[Cathar]

Okay, to answer a few question raised here about my block.

1) It was produced on a Pantograph mill, not a CNC mill. The production blocks will be done on a very high speed (RPM) CNC milling machine.

2) I really can't comment as to the curvature of the fins where they meet the base. The cutting tools is (of course) absolutely tiny and Dave is probably quite right about a slight curvature at the end of the mill bit. Given that the mill bit at this point has less strength bracing it, this would naturally be the case.

3) I did consider EDM but passed it over as a production method. If I were to consider EDM, then I certainly wouldn't use a 1mm wide channel given the machining flexibility that EDM offers. I'd go for something like 0.3mm wide fins and 0.25mm wide channels and a lower fin height (~3mm). I know I originally said 0.9:0.6, but upon further investigation of EDM I'd go for a even finer set of fins/channels.

4) The AMD die is actually 1.25mm off centre horizontally, but is centered vertically when looking at the socket with the latch arm to one side. I settled on making my inlet "nozzles" being 1mm off-centre.

5) My block was designed and drawn up using MS-Paint and I printed out the plans and presented them to the machinists who said "No worries - these designs look better than what we usually get given to us".

6) The "optimal" copper:channel ratio is anywhere between 1.5:1 to 0.8:1 from my research, with the actual ratio used dictating fin height. By optimal I mean that for a set width of channels and fins, a higher ratio means less channels but higher water-velocity to compensate, and vice-versa. It's a bit like plotting the bottom of a parabolic curve where the curve is basically flat for all intents and purposes within a certain range.

I considered radial designs but as Dave has mentioned, this does allow for a fair amount of "white space" above the critical central die region that is hard to address, not to mention the "as observed" manufacturing complexity. There was also the issue of how to bring down the nozzle inlet size to boost water velocity and still have that rapid impinging water flow distributed over the entire CPU die area.

Still, I admire the design and I hope you can get it made.

I suggest looking at key-cutter style cutting blades, which are a small radius allowing the blade to curve in and out without cutting up the edges of your block and can be resharpened easily and allow for fairly accurate depth. This was one manufacturing method that I considered. Blade widths down to 0.8mm are fairly easy to get a hold of.

Good luck!