Custom CPU water cooling block

[theMrDanielPetersen]

Hi guys!

We are two guys who are trying to design and build our own water cooling CPU block, and we would like to have your input on some things. Now the thought was to build all of it in copper. And shape it like a circle. Now that might sound weird, but we think we have gotten a good idea! Our thought was to make the chamber inside the block a spiral, so that the water enters the middle and spirals out to the side and then continues the loop. Now while we were designing the block we had to make a decision of how thick the bottom should be. What do you guys think? We will be able to post pictures of the model when we have chosen which shape we think is best.

[Ghost]

Aren't they typically as thin as possible so heat can conduct through to the water better?

P.S.: Would be great to see this as a blog style thing.

[MoeBrowne]

They typically tend to be as thin as possible on the base, while still retaining enough structural rigidity. Also thinner means less Copper which means less cost.

I have seen a few other custom water blocks using the spiral design, most notable are the ones with multiple 'arms' to the spiral increasing surface area and flow compared to a single channel.

I would also be very interested in seeing a blog/project log of your work.

[Salmiakbal]

Sounds like a fun design, dont know whether it will make a big difference in temps but it will probably look nice anyway :)

How are you planning on manufacturing this though? Machining curvatures is pretty hard, making straight slots (like a normal heatsink has) will definately be easier. Making it entirely out of copper is possible but not out of one piece of course, but you probably already knew that :p

I'll give some more input later, typing on phone is hard :p

[KyunG]

If you have the knowledge and machinery to produce a custom waterblock, might as well go silver :P Out of all metals, silver has the highest thermal conductivity. And once you are done with using your computer, just take the waterblock off and remove the silver and sell it! Good investment since silver will be worth more in the future :P

[Fordox]

With a CNC a lot can be done.

if you want a good block, try to "copy" an existing blok. a jetplate is required, and vertical lines are good.

with a spiral you will have a lot of flow restriction... try making it into a star (just something with straight lines)

[Ghost]

If you have the knowledge and machinery to produce a custom waterblock, might as well go silver :P Out of all metals, silver has the highest thermal conductivity. And once you are done with using your computer, just take the waterblock off and remove the silver and sell it! Good investment since silver will be worth more in the future :P

Haha, best idea ever.

Looking at these charts: http://www.engineeringtoolbox.com/thermal-conductivity-metals-d_858.html

Doesn't seem to be that much of a difference. But love your idea of re-melting and making a profit on the block.

It could also take the place of a silver kill coil.

[airdeano]

gimmickie at best on spiral. worked in air-flow dynamics for 10years+ and to make

the flow-pattern to do something its not naturally doing already starts a restriction.

and in an enclosed area, limited devices and endless other restrictions will make

it very challenging.

best idea is to mimick another style and pick-up off-market business. speciality

non-reference PCB GPU waterblocks, remote style CPU blocks, find a radiator

guy and make custom radiators (cross-flow instead of down-flow)

airdeano

[theMrDanielPetersen]

We tried drawing the spiral thing i SolidWorks, but we could'nt make use of enough of the surface on the cooler. The idea with two lanes sounds pretty smart, and we will look into that. We use copper instead of silver because we can more or less get it done for free in copper. We could'nt do that i silver. Also copper doesn't really loose its value that much. But we will try to make it as multiplatform as possible so that we can use it for a lot of sockets, and never really need to sell it. Now we had some problems with the spiral thing and the surface not being used as much as we wanted. But we ended up using a square for now, and drawing one lane with some fins in it. I will try to add the photos at the end of this post. Now we had some concern about ruining the flow with the fins, so we made a flow simulation, and they really doesn't seem to ruin the flow. And we believe that we will be removing a lot more heat with the fins than without. So we were thinking that the fins would more than live up for the lack of flow. The input thingy is 1/4".

[theMrDanielPetersen]

We were thinking about either constructing the block in copper, which would require CNC manufacturing and the price for the construcion would rise. Alternately in Silver, using additive manufacturing, but considering the strength of forged silver, we should consider a thicker design.

[Salmiakbal]

Looks like an interesting design, but probably pretty hard to fabricate, especially at that scale. Additive manufacturing would be a solution to that, but probably more expensive as well.

I thought you were meaning something like this:

[Ghost]

You want to avoid tight bends in your block as that slows down the water and can cause vortexes (or whatever you call water spinning). The worst thing that can happen is water spinning in a block as that would just slowly heat up. You may want to add banked corners in that design to help the water flow out a bit.

[Salmiakbal]

We tried drawing the spiral thing i SolidWorks, but we could'nt make use of enough of the surface on the cooler. The idea with two lanes sounds pretty smart, and we will look into that. We use copper instead of silver because we can more or less get it done for free in copper. We could'nt do that i silver. Also copper doesn't really loose its value that much. But we will try to make it as multiplatform as possible so that we can use it for a lot of sockets, and never really need to sell it. Now we had some problems with the spiral thing and the surface not being used as much as we wanted. But we ended up using a square for now, and drawing one lane with some fins in it. I will try to add the photos at the end of this post. Now we had some concern about ruining the flow with the fins, so we made a flow simulation, and they really doesn't seem to ruin the flow. And we believe that we will be removing a lot more heat with the fins than without. So we were thinking that the fins would more than live up for the lack of flow. The input thingy is 1/4".

Looks good, but dark blue means that the water's moving slow and green/yellow/red incrementally faster :p

[MG2R]

Looks like an interesting design, but probably pretty hard to fabricate, especially at that scale. Additive manufacturing would be a solution to that, but probably more expensive as well.

I thought you were meaning something like this:

Oh, how I want to have a CNC machine right now

[Art Vandelay]

We tried drawing the spiral thing i SolidWorks, but we could'nt make use of enough of the surface on the cooler. The idea with two lanes sounds pretty smart, and we will look into that. We use copper instead of silver because we can more or less get it done for free in copper. We could'nt do that i silver. Also copper doesn't really loose its value that much. But we will try to make it as multiplatform as possible so that we can use it for a lot of sockets, and never really need to sell it. Now we had some problems with the spiral thing and the surface not being used as much as we wanted. But we ended up using a square for now, and drawing one lane with some fins in it. I will try to add the photos at the end of this post. Now we had some concern about ruining the flow with the fins, so we made a flow simulation, and they really doesn't seem to ruin the flow. And we believe that we will be removing a lot more heat with the fins than without. So we were thinking that the fins would more than live up for the lack of flow. The input thingy is 1/4".

For a CPU block, you want the high surface area point to be directly over the core and not much anywhere else. Your block design would unnecessarily increase pressure drop in order to cool the area of the IHS that does not get very hot.

[Art Vandelay]

Looks like an interesting design, but probably pretty hard to fabricate, especially at that scale. Additive manufacturing would be a solution to that, but probably more expensive as well.

I thought you were meaning something like this:

That's pretty cool, but you need an outlet.

[Season]

Are you making a custom block for performance or looks? If for performance, I doubt you'll be able to get very good performance out of a custom block. Companies like EK hire people who are experts in fluid dynamics and spend several months designing blocks to get the best performance and flow rates. If for looks, I suggest taking an existing block and remaking the top instead. Good luck!

Edit:

I read you were looking to have it 3d printed in silver. This would be cool, but it would cost several hundreds if not thousands of dollars if you use one of the online print shops. My suggestion is CNC milling or 3d print in ABS, molding, and forging.

[Art Vandelay]

If you have the knowledge and machinery to produce a custom waterblock, might as well go silver :P Out of all metals, silver has the highest thermal conductivity. And once you are done with using your computer, just take the waterblock off and remove the silver and sell it! Good investment since silver will be worth more in the future :P

Wouldn't silver react with copper and/or nickel?

[theMrDanielPetersen]

Thanks for your feedback.. CNC milling is definetely the choice for making this in copper, mainly because we dont have any copper additive manufacturing machines available. We can however for the price of 20 usd per cubic cm, "3d print" a working model in silver.

I guess the main purpose is to build a block by ourselves an experimenting a bit with the design.

​

Yes, sharp bends is definetly a flow-killer, but its a dilemme between high flow or big surface area, and since the CPU often gets a lot hotter than the cooling liquid we are trying to optimize the heat convection. It was trying to find a solution giving both high flow and surface area, which led to the spiral idea.

(We are not saying that we think we can beat the current market products, just trying how well we can do).

Salmiakbal, We are not sure we understand your concept, as we only see a single inlet/outlet for water on your model, and if it was to work, we guess the outlet should be formed all the way around the outer circle perimeter, to ensure an even flow in all directions. Interesting idea. We would be very interested in working on your model, if you would allow it.

Sorry about the misinformation on the flow simulation. Its a vorticity analysis, and true, the blue symbols an slow flow and yellow a higher speed.

[Salmiakbal]

Thanks for your feedback.. CNC milling is definetely the choice for making this in copper, mainly because we dont have any copper additive manufacturing machines available. We can however for the price of 20 usd per cubic cm, "3d print" a working model in silver.

I guess the main purpose is to build a block by ourselves an experimenting a bit with the design.

​

Yes, sharp bends is definetly a flow-killer, but its a dilemme between high flow or big surface area, and since the CPU often gets a lot hotter than the cooling liquid we are trying to optimize the heat convection. It was trying to find a solution giving both high flow and surface area, which led to the spiral idea.

(We are not saying that we think we can beat the current market products, just trying how well we can do).

Salmiakbal, We are not sure we understand your concept, as we only see a single inlet/outlet for water on your model, and if it was to work, we guess the outlet should be formed all the way around the outer circle perimeter, to ensure an even flow in all directions. Interesting idea. We would be very interested in working on your model, if you would allow it.

Sorry about the misinformation on the flow simulation. Its a vorticity analysis, and true, the blue symbols an slow flow and yellow a higher speed.

Yes, I hadn't constructed an outlet yet in that model, though I have made a simple one now:

http://i48.tinypic.com/sdfnk9.jpg

And a flow analysis:

http://i47.tinypic.com/28l83gp.jpg

As you can see flow is okay-ish. About half of the fins get used in this scenario. If you want the files they're yours to have :) Change the design as you want.

http://speedy.sh/cSKVJ/Assembly-waterblock.SLDASM

http://speedy.sh/8CVd2/Waterblock-base.SLDPRT

http://speedy.sh/qYdEe/Waterblock-lid.SLDPRT

http://speedy.sh/FaEMX/In-outlet-cover.SLDPRT

Do take in mind though that I didn't take any specific dimensions in consideration so you may want to change that to a more realistic setting:)

[theMrDanielPetersen]

Are you making a custom block for performance or looks? If for performance' date=' I doubt you'll be able to get very good performance out of a custom block. Companies like EK hire people who are experts in fluid dynamics and spend several months designing blocks to get the best performance and flow rates. If for looks, I suggest taking an existing block and remaking the top instead. Good luck! Edit: I read you were looking to have it 3d printed in silver. This would be cool, but it would cost several hundreds if not thousands of dollars if you use one of the online print shops. My suggestion is CNC milling or 3d print in ABS, molding, and forging.[/quote']

Im sorry, but your price on "3D printing" is not accurate, Shapeways offer a very reasonable price for 925 Sterling silver Additive manufactured models.

However I am standing by for a new additive manufactoring center to open which I can access, but it doesn't seem likely that they will be able to make copper models

[theMrDanielPetersen]

After experimenting with several design concepts we decided to try a bit of an unconventional design, with a split flow. The main idea is to have a flow entry after the pump, then again after a radiator.

We have put up a simulation, with a heat source, mainly for the purpose of optimizing the heat conduction, we have not compared to commercial blocks.

Feedback and ideas are appreciated.

[Ghost]

After experimenting with several design concepts we decided to try a bit of an unconventional design' date=' with a split flow. The main idea is to have a flow entry after the pump, then again after a radiator. We have put up a simulation, with a heat source, mainly for the purpose of optimizing the heat conduction, we have not compared to commercial blocks. Feedback and ideas are appreciated. [/quote']

I suppose you know this but water flows VERY fast through a water cooling loop. The temp difference between any two parts of a loop is typically within a couple of degrees.

A side effect of splitting the block in two is one side might be warmer than the other by 1 or 2 C which means that hotter side it will deteriorate at a different rate. Also its worth noting that the chip is long and narrow in a 3#70k so you need more cooling in that area. You have a divider there right now.