Homemade Waterblock Questions

[Craptacualr]

Hey all, longtime reader and WCer, but first time poster here with a few questions on homemade WBs. I've been watercooling for the past 4 years with various retail and custom components, culminating with my current setup:

Maze 3
Dtek chipset cooler
DangerDen GeForce 4 cooler
Danner Mag5 pump
Oversized Cessna heatercore (12x5x2" area)
T-Line for filling/bleeding
All buttoned up in a standard Chieftec case (old alienware style)

This setup has suited my needs for respectable OCing on a near-silent rig for about a year now but I'm on the verge of a major upgrade from the venerable Socket A to a new 939 rig and of course my old maze 3 and perhaps vid card cooler dont have the correct mounting. dangerden of course doesnt make mounting brackets for a maze3 - 939 setup, so I've been examining my options on a new WB.

In the past I never had access to or proficiency with a mill, but over the last year, we've added a manual mill to our machine shop at my place of employment and i've become relatively profficient doing simple milling with it. Therefore I've been examining the idea of making my own blocks to fit my new 939 rig.

I've been looking at the waterblock gallery sticky for a few design ideas here and have come up with some questions. First one is what kind of minimal base thickness should one have on an all-copper base to prevent warping around the edges of the core? I'd imagine with the new heat spreaders you can go quite thin on the lower base as the force of mounting is spread across a larger area.

Seond, about what channel depth do you guys use typically? 1/4" or 1/2", or something else entirely?

And third and finally, those of you that use O-ring seals, do you just use a standard circular ring stretched to fit? And if so or not, what kind of vertical tolerance do you leave between the lip surrounding the ring and the top of it for compression and a good seal?

I'm leaning towards designing something similar to the WW but not as complicated or anything as jet impingement. My sole reason for wanting to do a WW clone is that its a relatively simple design to mill and something that can be done without the benefit of CNC.

I'm still toying with design ideas and my own proficiency with the mill and the time i'm willing to spend somewhat limits me, but I should be lookin at a final design sometime next week if you guys can help me out with those tolerance questions. Thanks in advance for all the help.

[Diggrr]

Base thickness minimum should be around 1mm, but that's only if there's enough structure in the block to support this.
What I mean is, if it's a fully open chamber above the cpu, don't go to 1mm but stay 3mm or more.
If it's got micro channels 1/16th wide or less, the bottom of the channels can be as little as 1mm thick.
If your channels are cut using a 1/4" mill, I'd stay on the thick side, like over 2mm or even more to keep it strong.
This ensures there's enough metal around the channels to keep the block bottom from warping under mounting pressure.

Channels are cut depending on the begining stock thickness. If it's a 1/2" thick piece of copper, cut them (1/2" - 1.5mm) deep...lapping will take care of that half millimeter.

The O-ring channels are the same circumfrence as an available size ring. Don't stretch an o-ring to fit, cut the channel to match what size you have available.
Cut the channel's depth to half the o-ring's thickness.

Have fun.... :welcome:

*BTW JFettig would have more experience than I machining waterblocks, maybe shoot him a PM..He's put together a pretty snappy mill setup.

[Craptacualr]

Thanks for the helpful reply Diggrr, that helps my designs out a lot. I forget what the smallest bit size we have for the mill is, but I wanna say 1/8" so I'll prolly keep it to around 2mm thick, especially with the somewhat wider heatspreader. Have to check when I drag my tail back into work mon ;)

Yeah, I've noticed JFettig has lots of manual mill experience in that thread, and Cathar seems to be really into the science of thermodynamics and design, their stuff has been great help in general thus-far.

Your post does bring up another question... How do you lap or polish the inner surfaces? My only lapping experience is with 200 then 600 grit on my older blocks on a nice level table. Seems difficult to polish the inner surfaces that way, especially with microchannels, so what kinda tricks are there?

[birdmun]

damn it :P I would hope JFettig could answer in this thread if the question were asked :)

[Diggrr]

You don't want to lap the inner surfaces. The mill, when run slowly through the last pass (using plenty of coolant) should be clean enough.
Any rough surface that is left simply promotes turbulence in the waterflow aiding heat exchange to the water.

Pics you've seen aren't really detailed enough to show that the passages are a bit rough looking, and show tooling marks.
Cathar's microchannels are cut by a different type of tool similar to a small circular saw blade, with the block sitting parallel to the bit axis (the block is stood up on edge in the vise). That makes a smooth looking finish by itself.

[Craptacualr]

Thanks again for the quick reply Diggrr. What you're saying makes some sense. I've only thus-far been milling alum and come to think of it the surfaces are relatively smooth minus the tooling marks when cut out. Anyways, thanks again for the advice and I'll post some pics when i'm done :)