Ultimate block? Theory.

[bigben2k]

Originally posted by unregistered
well speak of the devil . . .

you may be interested to know that a number of slightly similar ones are being EDM'd
cross flow, parallel flow, counter flow, square channel, round channel, etc. etc.

should be interesting, hopefully somewhat definitive

Like that wasn't a coincidence... Thanks, Bill!

[BillA]

strange place this
damned for what I do
lauded for what I don't

hi ho

[bigben2k]

I've been looking into turbulators, following an article I read, (I'll post a pic next week). Here's some other things I've found:

A link to Hayden industries
(not new to us, but an interesting place to start)

A Finnish site
with an interesting idea about throwing in a coil. (would help break up boundary layer)

A tiny bit of theory, with an Excel sheet

A PDF from Trane about turbulator (again!)

[bigben2k]

Originally posted by unregistered
well speak of the devil . . .

you may be interested to know that a number of slightly similar ones are being EDM'd
cross flow, parallel flow, counter flow, square channel, round channel, etc. etc.

should be interesting, hopefully somewhat definitive

Actually, I am interested in the square channel: I've even thought about irregular shapes, like a multi-point star (near impossible to mill though). The idea came to me when I remembered that a circular pattern has the most surface for the absolute smallest perimeter, so why not increase the perimeter (and the surface area) by having square channels, heck, let's use something irregular!

[sunblade]

If you want more perimeter for your area, you could try using long and short rectangles like this:

___________
|__________|

Of course, I still don't know how one would drill rectangular holes easily. Any mill-masters want to comment?

[V12|V12]

I agree... forget all the tap plugs and ish, just get a small piece of copper and solder/braze the holes, it's quick, perm and you can always unbraze the piece of need be.... I need a drill press!

[bigben2k]

Actually, I was thinking about bolting together multiple layers of CNCd parts.

[dream caster]

Originally posted by Xjinn
. . .

so if you guys wanna put together a list of questions i'll be happy to answer later today

EDIT: in the meantime, this should keep you guys entertained:
http://xjinn.procooling.com/
lots of goodies from times past

that is not a big deal!
playing with shortened web adresses in my browser had got me that.

More seriously I remember a thread where you explained the making and thinking behind that block. I searched for it but could not find it. Can you give us the link?

[mr. phantastik]

Howdy doody. I've been reading this thread for quite some time, and I'm blown away by the amount of knowledge here. I've been interested in milling my own block (or having the machinists at work do it for me ) for a long time, so I'm trying to follow along with the theory.

My question is: Is it only path diameter that determines wheather a block is low-flow or not? For example, the block pictured at xjinn.procooling.com would be a low flow, correct?

And as for turbulators, I came up with the idea of using taps, and just tapping the entire length of the cross sectional paths. The threads are perpendicular to the flow of water, and would thus cause turbulent flow, at least imo.

[Limeygreg]

That's a very interesting idea mr phantastic. I do however see a problem in that tapping small thread diameters into copper is fraught with the danger of tap breakage and the lack of resources to remove said broken tap(s) without destroying the block. The deeper the holes (as in these cross-drilled blocks) the more likely one is to breack a tap. #4 and #6 taps are quiet small and easilly broken in steel let alone copper, which is an absolute BITCH to tap. A coarse thread would best.

[mr. phantastik]

good point. what about using a diamond dremel bit, a rather large one, and inseriting it into the hole, and jiggling it around at certain spots (not bring it up all at once, as that would make grooves parallel with water)?

i think the spring idea is a good one, but what kind of flow restriction does that give?

what about using something corrosive to copper, and letting it sit in the channels for a said time, causing pits to be eaten away. would be hard to control

just throwing out ideas.

[ECUPirate]

RE: tapping... I was thinking that most taps would be threaded to fine. Perhaps water would get "stuck" in the threads and create hotspots throughout the channel. I dunno. I was wondering if there was some way to "rifle" the channel like a gun barrel. ....much longer twists may promote quicker flow through the channel, and maybe the spirals would help or something. Just a thought.

[mr. phantastik]

what about taking a piece of say, .5mm to 1mm thick copper sheeting, cutting a strip, and wrapping it around a dowel the same diameter of your hole minus the thickness of your sheet (would have to sand down the dowel to the exact size).

you could control the width of the gaps between the copper, and then you could put the tube into the drilled hole. i doubt you'd have to secure it if it was snug enough. perhaps you could tin the outside and then heat up the block when the loop was inside, melting the solder?

again, more ideas :]

[ECUPirate]

I've suspect that this would greatly reduce flow, and I seriously doubt what good it would do anyways in such a small channel.

[Limeygreg]

How about some very thin stainless steel wire, a guy I used to repair flexible endocopes for used to have it for surgical instrument repair. Just wind your own coil, a little bigger than the hole then as you insert it into the hole you rotate it such that the diameter reduces, when it's in rotate slightly back the other way to lock it in place and trim to length. They'd br like the flow disruptors show in the links BB2K posted earlier.

BillA - a question about the parallel drilled block. Isn't it preferable to not have the channels too narrow (or diameter) as it will promote and maintain a stronger laminar flow region. I seem to remember reading that on a site discussing laminar-vs-turbulent flow, possible from a link you posted somewhere. Isn't it the case that the boundary layer is a greater percentage of the tubes diameter in small diameter pipes and that is why laminar flow becomes more dominant as the diameter decreases?

[mr. phantastik]

so then my guess that low flow blocks use tiny diameter holes is wrong?

I'm curious, because I'm going to be running 3/8" tubing and won't be able to blast the fowrate sky high. I figure though that with smaller tubing, you raise your velocity, which is good for added turbulence.

[BillA]

Limeygreg

based on what I can mine via google, it seems that the conventional convection 'model' is not too apt for describing the heat transfer in small holes
apparently when small enough the boundary layer is effectively removed

the gist of it seems to be that smaller is better
the optimum solutions are integrated into the silicon on the top of the die and quite small indeed

as we have an externally applied heat exchanger the mini channels are the best possible, even with the pump pressure requirements

aquarium pumps are not the way to do this, eh

[Limeygreg]

The way I understand it is that high flow rate through a low resistance waterblock is better.

While small channels will cause an increased pressure drop across the block due to frictional and orifice restrictions, to overcome this a higher pressure is required. A bigger pump is usually the first step to overcome this and with that comes more heat dissipation to the fluid from the pump.

The aim is to optimise base thickness and have high flow with low resistance (so you can use a less powerful pump). That is not to say that you should run a bunch of 3/8" holes down the block to achieve high flow and low resistance - that would reduce the thermal transfer rate and hence the efficiency.

It's one big juggeling act to get high flow, for turbulence and low resistance for reduced pump sizing. Turbulence is necessary for increased thermal transfer of heat to the fluid and is achieved either by flow rate or by physical dissruption.

So, in a narrow bore tube, where the boundary layer maintains laminar flow to a greater proportional distance into the tube you would probably need to use a physical means to break the boundary layer and promote non-laminar flow, especially in a low flow block where laminar flow will be the predominant domain within the tube.

[BillA]

not the way I see it at all
to me it is about temperature gradients
we know what the coolant is, that is a limit
what is necessary to get the wb bp to as close to that temp as possible ?
more precisely, to get the bp surface in contact with the die to that temp ?

I believe, and there is an abundance of research suggesting this, that low flow - high resistance (turbulence) is the way to go
(as outlined previously)

but not the easy and cheap way, naturally

[mota]

Originally posted by unregistered
google is most informative, see "micro channels", etc
(ni the first 300 listings there are about 20 good ones, and 3 or 4 quite good)

http://www.google.com/search?q=coppe...8&start=0&sa=N

Here is another interesting variation of microchannel design by Thermotek that they are calling ThermalPlane. Hmm, looks like the page with the info on that block has been removed from thermoteks site.



Dual pass microchannel designs are also being developed.



You can forget about making a better waterblock with a mill, drills, saws etc.

[BillA]

Hi mota

ah, but with EDM we are free
if one has the loot !

the ThermalPlane is interesting, but why the stub fins on the 'off' side ?
(different kinda app probably, as in a radiator tube a la Lytron)
- too thin for much lateral dispersion

[V12|V12]

Looks great, but talk about TERRIBLE flow rates@ You'd have to have a herculinian pump for that and for all the hassle of cooling and charging, pelt cooling simply owns it...

[myv65]

Originally posted by unregistered
the ThermalPlane is interesting, but why the stub fins on the 'off' side ?

Just like normal fins/pins, to increase surface area for convection. I'll agree that they won't do ya much good here as the solid columns connecting the top to bottom would need to be thicker in order to transfer any appreciable heat to the top side.

[mr. phantastik]

i think the most economical way to test these theories would be to make blocks out of aluminum. its cheap, readily available, and easy to mill/machine/drill/do whatever too :]. Comparisons of temp readings using diff aluminum blocks could be compared, and then a final design could be sent to a machinist to make a copper block.

and would a high pressure drop really matter if you're using a system with a resevoir? the resevoir isn't maintained at the same pressure as the pump, so while the water pressure is reduced going out of the cpu block, wouldnt it all return to normal as the pump takes from the res?

I think there isn't just one design to the ultimate block, the real consideration lies in what other components you use. Of course this is all theory anyway, but I thought I'd throw in another opinion.

[Can O' Beans]

I was going to tap my holes, like mentioned above.

BUT, I was going to use a tap that was only very slightly bigger than the hole itself so that shallow "threads" were cut instead of a full thread cut that would normally occur.

Before I did that, I was going to try a longer "swirl" cut, but didn't know how to do it. I do have some long screws used in plastics - they only have about 2 threads per inch. I don't know if they would cut into the copper though, perhaps if I modified them...

I originally thought about putting a spring inside, but thought the cutting of very shallow threads would turbulate just as well, AND add slightly to the surface area.