WW kicks ***: BillA test results.

[EluSiOn]

Yeah... WW start popping out everywhere... right after christmas, WW was only #75 out there.. but by now it is like over #110 out there...

I posts at AOA forum about Epox 8RDA mobo and over there already 3 other people posting their WW pics and all with very success OC results

I think soon we need to have WW owners Year Books and WW owners reunion party etc.

 

[bigben2k]

Now that we can finally see just how well the WW performs, it's time for the "Attack of the Clones". I'm looking forward to seeing how well the DIY versions compare to "The Master".

The clones are out there...

Here's my variation!

 

[SysCrusher]

The clones are out there...

Here's my variation!

I was just thinking of doing something like that. Looks like you beat me to it. I wasn't going to run my channels though the center though and try a wacked out way for a jet impingment. Using three .25 inch copper stacked, an adjustable nozzle thats threaded in the second layer and two outlet barbs and inlet barb on the top layer.

 

[bigben2k]

Sounds promising! Keep us posted!

 

[Hoot]

I am very impressed and my whole attitude has undergone the proverbial "paradigm shift" regarding flow volume versus impingement in a water block intended to cool just a CPU core. The whole impingement approach to water cooling has yielded a lot of interesting reading in the past 6 months or so.

Kudos to Cathar for merging the factors to yield a water block that is knocking on the door of the TIM junction barrier.

Now, to scour the various pump manufacturers for an affordable, high pressure, not-too-loud, best match for the White Water. It troubles me to think that most pumps we have come to associate with water cooling are probably not the best design for impingement based blocks if we care about efficiency. I'm already eyeing my power steering pump

Hoot

 

[bigben2k]

Agreed, 100%!

In designing Radius, I've come to many conclusions, but the most important one is that a jet inpingement (or something similar to it, like in Little River - White Water) is the only way to go: it increases the cooling ability far more efficiently (cost wise) than any other solution. Forget about turbulators, and high flow rates: there will never be a pump affordable enough to provide the kind of pressure drop/flow rates where turbulators would be beneficial. Pins or fins are a must (fins are better). Baseplate thickness is critical. Aluminium is still a viable option, as long as the galvanic corrosion effect isn't there (it must be avoided at all costs).

 

[SysCrusher]

Sounds promising! Keep us posted!

LOL I lack the proper tools so all I got is a dremmil and drill press but I have been trying. I'm just worried about the center being strong enough to hold the clamping pressure. Works ok with aluminum with a slight taper to the base, gives it a little extra strength. The nozzle is the hardest part for me. What I have tried with some success is to position my nozzle so it's maybe 1 mm off the base. Seems to work until I get more than 3-4mm away. With just a open flat base, no fins and a slight taper to keep the bottom from warping, 6 mm up to 10mm works best. With out the nozzle I lose 2 - 3C. With the fins and channels they build pressure so once you add a nozzle you have to find the right size that builds more pressure and velocity than the fins and channels would. It's kind of like phasechange where there is a low pressure and a high pressure side. If both sides are the same or the pressure at the outlet end of the nozzle is higher than the pressure being built in the nozzle it will not work correctly. Imagine a water pistol with a micron hole.

 

[Penguin4x4]

So I'm guessing an Iwaki MD40RLT would be perfect?

http://www.marinedepot.com/a_pu_iwaki.asp?CartId=

 

[Penguin4x4]

Hmm, wait, kinda confused here...to get the best results you need a high head pump with moderate flow or a low/medium head with massive flow?

 

[Cathar]

Hmm, wait, kinda confused here...to get the best results you need a high head pump with moderate flow or a low/medium head with massive flow?

Depends on what you have in your setup. The WW block will work best with pumps that have high head/moderate flow, than low head/high flow pumps.

It's all relative though. We're talking less than a 0.01C/W difference between a 2m head pump like an Eheim 1250, and a 7m head pump like an Iwaki MD20-RZ. By the time you get to 6lpm (which an Eheim 1250 will do for you quite happily) then you're starting to approach the point of rapidly diminishing returns.

 

[wildfrogman]

That sure is great performance even at pretty low flowrates on the WW block. I bet most people with the 300ish gph pumps should get really good performance even at a restricted gph flow. Higher pressure pumps like the Iwaki MD40RLT would be perfect except for its price for some but yeah, high head pressure is whats needed. You also would need high gpm and high head to get the best possible performance, although you may not notice the difference sooner or later.

 

[Rainman]

Cathar.. I have just ordered..or should I say..enquired about purchasing two of these ww blocks for testing here on my own test rigs.

Is there anywhere I can see a pic of the inside of the block?


Thanks,


R.

 

[Frodo Baggins]

Congrats cathaar

 

[Penguin4x4]

Depends on what you have in your setup. The WW block will work best with pumps that have high head/moderate flow, than low head/high flow pumps.

It's all relative though. We're talking less than a 0.01C/W difference between a 2m head pump like an Eheim 1250, and a 7m head pump like an Iwaki MD20-RZ. By the time you get to 6lpm (which an Eheim 1250 will do for you quite happily) then you're starting to approach the point of rapidly diminishing returns.

Wow, a CPU block optimized for the Eheim 1250, Kewl!

 

[Les56]

More ramblings:
Implicit in the production of the graph in 1st post is the placement of the the White Water on my "T/W" v "h(bp to water" graph*.Whether would obey the graph if bp dimensions were altered is a dunno**.

This is a consequence of using the Jet Impingement Heat Transfer Coeffs "h(jet)" as the Convection Coeff.These are 2.5x those obtained at the same flow by conventional flow in channels calculations(e.g Sieder Tate)

That it may be possible to get Jet Impingement Heat Transfer values on the walls of a channel parallel to the Jet is to me a revelation.
However this is the only way I can predict the experimental (C/W) v Flow graphs. Perhaps someone can find an alternative interpretationof the results.

* First shown here: http://forum.oc-forums.com/vb/showthread.php?s=&threadid=159340
** Thickness : possibly would. Area : probably would not because of the "r" dependence of "h" in Jet Impingemet.

 

[Cathar]

Les, I have always loved your graphs, but I always suspected that something else is happening beyond what can be simply predicted.

There are multiple regional "zones" going on within each channel. The water moving down the channel walls is typically travelling at 3 to 6 m/s (for 5 to 10lpm flow rates). That in itself must be taken into account.

Then the impingement takes place on the base. I suspect that the impingement effect on the walls towards the bottom of the channels is acting at close to direct impingement convectional values due to the pressures and turbulence in this region. The water flow hits the base, and then wants to go in all directions but is constrained by the channel walls. This, I believe, is giving a secondary impingement effect acting on the walls close to the base, perhaps affecting the lower 1.5mm portion of the walls (which is the hottest section of the walls - which is exactly what we want).

Then after our "stagnation zone" (if it can ever be called that given what's going on) the block basically acts like a regular channelled block, but I believe that this takes place after about 6mm each size of the centre of the inlet nozzle.

The trick here is to try to figure out how all those separate effects are working together, and then work that into a sane model that the calculators you're using can predict something from. I suggest perhaps a limitation of the calculators for predicting something of this nature.

 

[Les56]

Les, I have always loved your graphs, but I always suspected that something else is happening beyond what can be simply predicted.

1) There are multiple regional "zones" going on within each channel. The water moving down the channel walls is typically travelling at 3 to 6 m/s (for 5 to 10lpm flow rates). That in itself must be taken into account.

2) Then the impingement takes place on the base. I suspect that the impingement effect on the walls towards the bottom of the channels is acting at close to direct impingement convectional values due to the pressures and turbulence in this region. The water flow hits the base, and then wants to go in all directions but is constrained by the channel walls. This, I believe, is giving a secondary impingement effect acting on the walls close to the base, perhaps affecting the lower 1.5mm portion of the walls (which is the hottest section of the walls - which is exactly what we want).

3) Then after our "stagnation zone" (if it can ever be called that given what's going on) the block basically acts like a regular channelled block, but I believe that this takes place after about 6mm each size of the centre of the inlet nozzle.

4) The trick here is to try to figure out how all those separate effects are working together, and then work that into a sane model that the calculators you're using can predict something from. I suggest perhaps a limitation of the calculators for predicting something of this nature.

1) Taken in to account.(Different velocities for different slot widths - velocities tranlated in "h" as in Graph1 of 1st post )
2) Use(and need) the full wall height in calculations. However since "Finning Efficiency" decreasing as the Convection Coeff increases the higher regions contribute less. This taken into account by Kryotherm
3) Calculations for the 3mm slot are based on the average for radii from 0 to 7.5mm( the average beyond 7.5mm decreases for 3mm Round Nozzle - from another thread* - http://www.jr001b4751.pwp.blueyonder.co.uk/JetArea.jpg ). If decide to refine model will have to consider the full h versus r curve.
4) Yes, use of mathematics or (god forbid) CFD is probably needed to consider the full h v r curve. I only do sums and would limit to using a 2 parallel heatsinks model.How does your model cope with the Convection Coeffs.? The last I heard you were still looking - http://forums.overclockers.com.au/showthread.php?s=&threadid=91695 .
However the crude model does give(in my opinon) excellent if coincidental agreement.

* http://forum.oc-forums.com/vb/showthread.php?s=&threadid=161124

 

[UnLoadeD]

Wow all this tech talk has my head spinning. As I try following the convo something comes to mind. You'll have to excuse my non technical way of posing it. I'm wondering how much impingment is taking place at the tops of the fins? There's got to be some going on, right? This brings up 2 issues that really have me curious. I gather that the tops of the fins are too far from the heat source for any impingement taking place there to really be much benefit, but would the impingement at the takes reduce the effectiveness of the impingement where its most useful (closer to heat source)? To me it seems there are 3 areas of impingement: 1-tops of fins 2-bottom of channels 3-walls of fins. Would the turbulence created at #1 interfere with #2 and #3? If all of this is so far reasonable I wonder what would happen if it were tweaked a bit by cutting the height of the fins directly below the nozzle? I don't know what optimal height for them would be, but I'm sure you guys would have an idea. It seems to me like this would give you all the impingement taking place where it's most useful, also it would increase the suface area at the hottest spot. If all this is way off I'm sorry. I know my "logic" has let me down before when it comes to this stuff. If it's possible this might help, I would like to see it attempted by somebody willing to mod their WW, which I am not 8)

peace.
unloaded

 

[Les56]

Fin height down Pressure-drop up.
For 5mm down to 3mm Pressure drop in Channels treble.

My analysis
According to my model,at 8LPM no detectable effect on Impingement cooling contribution(h~ 93,000kw/m*m*c)
In neither case would cooling from conventional flow(36k for 5mm and 42k for 3mm) kick-in to beyond the 30mm long channels(Impingement still 60k ish).Of course model could crap. Any other offers?

Net effect Presure-drop up.

 

[Les56]

Unloaded
Sorry mis-read.
You are suggestsing cutting height(or even elimiating) below the inlet.
I was open minded as whether the fins aided or hindered Impingement until I could not account for Billa's results by any other way than equating the Heat Transfer to "Impingement Heat Transfer Coeff" x "Fin area(Fin efficiency adjusted)". Now until I learn otherwise I consider the full fin height to be an advantage.
My initial answer still stands for the overall reduction of the fin height.