My waterblock design

[Dieter@be]

Hello guys,
I'm designing a copperblock to cool my dlt3c 1800+
It 'll be made with a CNC machine, but for now, I'm still in the thinking/designing fase.
I'll show you some drawings I made, and explain how it works. (The block has a top part, and a bottom part,
both in copper, i made a drawing of both, viewed from the side and the top)
(btw I dont know yet which pump, what fittings and rad i will use)




So this is the top, on which you see the holes to mount the block to the mobo, and to hold the top part and bottom part
togheter. But what's more important: Below the inlet, there is an "adapter" to change the waterflow. Just below were you screw in the fittings (fittings are drawn above the block, so not screwed in) the diameter is the same as the inner diameter of the fitting and changes from a round opening to a square opening, with the dimensions of the core. So the water will strike the bottom part of the block in the shape of the core. The "gateway" to the outlets is nothing special.




Here you can see how the "adapter" works.
Btw: The distances between the holes are 66,09 and 36,14 actually




note: the arrows and lines above the fat line aren't offcoure part of the block ;-)
So here you see how the water comes down, and strikes the finn's of the block, and is led to the outer outlets (i made
the bottom profile in that rounded shape to improve the water flow, you notice i marked the |out-to-in| distance, i think that the bigger this distance is, the better the water can flow (since the profile is better then))
The fins end just below where the outlet starts. (which is more clear ind the drawing below)




Here you see in - - - lines the shape of the core, so the area where the water strikes on.
The water is led away to the sides and go to the outlets.

What do you think of the block?
Btw I need to know how the core is placed in comparision with the 4 mounting holes. I heard the size is about 11x7mm so that will do.


Topic at Xtremesystems

[Pears0]

Hi Dieter@be,

Im in the same boat. I not long started a thread called A noob water block design. Input and ideas welcome! Design ideas aill b posted on there soon, hopfully with a nice CAD interpriment as well thanx to jaydee116. If u want to have a look at them for ideas ur mor than welcom. Though it is a bit elaboret of a design than yours and probly wont even need to look at a mill. All the best with everythin neways

Pears0

[Pears0]

i other thing i forgot to say, i found all the core specks on the AMD web site though i cant rember eactly which artical, and i dont have a pdf reader to find out which 1 it is at the mo so i can't link u to it, soz, but i hope this helps!

[Dieter@be]

I'm not afraid of some research

[Dieter@be]

Yo I'ld like some feedback about the design, to discuss about the principles and flow and such

[rocketmanx]

I'm new to this design thing as well. But I'm science nut so this stuff really gets me goin'

The principles are pretty easy and the formula for heat conduction will show you what makes heat transfer better. The formula explains why Cathar's design works so well. Here's the formula, beyond it it's a matter of inovation, and Cathars looks hard to beat, but all designs can be made better-

Heat Conduction Formula

Anything that makes the left side of the equation bigger (ie: thinner barrier, more area) will help.

[Dieter@be]

So that makes:
Heat transfer/time = k*A*tempdifference/copperthickness

with k the thermal conductivity of copper.

That means, to increase the performance of the block:
1)make the copper barrier between water and core as small as possible.
2)increase surface area.

Pretty logic aint it. Although you'll see in my drawings that there is a little "hill" on the bottom piece, above the core, which is not according to 1), but i believe this way the water can get away much better.

[rocketmanx]

The barrier, after I did some reading on jet impingement cooling, is actually a very fine "film" or area at the very surface of the material where the liquid is actually not moving. The jet impingement uses a high pressure jet to blast the surface which minimzes this barrier. Kinda like "approaches zero but never totally reaches it". I assume thinning the material barrier will help as well as it is a barrier.
I'm new to this too and did a lot of searching yesterday after seeing the cascade thread. There are a lot of universeties etc... doing research on microprocessor cooling I found out. So far it looked to me like jet impingement is the best thing they've come up with.
Hope this helps, I was trying to make my own design as well and found out I was going down the wrong path. Seems the science behind this stuff was already pretty well established, now it's down to who can design the one with the best preformance. Kind of a bumber IMO.

[Dieter@be]

1) I think that my system resembles the jet impingement, in my design the water is also pressend on the bottom, but in my design the water can get away better, then in jet impingement blocks...so I think.

2) Is it true dat copper can deform when in use (because the center is hotter ?)

3)Could I do without an O-ring? I think I would just do some silicone between the top and bottom part and then bolt it tightly together...

4) What kind of fittings should I use? (considering the water is pushed on the block in the shape of the core-->need good pressure)

5)What kind of pump should I use?
Eheim 1048 (600l/hour), Eheim 150 (1200l/hour), Eheim 1046 (300l/hour),... so much to chose from

5)Rad:
would one like this be overkill? I think she's so hot
(remember I'm after great performance )

[Dieter@be]

Hellooo.....echooo...

[Cathar]

Originally posted by Dieter@be
1) I think that my system resembles the jet impingement, in my design the water is also pressend on the bottom, but in my design the water can get away better, then in jet impingement blocks...so I think.

The "water getting away better" is not what you want. Let's think of it this way. The easier it is for the water to "get away", means that the water isn't really smashing into stuff and helping to strip away the boundary layer.

By very nature of physics, one simply must make the block restrictive in order to get the water to attempt to break-down the boundary layer.

The challenge of course is making it restrictive in the correct places where the heat is.

Originally posted by Dieter@be

2) Is it true dat copper can deform when in use (because the center is hotter ?)

Yes. When the copper gets warmer above the CPU die, it will very slightly become more convex.

Originally posted by Dieter@be

3)Could I do without an O-ring? I think I would just do some silicone between the top and bottom part and then bolt it tightly together...

Design choice. I like O-rings. O-rings are harder to implement than a simple gasket, but they make for a better mating between two surfaces and are easier to work with when disassembling and reassembling.

Originally posted by Dieter@be

4) What kind of fittings should I use? (considering the water is pushed on the block in the shape of the core-->need good pressure)

Your choice. I favor brass barbs.

Originally posted by Dieter@be

5)What kind of pump should I use?
Eheim 1048 (600l/hour), Eheim 150 (1200l/hour), Eheim 1046 (300l/hour),... so much to chose from

Again, this is a design choice. The pump you choose will dictate the optimal design parameters for your block.

Originally posted by Dieter@be

6)Rad:
would one like this be overkill? I think she's so hot
(remember I'm after great performance )

There's no such thing as "overkill" when it comes to radiators.

[TerraMex]

That's a "multipass" radiator, same as the Maxxxpert cubes.
(One long copper tube with several passes)
It's 1/4" and exceptionally restrictive.
I'd recommend not to use it if you're going to a 1/2" system ou similar. Use a proper heatercore.
It's good for 1/4" and 1/8" (innovatek and aquacomputer) but not more.

And if going for larger ID , larger pump, 1048 minimum.
Also better for the impingement effect.

[Dieter@be]

Then what kind of heatercore? (most performant )

[Cathar]

Originally posted by Dieter@be
Then what kind of heatercore? (most performant )

Perhaps the Thermochill 120.3 if you were already looking at 3-fan cores.

Sorry, I thought that was a Thermochill 120.3 as the picture was so small that I couldn't make it out clearly.

[mad mikee]

Go check out my 'concoction' (search hint). Somewhere in here (forums) is a cross ref to get the diff. manufacturers #s. Just got another for the beginnings of mikeeputer rev 3.0 from Delco, and both inlet and outlet are ROUND (old one had oval holes into the tanks) which I consider GOOD (no restriction) )

[Dieter@be]

I found this log of yours http://forums.procooling.com/vbb/sho...n&pagenumber=1

Remember i'm from Europe (Belgium), we have no Chevy rads here

[Dieter@be]

Originally posted by Cathar
Perhaps the Thermochill 120.3 if you were already looking at 3-fan cores.

I found this review but Bill doesn't compare them with other brands/types. Or is that the only brand that makes such big cores?
(I want the biggest, meanest available )

[bat]

Originally Posted by Dieter@be

I found this review but Bill doesn't compare them with other brands/types. Or is that the only brand that makes such big cores?
(I want the biggest, meanest available )

Do you know what a Citroën C15 is ? It's a french van, as you are in Belgium you should be able to find a core like this. It's huge and with a proper shroud should be an "overkill" core.

[Dieter@be]

plenty pictures of the car, none of the rad ...yet (I keep on searching)

You are talking about the heatercore that is in front between the two first seats to give the people in the car warm air, right?

[bat]

Originally Posted by Dieter@be

plenty pictures of the car, none of the rad ...yet (I keep on searching)

You are talking about the heatercore that is in front between the two first seats to give the people in the car warm air, right?

Right this is it. It's really huge, i got a friend who got one but does not know what to do with. If you were in France it would have been possible to send it to you, but as you are in Belgium it would be complicated. I will ask my friend if this is a C15 or a C25, because I'm doubting but I'm pretty sure this is a C15 one.
If you are francophonic we could talk about it in french

[Dieter@be]

you got a pm