As Cathar pointed out, the block isn't flow restrictive, but works best at > 3 lpm (liters per minute). It'll run poorly under 2 lpm.

Yes, it's evident that this kind of WB need a certain flowrate...;)

But, how could we improve its efficiency ?

The first idea, is to drill the holes straight through the baseplate, for a direct die cooling approach.

Otherwise, all I was ever able to come up with, is this:

Allow me to modify the "poorly" word to "less gracefully". It'll still perform well, but it would appear that the "drop off" in performance is sharper than with an impingement/micro-channel mix design, ala the White Water. I also have yet to properly quantify the extent of this effect, if even it's possible given my testing equipment. I'm just extrapolating on what I'm seeing.

Cathar

As we can see for the LRWW in the Bill Adams' test ?

What do you think about my 2 nozzles ?

http://perso.wanadoo.fr/webzeb/Buse2.png
61Holes (6 partialy hidden)
Hole diameter : 1/0.8mm
All the holes cover an circular area of 12mm (I use 12mm tubing)

http://perso.wanadoo.fr/webzeb/Buse3.png
19Holes
Hole diameter : 2/1.5mm
All the holes cover an circular area of 12mm

http://perso.wanadoo.fr/webzeb/Buse3.png
I made this one today, but with 1mm holes :
http://perso.wanadoo.fr/webzeb/19-1.jpg
http://perso.wanadoo.fr/webzeb/19-2.jpg
I also made the same with 1.5mm holes.;)

And I made my firt model with 61 1mm holes, but I missed several Holes...:(

The holes area is only 14.9mm² !:eek: (if we considère that the real dimensions are exactly the same as in theory).

Is it enough to keep a correct flowrate ?:confused:

My system : BigMomma, Eheim1250, very long tubing (approximatly 6meters of 12/16mm tube)

You shouldn't worry about how the inlet is going to fit over the hole area: don't let the barb decide your design.

If you need to, make the inlet 3/4", and use a 3/4 to 1/2 reducer.

In fact, I want to use only a very restrictive area, just over the core.;)

With a very thick base (Direct die cooling seems to be dangerous and inefficient...)

Thats the reason why I use a low area.

But I designed several version of my nozzle : holes surface between 14.9 and 47mm²...

What's your opinion about this choice ?

Thanks

Honestly, the cups needs to cover the whole cove, and then a bit more. this design was specifically made to reduce the resistance to the heat transfer, so if you don't cover the whole core of the CPU, then you're adding resistance to the heat transfer. Since there's little flow otherwise, that part of the block would perform badly, and will result in a block that doesn't perform as well as it could.

Do you have any core measurements?

ok, so you think that I should cover a bigger area ?

Yes, and don't worry about the pressure drop: the design will work nicely with an average pump.

ok, which area would offer the best results ?

Considering Copper base thickness of course...

The AMD Barton core is ~7.5mm by 14mm. You'll have to dig up the other ones (P4, Opteron,...)

ok, but I can change my nozzle...

actually I have a TBred....

p4 is 30mm x 30mm

little update i changed to my amd rig today and found the cup block is 4.5C lower than than a maze 3 instead of 1.5 lower on my p4 rig.
the p4 is an older 1.7 with 1.7v volts and the amd is athlon 1100 both give simular power outputs about 60W

could the temp probe attached to the underside of the p4 be my problem ?

has anyone looked into the differance in core temps and the underside of the p4.

Maybe the contact between cpu/wb is bad, and so you have a better result with the P4 because it offers a bigger contact area !;)

Do you have the temp probe attached to the bottom of the AMD? If not you should.

yes i do have a probe attached to the under side of the amd.

i have been working on a joystick / thermistor probe system with some software i have written in vb.

on the p4 i have tested the maze 3 and my diy cup block for the past week with this system. and found the cup block 1.5C lower.

today i put the amd rig on and tested both again 3 times.

here is a link to a graph with data i captured today

http://homepage.ntlworld.com/lee_smith/graph2.jpg

it shows the cup block over 4C lower than the maze 3

this is good news but i want to know if the probe is the problem on the p4 or the block design.

i spent over 4 hours calibrating the 4 probes today including the one attached to the amd so i know the 4 probes are reporting temps true to eacn other. It does seem there s a problem with the probes that are sitting in air they fluctuate more than the ones in water and attached to the cpu.

the thermistors have a tolerance of -+0.2C between 0 and 70C.

Aww yes, I remember from the other thread now. :)

P4's are very strange beasts. I'm not sure what the exact deal is. Volenti and I have both seen weird temperature behavior on them. You can get one P4 and its temperatures will be greatly lowered by one block over another. Typically this is for a P4 that was reading fairly high temperatures (>40C) in the first place.

Then you can get another P4, once that tends to read no higher than 35C or so no matter what you do, and the temperatures don't seem to be effected as much.

I'm unsure whether this is due to processor yield creating different chip resistances and hence lower heat, or due to bogus temperature reporting, or what. All I know is that the P4's are extremely hard to get any form of reliable relative information out of them from CPU to CPU.

For the AMD CPU's, it seems to be much more consistent, although for certain types of motherboards and setups, you can still get some weird "non-difference" behavior as well, but this behavior seems to be rarer for AMD's.

i am not using the motherboard for temp readings only the thermistors stuck to the bottom of the cpu's on both amd and p4.

it could be the underside of the p4 is not the best place for a probe.

As long as it's the same for each block you mount it should'nt affect it. Nice temps on the AMD block BTW ;) ...

The temp differential does get bigger as more heat is introduced, so you would expect to see less of a 'gap' between a P4 and a hotter AMD. AFAIK that is :) ...

I'd like to see the difference (C/W) in them in a BillA style test, I think you'd see a bigger gap when it's not 'system restricted' so to speak...

Have you got any pics up of your latest block finished lee?..

yes i have pics on my thread or back a page on this one. I only have 1 block that will fit both amd and p4 just lucky the gigabyte mobo i have has loads of space around the cpu socket