DIY copper sinks from scratch - "Alpha2" prototype finished ...
 

Hi, just now joined the forum.
I am a Machinist/Metalsmith, & at this time I am in the process of designing my own
heatsinks from scratch. Even smelted the Copper myself.

http://server2.uploadit.org/files/grxgrx-cp1.jpg

--------------------------------------------------------------------------

Have a couple design ideas floating around in my head, but it would be great to
get some thoughts from others to help spark new ideas.

How about some custom made heatsink images floating around the net?

it may be easier to cast a solid rectangular prism then use a slittling saw or soemthing to cut out some fins.

First two pics remind me of my uncle's gold workshop. Man is it fun to play with fire.

Playing with fire and melting things is indeed fun. :D
We work with Gold (18k & white) mostly, sometimes Platinum. And we make our own
Mokume Gane as well. Pure Cooper has been nice, great to have a big hunk to work with for a change.

ThanX for the suggestions DryFire .... keep 'em coming. :)

You could also go the swiftech way and screw in aluminum (or perhaps copper) screws.

This is to be a passive cooler correct?

perhaps if you have a small cylinder cast you could make a bunch of small cylinders and then mount them in styrofoam or something then pour the bas on it.

I still think a slitted or cross slitted prism would be easier (like the zalman passive nb cooler).

I was wondering... How pure is the copper used for copper wires? Maybe it could be used for the pins on a pin matrix wb (MCW6000-style), and you could cast the base as DryFire suggested above...
Anyway, there was a thread about melting copper for wbs a while ago, don't know if it's still active...

Yeah, it will probably be mostly a passive NB or SB sink DryFire. Hope to have the side panel
fan blowing onto it. I am also working on a set of small gfx card memory sinks too.

By law Copper pipe is made from 110 Copper I think. 9= 99.9%)
Not sure about wire. I know the more expensive audio/video cables are advertised as "low oxygen".
Making wire in the shop is quite easy as we have rollers just for that purpose.

Meaning it's been kept melted for a while in a sealed vessel with a charcoal cover on the melt, which draws the oxygen out of the metal. See here. I would imagine you could do the same thing in an open vessel if you used one of the welding "shield" gasses.
Any idea if low-oxygen copper has higher heat conductivity than the "standard stuff"? I would guess not just based on the fast that it hasn't become a marketing thing for the heatsink manufacturers.
Note I am not a metallurgist or even metal worker, other than annealing head gaskets and brazing frame tubes in a motorcycle shop a very long time ago...

During the melting, one can easily create a "reducing flame" to help reduce the oxygen in the
metal. That, along with Borax flux, is how I keep the Copper from oxidizing
while making ingots in the shop.

Your mention of Copper wires has sparked an idea. :D
How about a solid Copper heatsink like a little porcupine? Should be fairly
easy to imbed some wire lengths into a molten ingot. Would require some
experimentation with the firecoat & heatsinking the wires I am sure. But if successful the
Copper spines should significantly increase the surface area to the air. And if all
works out should net a solid Copper piece instead of soldering on fins.

In the early '40s there was a British bike manufacturer (I think it was AJS) that had a head with cooling pins coming out of it like that - and it was called the porcupine. :)

the MCW6000 bp is forged, so the monolithic base/pins is already being done
casting copper is a dead end street, but fun melting stuff I quite agree

Yeah, it's mostly for the fun. :cool:
Don't think I will be able to come up with any radical new designs in my small shop
that the big plants haven't already explored.
.
On a lighter note ........ I made a set of mosfet sinks last night (fins not pins).
Best to start small, especially when you are working from ingots poured yourself.
Will share a couple images later once the final "bling-bling" is done. ;)

I'd say it's almost soley for the fun as buy a nb cooler would probably be cheaper, at least that'd be the case for me if i had a workshop or enough space for some larger power tools.

Introducing ... the Cuprinnock "Alpha 2"®

These prototypes were designed to clear the caps & toroidal coil around the mosfets, that is why they have open sides.
This design could easily be upscaled for use as Gfx memory or Northbridge
sinks. And could also be changed to accomodate more fins as well. Another smaller fin
piece could even be placed in the middle area. Which is the next design most likely. ;)

http://server2.uploadit.org/files/grxgrx-alpha2c.jpg

In case you think that's solder joining the two pieces think again. That is pure Silver.
Took a bit of patience, but I was able to pull it off quite successfully. The
VERY small difference in melting points between Copper & Silver made this
a fun challenge. :D

Among the premise behind this design are .........

1] very simple - both to produce and to upscale for cooling of larger chips

2] effective - to maximize the contact area from the base plate to the fin piece

3] durable - had to take high torch heat because of the use of pure Silver as a joining medium instead of solder

============================================

Took a small amount of experimentation with torch techniques to get the
Silver to flow evenly without heating the Copper too much. Only 220 degrees (F)
between the two melting points.
Must admit however, Copper has a wider MR (Melting Range) than Silver, which was to our
advantage here.
After this the pieces were cleaned in acid to remove all traces of the firecoat
and oxidation. Clean surfaces are extremely important! Even a minor fingerprint
can keep the two metals from joining properly.

interesting, i must say first time I've seen that design.

Yeah, it's far from the normal type sink you see in PC's. So ... do you like it?

Mind you, those are just early prototypes to iron out some design & production issues.
By far the most time was consumed by working the small billets into usable
sheets. Just found out my "toolgirl" friend has access to 110 Copper flat stock at cost where she works.
Which should make the production process a lot easier. No need to go from
molten puddles. ;)

The next generation of sinks will be more elaborate. And no solder either.
I always knew solder was a poor thermal conductor. But this info source makes me think
it's much worse than I thought.
http://www.engineersedge.com/properties_of_metals.htm

Hard solder only 94 W/c-m?? :eek:
And I highly doubt makers like Thermalright use "hard solder". Most likely some
kind of "easy9soft) solder" with a low melting point.
Using pure Silver as the joining medium should me much, MUCH better.

I like the idea of a small heatsink that is easily bendable for those types of applications.

However there is a lot of wasted space in teh center, but you said you were planning to add some more to it.

Looking forward to it as always.

My two main objectives were to have the least number of production steps,
and a design that is easy to adapt. Also had to use pure Silver.

Not sure how accurate it is, but the webpage below lists the thermal properties of hard
and soft solder. All of which have abysmial numbers.

http://www.engineersedge.com/properties_of_metals.htm

Always knew solder was inferior, but that is much more than I expected. Makes one
wonder how efficient the bond on pre-made offerings like Thermalright etc are at
thermal transfer.

Thanks for the link - very nice!

Glad you like it. I'm wondering how accurate the values for solder are though. Those
seem pretty low to me. Can't seem to find any other resources for a confirmation
as yet.

edit :.......

ah ... here we go ... http://www.matweb.com/search/GetSubcat.asp
properties of 178 different solders. Basically solders suck far as thermal
properties go. Now what I need to do is build another exact sink with solder
instead and compare it with the Ag bonded ones. That means I need a test bed too. ... sigh. :shrug:

Maybe it's time to play with lost-wax? I know Bill says casting is a waste of time, but I wonder if you could make up some of the deficiencies by shot peening or bead blasting ...

You seem to be pretty good at metalworking. Think you could forge weld on some copper fins?

Your design seems like it might work for that.

bobkoure - -

Yeah, lost wax might be cool. Don't have easy access to any of that equip though. Mainly
a "forging shop". And cast isn't as dense as forged (cold rolled stock). Shot peening would
take care of the surface, but the internal will still be the same.

Matter of fact, I attended an MJSA jeweler's expo in NYC earlier this week. One of the exibitors
there had some nice CNC machines set up just for doing precious metals. They were drilling
out 2 inch round 18k Gold stock to make rings.
What I am getting at is their display case where they had two rings displayed.
One cast, the other machined from round stock. The cast had significantly
lower density. And as a result, the tensile strength & hardness was lower as well.
Likely thermal efficiency follows.
All this I knew years ago. Still, the display really drove the point home.

Ls7corvete - -

Forge welding is indeed an option. Have done it with Gold & Silver many times. The problem
with Copper is that it's so reactive though. Hard to keep it from forming an oxide layer. Which would
inhibit the merging.
And forge welding takes a fair amount of work compared to what was involved
with the 'Alpha2' above^. If I use pre-made flat bar instead of making the sheet from
scratch, that would eliminate over 75% of the work. Maybe more.

I think my next step is to do another prototype with more center fins. The 'Alpha2'
was mostly an exercise to see how the techniques & ideas applied. Now
it's time to step it up a notch ...... "Alpha3" here we come! :D

What demand you think there would be for high end, hand made heatsinks? (pricing?)

not a lot unlessyou can come up with soething amazing.

I think that you've got an opportunity to make blocks that are artistically more attractive; something which is rarely done, even to this day. As long as it's functional, and has a decent performance, I don't believe that the price is too relevant; it's art... ;)