I designed a waterblock

[AndrewZorn]

Here's my threading. It is exactly G 1/4" spec and is deep enough for even the deepest of fittings. After the threading, the plastic returns to the inner diameter of the threads, not of the fitting: this means that even the thinnest-walled fitting will not be restricted by the inlet/outlet.



I shortened up the whole thing a bit after reducing the height of the fitting housing and deciding that 4mm is probably excessively thick for the top. Now the screws are too long: the block is 18mm high and M2 only comes in 16mm or 20mm.

The distance between the centers of the fittings is 19.53. So that means the radius of an individual compression fitting would need to be ~9.75mm or less to fit. I don't think it will work. It really seems like you'd need something as wide as a Supreme to do it. Why are compression fittings getting popular?

BUT NOW I'm getting a bit discouraged. I didn't plan on getting this made when I started, but somehow the fact that it's going to be impossible isn't good. If the base can't be made I can't imagine the middle or top can't be (or can they be?). If the base is all I have to change, that'd be fine I guess.

[Martinm210]

Here's my threading. It is exactly G 1/4" spec and is deep enough for even the deepest of fittings. After the threading, the plastic returns to the inner diameter of the threads, not of the fitting: this means that even the thinnest-walled fitting will not be restricted by the inlet/outlet.



I shortened up the whole thing a bit after reducing the height of the fitting housing and deciding that 4mm is probably excessively thick for the top. Now the screws are too long: the block is 18mm high and M2 only comes in 16mm or 20mm.

The distance between the centers of the fittings is 19.53. So that means the radius of an individual compression fitting would need to be ~9.75mm or less to fit. I don't think it will work. It really seems like you'd need something as wide as a Supreme to do it. Why are compression fittings getting popular?

BUT NOW I'm getting a bit discouraged. I didn't plan on getting this made when I started, but somehow the fact that it's going to be impossible isn't good. If the base can't be made I can't imagine the middle or top can't be (or can they be?). If the base is all I have to change, that'd be fine I guess.

Don't get discouraged, it's really cool how you've taken and shared what solid works can do. Unfortunately I'm no longer a student, so the thought of a $7K piece of software is enough to put it out of my reach. I use microstation as my day to day CAD program and used AutoCAD for many years in a previous position, so I think I could have alot of fun with it, but cost is prohibitive.

I like your concept of varied midplate orifice openings to balance flow. I suspect D-Tek avoided that idea for both simplicity and the real possibility of putting the block back together wrong after dissassebly, but it's a cool idea.

I'd suggest taking a few of your ideas down to some local machine shops and just get a flavor for rates. Preferrably you can find some buddy connections or university types that may have a CNC available for educational uses.

My block didn't perform better than the top dog commercial blocks, but it did better than I expected and quite the accomplishment I'm still proud of and happy about.

I chose the manual route though. For about $500 you can buy a minimill, and maybe 3-400 later you'll have enough tooling to tinker with. Then it's just a matter of your own personal time, if it's fun...then who cares how long it takes..

Someday I might try to build my own mini CNC, it would be alot of fun and quite the challege getting it all to work, but imagine the possibilities..

[AndrewZorn]

I appreciate it [again] Martin. I probably will keep on with at least just playing around in SW. I'm going to redo a base that has cross-cut pins. I can still do the loft-cut base by milling that out before doing the cross-cuts. I just need to find out if a CNC will cut Delrin like the midplate.

A CNC in the garage would be amazing. I'd make stupid stuff ALL THE TIME, like absurd cupholders and 3D art.

[Waterlogged]

For anyone interested, I found the post stating how the Fuzion's are made.

http://www.xtremesystems.org/forums/...4&postcount=33

[AndrewZorn]

I can't believe I've been checking the forums this much to see if they're back up...

I'm just about done with the redesign. I think I like it more than the original, despite it being a compromise. I have a couple more questions:

- What is the hexagonal wrench size on a standard G1/4" fitting? What is considered to be a 'wide' fitting when concerning the size of the 'base' (I have to find a minimum about how far to keep these from the edges of the inlet/outlet housing, and I don't want to increase the size of the block itself, but might have to!). As in, I have the thread sizes and all, but the outermost circular part of the fitting that you wrench down on is starting to worry me, like you might have to take the fittings out to remove the hold-down plate.
- What is reasonable as far as cross-cuts through the baseplate? As in, what's the smallest cut that can be done reasonably, and what is the smallest distance between these cuts that can be accomplished? By taking in both of these, I can find out just how small I can make my cross-cut 'pins'.
- Base thickness... the Fuzion V1 is ~1.5mm thick, and the V2 is even thinner. I thought 1.5mm should be possible. Can I accomplish this?
- With a thin base like that, is it reasonable to thread the BASE and have the bolts go top-down like I do, or will the bolts not have enough 'teeth' and thickness to be able to hold the whole thing together well? I don't know how there are chipset and full-coverage blocks that are Xmm thick and have bolts that are threaded into the base to hold it all together.

[AndrewZorn]

Revised base, realistic construction (I hope) and everything meticulously threaded:








I can reduce the cut widths if possible. These are .5mm cuts .5mm from each other. Is this too fine?
Still having trouble with this stuff:

- What is the hexagonal wrench size on a standard G1/4" fitting? What is considered to be a 'wide' fitting when concerning the size of the 'base' (I have to find a minimum about how far to keep these from the edges of the inlet/outlet housing, and I don't want to increase the size of the block itself, but might have to!). As in, I have the thread sizes and all, but the outermost circular part of the fitting that you wrench down on is starting to worry me, like you might have to take the fittings out to remove the hold-down plate.
- What is reasonable as far as cross-cuts through the baseplate? As in, what's the smallest cut that can be done reasonably, and what is the smallest distance between these cuts that can be accomplished? By taking in both of these, I can find out just how small I can make my cross-cut 'pins'.
- Base thickness... the Fuzion V1 is ~1.5mm thick, and the V2 is even thinner. I thought 1.5mm should be possible. Can I accomplish this?
- With a thin base like that, is it reasonable to thread the BASE and have the bolts go top-down like I do, or will the bolts not have enough 'teeth' and thickness to be able to hold the whole thing together well? I don't know how there are chipset and full-coverage blocks that are Xmm thick and have bolts that are threaded into the base to hold it all together.

I also still REALLY want to get rendered/realistic images to work how I want them to. This is the best I can show off right now!

Those are not some fake threads either... I actually sketched one thread 'tooth' to spec and revolved it around a helix which is also to spec. It's the real thing!
EDIT except that right now those are 17mm screws. I'm trying to find a medium between real screws and overall heights of the block. Sure I can always use spacers or cut down real screws or simply do it different, but since I'm not totally sure that this is final, I won't do that yet.

[iandh]

This is a much more realistic design. The only changes that I would make (probably tiny improvement) would be to shorten the pins some more in a concave pattern right at the inlet.

.5mm pins x .5mm spacing is not particularly hard to do. Now we are talking mainly machine time.

If I had to hazard a guess I would say somewhere around 2-3hrs to cut the pins for a prototype unit, and several more hours for the rest of the block. You would be looking at somewhere around a day of machining time for a prototype unit; normal shop rates are $65-85 an hour depending on machine type.

This could be whittled down in mass production, although it still would be quite a challenge even if sent overseas.




The CPU block market is kind of an "all-or-nothing" type thing as far as profitability goes nowadays. It isn't like the old days when what a clever person could do in a small shop was good enough to profit.

[AndrewZorn]

I wasn't trying to make money off of it, I just thought it'd be neat to have something I made.

I was thinking about shortening the pins after looking at that cutaway view.

Is a 1.5mm base thin? Can I go a little less? Do you think that there will be enough 'thread' to catch an M2 bolt on that to hold the thing together (it's the only part that's threaded)?

$65/HOUR? It sounds reasonable, but when you say hourS... I didn't want to shell out more than $100. I know that's not realistic now!

The fun part was designing it, and maybe I can make some neat chipset block or something that someone else will want to use/make... because the fancy designs are cool and all until I hear the rest of the stuff will be "several more hours"!

I have to think all this over.

EDIT and of course, iandh, the concavity would be cut out of the copper FIRST by cnc and THEN cut the rows with some sort of micro-saw... right?

[gillbot]

I made a block for my old PIII, if I had to tally up the cost of it.... GAH! I'd probably faint. Sadly I didn't even own a digicam then so no pics, but it did work well.

[Martinm210]

I've seen pins as small as .3mm square x 2mm in hieght cut with .4mm saws, but I that's getting pretty small. Although I don't thin pins outside of the critical 24mm square area do very much, I think it all depends on base thickness.

Looking good!!

[tw33ter]

stepped or bowed base?

[AndrewZorn]

I've seen pins as small as .3mm square x 2mm in hieght cut with .4mm saws, but I that's getting pretty small. Although I don't thin pins outside of the critical 24mm square area do very much, I think it all depends on base thickness.

Looking good!!

That's why outside of the 'hot area' (which I made a liiiitle larger for upcoming CPUs) I undensened the pins. There's also the concave factor to it: the taller pins allow more flow anyway the further out it gets.

stepped or bowed base?

I don't know if I'll do either. Bowing is irreversible and I don't know how reliably one can just guess at throwing in O-rings (though my design is capable of this). I might step it. Isn't stepping it just putting a small square on the bottom, diagonally cut outwards from the CPU? I don't really see the point...

EDIT: to shorten the pins, should I simply drop the whole base height and keep everything else the same, or increase the concavity of the array? Or both?

EDIT I'll probably see if it's viable to drop it low enough to use standard M2x16 screws, if not, then I'll increase the ceiling to use x20s.

And any opinions on nozzling? It's at 6mm right now if I'm not mistaken. I know the largest Fuzion nozzle is what, 4.5mm? It's a flow vs temps thing, I know, but trying to find a flow-favored balance.

[iandh]

I wasn't trying to make money off of it, I just thought it'd be neat to have something I made.

I was thinking about shortening the pins after looking at that cutaway view.

Is a 1.5mm base thin? Can I go a little less? Do you think that there will be enough 'thread' to catch an M2 bolt on that to hold the thing together (it's the only part that's threaded)?

$65/HOUR? It sounds reasonable, but when you say hourS... I didn't want to shell out more than $100. I know that's not realistic now!

The fun part was designing it, and maybe I can make some neat chipset block or something that someone else will want to use/make... because the fancy designs are cool and all until I hear the rest of the stuff will be "several more hours"!

I have to think all this over.

EDIT and of course, iandh, the concavity would be cut out of the copper FIRST by cnc and THEN cut the rows with some sort of micro-saw... right?

I was just quoting you an average rate... there is a chance that you could get a shop to try it as a "for fun" thing for a few hundred dollars, but I wouldn't expect it in a hurry.

Some shops charge a setup fee for each part as well, so if they were really doing it "by the books" having one made would cost $500-1000.


I've done stuff cheaper in the past just to help folks out. The last block I tried to make for someone I ended up losing the entire amount they paid me in broken tools (tricky machining) and ended up having to pay them that back when I couldn't produce the block, with nothing to show but wasted time and money out of my pocket. I'm not upset about it or anything but that is just how unprofitable this is when not in quantity.

Even if I did manage to get it made, I still would have been working literally for free. Since that I have vowed not to take on any block projects ever again, other than for personal use.

[tw33ter]

I suggest reading up on stepped bases (pefect step would be 23mm x 23mm)... stepped bases outperform bowed bases

http://www.xtremesystems.org/forums/...1&postcount=26

http://www.xtremesystems.org/forums/...d.php?t=161963

http://www.xtremesystems.org/forums/...d.php?t=160566

[AndrewZorn]

Okay, I put this together really fast:



It's 23x23mm, .3mm deep, and cut into the base. It is chamfered to a 45* angle all the way down, so technically the contact area is a little under 23x23mm. I was wondering if this would be a good idea, just in case the next and next series of CPUs are not exactly the same size. I left the outer edges uncut-into because I'm still worried about sufficient threading, and .3mm is almost one thread tooth for an M2 bolt. The base is now 1.2mm thick outside the step, but I think this will be okay because that's the thickness of the new Fuzion. One could THEN say that since the step-area is a CUT, I could start with something thinner, like 1mm, and cut the edges down to a technical .7mm or so... how thin is too thin?