My waterblock design

[socool]

I have made a waterblock design that i would like to have som opinions on, it is for a school project. So i have access to my schools CNC-controlled mills and stuff like that.

I have looked around a bit end it looks like the best solutions are with fins sticking combined with one inlet and two outlets. So thats what i heve based my own design on.

This block is for the CPU and will probably be made of copper. Let me hear some opinions on what you think, its always good to here others opinons.

[bigben2k]

Looks fine. Should be easy to make, if done with care.

There are many ways of refining it. If I was you, I'd look into various designs of inlets. Since you have access, I'd even try out a few designs.

Report results!

[socool]

Yeah, i plannes on making prototypes and testing them. But i dont have so much time, i am already almost two weeks behind my time-schedule.

[Susquehannock]

Looks good. Best of luck with it.

Just a few notes on copper .... many know the advantages of 110 copper. A better alternative, although more difficult to get, is OFHC (oxygen free high conductivity) Copper. As a retired machinist I can say from personal experience that stuff is more dense, thus harder to machine than 110 Copper. Yet it's properties will net better thermal transfer.

[davidzo]

fin spacing is really wide. reducing the height of the fins will increase the velocity and aid cooling.
i would let the water also flow directly to the outlets without the 180° bend, but thats only for flowrate and won't affect performance much. many people also prefere twobarbed versions which does not affect performance as the outer channels are big enough compared to the more restrictive pin channels.
if it really should be done on a cnc mill consider a round or at least oval oring, it will be easier to mount and grant you with longer sealing lifetime. but its not that big of a deal, the oring in you rendering would also work ok.

currently it looks like the block is extremely hughe or does only have 1/8" Inlet Threads, which both wouldn't make sense. Try using a more rectangular and not square shaped form if you want a 3barbed block. width should be at least 38mm and lenght is good between 70 and 90 for an 3barb design. a two barb design is basically the same but a can be a mere 15mm shorter and mounted asymetrically (like Dangerden TDX).
Its always to have good and big outer channels for lower pressuredrop and to prevent recirculating of the water, but too big blocks are neither practical nor better in flowrates, thus keep it in a good balance.
It also doesn't help cooling to make the square pinfield in the middle bigger than current cpu dies as conroe or Kentsfield, it only harms flowrate.

As for the milling, dipping in the material is always connected to indreased mechanical wear of the milling bit. milling time also increases when a lot of turns and direction shifts are needed to mill everything. if you plan a lot of holes, better use a drill bit instead of a milling bit, cause especially the moderately priced high performance milling bits with a grinded fischtail which can go really fast in normal scenarios, don't like to be used in Z-Axis a lot.
Think over using the smaller bit which you also use for the o-ring, for a more common cross pinsection insted of using the bigger one as a drill(maybe 45° rotated). That would also solve the problem of too big spacing between the pins which you wouldn't have to solve with a decreased height. a bit between 1-2mm is still ok for axial milling bits. if you want smaller channels, you better use a radial slitsaw.

[davidzo]

Also your top still looks a little risky for failures.
Also Plexiglas (Acryl) may not be the best material for the top as it can crack easily. polycarbonat is a lot better, but is less hard what evolves in problems with leaksealing at higher pressure at certain screw distances. its always better to have a little thicker top and a bigger radius to the side aroung the screws. dangerdens radius around the screw is too small, thats why those blocks often crack even when the plexi is relatively thick. Adittionally they have a really bad milling quality of the plexiparts and to not temper them long enough what makes cracking acrylic much likelier. When Plexiglas is handled in the correct way (e.g. not in dangerden way) it can still be a reliable material, its hardness is even superiour to non translucent very popular delrin(pom, acetal), what makes constructing with less screws (=better looking) easier and when it is correctly tempered it can withstand enough to last longer than every normal pc watercooling setup.
There are just a few musts and no nos that have to be allowed for:
- milling with good cooling and a special plastics bit, even better lasercutting
- at least 8mm thickness for threads and screws, better 10 or 12mm
- before tempering the surface should be flat and even and not rough like a rasp on dangerden blocks, polishing helps to make tempering more effective
- tempering should be done after all other work, the material should be handled with care at every point and should not be brought in contact with chemicals as Aceton, Benzine or other agressive chemicals used for cleaning
- never use teflon, be aware of orings which seal inside the thread or other thread sealants, only use orings which seal on the top surface, not in the thread
- use a minimum radius around corners of 5mm to prevent cracks around screws
- never use counter sunk screws in plexiglas or any other acrylic plastic
- do not use threads inside the plexiglas other than those of the barbs, better make the threads in the copper (e.g. do not use small therads with high loads)
- do not overtighten screws or barbs. sometime sit is useful to washers which spread the applied forces and do not harm the surface while tightening.

If you can maintain these important rules for use of plexiglas in PC-watercooling it won't make any problems.
A much more tolerable Material also arguable not as good loking as acrylic materials, is POM also known as delrin or Acetal. Acetal is easy to mill even without special cooling and it has superiour mechanical capabilities. its dynamic friction is really low what makes screwing in the barbs really easy, but does also make polishing impossible. you can either snadblast and oil it which makes for a nice dull black finish or you have to optimize the millingprocess that the material looks good enough when it comes out of the mill (like the swiftech storm blocks).

Think about that and oh yeah, the S939holes on your picture should be a little bigger otherwise even a tiny M3 screw woulnd't fit them and it would also be wise to add S775 and AM2 support.