could you use a laser to lap a heatsink?

[Crash893]

i heard on the news that they devloped this machine to laser glaze railroad tracks to make them ultra smooth to reduce drag on rail cars


i was wondering if you could in theory flash melt the to 1/10000 of an inch of the hsf and have it cool smoother than it was?

[Crash893]

http://www.techtransfer.anl.gov/highlights/9-2/laser-glaze.html

[blackhabit]

interesting but who has a lazer?

[Crash893]

im just wondering if it could be done

[Since87]

The high thermal conductivity of copper makes it difficult to heat a small portion of it to melting. Steel has much lower thermal conductivity, making this much more practical.

Lasers are commonly used for cutting sheetmetal like aluminum, but only very thin copper can be cut this way. because of how quickly the heat spreads.

[RadRacer]

What about casting? I read an article online about using a microwave to melt metals (its no joke, it can be done, but that doesnt mean you should go sticking forks in your microwaves just grins). If you were to melt copper in an open top crucible (say 1" diameter) you should be able to get a perfectly flat surface. Oh well, just a thought.

I don't think lasers would work too well. Copper is a very soft metal and would probably deform under the heat because the heat spreads out so quickly

[cV]

You could do it with quick successive passes and some serious heatsinking on the copper, but making it so smooth would be pointless because A) there would be no unevenness to grip the CPU and facilitate contact and B) tarnish would more adversely affect performance.

C) It would be way too costly even for HS corporations to implement on a mass scale.

[Crash893]

i dont think it would cost that much

you would just need a few highpowered lasers

if they can do it to miles and miles of railroad track then it can tcost that much

im not entirely sure what you mean about "1) there would be no unevenness to grip the CPU and facilitate contact"

you want cpus and hsf to be as flat as possible so tehre is more surcface to surface contact to excange the heat

[SpliT71]

that's a good idea but would cost a fortune for a laser that could hurt someone also...

[bigben2k]

As Since87 pointed out, the use of lasers is limited. In the kind of power range that we're talking about, I would have to assume that this is already a pulsed laser, which would have been the key to using it with Copper. A cooling solution would be required, but it would have to be liquid, and not interfere with the laser. I just don't see it happening.

BTW, the article linked is about resurfacing a steel beam, not cutting it.

As for casting copper, it is quite possible. It has even been suggested to use a steel mold! Not a DIY project.

[Crash893]

im not talking about cutting it

im talking about resurfaceing it

lay the hs on its back

have a laser zap it for like a millionth of a second

it melts the top 1/100 of an inch of the copper and it cools flat

[bafbrian]

Sounds like a plausible idea, but like what is further up, who has a LASER you can borrow ?

Besides, will we never know till one of us tries it out.

I think it is genius, pure genius.

LOL

[Crash893]

well if i get my hands on a high power laser i might get around to doing this allthough i can think of like 1000000 things ill shot it at first

[Since87]

Originally posted by Crash893

have a laser zap it for like a millionth of a second

it melts the top 1/100 of an inch of the copper and it cools flat

It might 'cool smooth', but what makes you think it would 'cool flat'?

[TollPhree]

hmm =P What about water cut.. =P serisly I saw they cut steel with water howz that work?

[RadRacer]

I'd like to know where you saw that. I don't think that would work at all. Are you sure they simply weren't cooling the steel with water?

[GigaForce310]

You can cut steel with water. It takes very high pressure [thousands of PSI] going thorugh a tiny needle. It puts so much force on a concentrated area that it blasts a very fine section of steel out. The problem is that it only works with relitivly thin metal. Something along the lines of 1/2"-1/4" depending on the hardness of the steel.

This won't give a perfectly flat surface any more than lapping will. I think the best way to cool a processor is to take a heatsink or waterblock and make the processor right on it. Lapping a heatsink with 600+ grit sandpaper is just about as good as it gets.

[donny_paycheck]

Copper can be sprayed with an anti reflective coating to reduce its albedo and increase its fusion. An AR coat would make this MUCH easier, as copper is much more difficult to process using laser light without an AR coat. This would take care of your heat dissipation problem in the instant that would be necessary for surface refinishing.

[Diggrr]

It probably could be done, as the laser would heat it much faster than any metal could dissipate it.
But my question is, why? As I understand it from BillA's flatness tests, the cpu isn't that flat, so it would be a superfluous step.

One thing I'd like to see tried is silicon vapor deposited onto a 2" square heatsink (or waterblock, better still), and the circuitry etched onto that. Then add a substrate ring and contacts...violla! No more tiny hot die, and no more frail surfaces, or ineficient heatsinking interfaces.
This would also get rid of the need to make MONDO expensive silicon slabs. Could save a bundle of money (both theirs and ours).

Just my idea.

[madcow235]

If it gets to hot just keep the thing submerged in LN2 and there will be no problem the laser will heat the metal it touches but nothing else will warm up. Now i must find a laser

[Lt. Max]

diggr i think why the manufacturers are not doing what you suggested, because it doesnt help keep costs down, the manufacturers try to put as many cores per waffer as they can and thats why they wanna keep it tiny tiny. and afterwards the cores are put onto the chips i guess.. so i think your idea isnt really as efficient as it sounds. although you could send them an email with the idea.. sorry we drifiting a bit off subject here, but on the heat conductivity point i guess your idea would be the best, the manufacturers just dont support our cause :)

max

[Daemonfly]

Well, with the way the current CPUs are heading, bonding that little silicon chip right to a nice heatsink might be a good idea - i.e. heatsink built right into the chip itself.


Kinda makes after-market heatsinks obsolete though...

[Numbstruck]

My question would be.......

would this(laser smoothed) actually give enough noticable performance to make it worth while?

[Daemonfly]

Most likely not ;)

Unless you're a buisiness and can easily sell "laser-polished" heatsinks to gullible cooling newbies.

[KEBunny]

hello thermaltake....

[futura2001]

First off, define flat, is flat:
all molecules on a surface aligned to a certain variance of height
or
wow... I shore do look purdy in that mirrored reflection
Now if it is the former, then you are talking about overkill. To have something that flat would not see a noticeable difference in temperatures. Oh sure, there would be a difference, but we are talking about tenths or even hundredths of a degree. That will do nothing in the real world. At that price it is not worth it. You could probably manufacture an slk900 out of synthetic diamond for that kind of money.
If it is the latter, just buff the thing with a drill press. Eventually you will get one smooth surface...

[Xtc4u]

after reading the article it could be possible but why would you spend the money on that? If it was economical sure but until the common joe can buy lasrs that powerful for decent prices forget it.

[cV]

Originally posted by Crash893
i dont think it would cost that much

you would just need a few highpowered lasers

if they can do it to miles and miles of railroad track then it can tcost that much

im not entirely sure what you mean about "1) there would be no unevenness to grip the CPU and facilitate contact"

you want cpus and hsf to be as flat as possible so tehre is more surcface to surface contact to excange the heat

It has been shown that a mirror finish impedes thermal conductivity, why is not known but it is theorized that there is less surface area for thermal compound to "grip" to on a mirror surface :p ;)

[Crash893]

then why do people get better results with lapped heatsinks?

[cV]

Lapping a heatsink to get a near mirror finish surface increases performance, but not a 100% reflection.

e.g. the reason why swiftech's bases are now duller than they used to be...