[Some_Idiot2]

Looks great, but how are you going to mount it.
Also, I would locate the outlet barb on the opposite side of the block, instead of right next to it, like this:

This should allow for better temps.

 

[confusi0n]

Quote:

Originally Posted by pyr0m1 Would this be a good CPU block if I were to have it cast? The measurements are correct for 1/2" ID tubing and an Athlon64 CPU. Thanks

Looks like a good idea to try to me, but then im certainly no expert. Have you made sure they wont restrict the flow too much? I think it would be a really good idea to aim the inlet at the cpu die, and provided the base is thin enough it would be really interesting to see the results.

Good luck dude,

John

 

[Ty Auchter]

I agree with Some_Idiot2 -- putting the in / out at opposite side would be the most efficient.

 

[confusi0n]

Quote:

Originally Posted by Ty Auchter I agree with Some_Idiot2 -- putting the in / out at opposite side would be the most efficient.

Forgive me if im being an idiot, but wont the flow at the other two corners be reduced?

 

[pauldovi]

Here is some actual data and equations on heat transfer that I put together. (I'm a physics student) Hopefully this will help you in your project.

Thermal Conductivity of Substances (k in the equation):
(J/m*s*C)
Aluminum: 240
Copper: 390
Iron/Steel: 46
Silver: 420
Water: .57
Air: .024

 

[pyr0m1]

so, for the most efficient block, I would want to use silver?

I would also want the block to be thin?

Also, would I want to set up s-curves to maximize flow from one corner to the next?

Thanks

 

[confusi0n]

Quote:

Originally Posted by pyr0m1 so, for the most efficient block, I would want to use silver? I would also want the block to be thin? Also, would I want to set up s-curves to maximize flow from one corner to the next? Thanks

I'd certainly consider the s curves, but i'd also direct the inlet right over the cpu core; perhaps you could put to outlet barbs in there. I was thinking today of something like fstfrddys second waterbock design, but with a 'honeycomb?' type water path like the one you suggest. All you'd need to do to make sure the flow wasn't reduced is make the stacks (and therefore the block) a little taller. The more surface area covered by the water (provided waterflow isn't restricted), then the better the heat transfer per unit time. I dont think raising the height of the block would have such a negative effect as to outweigh the positive effect of having a greatly improved surface area for the water to 'hit'.

How are you cutting this? I was wondering if perhaps an array of stacks like that could be cut simply with a small drill bit in a drill press, or would the copper be too soft to do it this way?

Good luck anyhows

John

 

[ShawnMcGrail]

Quote:

Originally Posted by pyr0m1 so, for the most efficient block, I would want to use silver? I would also want the block to be thin? Also, would I want to set up s-curves to maximize flow from one corner to the next? Thanks

Silver IS the best, But for the price

erformance ratio isn't very much worth it.

The bottom of the block is best between 3-5mm.

 

[pauldovi]

Quote:

Originally Posted by pyr0m1 so, for the most efficient block, I would want to use silver? I would also want the block to be thin? Also, would I want to set up s-curves to maximize flow from one corner to the next? Thanks

Yes silver is the best, but you need a big wallet! Just come up with some designs and figure out all of the variable. The put them into the equation and see which one comes out the highest.

Or you may not care about getting that technical... Its up to you!

As I think about it, larger surface areas would not help if they are only on one side. Look at the equation. If the surface area is larger on one side that it is on the other, the net effect is going to be as if both surface areas are that of the smaller one. But, there are probably more complicated equations, because that is how heatsinks work...

 

[Higgins]

Looks good, You might want to make the base shallower though.

 

[CrimsonMango]

why not just buy a pre-made block that will probably perform better than that?

 

[thenut]

Quote:

Originally Posted by CrimsonMango why not just buy a pre-made block that will probably perform better than that?

Maybe he just wants to experiment like the other couple of people on overclock.net who made their own block

If theres something that interests you, odds are that doing it yourself, building it yourself, etc. will make you feel 100 times better then relying on someone or something else

 

[pyr0m1]

1 more question for now:

Is it better to use s-curves or a series of thin "fins" or a honeycomb?

 

[confusi0n]

Quote:

Originally Posted by pyr0m1 1 more question for now: Is it better to use s-curves or a series of thin "fins" or a honeycomb?

Whatever has the greatest surface area covered by the water without restricting any flow, and with the coldest water directed at the cpu core. That should be the best bet i would imagine.

John

 

[TylerBello]

Quote:

Originally Posted by pyr0m1 1 more question for now: Is it better to use s-curves or a series of thin "fins" or a honeycomb?

It really all depends. Leading blocks use all 3 methods.

The Swiftech blocks though,rely on the pins method,and they tend to be the best performers(G4,Apogee,MCW series).

The easiest to manufacture at home though be honeycomb,followed by S curves. My suggestion would be to do a mixture of the two(honeycomb in middle of block with S curves branching outwards)

Since you are casting it though,why not go with the pins? They look like the best performers to me.

 

[fstfrddy]

Yeah the honeycomb won't be too good, there is too much metal there and no flow, the coolant would just fill the holes and stay there and flow over the top, you would have to cut between them to create pins.

 

[fiendofmine]

then you should get 2mm of industrial grade diamond and remove the IHM(i think thats what its called, its the metal plate on the cpu)put that in contact with the diamond then make that custom cpu block a LN2 block.
about -40C i think

 

[pauldovi]

Quote:

Originally Posted by fiendofmine then you should get 2mm of industrial grade diamond and remove the IHM(i think thats what its called, its the metal plate on the cpu)put that in contact with the diamond then make that custom cpu block a LN2 block. about -40C i think

No ambient cooled water is going to give you -40C!

 

[confusi0n]

Quote:

Originally Posted by fiendofmine then you should get 2mm of industrial grade diamond and remove the IHM(i think thats what its called, its the metal plate on the cpu)put that in contact with the diamond then make that custom cpu block a LN2 block. about -40C i think

LN2 boils at -190 or so degree's C, so you could expect cpu temperatures a little lower than -40

lol

John