Clip-0n Athlon WB design (another one)

[Gooserider]

I'd like to get some opinions on the following WB design which I'm planning for my next system.

The system I'm designing for has to use a clip on type mounting, using the lugs on the CPU socket. I can't use a 4-bolt style because the mobo I'm using doesn't have the holes for it.

I see this as constraining my design in some ways. I don't see an easy way to do a center input for instance, since the clip will be using that real estate for applying mounting pressure.

Instead, I'm planning to do a high flow serpentine pattern with a bunch of turbulence pins in the center part of the passage over the die area. I'm using 1/2" inlet and outlet barbs. My mill is manual, so I can't do the level of fancy curves and such that the CNC guys can put out. This design is all straight lines, so it should be pretty easy to make.

As I interpret the AMD Athlon cooling design docs, I have a maximum width (between the clips) of 63mm (2.44") by up to 80mm (3.2") long. Are these dimensions right?

Assuming that they are, my block will look about like the following <UGLY ASCII ART ALERT!!>

--------------------------------------
|.....................................|
|..|-------------------------------|..|
|..|.........................|.IN..|..|
|..|.........................|BARB.|..|
|..|.....|..X..X..X..X..X..X.|.....|..|
|..|.....|.X..X..X..X..X..X..|.....|..|
|..|.....|..X..X..X..X..X..X.|.....|..|
|..|.....|.X..X..X..X..X..X..|.....|..|
|..|.....|..X..X..X..X..X..X.|.....|..|
|..|.....|.X..X..X..X..X..X..|.....|..|
|..|.....|..X..X..X..X..X..X.|.....|..|
|..|.....|.X..X..X..X..X..X..|.....|..|
|..|.OUT.|.........................|..|
|..|BARB.|.........................|..|
|..|-------------------------------|..|
|.....................................|
--------------------------------------

<END UGLY ASCII ART>

The block will be about 3/4" thick, with a 1/4" copper top. The bottom will be about 3/16" thick (3-4mm). The walls will be about 5/16" thick, mostly so I have room to drill and tap for #6-32 screws to hold it together. (I'm thinking 3 screws on the short sides and 4 on the long ones (mostly to clear the clip hardware) should be appropriate.) Side question on the screws, Is Stainless OK to use for fastening screws, or should I look for brass or copper screws?

The channels on the outside will be about 1/2" wide to minimize flow restriction.

The X's represent the turbulence area, which I plan to make by using my smallest mill bit (1/8") to cut as many passages as I can at a 45* angle to the block in each direction with about equal amounts of width for the channels and the pins. I think I should be able to get at least 5 sets of channels into the space which will give me a bigger flow area than the side channels, but a lot more turbulence.

What do people think? Will it work reasonably well?

I know it probably won't even come close to some of Cathar's blocks, but I don't think I need to. I would like to think I can match some of the commercial efforts. FWIW, I am planning to use a high volume pump such as an Eheim 1048 or 1250 or equivalent, so what I don't get in cooling quality, I should make up for in volume.

Comments and suggestions are welcomed and appreciated...

Thanks,

Gooserider

[#Rotor]

well then allow me to show you how i got that problem to go where no problem has returned from, yet...

[Gooserider]

NICE! that is a neat solution, though it gives me some questions that I hope you don't mind my asking...

1. What type and size of wire?

2. How do you get the tension uniform and ensure the cooler is square on the CPU?

3. Judging from the holes, I assume you originally intended to grab all 6 lugs, is there a reason you didn't?

4. What are those red and black wires that look like they're attached to the blocks?

I'm involved in a discussion about cooling flows and plumbing patterns over on http://forums.procooling.com/vbb/sho...&threadid=6385
I notice you have a dual CPU mobo, plus other things. What is your plumbing pattern and why? I was originally planning to have several circuits, including the CPU's in parallel (w/ 1/2" tubing) and a third circuit for drives, NB, etc. w/ 1/4" tubing. The suggestion on the other thread was that there was no real advantage in this as running the CPU's in series would only result in about a 1/4th degree C difference between them. What is your opinion? (FWIW, I will not be doing pelts, just WC to ambient)

I'm also discussing cooling hard drives, and noticed the nice hard drive block on your website. I think I mentioned it, but I hadn't noted your site, so didn't give as much credit as I should have. The thread is http://forums.procooling.com/vbb/sho...&threadid=6149

On the block itself;

1. How do you make the water loop over the entire plate and keep the water from just taking the first couple of cross passages between the inlet and outlet?

2. How do you avoid problems with the barbs leaking at the seam between the two block halves?

3. I assume the block goes on the drive cover, how do you attach it? Do you use any TIM, if so what kind? Do you take the labels off the drive?

4. It certainly won't be as pretty as your block, but I was thinking of just soldering a maze of 1/4" copper tubing onto a 1/8" plate. Would that work reasonably well?

On a more general note, there is a debate between those advocating putting the cooler on the sides of the drive, and those (like me) who say it's better on the top cover, or possibly on the bottom over the controller card. What is your opinion?

Hope I haven't bothered you with to many questions...

Gooserider

[#Rotor]

I used stainless steel fishing tackle wire, the kind you use in making spinner baits and stuff....

getting the tension uniform is done by a process called "tuning the block" it is a wee bit time consuming, but it is recommended, whether you use this method, or the through hole method.

those blocks where designed in the good old days of slockets and K6-2 CPU's, where the 6tab socket-A's just started to role off of production, I had to provide for both. using all 6 is not required at all, the corners is way more than enough....

those red and black wires are the electric leads of two 172W peltiers, one in each block...

heheh ya, those hard drive blocks are something else, they sure way a ton, (almost)


for the Block

I am actually bargaining on the water doing exactly just that, Taking the shortest path across....... Why, because the faster I can get each molecule to get in close proximity to the hot sides or pins in the block, and then the hell outa there, the faster a fresh cool molecule can come in after it... I also use a design principle I call VDFC.


leaking at the seam, that problem gets solved by having 2 bolts on either side of where the barbs go, to insure mechanical rigidity, also that I thread the barb holes a tad bit deeper, than usual, and then, by using Teflon tape, one do not need to have as much torque on them while still getting the perfect seal....

I have some custom made brackets, enabling the block, or rather, the drive, to be mounted on the block, I do not remove anything from the drive, nether do I use any paste, It is really not necessary, a hard drive do not produce it's heat fast enough to warrant such measures. The sheer size of the transfer junction, even with no thermal compound, is way more than enough to do a perfect job at cooling the drive....


the sides of the drive is good.... yes, but you then need to have two blocks, to insure, thermal uniformity.

I prefer the top, remember that the insides of the drive, still get cooled by the air that circulate while it is spinning, and the top plate is as good as one can hope for, to get to that air. But to be honest, you can even just slap your maze block on it as is, anywhere you can find a flat spot.... you will be amazed at how much of a difference it will make.....

The only reasons I make my HD-blocks that big, is so I can fit two drives to one block, and for mounting purposes, the block is treated like just another drive, even has the holes tapped to the same spec as those in a drive itself, makes installation a breeze.

[Gooserider]

I used stainless steel fishing tackle wire, the kind you use in making spinner baits and stuff....

OK, that sounds plenty strong enough, any particular size, or does it matter?

getting the tension uniform is done by a process called "tuning the block" it is a wee bit time consuming, but it is recommended, whether you use this method, or the through hole method.

I don't recognize the term, could you point me at a definition or explanation?

those red and black wires are the electric leads of two 172W peltiers, one in each block...

Out of my league though, I don't think I need pelts for what I'm trying to do.

I am actually bargaining on the water doing exactly just that, Taking the shortest path across.......

Makes sense, but I'd think the water in the far end would tend to get hot and stagnant (comparatively speaking) while the cool water took the short cut and didn't do much work. I probably would have left one row of holes uncut down the center for most of the length of the block to force the water to make a trip up and back. However whatever works, and as much mass as those blocks have it probably doesn't matter that much.

I also use a design principle I call VDFC.

Care to define? I know you advocate high flow volumes, is this something to do with it?

leaking at the seam,

Nice design, I hadn't realized the blocks were so thick.

I think I may end up sticking with my idea of doing a tube on a plate though. I think I've found the case I want to use, it's a dual chamber server case with a layout like the YYC case reviewed here: http://www.procooling.com/reviews/ht...ase_review.php
It will be big, but at least it has wheels

I figure that I can hang the rad off the back over the exhaust fans on the drive side. The upper fan will pull accross the drives. If I use a block like yours that flow will be mostly blocked. But if I put a thinner plate (I'm guessing ~1/2" if I put 1/4" tubing on a 1/8" plate) then I can space the drives in every other bay, with a gap between them. Then the water cools the top and inside of the drive, and the air will funnel through the gap and cool the controller board. Some might say I wouldn't need to cool the drives at all; but I think it would be better to have the heat in the radiator with cool air blowing through it than to be blowing drive warmed air through the rad.

Gooserider