homemade water cooling for everyone ;) - january, 2006.

[turpija]

I didn't wanted to buy finished products and just put it together ... that's not fun, it's boring. So i've come up with few ideas, and was thinking, why not watercool everything when building all from scratch. And this is how it's suppose to look ...

discussion log



and a little playing in Rhino3D



As you can see, there would be six blocks, reservoir which would be also a window, and homemade radiator, yep that's right, why not make your own rad that fit perfectly in case, and nothing would be stickin out of case.

There would be probably some problems, like how to make an equal flow for all blocks, but i'll give it a shot and we'll see what happens.

and here's some info about machine that we'll be essential part of it
AOpen H600A case
Soltek 75FRN2-RL (bad thing is that mainboard doesn't have those four holes around cpu socket, so i'll have to figure out some other way to hold block)
Barton 2500+
2x256mb infineon ram
Sapphire 9600Pro
and other little pieces, that are pretty much alike
... it's not one of those super expensive fancy machines, but it's good enough for the money.

[turpija]

First thing is cpu block, it have to be enough simple to build with my el-cheapo tools, and enought efficient to actually do some cooling.

a little print out for drilling ...



and let's start some real work ...



but the problem has occured quite early, because i didn't know how to properly do it i've gone straight to biggest dril bit and f*** this one ... my el-cheapo dril is not very precise so this one is waste ...



Without too much crying, i've printed two more schemes, and stick it on. As you can see these two part would be sandwiched later on so actual dimensions of cpu block would be 50x50x20mm ...



With a previous example how not to drill, i've done this one by few stages (slowly increasing size of drill bit) and make this interesting holes













This is all for now, cause i don't have anything to cut it, it'll stay like this for some time. Next i'll upload some pictures of building radiator, and now i have pretty much all stuff needed so i'll hope things will flow from now on

[ReFredzRate]

Looks nice. Bit busy though...

[jokkos]

I'll be following this one. Huge reservoir that is also a window, hmmmmm...

[Tulatin]

Won't work. Water, like electricity and humans, will take the path of least resistance. Meaning, your cpu will end up with about .3GPM, while your memory, with a straight through flow, will get about .8-1. Sort of an issue, no?

[turpija]

Originally Posted by Tulatin

Won't work. Water, like electricity and humans, will take the path of least resistance. Meaning, your cpu will end up with about .3GPM, while your memory, with a straight through flow, will get about .8-1. Sort of an issue, no?

Well, i've been thinking about that, and that's why i've decided to use 8mm ID pipe for cpu, and 6mm ID pipes for all other blocks. That should solve the problem with water supply for cpu, and hopefully other blocks would have decent flow because main pipe, from pump is much larger, 13mm ID.
But i'll know for sure if that works when it's all completed and tested, everything else is just guessing.

[frodo]

if you need to control how much flow goes to each loop you could use a manifold with ball valves, therefore you can control how much flow goes to each loop

ive thought of doing the same thign for a very long time!

looks promising

[turpija]

Pipes cutted for radiator, 15mm OD.



And this is about 30-40 thin copper plates (0.3mm) cutted in 140x120mm, and pressed together with few mm thick steel plates. Later, this is proved to be a bad idea cause copper plates are not pressed and there is alot of air in between, so the driled holes are not parallel to eachother, and overall - it sucks. And yes, about few kilos of copper wasted ... again





Much better now, much thicker steel plates (4mm) pressing already straighten copper plates without gaps and air between, this time here's 50-60 plates. Drilling is pretty much straight-forward. Drill bits used - 3mm, 6mm, 10mm, 12mm, 14mm, 15mm ... alot of drilling with same el-cheapo drill, but it looks quite nice









Test fitting all the pipes ...
it looks promising



And finally, cutting it to exact dimensions ... as said before it's 140x120mm.



Atfer disassembling whole package, it's all full of WD-40, and copper "dirt". Time for some cleaning ...



And while cleaning that, time to prepare pipes and elbows for soldering.





Let's assembly that to see how it looks ...



... looks like it's gettting nice shape ...



And with friendly help of "blue flame" it wouldn't be possible ...



... to solder this bits together. Looks atractive, don't they



And that's pretty much all for now. Soldered connections seems pretty water-proof, but that will be tested when whole thing is completed. I hope it won't leak
Next i have to get some thermal counductive glue to fixate those plates on equal distance, and it's kind of inacessible to solder them all, and hopefully glue should be more than enough.

[turpija]

Originally Posted by frodo

if you need to control how much flow goes to each loop you could use a manifold with ball valves, therefore you can control how much flow goes to each loop

ive thought of doing the same thign for a very long time!

looks promising

Don't quite understand what ball valves do you mean ? But until it's all finished we won't know will it work OK, or just flow in few blocks.

What frodo was saying was basically using valves for each output of your "manifold" (the long block you have several pipes stemming off of). By using these valves you could directly control the resistance of the water coming out of each pipe. No need to hassle with different diameter tubing.

I mean this in a constructful criticism type of way: I highly suspect your water cooling system is going to be very ineffecient because there is too much resistance/too little pump power and your radiator will need to be constructed carefully to operate well. Just don't expect to overclock very high and you'll probably be fine...You'd fare better with two cooling loops that have higher flow rates...

I'm interested in seeing how this one works out. Your project is unique.

[frodo]

unique yes, but not 100% original

its been done before, but most people just scrap the idea as it hardly ever works that well

best way to do multiple loops is to have an equal pump per loop after the res - that way they all get the same flow, but in your case that means like 6 or something different loops....

[turpija]

Originally Posted by SensesFail

What frodo was saying was basically using valves for each output of your "manifold" (the long block you have several pipes stemming off of). By using these valves you could directly control the resistance of the water coming out of each pipe. No need to hassle with different diameter tubing.

that's nice idea, but what type of valves ? do you have maybe some pictures, or links what should i look for or how it looks ?

Originally Posted by SensesFail

I mean this in a constructful criticism type of way: I highly suspect your water cooling system is going to be very ineffecient because there is too much resistance/too little pump power and your radiator will need to be constructed carefully to operate well. Just don't expect to overclock very high and you'll probably be fine...You'd fare better with two cooling loops that have higher flow rates...

I'm interested in seeing how this one works out. Your project is unique.

I don't know how good will it be, or how much flow will it be in loop, but after i assemble whole thing i'll have to test few pumps to see how it works. I have zero experience with building water cooling, and especially with how pumps perform in different situations, so only way is to test it in real life.

btw what you think that i should carefully construct radiator ? there is few pictures where you can see how radiator should look. Do you have some idea or sugestion how to improve it ?

Originally Posted by frodo

unique yes, but not 100% original

its been done before, but most people just scrap the idea as it hardly ever works that well

I don't say that it's first time anyone has ever build something similar, but just didn't see it (do you have maybe link/picture where can i see how others done their), and i think it could look nice when completed, so why not give it a shot

Originally Posted by frodo

best way to do multiple loops is to have an equal pump per loop after the res - that way they all get the same flow, but in your case that means like 6 or something different loops....

Well i was thinking about connecting it all in series, but this way is far more attractive and only thing that bothers me is equal flow for every part/block (which might be fixed with valves if necessary), overall flow will be (hopefully) sorted with decent pump.

Thanks you all for critics and sugestions, that's what this topic is all about

[turpija]

I've worked a little more on block for cpu so it's practically finished now.

First, one side is cutted and prepared for soldering.



Then it was heated and solder is applied to both sides.







It is then "sandwiched", heated and at same time pressed so exces solder will squeeze out so there is not air gaps between copper plates and whole thing is waterproof.



The holes are enlarged and threaded for connectors.



This is alot of connectors, all hand made (offcourse not by me, i don't have that advance machinery ). It needs a little bit of finishing touch, but it's just right dimensions and it's cheaper than commercial ones





And test fitting it on the block. Block is still pretty rough but i'll polish it later when all block will be finished.



This is all for now folks

[turpija]

Here is final assembly of radiator.
distance between copper plates is around 4mm.



Elbow connectors fitted on the other side and soldered.





And comparation with CD. Radiator will be fitted with two 120mm fans.
Overall dimensions of radiator is 12x14x24cm, and it weight a little less than 3kg.
Notice how tiny cpu block look in background



Now i have to test it if it's really waterproof and fit a pair of connectors for tubes. btw have anyone has a good idea how to pressure test it ?

[AJB2K3]

Looks good but as to pressure testing its easy.
Take a bucket of water to a car tyre pump Block one end of rad make an adapter to to connect pump to rad, put rad in water and pump for all you worth. (a pump with a guage would be of use.

[turpija]

Originally Posted by AJB2K3

Looks good but as to pressure testing its easy.
Take a bucket of water to a car tyre pump Block one end of rad make an adapter to to connect pump to rad, put rad in water and pump for all you worth. (a pump with a guage would be of use.

Yes, i've found few more guys that pressure tested their block with that method, and if it doesn't leak air, than it definetly won't leak water which is circulating in much less pressure.
Now i just have to figure out how to make adapter for pump.

thx AJB2K3

[Ener]

like the custom radiator, i like the way it looks all big and bold, hope it all works out for you

[turpija]

Originally Posted by Ener

like the custom radiator, i like the way it looks all big and bold, hope it all works out for you

But the best thing is that it will fit in midi tower like a glow, and available space in case is 100% used, not like with commercial radiators, where you have to have big tower in order to use some serious radiator. And with little polishing this baby will look like even better

[AJB2K3]

Just a minor but what would it have been like with a CCFL strait down the middle of the rad?

[turpija]

Originally Posted by AJB2K3

Just a minor but what would it have been like with a CCFL strait down the middle of the rad?

Don't quite understand ... ccfl ?
I suppose it's neon light, if is, I'll test it when I buy one (or two) ...