This may sound crazy, but it just might be the case...

[jungle]

Ok, i have no idea if a lot of yall have masters, doctorates, or just a lot of experience in the field of heat dissipation, but i was talking to my grandad the other day, who did his masters in thermodynamics and heat transfer before going on to get a doctorate in chem.e. anyway, he's extremely smart, but also very stubborn, so there's a chance that what he said could be improved upon.

basically, he said that the best heatsink would be a copper block with many small diameter holes drilled straight through it, no maze or anything. as long as the block had a high flow, this would be ideal. the problem, he mentioned, was that you would have to have a pump that could put up enough pressure to maintain a decent flow through a pretty resistive block. the way i look at it, the fairly recent jet impingment movement is very similar to this idea in many respects. now before you jump all over that statment, it obviously is quite different. but i've noticed a trend, i believe, in waterblock construction that might suggest that he could indeed be correct.

start off with the early waterblocks (the ones i know about anyway), say the early dd mazes and d-tek tc-4. these were basically large diameter holes drilled in a switching patter through a block. a big step came from the white water, which utilized micro-channels (if i'm not mistaken), which is closer to the small diameter many hole approach. huge gains in performance. then the cascade/ storm, which both apparently use the jet in cup method.

from what i can reason, the main difference between this jet-cup method and the straight through hole method is turbulence. also, the actual contact area of the heatsink (since the jets are not conductors of heat) is the cup area, as opposed to the much longer contact area of the drilled hole. i realize that turbulence is a great promotor of cooling because more water comes in contact with the actual heatsink. however, when i think about it, it seems like this turbulence would be simliar to the added surface area of the straight through hole. granted, the turbulence could bring more water in contact with the heatsink, but i think that the straight through block could rival that with increased flow and increased surface area of the block.

one more thing, my grandad explained to me that (in a tube at least) there is a layer of water all around the circumference of the tube that is essentially unmoving, and therefore insulating. increasing the flow rate decreased the thickness of this layer of stagnate water, so increasing performance two fold. it could make sense to me that such a condition might exist in the bottom of the cups of the cascade etc., in which case, perhaps it would be beneficial to attempt to put a small cone inside the cup. anyway that is just my random thought and purely unfounded.

i really hope that i haven't offended anyone with this post. i have the utmost respect for everyone that has put so much effort into such an interesting area. please do not think that i think i know more than you or anything like that. my mind just came up with the idea that, as we improve designs more and more, we are actually coming nearer to this idea that my grandad told me about. is it possible that something so simple could be the way to go? maybe something like this could be improved with turbulence or something like that. or maybe this could be an easy diy job that wouldnt require much heavy machinery that could perform somewhere around the level of the top-end waterblocks, if perhaps actually not as well. what he said just threw me for a loop and i thought i would share it here and see what the other experts thought. eagerly await opinions.

jungle

[BillA]

[jungle]

i see i inspired you so much you went outside and made one jk

yeah i guess that looks like what i was talking about..... but what do you think? lol

jungle

[bigben2k]

I don't think Bill made that one; wasn't it XJinn's?

Here's a thread where I posted my discovery of the jet method:
http://forums.procooling.com/vbb/showthread.php?t=6101

[pHaestus]

I wouldn't want to drill all those holes through copper. Would make the Storm look like a bargain in machining costs...

[Remote Man]

It's a lamiflow block isn't it?

[Cathar]

Originally Posted by Remote Man

It's a lamiflow block isn't it?

Nah, this is a "Lamiflow"

http://cammo.kangaweb.com/design.html

[Remote Man]

hhhmmm I remember someone making the block Bill has a photo of there recall there being some problems with flow rates? or am I getting confused/dreaming stuff up?

[Kobuchi]

Lamiflow. I'm glad to see someone managed to make one, and well. I'd tried to build such a block but with a configuration too ambitious (different port locations, tapered channels). It would be cheap to produce in quantity though as the parts could be stamped.

***

Your grandad ought to check into these forums, jungle.

He's right, I think - the main thing is to rub that boundary layer off. You do that with restriction, one way or another. So long as people use impeller pumps to push water through blocks of the tolerances we're used to, that means high flow rate translated briefly into higher pressure in the block.

[Les]

Originally Posted by bigben2k

... wasn't it XJinn's?....

Think is Hoot's creation.
Link only showing "red crosses" for pics, so Bill would have to confirm.

[lolito_fr]

speaking of lamiflow : a commercial offering
(the "copper strip" lol)

[Jag]

Better still:
1A-cooling new block HV4

[BillA]

"wasn't it XJinn's?"
yes, click on the image, properties, and its the image name

Joe
you still have the wb ? pass it to pH ?

Jag
what's the fin gap ? looks like another filter to me

[Roscal]

Fin gap is 0.5mm

[Jag]

0,5mm fin gap, are you sure Roscal?
Comparing with the picture below, it gives the impression of being less that that (only visualizing), and it also seems a remake of a previous block, showed before in this forum: 1A-SL.
Even the water trajectory seems to be consistent to the aforementioned version.
Data that i gathered from the manufacturer:
1A-HV4 = 58x58x15mm and 1,6lpm with a Eheim 1046.

Former model:

[Roscal]

It's the same than 1A-HV3, there are 42*0.5 cooling copper sheets

[pHaestus]

Yes that's xjinn's wb for sure. I am almost certain that Joe either (a) never physically had that block or (b) returned it to xjinn long ago.

I am not sure how to get in touch with xjinn; was he in engineering at RIT? I was told that he dropped in Pro/Chat one day months ago when I was not in there and wanted my contact information to send me that block to test. I never got an e-mail or saw him again though.

[jungle]

so are all of you saying that it will be best? i'm a little confused. also, i forgot to mention this. when i told him about my idea to make the waterblock base thinner to reduce the resistance, he was erm quick to correct me. he he. he said that doing that would make little to no difference in temps because the internal resistance of the copper is miniscule compared to the two other contact resistances:

a) the interface between the block and the die. for this he suggested that, if it were possible, soldering the two together might be good. he doesn't know a whole lot about the chips themselves, so i'm not sure if it work. i guess you would have to have some pretty low melting point solder. really, i bet it trasfers heat better in its liquid state. so if there would be a way to keep it liquid when the proc is running, i bet that would make a hell of a thermal joint. on the other hand, would simple arctic silver paste be better in the first place?

b)the resistance of the described stagnate layer of water in the tubes. the solution to this is merely higher flow. bascially, the best would be extremely small crossectional area and many many holes, providing that the same flow could be kept up through the block regardless of the number of holes and their diameter. i have a question though, if you were to drill enough small diameter holes to have the same crossesctional area as the inlet, would the resistance be essentially the same as just putting it through a tube? if this is the case, would it not be a good idea to just get a large block of copper and drill enough holes to equal, or even possibly exceed, the area of the 1/2" inlet and outlet. i think that would make one hell of a block. but would added weight be a problem? also, it would be best if the block were continuous, not layered. anyway. tell me what you think about this.

jungle

edit: pH, i was really thinking that someone could just drill through it with their own drill press, so no cost, or did you mean for mass marketing?

[mastermind2004]

a) Arctic Silver is more than sufficient. Any type of solder will ruin the chip. And, you'd need a solder with a melting point below 70 C or so to avoid destroying the chip. It would also have to be non-electrically conductive. That makes it very difficult to find anything that would work.

[BillA]

chips can be soldered with various alloys, max transient is ~150°C

[mastermind2004]

Is there any real performance difference though?

[killernoodle]

Isnt this how the P4 cores are "glued" to the heatspreaders now?

[BillA]

everything has a difference, pros and cons

some soldered, some with paste (not AS); always done deliberately

[Cathar]

jungle, your grandad is a smart guy who knows his stuff, but waterblocks are starting to progress beyond the broad-scale statements he's making, with the newer ones really starting to fine tune and eke out possibly more performance than he's decribing.

The Atotech MC1 is perhaps the closest commercial example to what gramps is talking about. It is basically a stack of copper plates that have been soldered together with chemically etched 0.2mm micro-channels in the plates. When the plates are all soldered together it forms a "block" of copper with hundreds of 0.25mm squared pin-holes through it.

The MC1's performance, about 2 years back now, arguably matched by the combined early micro-mini channel/impingement blocks (ala. White Water) for which the same basic design is a fairly common design in use today exemplified in a whole host of blocks, most of which outperform what the MC1 offered.

The copper's thickness is now an issue too, and cannot be trivially dismissed as being a good enough conductor. True, the resistance of the copper is typically smallish in comparison to the thermal paste interface, but with the advent of properly lapped waterblocks and high quality thermal pastes, it is no longer something which can be dismissed off-hand. Thicker copper base-plates allow for wider heat-spread to the convection area typically resulting in better low-flow performance but sacrificing somewhat the available gains which can seen with higher flow-rates. The reverse is true for the thinner base-plates, where low-flow performance tends to fall away, but higher flow rates generally allows them to surpass the thicker base-plate designs.

The simple side to side high-water-flow-through-channels approach will provide highish end performance compared to today's top-end blocks, but it is seeming more clear that some degree of impingement is a more efficient way of stripping back the boundary layer to eke out even more gains. This can be witnessed in that there are almost no top-end blocks on the market today that do not utilise some form of impingement.

Some of the super-refined micro-channel blocks that very carefully balance off pressure-drop vs performance are holding their own very well though. I'm personally of the opinion thought that pure micro-channel blocks, at least in terms of the balance of presssure-drop vs performance, are getting pretty close to the limits of what they can offer, especially for very high heat load modern CPU's. There do exist a number of exceptional examples though that show close to the full potential of what micro-channels can offer, and it is very impressive.

[DrMemory]

Does any of this sound vauguely familiar to anybody else? Like PolarXstream.