My Water Block is Finaly Done

[IronFire]

Right it is here, the water block I have been asking feedback on for the last 8 months or so. If you do not know the work behind it or seen my other posts on the OCUK and other forums here is a quick run down.

Last year roughly in June, I started my Design A-Levels final year project. I decided to design, if possible, a more efficient water block design for water cooling of computers. This project was what brought me to these boards as I needed to ask potential customers questions. Since then over the months I have come back and asked for feed back on models of my final design. Well it is now finally finished and I need a final evaluation from my customer group into what they think.

Here are some pictures of the final block from the different angles:


Rough Specifications of the Block:

For use on AMD CPU’s with motherboard mounting holes using the Danger Den Maze 2 mounting mechanism.

Single Barbed Brass Inlet 5/8" internal Diameter

Dual Barbed Brass Outlets 3/8" internal Diameter

Unique Dual layer design with a 3D water flow pattern, the water moves in the X Y and Z dimensions.

Maze design combination with direct die cooling for maximum cooling.

Suitable for use with Single High Powered TEC’s i.e Greater than 200W

Machined from Pure Copper for its greater heat transfer properties than all other materials apart from Silver and Diamond.

Soldered joints for leak proof finish with no perishable seals to be replaced.


Now I hope I have you interested I shall describe in more detail how the block works or how in theory it should work.

[IronFire]

The block had under gone many changes through the modelling process but this was the final design I settled on after much work.

Here are some Pictures of the bottom layer of the block:













The water comes into the centre of the block, which is directly over the CPU die. The block is at its thinnest here with the copper thickness over the die being only 2.5mm. Connected to this is to smaller channels that are 10mm deep from the top of the layer which spiral outwards in a smooth curve to help keep water flow levels high. This channel is 3mm from the base and moves out round the block in the bottom of the main channel. The water in the this channel will heat up quicker than at the side of the channel due to it having a base 1mm thinner and so will create convection currents. This will make the water move up in the block and bring cooler water from the sides down into this lower channel; this mixes the water in the main channel. The channel depth is 1mm and width 4mm.

The centre chamber in the block has a diameter of 20mm and this is filled with the fresh cold water from the inlet of the block.

Once the water has been in the main chamber it moves out through the arms spiralling into the main chamber. These main channels have diameter 6mm and depth of 9mm and together make a combined water channel out of the block of over 12mm width. Along the side of these main channels there are turbulence inducing "dimples" which are on both sides of the channels. These have a depth of 0.5mm and have chamfered corners of 0.25mm for smooth water in and out flow, they are 9mm in length not penetrating to the depth of the lower channel. These help create turbulence in the water flow as when the water hits them it goes into the dimple then directed out at angle into the main flow of water. These devices also increase the surface area of the inside of the block, which the water has contact to as well as increasing the volume of water that the block can hold.

Due to the "dimples" the need for pure 90 degree bends to induce water turbulence has been reduced. They have been partly eliminated with the bends in the blocks water channels being very smooth and wide with heavily chamfered ends. This reduces the effects of flow constriction caused by corners. The corners themselves also contain dimples.

The copper of the channels has not been machined to a smooth finish and has been left with a slightly rough finish again to help create more water turbulence and to increase the surface are of the copper in contact with the water

When the water comes to the end of the channels the lower channel stops forcing the water up into the main channel only. Above this the layer has a hole in its base at the start of its channel. The water is then forced up to the next level of the water block by the water below it.

[IronFire]

Here are some pictures of the second layer of the block:










The design of this layer is a simple due maze with each side of the block kept separate from the other. Again the same principles for the channels with a channel within a channel incorporated as well as dimples and smooth large corners.

I cam up with this idea when trying to work out a way to have as man channels as possible in the space I could use over the CPU described by AMD’s specifications for thermal dissipation devices. I tried using lots of small channels but the flow rate in the block and the difficulties of machining it would have made it useless. I then came up with the idea after some one said something to me and I remembered about new CPU architectures based on multiple layers linked through them middle.

The large hole that is not connected to the channels in this layer is for the water to pass through from the main inlet with out getting heated before it reaches the area over the core.

In my final ideas I had thought I would mirror the image of the layer below this one on the base of this second layer again increasing the height of the channels but it became to difficult due to inaccuracies in the machine that it was abandoned.

[IronFire]

Pictures of the 2 Layers:





The final layer of the block that has the connectors for water in and water out. These are barbed so that pipes can be used with out worm drive clips if wanted and to stop leaks or the pipe popping off. The centre inlet is a 5/8" internal diameter connector that has been bored out to give an even greater inlet size. The outlets are dual 3/8" internal diameter that also has been bored out. The top layer is 10mm thick where the connectors go in and 5mm ad the edges to help reduce the wait of the block.

Here are some pictures of what it will look like, the actuall one is still being soldered together and couldnt make time for a photo shoot:






Over all the wait of the block is over the AMD specifications of a maximum wait of 300g but is near the weight of the giant Swiftech MC 462A heatsink.

[Fixittt]

where are the pics?


edit
There they are

[IronFire]

Right there it is, finally finished any feed back good and bad would greatly be appreciated for my A-Level project. When I finally get it back from marking I will have time to actually get some one to test it in a PC so until then its true cooling potential wont be known but in theory it should work reasonably well.

The ideas here are my interlectual propertie and so are the pictures and the text, if you wish to use them in a website please e-mail me first thanks

[IronFire]

Right the block is actually finished now and here are some pictures of the final completed block.



The layers where soldered together and the pipe connectors where soldered into the top layer which has had copper removed from the corners to lighten it. The entire block weighs 730 grams which is almost exactly the same as the Swiftech MC 462A heatsink and fan.

Once the layers where soldered it was tested for leaks using compressed air @ 5psi with the block in a bucket of water. A leak was found so the entire block was re soldered and re tested. This time there was not a single leak of air and as the water will not be under any pressure I am now confident it is totally leak proof. The block then was milled to remove excess copper and solder then re-tested to make sure it was still leak proof.

It was then pickled in acid for 2 days to remove all the flux and copper oxide from inside the block in the channels and to leave it bare pink copper. It was again tested with compressed air @ 5psi to make sure it was still water tight and it was. It was then finished with a steel brush to give the nice shine it is now.

Now it is totally finished and as I got not that many reply’s to my last post I am again going to ask for you to say if you like it or if you dont. What you would change if you could and why and what you would leave and why. If you could ad anything what would be it and why ? What would you be prepared to pay for it, around £60 ? If not why not.

Thanks for your help as my A-Level course work has to be done my May the 1st so get replying!

[Fixittt]

a few concerns.

The weight. that thing must weigh a ton, I know the blocks I make are really heavy

also, doenst look like you leave much clearance for the mounting nuts. Is this going to be fixed in the near future? Also with the weight of this monster, I would also recomend including some sort of solid backlane. to give the "METAL" bolts something to use for a backrest instead of the motherboard.

other then that, I dont really see much heat at all getting to the top 2 layers.

looks really good! I would leave the burrs in the block to add turbulance.

[IronFire]

To address your concerns, in my Last post I have put that it ways 730 grams, that is almost exacterly as much as the giant Swiftech MC 462A Heatsink.

The clearence for the nuts isnt great but it shouldnt be to bad as its to be used tihe the Danger Den Maze 2 style mounting kit and the nylon bolts are flexible and allow for bending.

The roughness of the channels was cuase by me having to use a CNC machine with outlube as the other machine was broken by peopel cutting wood on it and jamming the lube pump I have left it as it was to help give more turbulance and surface area.

[beav]

it only weighs as much as the swifty? It looks 3x heavier than that OCH reactor block!

[IronFire]

As the block has had so much copper removed from it to give all the space in it it actualy doesnt weigh that much. 730 grams as I said so its safe

[Fixittt]

GREAT, then I look forward to seeing how it preforms!

[Jessfm]

Hey can I preview my latest block in here too ?
This has been a long project for you iron, I just hope you get the grade you deserve.
Anyway, the weight issue.
It is an issue, as we getting heavy block, and since the mobo's are verticle, the CPU's top edge has less pressure then the bottom edge. Not to mention pipework and the water(dependant upon volume) all pulling on the top mount springs. at a guess 25% more force needed on top edge to creat an equal force applied over the die.
I have not read anything about these issues.

What i do find puzzling is the need for the mid section ??

[Adam]

I'd clean up the milling a little bit. Those little burrs sometimes break loose and like to eat up pump impellors.

[Brad]

great block design, sort of overdesigned though.... what about using 2x 1/2" barbs?

how well does it perform? (you obviously have a maze2...)

[Marco]

Haha did brad just say overdesigned? Never thought i'd heard that from him!!

Aside from the fact that it's like the size of a brick it looks mean! Looks to me like you will have to be really careful with it in order to avoid a crushed core. Haha one odd thing i noticed: your two smaller barbs are different! Top layer does look superfluous to me as well. Maybe you should have used it to combine the two outputs into a single unified one?

[IronFire]

To awnser your questions again:

The mid section was an idea I came up with to over come the AMD specifications for the size of CPU cooler to stop it interfering with motherboard components. As I wanted more surface area and water volume I decided to build up. I have no proof that this works or if it actually decreases the effiecny of my bottom block which would probably be fine on its own but as I have had only time to produce one block I decded to include it.

[IronFire]

Gah it decided to post my post half way through :/

[IronFire]

One thing I would like to point out is the fact that this was an a-level project and as such I nearly didnt have time to build what you see now. I has not been possible to test each combination in copper only with wax models.

To awnser Marcos points, one barb is different as I dropped it and one barb got bent and snapped off, it has since been soldered back on as I didnt have enough time to replace it.
Also the block did have a third layer which combined the channels but this was abandoned becuase the block would have been way to heavy and complex.

I have not had time to test and will not untill I get it back from the exam bords. I will not test it as I just sold my H20 rig as I found it to hard to make it easily LANable. When I do get it back it will be up for grabs to be reviewed by some one with a large TEC.

The weight of the block for mounting will not be a problem, the only thing that will be is the large inlet pipe which will be stiff and if not suported could drag the block off the core, if suported the block should mount fine. If not stiffer springs can be used on the top over the bottom bolts.

To beavs point, I have read the PRO coolings block round up and reading abut this block I wanted to make sure my block would not end up the same way. As man weight saving features have been used as possible with out counting removing the second layer.

Any more questions fire ahead but dont make negative remarks that are not constructive.

[ondaedg]

did anyone mention that it is big yet? =]

The design is nice. Right now however, I think blocks are heading into a direction that produces more turbulence, high surface area, and high flow rates. I must say that you did a good job if this was your first venture into making waterblocks. The middle piece probably did not have to be so large. Reason being is that I am assuming that you are using plumbers solder which is typically a 95/5 ratio. Plumbers solder is not the greatest when it comes to thermal conductivity. The heat has to fight through that layer of solder to get to that very large center piece. However, I could be completely wrong and after testing you could have produced the most powerful waterblock to ever hit the planet. :P

The block is nice, but we want to see testing results!

[Brad]

onda, I think we are getting to the point where the cpu --> wb interface is the major limiting factor, not the block itself. we are starting to see less and less gains in both air and water

[morphling1]

IronFire,WOW great looking wb, you have almost the same idea like the one I'm doing now, only that I'll use second layer to connect both outlets, cos' I want my block to be as low restrictant it can be. The only problem I see (beside the weight) is the thickness of the block, as I read on overclockers forum, one guy is numerical modeling simple water block and the greater thickness helped. And as I like to verify that with test, I'll make my block with starting base thickness of 7mm, than test and mill away bottom in 1mm steps, that way I'll find close to optimal thickness for my particular block. It would be a shame to lose performanse of otherwise grear block because the thickness isn't right.

[bigben2k]

This block might be hefty, but it has a definite advantage: great flow!!!

You should be able to pump a lot of coolant very, very fast, the only problem is...

Your rad works better with a low flow rate (giving more time for the coolant to cool!). So you'll have to either use a lower flow rate, or a very large rad.

Also, this design, like the Swiftech, might have this quirkiness where you have to hold it upside down while filling it up, otherwise you might have air trapped in your WB. It's just something to keep in mind.

In any case, you won't have to worry about a leak because of the coolant pressure!

Nice work...

[Divided]

i'd like to see screws holding the blocks together and it also looks like a big sucker.. eh.. oh yea.. no wait, i forgot...