Which of these two base-plate finishes would you rather have?

[greenman100]

One thing that does bear investigation as well, and I don't know the answer on this one.

Take two small pieces of smooth & flat glass, such as what would be used to make a mirror.

Add a drop of water to one (perhaps with a non-particle based dye to assist), and put the two pieces of glass together.

Do the two pieces of glass even touch at all? What would be the effect of this on the efficacy of the CPU thermal contact (or lack thereof)?

define "touch"

the nucleus of an atom never touches another. or are you asking if the water will be squeezed out totally?

 

[aldamon]

I hope you're kidding...

Of course.

 

[Graystar]

Take any object lapped to 400-grit or finer, hold it up at a sharp enough angle (45-60° from the perpendicular is usually sufficient) and you'll get a near mirror-like reflectivity.

There ya go...mystery solved.

 

[Cathar]

or are you asking if the water will be squeezed out totally?

Yes.

 

[Sean Lindstrom]

define "touch"

What an engineer would call "press fit" - not "loose fit" or "interference fit" - I assume.

A water barrier between smooth, matching surfaces maintains a snug, but loose, fit. That might explain some observations that plates not finished very finely conducted better than the mirror-lapped ones of equal flatness.

Personally, I like to put the final cut on blocks by twisting, forming concentric rings. I like to think this helps against pump-out from expansion cycling, while leaving the faint ruts that improve or at least don't hamper performance.

 

[Styyn]

One thing that does bear investigation as well, and I don't know the answer on this one.

Take two small pieces of smooth & flat glass, such as what would be used to make a mirror.

Add a drop of water to one (perhaps with a non-particle based dye to assist), and put the two pieces of glass together.

Do the two pieces of glass even touch at all? What would be the effect of this on the efficacy of the CPU thermal contact (or lack thereof)?

What first comes to my mind in this kind of situation involving extreme "flatness" is the molecular structure of the surfaces. At some point the two surfaces to be mated--of identical make-up, purity, flatness and finish--reach the point where the chains of molecules that compose each block will line up such that each mating surface presents itself in a "final" arrangement. Best way I can describe this is with the pictures that follow:

Pre-"flattening":




Maximized "flattening":




Diamond:




Given that the actual chains and molecules themselves are in 3d, that these "chains" are in no way accurate models and the difference in scale between the orange "copper" and the grey "diamond", this is a far from perfect representation--I'm only trying to serve an example. In the "copper" pictures imagine looking at the surface "side-on" with the right side being the mating surface. Initially you have a hodgepodge mixture, but after the "flattening" you've pushed them into a maximized grouping. An analogy that comes to mind would be having a 500 piece puzzle where all the pieces are identical in shape. Initially you have them in a pile, but with a little work you can snap them all into an ideal position.

In the "diamond" picture imagine you're looking at it from the top-down. Obviously you'd see a smaller hex shape in the center of each hexagon if it were in 3d, but you get the idea. To me, this also explains why diamond is an ideal conductor--contact area is nearly perfect from one carbon chain to another.

Moving on... Now imagine two of the idealized copper surfaces (or glass if I'm to stick with Cathar's original question). They'd look something like this:




To finally get to my hypothesis for Cathar's question, whether or not the water molecules would be squeezed out entirely depends on the size and shape of the molecules relative to the gaps created by the mating surface material's final or "ideal" structure. Apply pressure to the two surfaces and the thermal interface--water in this case--will be forced into the gaps. A problem now arises, however, in the way the maximized surface "texture" of the gaps lines up between the two surfaces. If they line up as in the picture above, you have a fair sized gap for the molecule(s) to fall in to. Let's assume one molecule of water just fits into the gap. Now imagine if the surface on the right were shifted up so that the two small triangles didn't meet to create the larger triangle. In that situation the water molecule would no longer fit, and all of the water would be forced out the sides. Perhaps this is the kind of situation that accounts for some of the variance people see when re-mounting waterblocks and heatsinks?

Now for some conjecture of my own . Imagine a waterblock and die mating in an idealized fashion like the above picture, but where the edges of the surfaces are magically sealed, preventing any thermal interface material from escaping--essentially creating a fixed volume. Now imagine that there's exactly the right volume of TIM (let's say water is the TIM for now) for a few molecules to fall into each gap. Further imagine that the die is capable of withstanding a few thousand PSI without cracking or warping. If you were to apply enough pressure for the water in the gaps to vaporize, would having it in this state affect the heat transfer from die to waterblock? Taking it further, if you could rapidly apply and remove pressure, causing the water to vaporize and then recondense, how would that affect things?

Anyway, everything I've said here is conjecture based only on some highschool chem, and may be grossly inaccurate, if not flat-out wrong.

My apologies for the length, as well. If this is a total thread-crap, let me know and I'll take it down .

 

[ghettocomp]

My bet is for the one on the right, the left one is all shine and flash, but, I can see that it is not very flat, just smooth.

 

[Sean Lindstrom]

If you flatten down to the molecular level, I think you'll have problems. Not everything between the two planes will be squeezed out. A bit will get stuck, like a hair in a book. A tiny invisible mote of dust will sit between the perfect surfaces and perhaps cause worse contact than if there were porous surfaces to absorb such dust and oil and so forth.

But then the metal does deform and bite somewhat with a little pressure, as does the die. You can see a mark after mounting even when no interface was used.

*Throws up hands*

 

[feyd83]

Just read Xbitlabs' article on the Reserator, they're stressing flatness over shinyness.

The sole of the water block is not just smooth, which is easily achieved with chemical polishing and not necessarily an indication of high quality, but also almost perfectly flat, which is more important than nice mirror reflections on the smooth surface.

 

[9mmCensor]

Just read Xbitlabs' article on the Reserator, they're stressing flatness over shinyness.

Reviews like that are rare.

 

[ZachM]

Uhh, 9mm, this thread is 4 months old. Your reply wasn't worth bumping it.

I'm torn between thread necro, or thread resurrection batman.

*Edit*
Thread necro it is.

 

[alinosa]

dang and i was ready with my snappy "the flatter one of course"
oh well.

 

[9mmCensor]

Uhh, 9mm, this thread is 4 months old. Your reply wasn't worth bumping it.

I'm torn between thread necro, or thread resurrection batman.

I know, but I missed it the first time.

I like both, toss a coin.

 

[nweibley]

Considering the laws of physics dictate that you could, in theory, float an oil tanker on a gallon of water if you created a container that contoured perfectly to the shape of the boat.... I would say, with two extremely flat pieces of glass that water will seperate the two untill a rather substatial force is applied. None the less... without some sort of resistive force at the edges to counter the water being force out from the center and away, I do think it is possible to completely evacuate the water from between the two pieces of glass, however... the glass would probably break before that point is met.

Then again, the is the properties of cohesion and adhesion to account for, but I still posit the same theory.

(Yes it can be fun to dig up dead threads)

 

[Kevin007]

the right one looks very scratched..I'm confused lol

 

[Sneaky]

why'd you have to go digging this up

 

[9mmCensor]

why'd you have to go digging this up

Because people need to see it

 

[g0dM@n]

I didn't really read all the posts, but I can see that the left picture is a little wiggly with the reflection of the table. I never thought that a mirror finish really meant flat... although a flat finish is usually more reflective than non-flat. Curved, on the other hand CAN be shiny. A curved surface that is smooth can easily be shiny, such as a metal mixing cup that a bartender uses. Unfortunately, it is not flat. Therefore, "smooth" and "flat" are entirely different things when it comes to our type of cooling.

 

[crimedog]

why revive it if you're not gonna read it?

 

[SpaceChicken]

This is probably a stupid question but how do you know when something is flat? Cathar says that the poormans flatness test doesn't show how flat something is, so if I was to hand lap my wb how would I know when to stop without a very expensive flatness testing machine?