• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Waterblock Design

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Superami

Registered
Joined
Jul 5, 2002
Location
Germany/Texas
OK I'm new to water cooling. I'd like to make the water blocks myself. Luckily I have a friend with a mill so that gives me several options. I was considering the following design as a top milled copper block with a lid and water entering through the lid in the middle area and leaving through the lid from the lower right. Now my thoughts with the water path were to make it as jagged as possible to increase turbulence through out, but keep it wide enough as to not impede water flow. I figure if I'm using 8mm hose then a 10 mm bit should be able to accomplish this.

For a lid I was considering using a clear acrylic or lucite so that any corrosion or blockage could be seen. My concern is just wether or not the lucite will be strong enough and that it won't conduct any heat, reducing the ability of the water block to convey heat into the air. I'm concerned with sealing the top. My plan was simply an O-Ring and screws. I was also going to try and mount it using 4 more screws and the Socket A mounting holes (thus the extra width).

My other main concern is the thickness of the floor. I mean if I make it too thick then it won't get as good of a transfer to the water, but if its too thin then the heat won't transfer throughout the block as well, thus reducing the overall surface area for water/copper heat transfer. I was also considering not using the large bit to mill the area directly over the CPU, but rather a small bit 2mm or so to make short spikes (similar to a traditional heat sink design), so that this area has a maximum surface area.
 
Here is another design I'm considering. This would be one solid block with holse drilled from the sides. Basically I took one of the simple designs from the article section here and modified it so that the water exits on the opposite side. I figured this would assure the most water flow an that all of the channels were used. The middle channels would also be smaller to assure that all of them are used. I figure the four connecting channels would also be closer to the center where they are most needed.

I like this design for its simplicity and ease of construction, but I feal that overall there would be much less turbulence than my milled design, and thus less heat trasnfer.

The big advantageous is that I would only have 4 holes to plug, and so I wouldn't have to worry about leaks as much.
 
Here is my last design I'm currently working on. It would require drilling seven 10mm holes from the sides. Basically the 2 water channel holes would be drilled from each side, with the center two as close as possible. Then one hole is drilled from each side to connect the left pair of channels and right pair of channels. Then finally one more hole is drilled from the top, but only far enough down to connect the two channels. Hole 1 is then used for water intake and hole 2 is used for outtake.

Again this has the advantageous of being easy to seal and even hold a lot of pressure. Constuction is also easy. But it basically allows for a traditional maze WITHOUT a two piece block.

I don't like that the water doesn't start at the center. I think this would reduce efficiency, and the smooth water ways would reduce turbulence.

The one advantageous I can see with the round paths is that it allows the heat to travel up the sides of the water pathways better, because the pathway walls are then thicker at the bottom and then get thinner.
 
Woops forgot to attach the last picture.

Basicaly my thoughts are that a single piece construction allows for the greatest waterpressure and least leak risk. Of course you can always seal the 2 piece blocks together, but then you lose the advantagous of being able to clean the block from time to time.

The major advanategous of the 2 block system is you can make the water take anypath you want. I'm surprised more people haven't used jagged walls for the water path though, as I would think this would increase turbulence and thus increase heat transfer.

The other thought I had was taking a thicker water block and carving a traditional heat sink sort of design on top of the waterblock above the water paths. This could increase the overall heat transfer of the block a little, but more importantly, should the pump fail, it would give the block a better chance of keeping the CPU alive until you can shut down. This would of course also increase wheight. One could also always attach a traditional alluminum heat sink to the top of the block and get a similar (but not as effective) solution for less work and weight.

Well I hope I get some feed back to my designs. I'm off now to look at parts and see if I can find copper blocks to carve up anywhere in town.
 
Woops one more thing while I'm at it. Depending on the size of the water block, the last design could easily be modified to contain even more bends to the W.

Also the pictures are more or less to scale, with 1 pixel = .1mm.
 
Superami said:
Also the pictures are more or less to scale, with 1 pixel = .1mm.

First, Welcome Aboard!~! :D

Now, Im hoping you mean 1mm, .1 would be a little small I think. But, Im not sure I follow that exactly.

Now, for the base thickness, the swiftech? I believe it is has a base thickness of .25 inches. I would trust them with their results since they seem to make such good water blocks. But, I personally dont know exacts for better or worse.

The first design I think would be the best IMO. Mostly due to that the fresh water would come in directly and hit the CPU spot right on first.
Where the others, have been used, minus maybe your last, cant say for sure. I think the overall concensus has been that the first there is one of the better designs.
As far as sealing it, that has been discussed in many ways with many folks. I will be attempting an O-ring style as well. With luck that is, heh. That I know has been done, and with rather good results overall I think!~!

IFMU
 
well

the one thing you might want to think about is dead spots where water wont flow effectivly

thats the problem with makinga waterblock with TO many surfaces that are at od angles. you might creat turbulance in some good places but others might let water sit and degrade performance
 
1/4 base with a XP type processor works well ..
My homemade spiral maze block is doing GREAT !!..
And I wouldnt worry to much about a lucite top not radiating heat to the air .. I can touch all 4 sides and the top on mine and its cool to the touch ...
 
Re: well

thorilan said:
the one thing you might want to think about is dead spots where water wont flow effectivly

thats the problem with makinga waterblock with TO many surfaces that are at od angles. you might creat turbulance in some good places but others might let water sit and degrade performance

DKYoda -

How does your maze look? What do you use to seal the block?

The other thing I'm worried about with lucite though is mounting. So far the local hardware stores have only had 4 - 5mm thick lucite or acrylic. I'm worried that this might be too flexible, and cause leaks if I don't use a lot of scews.
 
well you should soldier the top on, thats what DKY did, he has a post like rite under yours, soldier is almost gaurenteed not to leak, unless you dont do a half decent job, and i have made a few blocks in my day, check my www below this text...
i would say make it smooth dont worry about roughing it up... it may add a little but it would be more flow restriction.

and those designs are pretty basic, is the mill a cnc or is it one with the hand cranks??

that will be interesting to see what happens,
and if this is for amd processors you mite want to integrate the 4 mounting holes
 
the overclocker said:
the base should be around 3mm, any thinner and it could bend, but if the base was thicker then heat transfer wil be impared.

actually thicker is a little bit better, because it can spread the heat farther, more heat can be sucked up
 
Well now I'm getting confused. For base thickness I've gotten 1/4" (6.35mm) and 3mm (1/8"). There is a big difference here. I could see a quarter inch being a bit thick, but at the same time thicker would conduct the heat faster through the copper. Maybe 4mm or 5mm would be good?

BTW- to an earlier thread, I did mean 1 pixel = .1mm. Basically if you blow it up 100 pixels represent 1 cm.
 
maskedgeek said:
well you should soldier the top on, thats what DKY did, he has a post like rite under yours, soldier is almost gaurenteed not to leak, unless you dont do a half decent job, and i have made a few blocks in my day, check my www below this text...
i would say make it smooth dont worry about roughing it up... it may add a little but it would be more flow restriction.

and those designs are pretty basic, is the mill a cnc or is it one with the hand cranks??

that will be interesting to see what happens,
and if this is for amd processors you mite want to integrate the 4 mounting holes

Well I'm definately integrating the 4 mounting holes. It only makes sense with a heavy copper block.
The mill is hand cranked from what I understand. I haven't seen it yet, but it is in my friend's basement so I'm not expecting too much.

Another thought I've had is to step down the high at the intake a little (and back up at the outake) so that the water doesn't have to make a 90° turn.
 
i just made a block similar to your top design, and it doesnt perform all that great, its good, like maze 2 performance but not like gemini performance... if you think you have the skills i think i have a wb design for you to try out...

and the base thickness... we have our own opinions.. with a high wattage overclocked cpu a thicker base is better. if this is a celeron or something it doesnt matter
 
maskedgeek said:
i just made a block similar to your top design, and it doesnt perform all that great, its good, like maze 2 performance but not like gemini performance... if you think you have the skills i think i have a wb design for you to try out...

and the base thickness... we have our own opinions.. with a high wattage overclocked cpu a thicker base is better. if this is a celeron or something it doesnt matter

I think I have the skills. What does your design look like?
 
Some Brainiac somewhere did some pretty deep test on base thickness .. and with the xp core 200 Mils was optimal ..
That is 5 MM or 1/4 inch give or take :)
It made my milling pretty easy actually ... The end mill I had was 3/4 before it flared out to shank size ..
With a 1" think block it was a cakewalk !!!

Hey Geek ... I got that sliding Table yesterday ..
I wouldnt use it on an actual mill .. Has a tiny bit of slop in it ..
But for a drill press and just poking ACURATE holes ..
Its PERFECT !!!
And as usual ... the price went down AFTER I bought it hehe
Its only 25 Bucks now !!!
 
Superami said:


I think I have the skills. What does your design look like?

ill take it to the mspaint and draw it up for ya'

and dky, the table isnt all that great for milling then? so its just not strong or whats the deal?? if not i have another table lined up that gonefishin uses, 80 bucks...
 
allright, here we are.. channels are 1/8 and 1/4inch, lots of surface area and good flow
 
Yeah ... Ive laid hands on the one GoneFishin' uses ..

I have one of those stores real close to where I work ..
It would be alot better for milling ..
This one is sturdy as heck .. and has a nice vise on it ..
But it has that tiny little bit of play in the controls ..
Im gonna work on Tightening it up some ..
If the extra cash isnt an issue ..
I would say go with the one GF uses ..
I just didnt want to dip too far into my Beer budget !!!:beer:
 
Back