Direct-Die water cooling - Page 5

Gizmo · #81 (**permalink**) 2nd May, 2004, 09:12 PM

Well, that lasted all of what, a month?

I can't help it; I like to tinker.

I took my block back out, tweaked it, and I'm back to running my Direct-Die block again. Main reason being that, although the DangerDen RBX block works VERY well, my Direct-Die block works slightly better (with the tweak that I made). Number two reason is that with my Direct-Die block, I can't chip my CPU cores the way I did yesterday with the RBX. It's not physically possible. While it <IS> a pain in the behind to change CPUs, I can live with that.

Daniel ~ · #82 (**permalink**) 2nd May, 2004, 10:28 PM

You lft out "because I can":O}

Gizmo · #83 (**permalink**) 2nd May, 2004, 10:45 PM

Quote:

Originally Posted by Daniel ~

You lft out "because I can":O}

Oh, that. I figured that was self evident. Thank you for proving me right. LOL

purplemohawkman · #84 (**permalink**) 3rd May, 2004, 12:46 AM

Can't make mods or tweaks with out pics, hint hint ;-)

Gizmo · #85 (**permalink**) 3rd May, 2004, 01:22 AM

Yeah, yeah. I know. I'll tear it down far enough tonight so that I can get a couple pics.

Gizmo · #86 (**permalink**) 4th May, 2004, 10:01 PM

Here's a pic of the new nozzle configuration. I'm not quite sure why this picture is in black and white, and I know my photography stinks, so just cut me some slack on that, ok?

This design is about 5C cooler than my RBX under the same conditions, and is MUCH easier for me to make than my previous designs. Given the amount of power I am pushing at this point (about 135 watts), that means that I am able to handle another 50 watts or so of power.

Basically, what I did was I took my 1/2" tubing, and shaped about the last 3/4" of it into a rectangle that was approximately the size of the core (rather similar to my first design, actually). Then I cut the corners with my dremel and folded the long sides into the center until I had a slit about 1/16" wide, while leaving the short sides untouched to form 'ears' which help me seat the jet at the proper distance from the base inside my plastic cube. I then brazed the seam between the folded part and the straight part so that water had no choice but to exit through the slit over the core.

I came up with this approach because I was sitting one night thinking about my results up to this point, and realized that at least part of the reason I had gotten better results with the 4 jet job over the 8 jet job was probably because I had too many jets; the water was getting in the way of itself. I had lots of turbulence, but the warm water wasn't being efficiently removed from the path of the cool water. So I reasoned that the ideal jet would probably be something along this line in design; a stream that covers the center of the core and then spreads out along the surface to the edges, where it is removed.

I could really, REALLY use a simulator, but I don't know enough about hydrodynamics at this point to write my own (as Cathar supposedly did) and all the ones I've come across on the net are either written for Unix and don't have source code, or have some heinous licensing agreement to get the source code, or cost lots of money.

Daniel ~ · #87 (**permalink**) 4th May, 2004, 10:18 PM

Lets get wet! ":O}

http://www.aoaforums.com/

danrok · #88 (**permalink**) 5th May, 2004, 07:16 PM

Any chance of a pic from a different or wider angle?

Gizmo · #89 (**permalink**) 5th May, 2004, 07:38 PM

Mmmmmmmmmmmmmmmaybe.

What are you looking for, specifically? Bear in mind that I'll have to tear my rig apart to get the pic, so it may be a few days. I'll be happy to get you all the pics you want, I just don't want to tear the thing apart every three days.

At least, no so I can take pictures. LOL

Gizmo · #90 (**permalink**) 7th May, 2004, 03:07 PM

Well, I got two more procs; mobile 2400+ this time, AQYHA stepping. We'll see how they do. I'm gonna tear the rig apart tonight and see how they run, and I'll get you some pics while I'm at it, danrok.

Gizmo · #91 (**permalink**) 8th May, 2004, 10:55 AM

New pics. This time they are actually in focus. WARNING! These pics are about 450K each.

Daniel ~ · #92 (**permalink**) 8th May, 2004, 12:26 PM

Much better pics! Still not good pics, but much BETTER pics thanks! ":O}

Just remember. every guy without a camra,...becomes a critic!

danrok · #93 (**permalink**) 8th May, 2004, 12:43 PM

Thanks for the pics! I am now a little less confused.

How does the cube attach to the board/cpu?

What is it made of?

Gizmo · #94 (**permalink**) 8th May, 2004, 01:14 PM

The cube is made of 1/8" plexi. The four sides are each 2" square, and the top and bottom are 2 1/8" square. The inlet, outlet, and cpu seat are set in the exact center of each face. Originally, the inlet and outlet were both the gray PVC 1/2" MPT connectors. However, the inlet connector had to be torn apart when I changed the nozzle arrangement, so I replaced it with a piece of regular threaded PVC so that I could then thread in whatever pipe fitting took my fancy.

The cube attaches to the cpu via an extremely complex mounting mechanism

:

two pieces of 1/2" wide 1/8" steel, 3" long run the width of the block and attach via 3" 8x20 machine screws to the mobo (very similar to the way the RBX WB attaches, actually).

You don't see it in the picture because I didn't have it on there yet, but there is a water gasket made of 1/8" closed cell insulating foam which goes over the bottom of the cube. This seals the cube to the CPU so that there are no water leaks, without making the cube installation permanent.