apply only a film under the wb where the core will sit.
Of course mirror finish is better for obvious reasons...

I'll do the search. I know that it's an available product in Europe, somewhere.

I see where you're getting at. It seems to make sense that a mirror finish is best, when you look at it all under a microscope. Why use AS when its not necessary?

So, which "no-name quasi-fluid white 'goopy' paste" do you use, gmat?

(Je tombe moi meme dans #1 et #3, mais definitivement pas #2!)

Fixittt · 07-09-2002, 08:11 PM

Chip from Lufken tech (SP) that I beleive has long sold out. He made a direct die cooing block, and tried pumping the hell out of it. From what I read, a few people lost there rigs due to it. And even more were against the idea. At least with copper, the block can absord and move the heat away from the core, allowing the increased surface area of the waterblock to preform better.

Also on one of the forums, I read an article, where a guy who works with lasers, ran a comarison test lapping different heatsinks to different finishes, with different amounts of goo. He was cooling some hot parts of the laser. He concluded that a wet (Using water or WD-40) as a lube, worked better for him then a highly lapped surface.

my 2 cents

edit:
Ok I have 2 more cents.

Seeing as the cores are not perfectly flat, you need to use some sort of goo. No matter what. Actually the pits in a core are quite big, so artic silver of some sort is recomended. So if you have a gritty substance (Artic silver) and tried putting a mirroed finish cover over it, wouldnt that (For argument sake) lift the mirrored surface off the core? So why not have 2 roughter surfaces so that the artic silver could effectivly fill both surfaces, bringing them closer. Also I think it should be said that before clamping anything down, the block should be rotated on the core to work the paste into the cores surface as well.

But then again I could be wrong.

WireX · 07-09-2002, 10:27 PM

www.cpuwatercool.com used to have it, before they went outa buisness... it was called the direct die injector cooling system (or a combination of those).

They also had one on a pelt, but the results were less than impressive.

I really looked like a swiftech waterblock (with the open inside) with the barbs at a slant, and a large o-ring.

A guy in town who works at my favorite computer store is working on direct die oil cooling. I will tell more once I get more details.

-WireX

koslov · 07-09-2002, 10:53 PM

I personally think as flat as possible is the best way to do it. I don't have any numbers or theories to back that up, but it just seems logical that the closer you can have the two surfaces, the better.

Also, you guys see this article at Spode's Abode about Direct Die Cooling? http://www.spodesabode.com/content/article/directdie1

Also here he compares it to a Maze. http://www.spodesabode.com/content/article/directdie2

Organized_Chaos · 07-09-2002, 11:15 PM

Blocks

What about directly cooling a TEC, does that work any better because of the large surface area?

gmat · 07-10-2002, 03:00 AM

Koslov: mmh nice performance. Better than the test units produced by Leufken.
Fixitt: thx for refreshing our memories. My point is, if you have a mirror surface, dont use arctic silver. For one the "performance" of AS is questionable at best, and it's way too thick.
BB2K: i use 3 different pastes:
- Arctic silver i had for free from DangerDen
- 'Arctic copper' i had for free from a company next door where they have an extruder (this thing *badly* needs cooling... they use buckets of thermal paste, and by buckets i mean it... so i got a big container of it hehe)
- 'no-name' white goo i had for free with each heatsink i bought , each computer i mounted, and watercooling HW i ordered.
I use Arctic silver / copper for the GPU and NB where chip surfaces are rough. And white goo on the CPU. AMD cores are *very* flat (at least, mine is).

Hallis · 07-10-2002, 06:13 AM

Bigben2k, you have any spare waterblocks lying around? youcould always try thr theory,, lapping to 600--800 and test for a few days,, then lap to mirror and test just the same,, find out which works best for your application.

bigben2k · 07-10-2002, 08:52 AM

Quote:

Originally posted by Organized_Chaos
Blocks

What about directly cooling a TEC, does that work any better because of the large surface area?

It can't. The pelts have to be clamped, so direct cooling is not possible, but I suppose someone could get really creative and find a way to do just that...

07-09-2002, 01:59 PM	#5
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Well, I didn't want to get to talk about the other side of the block. Let's keep that for later, shall we? I agree, sanding the core is a big no-no, but sanding the surface of the block that touches the core is what this is all about. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

07-09-2002, 02:09 PM	#6
MeltMan Cooling Savant Join Date: Aug 2001 Location: KS Posts: 374	heh chipping the core is another big no no. __________________ MeltMan Lurker Supreme!

07-09-2002, 02:22 PM	#8
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Ok, so why do some people report better results lapping to 600-800, rather than polishing to a mirror? Random error? __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

07-09-2002, 02:50 PM	#10
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	I understand. Then it all comes down to the viscosity of the thermal paste. If it's too thick, then it won't fill an unpolished surface, it will leave air pockets. If it's too thin, then it still works very well, because it does still get rid of the air pockets. If the surface is sanded to 600-800, the paste will fill the gap, and if it's polished, then the gap is still filled. The problem may come from an unusual process by which the surface is polished, where the surface is like a mirror, but microscopically, it's a weird kind of flat, and the paste doesn't work very well. Does that make sense? __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

07-09-2002, 03:25 PM	#12
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Well, it seems that stuff like Arctic Silver is fairly liquid, but settles over a few days, so it's liquid for the purposes of the application, then solid for long term use. So... obviously, once it has settled, one would not want to move it at all. Also, as I opened this thread, I pointed out the importance of sticking that paste in there, with some pressure, from all directions. Some people have resorted to bypass the whole thermal paste issue, by using a waterblock that shoots water directly on the core. What's wrong with that? __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101 Last edited by bigben2k; 07-09-2002 at 04:08 PM.

07-09-2002, 01:44 PM	#3
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	I see your point, but then, it's not possible to get a perfect junction, so we have to have something, otherwise, all we have are microscopic edges touching the core. Maybe the paste's only purpose is to get rid of the air! __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

07-09-2002, 01:54 PM	#4
airspirit Been /.'d... have you? Join Date: Jul 2002 Location: Moscow, ID Posts: 1,986	I've heard just that: conductive paste is exclusively used to fill in the air gaps. If you were going for surface area, use a heat sink: that's what they are is increased surface area for air cooling. I reckon the idea with a water block is to absorb and xfer heat to the water itself, not to the surrounding air. You'd get better heat conductivity from perfectly smooth surfaces on the die and block. Since that is impossible, a tiny amount of thermal paste fills the air gaps and you're gold. In the meantime, lapping the block and die (I refuse to touch a processor die to sandpaper ... I've burnt one too many processors as it is and I don't want to tempt fate any more, which is why I'm migrating to water cooling on all my rigs) as smooth as possible will give you the best results with your silver goop. What you suggest would make a kickass heat sink surface, but not a good water block, methinks.

07-09-2002, 02:10 PM	#7
airspirit Been /.'d... have you? Join Date: Jul 2002 Location: Moscow, ID Posts: 1,986	Actually that brings up a good point: if the surfaces are rougher on the sides that aren't touching the cpu, it might serve as a mini-heat sink in and of itself. But then, we're talking about such small amounts of space, would it even make a difference? The guy above nailed it on the head, though: Copper conducts so much better than the silver goop it is obscene. You want as much copper as possible touching the cpu, and that means as flat of a surface as possible.

07-09-2002, 02:32 PM	#9
airspirit Been /.'d... have you? Join Date: Jul 2002 Location: Moscow, ID Posts: 1,986	That's easy. They're using too much silver compound. What happens is with a rough surface and large amount of compound, there are the "points" of the rough surface touching the CPU. With the smooth surface and large amounts of compound, it forms a film over the entire surface. When you have a mirror polished block or heat sink surface, you should apply the compound as thin as humanly possible, almost to the point that you can't tell it's there save for the sparkle on the metal. What many people are doing in error is following the directions on the box that says "apply an amount 1/4 the size of a BB and spread it evenly over the area that will contact the CPU." On a mirror polished surface that is about 10x too much compound. On surfaces that are that clean you need to use an extremely small quantity, since the air pockets are minimized. I guarantee that people that are reporting better results with a rougher surface are using way too much silver compound. Whew! That was just a little long winded of me ....

07-09-2002, 03:17 PM	#11
airspirit Been /.'d... have you? Join Date: Jul 2002 Location: Moscow, ID Posts: 1,986	EXACTLY! A thinner, more fluid compound would work better because there are fewer and smaller air gaps to fill. If you had a liquid compound you'd use almost nothing at all because it would practically soak into the cracks leaving the excess to run away. Same thing, though, you're basically using a smaller amount of compound due to more metal on die contact (which is ideal). BTW, if anybody knows of any liquid thermal compound that sets over time (like super glue), let me know, cos I'll be first in line to buy it.

07-09-2002, 05:16 PM	#13
airspirit Been /.'d... have you? Join Date: Jul 2002 Location: Moscow, ID Posts: 1,986	If you could get away with shooting water onto the core without giving all of the guts a bath, you have the perfect system. Really, all the block is there for is to transfer heat to the water. By bypassing the "cold plate" of the block, you are getting maximum heat xfer and you will get better cooling than any other pure watercooling system you can put together. Direct water contact on the die will give you maximum efficiency. I just don't think too many people have the balls to do it (I sure the hell don't).

07-09-2002, 05:26 PM	#14
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Yeah... a block like that, ideally, would be clamped to the CPU, not just the mobo. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

07-09-2002, 05:29 PM	#15
gmat Thermophile Join Date: Mar 2001 Location: France Posts: 1,221	Hmm you should search on this very forum. Actually there are quite a few ppl who tried a 'custom' direct core cooling. What they learned was, one can't fight fluid dynamics laws. Any system 'shooting' water on the core wont be efficient because of two main reasons: 1 - the exchange surface area is very small. Way smaller than in a WB. And you want the water to stay "longer" in the wb ie you need a longer path to optimize transfer 2 - the core is square and flat. The flow just over the core will be quite laminar, ie not efficient, whatever the way you 'shoot' water at it. Overall it wasnt so bad (because of direct heat transfer), but it didnt outperform a Maze2. Now about lapping. The finer the finish, the finer the past should be. Rough finish -> arctic silver. Thick, with big particles Mirror finish -> no-name quasi-fluid white 'goopy' paste. So fluid it will squirt on all sides. Apply only a film under the wb where the core will sit. Of course mirror finish is better for obvious reasons...

07-09-2002, 05:33 PM	#16
airspirit Been /.'d... have you? Join Date: Jul 2002 Location: Moscow, ID Posts: 1,986	You know, I've heard of people doing that too ... I've just never seen a picture or diagram of a block that allows flow on the die. I can't see how you could do that without a ton of silicone around the die ... and even then, how do you prevent the water from soaking through the surrounding card? I think that somebody that would go to that extreme has to be: 1) absolutely gonzo crazy 2) rich, and doesn't care about frying one of their 17 gaming machines, or 3) st00pid, and doesn't know that electronics and water don't mix I think it would be easier to peltier cool the thing ... it would be more reliable and ... uh, less panic-inducing to do it that way. It took me over eight hours to fully pressure and seal check my rig, and I still pop open the cover before I turn it on to make sure there aren't any leaks. Direct cooling like that would make me neurotic.

07-09-2002, 08:11 PM	#19
Fixittt CNC Beyatch Join Date: Aug 2001 Location: Tulsa Spell it backwards Posts: 721	Chip from Lufken tech (SP) that I beleive has long sold out. He made a direct die cooing block, and tried pumping the hell out of it. From what I read, a few people lost there rigs due to it. And even more were against the idea. At least with copper, the block can absord and move the heat away from the core, allowing the increased surface area of the waterblock to preform better. Also on one of the forums, I read an article, where a guy who works with lasers, ran a comarison test lapping different heatsinks to different finishes, with different amounts of goo. He was cooling some hot parts of the laser. He concluded that a wet (Using water or WD-40) as a lube, worked better for him then a highly lapped surface. my 2 cents edit: Ok I have 2 more cents. Seeing as the cores are not perfectly flat, you need to use some sort of goo. No matter what. Actually the pits in a core are quite big, so artic silver of some sort is recomended. So if you have a gritty substance (Artic silver) and tried putting a mirroed finish cover over it, wouldnt that (For argument sake) lift the mirrored surface off the core? So why not have 2 roughter surfaces so that the artic silver could effectivly fill both surfaces, bringing them closer. Also I think it should be said that before clamping anything down, the block should be rotated on the core to work the paste into the cores surface as well. But then again I could be wrong. __________________ Creator of the Spir@l Block Longest post ever http://forums.procooling.com/vbb/showthread.php?s=&postid=43808#post43808 Last edited by Fixittt; 07-09-2002 at 08:15 PM.

07-09-2002, 10:27 PM	#20
WireX Cooling Neophyte Join Date: Apr 2002 Location: Winnipeg Manitoba Canada Posts: 68	www.cpuwatercool.com used to have it, before they went outa buisness... it was called the direct die injector cooling system (or a combination of those). They also had one on a pelt, but the results were less than impressive. I really looked like a swiftech waterblock (with the open inside) with the barbs at a slant, and a large o-ring. A guy in town who works at my favorite computer store is working on direct die oil cooling. I will tell more once I get more details. -WireX __________________ I am a Canadian Cooler and proud of it If at first you don't succeed, Skydiving isn't for you - Me The world is run by the internet The internet is run by computers Nerds run those computers I'm one of those nerds

07-09-2002, 10:53 PM	#21
koslov Cooling Savant Join Date: Jul 2002 Location: Posts: 264	I personally think as flat as possible is the best way to do it. I don't have any numbers or theories to back that up, but it just seems logical that the closer you can have the two surfaces, the better. Also, you guys see this article at Spode's Abode about Direct Die Cooling? http://www.spodesabode.com/content/article/directdie1 Also here he compares it to a Maze. http://www.spodesabode.com/content/article/directdie2

07-09-2002, 11:15 PM	#22
Organized_Chaos Cooling Neophyte Join Date: Jan 2002 Posts: 3	Blocks What about directly cooling a TEC, does that work any better because of the large surface area?

07-10-2002, 03:00 AM	#23
gmat Thermophile Join Date: Mar 2001 Location: France Posts: 1,221	Koslov: mmh nice performance. Better than the test units produced by Leufken. Fixitt: thx for refreshing our memories. My point is, if you have a mirror surface, dont use arctic silver. For one the "performance" of AS is questionable at best, and it's way too thick. BB2K: i use 3 different pastes: - Arctic silver i had for free from DangerDen - 'Arctic copper' i had for free from a company next door where they have an extruder (this thing badly needs cooling... they use buckets of thermal paste, and by buckets i mean it... so i got a big container of it hehe) - 'no-name' white goo i had for free with each heatsink i bought , each computer i mounted, and watercooling HW i ordered. I use Arctic silver / copper for the GPU and NB where chip surfaces are rough. And white goo on the CPU. AMD cores are very flat (at least, mine is).

07-10-2002, 06:13 AM	#24
Hallis Cooling Savant Join Date: Oct 2001 Location: Dallas, Tx Posts: 469	Bigben2k, you have any spare waterblocks lying around? youcould always try thr theory,, lapping to 600--800 and test for a few days,, then lap to mirror and test just the same,, find out which works best for your application.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)