New Waterblock in progress...
 

After learning much from this site and all the educated people here, I decided to make an original high-flow design straight from my head.

http://www.magicalbox.com/~casey/New...ck%20Plans.jpg

Tell me what you all think. The block features one right angle instead of the traditional 2 right angles in a maze type design. The bottom of the channel will be roughed up with a dremel so that the water is turbulated. The block will be assembled in 3 parts: Left plate, middle channel, right plate. It will be soldered together vertically. More info on construction soon...

It will be clip mounted to the chip.

Interesting. I like ur design. The thing is, u might have trouble clipping that. U cant use a peltier with that. My understanding is u got the thing to work on the core so it will be significantly smaller from the regular joint. If you do that, u would need a strong pump to be able to suck all that heat energy.

That's a nice,elegant design, build it, test it, let us know how it performs.

and don't forget the pics:D

There was a block like that, without the 45° barbs (it was straight-through). It was called micro-something, cant remember. Very good performer.

Like that :

http://odemarie.nerim.net/maze/ddetec-V-Tune2.jpg

or this.... :D

http://3rotor.dns2go.com/images/h2o_...P02835_sml.JPG


taken from here.

Hehe, yeah! I actually got the idea from looking at a large brass 45 angle. The block isnt meant for use on a pelt. The waterflow is concentrated directly on the die, where during the 90, will turbulate on the rough surface. My block is significantly better for flow than this:
http://odemarie.nerim.net/maze/ddetec-V-Tune2.jpg

because this block pictured above has 2 90 degree bends. Mine only has 1 :D so it should be the highest flow design available unless you could fit a straight through design in a case, though the water wouldnt be concentrated on one spot.

the pictured blocks look like core killers, I'd like to see a block big enough to sit on the 4 feet athlons have

That is true, I could just make the base square. I was just trying to shead weight to prevent core crushage...

I like the design, one suggestion though:

This design has little surface area (not necessarily a bad thing), but since surface area is minimal, you will want to compensate not only with increased flow rate, but increased fluid velocity (the true factor impacting heat convection coefficient).

I would make the interior flow dimensions no bigger than the inner diameter of your barbs. The effect on surface area available at the core would be minimal, but the fluid velocity will increase with the square of the reduction in diameter.

As it is now, fluid will actually slow down for the trip through the block, then speed up on exit. Make the faster fluid work for you.

All you would have to do is expand your block a bit so that a plate will extend from the die contact area out to the four corner pads (just a 1/8" thick copper plate or so). It'll allow for better balancing, methinks, and shouldn't be too difficult to do. It might even make things easier if you extend it out further to take advantage of the board screwholes (you gotta admit, that'll be a biznatch to attach).

Good point... although I'm not sure that the original pic here is to scale.

In short, if you're going to have a point in your rig that achieves the fastest speed, and that point is NOT the area over the CPU core, then you might as well try to make it so.

Unless you want to run the numbers, to calculate if turbulent flow is achieved...:D

You might want to try this, and making the water channel at the base wider than the channel at the inlet. This would force the water across the block faster and allow better thermal xfer. Otherwise you'll have faster water flow at the top of the channel and slower at the bottom at the die contact point. Good idea with the dremel, though don't go too crazy with it or you'll provide hidey holes for stagnant hot water. Simple scoring will do it well for you. I'd score the entire inside, though, to allow for maximum heat xfer at all points ... this has so little surface area and water surface area that it will definitely increase heat absorbtion rates.

By the way, are you going to be making this or modding a pipe fitting? If you are going to mill a bunch of these, I might be interested in one. This looks like it can get pretty damn good results for a straight water block. Unfortunately it would suck for pelting (would it be possible to mod a variant for that purpose?). If anything, it sure will release a lot of flow (as if I'm lacking that, hehe), but it looks like there'll be no hot spots in the block near the cpu.

http://www.magicalbox.com/~casey/New.../rawcopper.JPG

Here is the raw copper for the waterblock. The thick 1/2 inch piece pictured will not be used. It will be replaced with another (better) piece of identical thickness. The block will be assembled in 3 vertical pieces (stated in first post).

http://www.magicalbox.com/~casey/New...rawcopper2.JPG

Just imagine another 1/4 piece of copper on the bottom, with the hose barbs drilled into the sides at 45 angles. This lets me "play" with my center "core" slug. I do like the design posted by airspirit. Very good idea, thanks for the contribution. I'll do it a little more curved than the square though. Im planning on drill pressing out the base of the water channel, so ill just use the circle curve of a drill bit to make the center "thingey" more unrestrictive.

any other ideas?

Let's see...

Let's assume that you're basically setting up a 90 deg elbow here, for practical purposes. What kind of pump do you plan to use? What kind of flow rate do you expect to achieve, or are targeting?

I'd follow up on airspirit's idea by sizing the little extension specifically for the flow rate that you're shooting for. What pump did you have in mind?

Of course your baseplate thickness should be optimized for that flow rate. Where on earth did you get that odd copper piece?

In another thread, I advanced the idea of having perfect rows of square pyramids (for another purpose though). I think you might want to try it, instead of the "drill bit circle": just use a file (if you don't have the power tools).

Are you going to braze the whole thing together?

Yep, it will be brazed together. This block is being made for my brothers "hairdryer sounding" computer. He will most likely use an eheim 1250 pump.

I have no flow rate numbers to shoot for. I was just trying to improve on my last home made block.

The odd copper piece is actualy a cutt off end of "bus-bar". Its from a re-vamped power plant. That bar carried in excess of 2400V at over 500A, so its about the highest grade of copper you can get.

I was going to try airspirit's idea, but I was going to smooth it out using precise drilling, dremeling, and filing.

Ok, an Eheim 1250 will top out at 317 gph, so if everything goes well, you could reach 200 gph, which falls in that range where anything more is insignificant, and anything less is pointless...

1.2 MegaWatts? Yikes!!!

Now, what's the ideal plate thickness for a 200 gph block?

I was thinking around 3 or 4 mm. Does this sound reasonable? Any experience from the pros would be greatly appreciated. What is the base on a maze 3 or a TC-4 or any other high performance waterblock?

If you're shooting for 200 gph, 5 mm would be best. If you only get 100 gph, then 12mm would be preferable.

Check it out

I would stick with the 5mm thickness.

This thickness will help spread the heat thrueout the block, and not have a concentrated center hotspot. I had a block with a very thin base. 2 to 3 mm. and got worse temps. My best guess is that I saturated the material with heat faster then it could move it. so in turn the core got hotter. But with an identical block with 5mm base I seen a very good sized temp decrease. My guess is that the heat was able to be pulled away buy the rest of the blocks material. but when I moved up to 7 mm i noticed that overall system temps ran higher.

just my observations.

That would be in line with a 200 gph flow. What pump did you use, Fixittt?

The pump I used was a HUGE rio powerhead.

I think it was a rio 800, but im at work, I cant check. its the pump I use now for the flood coolant system on my CNC mill

Dang, I thought my rio 1400 was small... :shrug:


Ill be going with the 5mm base after the kind suggestions of all you intelligent people. Ill keep you all posted with temps and pics of construction.

How will the 'turbolence-creator' look like? holes drilled in the bottom, or?