Wow no THAT is a useless picture. The only thing it really shows is the black delrin, and we already knew about that... We want to know the underlying theory!!! I can't think of anything involving storms (excepting thousands of tonnes of water moving) that would be applicable to cooling. Calling a storm the most destructive force of nature is really misleading, I must say.

lol Your probably right. I remember my first design and all I got was "So?" or "Why?". I wouldn't show it at all if it was me up until the last minute. Though I'm itching pretty bad to see how you done it. I know all those little details make a difference. One of those details is not much of a difference but add all those little details and your on to something.

Are you always this negative? Not really misleading, sometimes it's just better to keep the other side guessing.

LOL - Now, now AA. I know you're frustrated.

A storm is not one of the most (mass) destructive forces of nature (on the planet)? Guess you better head down to various states in SW USA and tell them that. ;)

"Storm" is just a code-name anyway. I already have a name for the block. "Storm" is not it.

Storms aren't destructive? Check today's news.
http://www.msnbc.msn.com/id/5054522/

That's a storm. A tornado, as would probably be found in the SW USA, is much more destructive, as is a hurricane. If THOSE are the effects of which you speak (I think they may be) that's an entirely different story. I suppose a storm in Peru (one of those crazy newage ones) would be really destructive as well, I guess, but once again, just a lot of water.
Mostly, but I'm on drugs to fix that.

The fact that I'm bitching about this indicates to me that I've got far too mcuh time on my hands in the morning...

While you've got the dictionary in hand, look up some of the words you just used and tell me that the word "storm" does not appear in any of the definitions...It certainly does in my dictionary.

Look, it's just word play. You're getting way too fixated on it. What? You want me to start spelling stuff out for people exactly what the cooling effect resembles and name it as such so all can run off to their mills and ferret away on it?

Well, I'm guessing it's unlikely you've replicated anything along the lines of an earthquake so I'm gonna guess "Tsunami".

Nice idea with the Delrin btw. :)

Derlin? We used to use it to make embossor dies at stamp works. We would use the Laser to make a design, then flip the design and make a matching die. Then put it in the crimper and it crimps the design into paper. It is a very stroung material although kind a flexable. Jfettig made a block with a derline top and couldn't get he O-ring to seal because the derlin flexed to much. Although the massive o-ring used didn't help either.

Unsure of the specifics of what JFettig was doing. Have no such issues here with sealing. The block clamps together very nicely, although I am using a 1/2" thickness of the material on the top plate. Fairly inflexible (for purposes of O-ring sealing) at that thickness. The stock material isn't totally flat from the shop though, and the pieces do get a run-over with a fly-cutter to flatten them out.

NO inside pics...EVER!
 

WARNING: Water Block will Self Destruct if opened! (you'll LOOSE fingers, not just CUT them, pHaestus!!)

something like that?

.

What about a little glass container with violtile chemicals which mixed together react producing massive amounts of heat (non-explosive to avoid shrapnel) to melt insides if tampered with? Gentlemen, meet Mr James B. ;)LOL
If derline (or however it is spelled) is so flexible why not make a little sharp upstand and use more than 4 bolts to secure tow parts together?

Looks good stew, (I sent you a pm at overclockers)

about the O-ring problem, I was using too thin of material, dont worry about it with the half inch stuff.


Jon

Updates? ?

Check the OCAU website.

Back in Australia again.

The Cascade SS and the G4 have been tested by Phaestus and should be heading back my way. Interesting results with the SS actually, and somewhat puts the regular Cascade to shame.

The G4 test was an interesting scenario. The G4 results, while excellent, weren't quite what I expected (I expected a little better).

Now I test with pure water. Phaestus tests with a 25% glycol mix. I am acutely aware that one of the principles in the operation of the G4 is more heavily affected by the kinematic viscosity of the liquid than some other designs. I had pretty much designed around kinematic viscosities typical of water/alcohol, but a 25% glycol mix is around 5x more viscous than water, and I strongly suspect that this is causing an issue. Certainly in the models, the kinematic viscosity is very important.

For example, in one of the accidental prototypes of the G4, in which the machinists had misprogrammed the block and left off the very things that makes the block do its job, and resulted in performance for me that was somewhat behind what Phaestus saw, basically falling to around regular copper Cascade performance.

As viscosity goes up, the effect of those elements is diminished as they are not "significant" enough to push the goop about in the way that they do with plain water. Get the goop thick enough, and they may as well not even be there.

So the plan is to very slightly modify the design to be slightly more goop tolerant, and also be a little less restrictive. These things were already planned before with a question mark, but with these results, will most likely go ahead.

I just need my blocks back again to emperically confirm the mathematical explanation to the results that Phaestus saw, and this will prompt the final design. As a result, I'll also have to include an explanation about not using >10% concentrations of glycol in one's mix to achieve best performance.

Once that is finalised, I can then get the G5 (silver) plans finished and get a prototype of that made up.

So yeah - slight spanner in the works until I can establish what the full story is. Nothing major, it just needs to be investigated, explained, and accounted for.

http://www.jr001b4751.pwp.blueyonder.co.uk/Cascade1.jpg

An example of the temperature effect on h of as predicted by Sieder-Tate and Flomerics.
Although have not isolated Kinematic Viscosity, this is probably partially responsible.
Have you anything similar for "one of the principles in the operation of the G4" ?

Nothing I am willing to discuss.

Actually the effect is not something that I've ever seen modelled mathematically and I do admit to somewhat fumbling in the dark myself with it through extrapolating and modifying similar models. Overall I seem to keep hitting the mark, with theory matching up with results most of the time.

The nature of the problem has more to do with the elements not impacting the fluid as much when the fluid is more viscous. The closest analogy I guess would be stirring a cup of water as opposed to a cup of honey. The effect of the stirrer on water has lingering effects, but when stirring honey the effects dissipate very soon after the stirrer passes. The problem is not so much directly related to how the viscosity affects the convectional process, but how the viscosity affects the turbulence enhancement down the track.

Sorry Les, I know that you really want some hard numbers and maths for all this. Sadly I am partially unable to explain it adequately in theory, and even for what I do have, I am unwilling to share at this point in time.

Deliberately not inquiring about the principle.
Not trying to educe which principle
Just interested in the effects

So what is this?

To see if my fumbling theory holds true in practise.

So, we will see some fumbling numbers eventually.
Excellent, look forward to playing with them.

If you're asking for a graph of the relationship between viscosity and performance drop, I don't have one.

With water, the turbulence enhancement appears to offer around a 20% improvement in h over the closest equivalent mathematical model. That 20% is mostly emperically determined.

I roughly predict that the 5x more viscous liquid would roughly reduce that gain over the standard model to around 5-8% above what the standard model would predict for h for that level of viscosity.

Again, not plotted. Am more interested in emperical evidence and formulating rough theory, than attempting to assume that I have a hard theory that works.

Or this is an interpretation of pH's results?

Why not just add 25% glycol to your system and see if the theory holds true. Are you measuring your temps through the socket or the internal cpu sensor?

Edit: I just realized you don't have the block back yet

What comes first? The results or the theory?

Am aware of the impact of viscosity. Am unsure of the scale of the impact. Have some thoughts. An uncorrelated data set of 1 (Phaestus's results) is hardly something to form a theory around (unless it involves sticking one's hand in fire). Guesswork at this level Les.

Thanks for the update Cathar. Some interesting developments.

The gains of the G5 vs other blocks will be of great interest.

Thanks for sharing what you can at this time. :D

I think almost totally reponsible Les

From properties here.

Some charts:
Density and Viscosity
http://w1.863.telia.com/~u86303493/C...rt_VD_vs_T.jpg

Giving Kinematic Viscosity
http://w1.863.telia.com/~u86303493/C...rt_KV_vs_T.jpg
Well, some fumbling.
http://w1.863.telia.com/~u86303493/C...Re_vs_T_kv.jpg

(how do you push 4 lpm through a 2mm diameter tube?)

The step from Reynolds number to Nusselt to "h" is fraught. I have no idea of the barrier thickness, no clue how to guess either. Empericality rules. Suffice to say that Nusselt (and thereby "h") is proportional to Reynolds. Ish.

Think it boils down to the "Ish"
From here, one of my fumbles with "ish"'
Now think "h"s are too high(over estimate of C/W(TIM)) but .....

Some interesting/useful data plotted, will digest.

[quote=Les]Think it boils down to the "Ish"
Earlier in this thread(page 4) one of my fumbles with "ish"'
Now think "h"s are too high(over estimate of C/W(TIM)) but .....

[quote]

Unsure of the exact numbers but your curves look like what I'm getting Les. Will check specifics later.

Here are some Nusselt curves,
http://w1.863.telia.com/~u86303493/C...t_Nu_vs_kv.jpg
more proportional to "h" than Re.# is.