The "Cascade" - mini-cup directed jet-impingement block design (56K warn)

[Cossey3]

Originally posted by Cathar
http://www.employees.org/~slf/lrwb/

Good profit? Not even close.

The middle plate with the jet tubes takes almost 30 minutes of machine time. Polycarb is easy to machine, but we're using sub-mm cutters here, requiring over 4x the total bit travel distance as was in the White Water base-plate per pass. Can't cut as deep per pass, and the feed-rate is limited due to the tiny size of the cutters used. The middle plate is almost literally worth it's weight in gold.

The copper plate's holes are plunge milled (not drilled), and "pecking" is used to clear the swarf for each hole as the bit descends. The copper plate takes a significant amount of time to machine due to the number of holes, then its base is machined flat afterwards, and tapped.

In all, there's actually more machining time per block than was in the White Water's, and you'll notice that I'm selling the blocks for less this time.

They are easier for me to finish off ready to ship though, requiring less of my own time and equipment, but right now I have enough debt on this to buy a small car outright.

Put it this way. I make more (gross salary) in a month with my real job, than I would make in a year of selling these blocks. I'm just looking for the "break even" light at the end of the tunnel, and right now it's pretty damn far away.

£60 to the uk thats very good seeing as a maze 3 or 4 is £40 and the maze isnt half as good. i wonder whether auspost tip off customs for the vat like ups do.

[utabintarbo]

Originally posted by GTA
Been thinking that myself. Current avenues of thought include :

1. Inducing turbulence in the water before it enters the block.

Why? It would seem that one would prefer relatively linear (laminar?) flow down the jet so as to have maximum impact with the bottom of the cup. I believe this might be accomplished using a larger plenum btwn. intake and jets, but Cathar was somewhat dubious on that.

Originally posted by GTA
2. Multiple inlet designs

Unnecessary with a plenum as above.

Originally posted by GTA
3. The 1" outlet thing above.

I think this might have possibilities in terms of performance, but as Cathar stated above, not economically feasible. Too much complexity for the mass market. It'd be cool, though!

Originally posted by GTA
4. Twisterbongs incorporated into blocks.

...

Huh?:shrug:

Bob

[bigben2k]

Originally posted by GTA
Been thinking that myself. Current avenues of thought include :

1. Inducing turbulence in the water before it enters the block.
2. Multiple inlet designs
3. The 1" outlet thing above.
4. Twisterbongs incorporated into blocks.

You got anything in mind?

I don't think #1 would beat jet inpingement, and that's most of the problem.

#2 isn't really relevant, it just applies to the design, as needed.

#3 again, only if the design calls for it. Otherwise, the product's appeal would be both function and form, and would be a high-end product, and that's a pretty small audience.

#4 Now there's something I've explored before (aka running the water in a circular motion), but gave it up when I realized that the cold water molecules would be thrown outside of a circumference, because of their higher density, and I didn't have (couldn't figure out) any way of taking advantage of that. The twisterbong itself can't be applied directly to a waterblock, because of the air involved.

There's also the issue of added heat, from the stirring action required to force the water to run in a circular motion.

I think it's clear that it comes down to reducing the thermal resistance, and with water, that means throwing it into turbulence. Since we can't do it with sheer pumping action, we're down to turbulators and/or jet inpingement. This is where I'm stuck.

Either there's another (third or fourth) way to get turbulence, or there's something radically different out there than what we've been thinking, to reduce that thermal resistance, between the copper baseplate, and the water.:shrug:

You know, there's another part of this that's so significant, that it might be worth approaching again: the TIM joint is now a very significant part of the temp difference, between the CPU and air temp. As Cathar recently wrote in AMDMB's forums, it's ~70% of the whole temperature gradient. It's also detailed in the OCAU thread.

[gone_fishin]

Originally posted by utabintarbo
Why? It would seem that one would prefer relatively linear (laminar?) flow down the jet so as to have maximum impact with the bottom of the cup. I believe this might be accomplished using a larger plenum btwn. intake and jets, but Cathar was somewhat dubious on that.



Unnecessary with a plenum as above.



I think this might have possibilities in terms of performance, but as Cathar stated above, not economically feasible. Too much complexity for the mass market. It'd be cool, though!



Huh?:shrug:

Bob

On the plenum. It is obvious by the pics Cathar posted (of the top two sections assembled and water squirting through the jets) that there is much greater force from the center jets then the outer jets. He stated somewhere that this was intentional. A more even presure spread may stagnate the central exit flow somewhat?

GTA's #4 lol. If you accomplish such a thing then get one of them micro spy cams to go along with it and usb it out to the monitor. Now that would be eyecandy.

[gone_fishin]

Originally posted by bigben2k
I don't think #1 would beat jet inpingement, and that's most of the problem.

#2 isn't really relevant, it just applies to the design, as needed.

#3 again, only if the design calls for it. Otherwise, the product's appeal would be both function and form, and would be a high-end product, and that's a pretty small audience.

#4 Now there's something I've explored before (aka running the water in a circular motion), but gave it up when I realized that the cold water molecules would be thrown outside of a circumference, because of their higher density, and I didn't have (couldn't figure out) any way of taking advantage of that. The twisterbong itself can't be applied directly to a waterblock, because of the air involved.

There's also the issue of added heat, from the stirring action required to force the water to run in a circular motion.

I think it's clear that it comes down to reducing the thermal resistance, and with water, that means throwing it into turbulence. Since we can't do it with sheer pumping action, we're down to turbulators and/or jet inpingement. This is where I'm stuck.

Either there's another (third or fourth) way to get turbulence, or there's something radically different out there than what we've been thinking, to reduce that thermal resistance, between the copper baseplate, and the water.:shrug:

You know, there's another part of this that's so significant, that it might be worth approaching again: the TIM joint is now a very significant part of the temp difference, between the CPU and air temp. As Cathar recently wrote in AMDMB's forums, it's ~70% of the whole temperature gradient. It's also detailed in the OCAU thread.

A radical approach would be to incorporate an actuator in the block itself. There are other considerations though. The cycling would wreak havoc on a paste style tim.

[hara]

I have a couple of ideas (probably useless)

Can a middle plate be made out of brass/al/copper (whichever expands most) and then heat it up and inser the eyelets or copper capillery tubing (after being frozen)? (shrinkfitting I think it's called). I don't know wheter you'll find this kind of tubing though.

Can capillery tubing made out of acrylic (like the ones used in chemistry labs for small samples) be used by manually cutting and pasting into the middle piece?

[bigben2k]

Originally posted by gone_fishin
A radical approach would be to incorporate an actuator in the block itself. There are other considerations though. The cycling would wreak havoc on a paste style tim.

Yeah... maybe.

We might as well look into the pumps too: a propeller, in line with a straight tube section might be more efficient, but I digress...

Maybe we could tweak the jet inpingement, by inserting a needle point into the small tube: it might help reduce the wide spreading of the inpingement area.

(Borrowed from LeeJSmith's thread, who's been doing excellent work, BTW):

[bigben2k]

Originally posted by hara
I have a couple of ideas (probably useless)

Can a middle plate be made out of brass/al/copper (whichever expands most) and then heat it up and inser the eyelets or copper capillery tubing (after being frozen)? (shrinkfitting I think it's called). I don't know wheter you'll find this kind of tubing though.

Can capillery tubing made out of acrylic (like the ones used in chemistry labs for small samples) be used by manually cutting and pasting into the middle piece?

Hum... this might help the construction, but I don't see how it'd improve performance.

I like the idea of using glass microtubes. Anything that's got a smooth surface inside might help: the drilling process leaves a restrictive mess, but probably not relevant here, because the flow is laminar.

[Cathar]

Originally posted by gone_fishin
On the plenum. It is obvious by the pics Cathar posted (of the top two sections assembled and water squirting through the jets) that there is much greater force from the center jets then the outer jets. He stated somewhere that this was intentional. A more even presure spread may stagnate the central exit flow somewhat?

This is exactly right. The water coming out of the middle jets has to force its way through the maze of tubes to get outside of the central area. I actually wanted an imbalance in "free-flow" mode to favor the inner jets. Further, the density of the heat is concentrated more in the middle of the heat source as at the edges, the heat can laterally spread out away from the heat source somewhat before it is cooled, whereas in the middle the heat really has nowhere to go but up, so it makes sense to give a little more focus/power to the middle jets. The plenum chamber is sized in a fashion that promotes this effect.

[t00lb0x]

Originally posted by Cathar
This is exactly right. The water coming out of the middle jets has to force its way through the maze of tubes to get outside of the central area. I actually wanted an imbalance in "free-flow" mode to favor the inner jets. Further, the density of the heat is concentrated more in the middle of the heat source as at the edges, the heat can laterally spread out away from the heat source somewhat before it is cooled, whereas in the middle the heat really has nowhere to go but up, so it makes sense to give a little more focus/power to the middle jets. The plenum chamber is sized in a fashion that promotes this effect.

OMG its a science? :shrug: j/k. How many tries did it take you to get it to do exactly that?

[GTA]

I wouldn't throw off #4 as an idea just yet.

It ties in with the turbulence idea. Think twister+pin heatsink over the core+method of water removal to give a semi closed loop.

Okay, we're not talking super performance, but it would look AWESOME

In my opinion, its an idea to be taken further.

[GTA]

Originally posted by utabintarbo
Why? It would seem that one would prefer relatively linear (laminar?) flow down the jet so as to have maximum impact with the bottom of the cup. I believe this might be accomplished using a larger plenum btwn. intake and jets, but Cathar was somewhat dubious on that.



Unnecessary with a plenum as above.



I think this might have possibilities in terms of performance, but as Cathar stated above, not economically feasible. Too much complexity for the mass market. It'd be cool, though!



Huh?:shrug:

Bob

These ideas are not necessarily concerned with the cup and straw design in Cathar's block. They are general ideas that I am considering as possible avenues to new waterblock design.

As you mention, okay, maybe not for the cup/straw design, but does the thought of pre-turbulence allow for other designs that way perform well, given that the furniture inside the block does not have to actually create the turbulence itself?

These were not really meant as ideas to improve the Cascade, more as suggestions on possible avenues of thought on where we go next.


Lastly, to Cathar, was Dtek releasing their WW a factor in getting this block to the market quickly?

[Cathar]

Originally posted by GTA
Lastly, to Cathar, was Dtek releasing their WW a factor in getting this block to the market quickly?

DTek was not a factor at all.

Spent quite a deal of money playing with the Hydra micro-channel GPU blocks, and could not get a way to machine them cheaply with the limited setup funds I had. Accrued some largish bills along the way.

Releasing the Cascade design became a financial necessity for me. Ideally I would've held onto it for a little longer, but I've got bills to pay, and a family to support with another child on the way, and a mortgage, and my machinists were getting antsy about when I was going to either deliver them some real business other than prototyping, or when I was going to cough up the bills and piss off and stop wasting their time.

I guess that's the unglamorous side of the game eh? Whether you buy all the tools yourself, or you emply some professionals to do the job for you, it the end it all comes down to needing the funds to do the research.

Still, I wouldn't say I rushed the block though. I took as much time as was needed and put a few hundred hours of research into it once I really started focussing on it. Up until 3 weeks ago I would say that even then I was still undecided as to whether to not I'd make them up, and would've just paid the machinists off out of my own money over time.

[t00lb0x]

You hired someone to make the blocks for you?

[bigben2k]

The pre-turbulence you mention would have to be induced, not pumped, and I'm afraid it would only help in distorting, and spreading the inpingement area further, which would reduce performance.

This design was created specifically for low flow rates.

I'd still love to experiment with the needle, perhaps on a larger scale, to try to straighten up that jet. Oh to have a workshop, how wonderful it must be

[Cathar]

Originally posted by t00lb0x
You hired someone to make the blocks for you?

Yeah, I seem to have a shortage of $250K CNC machine centers in my garage.

[t00lb0x]

Hmmm...so do I . Seems like a good idea to hire someone. How much does it cost to have someone do it?

[Cathar]

Originally posted by t00lb0x
Hmmm...so do I . Seems like a good idea to hire someone. How much does it cost to have someone do it?

That depends on how complex the design is. Do you want to make a one-off? A batch? A big batch?

Prototyping a single block is anywhere from $50-$300 US depending on the complexity of the design and CNC programming involved, and then on the resultant machine time. Each Cascade prototype variant was costing a few hundred dollars US to make up, and that's only because the machinists were being nice due to our prior working relationship.

To walk off the street unknown and ask for a Cascade prototype to be made up would be more like $500US, per variant.

If you know someone who's willing to do it for you for cheap, and you can draw it all up in CAD, you might save a few hundred dollars.

If you ask for more than one or two, then the prices come down. Ask for one hundred and the setup prices get amortised pretty well and it just becomes a matter of paying for parts, machine time and labour.

It's the setup time (also called "tooling costs") which is the big hit.

[t00lb0x]

WoW thats so pricey. $500 for a perfectly done block. One more question, did you sell your WW idea to D-tek or did you "donate" it?

[Cathar]

Originally posted by t00lb0x
WoW thats so pricey. $500 for a perfectly done block. One more question, did you sell your WW idea to D-tek or did you "donate" it?

Well, you have to come up with the design first...and $500 may be a conservative estimate for something as complex as the Cascade. If you wanted to just get a Maze style design done, that'd be fairly cheap, maybe $100 US for a one-off.

DTek's and my relationship is private, suffice to say that the White Water design was not "donated".

[t00lb0x]

Originally posted by Cathar
Well, you have to come up with the design first...and $500 may be a conservative estimate for something as complex as the Cascade. If you wanted to just get a Maze style design done, that'd be fairly cheap, maybe $100 US for a one-off.

DTek's and my relationship is private, suffice to say that the White Water design was not "donated".

Sorry, didn't mean to be nosy. How much would it cost to machine my "test block"

[Cathar]

Originally posted by t00lb0x
Sorry, didn't mean to be nosy. How much would it cost to machine my "test block"

Is your design finalised? Last I looked you were still working on it. The cost to machine your block is dependent on the dimensions of the fins/channels. The finer the channels, the greater the cost.

Just doing a some rough assumptions, and going on my past experiences, maybe $400US or so to get a professional shop to make it up for you as a one-off.

[t00lb0x]

No its not finalized now. But it will be soon. I haven't got it drawn up fully. Still waiting for some 3D drawing help (thanks utabintarbo.) Have some more basic things to fix but then I'll be done.

[bigben2k]

Ok, for those who missed it, I pulled (and I mean "painfully extracted") a bit of relevant info, from the OCAU thread:

For an 80W CPU, the theoretical limit for a "perfect" waterblock is ~4.4C better than the White Water at 10lpm flow rates.

For example, for an 80W 100mm^2 heat source on the White Water with 10lpm flow rates, the total temperature rise is ~14.0C. Of that 14.0C, 8.0C is due to the thermal paste, 1.5C is due to the copper base-plate itself, and 0.1C is the amount that the water would be heated up by. Those are all values that can't be altered. So the 4.4C that remains is a measure of the efficiency of the block at 10.0lpm.

For the Cascade, that 4.4C is being brought down to around 2.5C, and it's taken a LOT of refinement using the most efficient known way to use water to cool (impingement), to get it that far.

I've come to see the Cascade as the final destination in my water-cooling journey.

So to improve on this, I think I'd take another serious look at that darn TIM joint!

New thread

[CNYC]

Hi Cathar,

Is your relationship with DTEK exclusive or are you open to producing variants of the WW block you created? Looks like I might have missed a great opportunity if its an exclusive.

I would hope that you make a great profit from the time you spent designing the block.

I really am looking to get deep into watercooling since I think the timing is right. Do you think you can produce a lower cost, one inlet one outlet design? What do you think about silver plating a block? (I know a factory that can do silver plating very inexpensively)

I would like to take watercooling products (all in one unit(s) and waterblocks) to the mainstream; something that could be used by Dell or Gateway and something an enthusiast would be happy having in his system.

Andrew