In theory -What is the best CPU block flow?

[ZapWizard]

I have been planning a water cooled system for a long time.
I want to make my first watercooled PC project.

I want to make my own design for a water block.
So what would you consider the perfect design?

One thing I have always based my plans is that if you don't use active cooling (peltier, fridge) then the only thing that matters is how fast you remove the heat. (not how cold the water is)

Here are my theories:

---------------------------

Horizontal flow: (long ways across the chip)
This creates a slight gradient of hot to cold as the water passes from one side of the CPU to the other.



---------------------------

Virtical flow: (From top to bottom of die, sideways)
This has the same gradient, but not as wide of a tempature range



---------------------------

Center down flow: (from above die to the surface, and out to the edges)
This creates hot spots in the corners of the die



---------------------------

Outside-in then up flow: (from edges then up out of CPU Block)
This to me seems like the best option.
You get cold flow from four sides of the CPU, and imediatly remove the heat upwards away from the CPU, leaving just one hot spot.



---------------------------

Is this all even worth it?
I plan on trying some extreamly high overclocking.

I want to take my 3200+ from 2.2Ghz 400Mhz FSB to at least 2.8ghz (mabye 3ghz) with a 500Mhz FSB

[jaydee]

Originally posted by ZapWizard

So what would you consider the perfect design?

http://forums.procooling.com/vbb/sho...&threadid=6666

About as perfect as it can get at this point in time.

[ZapWizard]

Originally posted by jaydee116
http://forums.procooling.com/vbb/sho...&threadid=6666

About as perfect as it can get at this point in time.

Which way is what on that?
(Middle-in or Middle-Out?)

[jaydee]

Originally posted by ZapWizard
Which way is what on that?
(Middle-in or Middle-Out?)

Middle is the inlet. The water shoots through the pipes (which actually fit into those cups in the base) and creates an impingment effect. Best water block on the market today.

[g.l.amour]

and can be 'easily' reproduced #rotor style

[Cathar]

Originally posted by ZapWizard
Which way is what on that?
(Middle-in or Middle-Out?)

It isn't any of the methods that you describe above.

The cooling effect is distributed fairly evenly across the entire section that has the holes in it. The design can be scaled up (and I have done so privately) to cool peltiers and the like, and still provide a fairly uniform cooling effect.

I guess you could describe it a "a distributed array of in & out points"

[maxim]

Originally posted by g.l.amour
and can be 'easily' reproduced #rotor style

what did you have in mind? where would you put the jets in using the #rotor? and what about the cups?

[starbuck3733t]

I knew it. Eventually Zap would migrate here

Looking forward to your WC rig man, keep the BT crew posted!

[Bugaloo]

probably gonna get banned for this, but if you want som extreeme ocing you can buy a vapo or a prommie and make watercooling to nb and gpu

[tuxp3]

the wb 'Radius' (never bulit) would be (in theory) the best wb possible and it makes sense, the surface area is very high the thickness is 2-3mm base (decent) and the flow path is not overly good or bad and it leaves plenty of room for flow to outlets..
how ever that is impossible to bulid so a design like the one mentioned is a good wb design that uses impingment to cool, im going to design one thats somehting lieks radius's but more drimeled good-ness and beginner's designage, something a bit less complex. but of couse it will be top out..
thats just my 2 cents

[8-Ball]

Originally posted by tuxp3
im going to design one thats somehting lieks radius's but more drimeled good-ness and beginner's designage, something a bit less complex. but of couse it will be top out

As far as I know, the Radius is "TOP IN" rather than top out. In fact, most god blocks have a central inlet with outlets to the sides.

8-ball

[Dieter@be]

Originally posted by tuxp3
the wb 'Radius' (never bulit) would be (in theory) the best wb possible and it makes sense, the surface area is very high the thickness is 2-3mm base (decent) and the flow path is not overly good or bad and it leaves plenty of room for flow to outlets..
how ever that is impossible to bulid so a design like the one mentioned is a good wb design that uses impingment to cool, im going to design one thats somehting lieks radius's but more drimeled good-ness and beginner's designage, something a bit less complex. but of couse it will be top out..
thats just my 2 cents

Tell more about that block, although it has never been built, you do have some drawings or shemes or something?

[tuxp3]

errr TOP IN i meant. the design i was thinking of would be some think like radius's "star" fins but instead of fins they would be rows of holes and the center is a cluster of holes... the area between the rows of holes would probly be a channel to allow the water to escape to the sides (drimmel) but this is a idea i have been toying with, nothing in stone yet.

[bigben2k]

Originally Posted by Dieter@be

Tell more about that block, although it has never been built, you do have some drawings or shemes or something?

See the link in my sig, for info on Radius.

Yeah, there are easier ways to make something like it.

[Dieter@be]

Oh yes the radius, I already found out about your block a while ago

[pdf27]

The problem with the water leaving the the block in the centre is that you're probably going to get some stagnant/slow moving water on the surface of the chip, right next to the centre of it. Since this is the point where the heat generation is greatest, this strikes me as a bad idea.

[asfd]

Originally Posted by zapwizard

not how cold the water is

If I remember my physics class how cold the water is does matter because the rate at which heat is exchanged between to bodies depends of their temperature differencee: The higher the delta T the faster the heat exchange...


***I dunno why but this thread popped on top of the list even if it's an old one...

[maxSaleen]

The higher the delta T the faster the heat exchange...

That's correct. I think Zap was just trying to communicate that he doesn't want to use any sort of TEC/phase change to cool the water. He's shooting for extreme ambient.

IMHO, you are very hard pressed to beat the cascade's design, and even if you could, the complexity of making it wouldn't be worth for the .5C you (might) gain. Cascade can also be reasonably made with a saw and a drill press (from what I've read, still haven't had the time to "roll my own" though I want to some day).

For GPU and NB #rotor style scaled to an appropriate size will provide all the performance you need. Once again, reasonably easy to make.

The only idea that I have for you is to mess with the angle of the inlet/outlet. I've always believed that having a 45 degree inlet and a 45 degree outlet will lead to the smallest pressure drop and may actually increase heat dissipation across the block's surface. This wouldn't benefit the cascade, as, from what I can recall, much of its cooling prowess comes from a "re-circulation" effect of the water exhausting off of the cups in the base (the water hits the cups, curves upward, and then circles back to hit the block again at the wall between the cups).