Why a large waterblock?
 

Why should a waterblock be so large?

I've thought of this a few times, and I've also been working on and off with the microchannel concept for about 1½ year since I started thinking about why radiators had so many thin fins, and with so little space inbetween.

Maybe I should post a small introduction first. I just found this forum a couple of days ago, and it seems the discussions stick to the subject here more that most other forums, and you guys seem to have lots of good ideas, and useful knowledge. So I'll probably hang around here for a while :)
I'm 27 y.o and I have 6,5 years of university level studies behind me (mechanic/electric engineering, psychology, networking)
___________________________________________

Ok, that's the very off topic part.. Back to business.
Almost all the blocks I've seen are about 40 x 40 mm or larger.. the first I've seen that resembles my ideas (if not completely, at least close enough) is Cathar with his microchannel concept block.

The only reason for a block wider than.. say 20 mm would be either if you used peltiers, or for fastening purposes. Of course this calls for microchannels.

I've inserted a picture here to show what I mean. The green square is a generic cpu, the size of it is 39 x 39 mm (for no special reason) I only included that to show the proportions. The block is quite easy to make (takes no cnc-mills or anything fancy) You just need a drill-press, file, saw, sandpaper, torch and a thin sawblade.
The block is made for 1/2" hoses, the internal flow-area is slightly larger than the area of the hose to make up for the resistance when the water is disturbed. The flow is quite straight, and there are no bends and stuff that will stop the flow.

Why aren't more blocks constructed like this? This is even 1,5 mm thick fins, which could easily be reduced to 1 mm, to gain even more contact area.
Any questions? comments? please ask and I'll try to answer as good as i can :)

People have done similar sized waterblocks, but the larger ones are easier to mount and IMO they have less of a chance of loosing their footing on the die itself.

on OCAU here

Ahh, I knew I forgot something.. the yellow circles on the side-view are bolts that holds the block steady, and also are part of the clamping mechanism...

Apart from the mounting problems (which are obvious if I don't show the clamping) wouldn't this be a quite effective solution?

i'm proably out of my league in this part of the forum :)

But it should work...mass production and consumer level usage is a different story.

Cause its been done in the past a few times and is currently being done by someone on that forum BillA posted(which requires registration...glad I already was). Ease of construction and mounting is what most people aim for when building a block.

Actually I thought was was being described also sounded a lot like sciencewhiz's "lamiflow".

Bit of a cross between the ghetto concept and the lamiflow.

Hi Axly!

If it hasn't been done before, Welcome to Pro/Cooling!

I found your block originally on OCAU, but as I remember, you presented it as a copper elbow soldered to a plate.

It would work, not argument there.

I think you'll catch more flack from lack of originality, than praises for simplifying Cathar's Concept block. I for one, believe that the barbs at 45 degrees is an excellent idea.

The elbow has been done before, but not with fins in the center area. The point to remember from Cathar's concept, is that the flow must be forced through the fins, and not be allowed to flow over them. I believe that the original idea that you presented didn't make that clear.

The baseplate does provide stability. It may be called "lack of effort" to try to minimize the size of the baseplate, to its very minimal values, but then the scraps may be useless, and would only involve more cutting.

The thing to keep in mind is that the block may also be used (or adapted) to a TEC, and your design most certainly will not allow this. Some people may also put on their own clamping devices, and again, your design limits this, to a certain extent.

Still, nice work.

another thing I see about such a small mass in a block, is that it would be awsome if we never had pump failures. But we do. One thing I liked about the Original Spir@ls is that with all the mass in the block. A full .900 tall. My ferret had a bad habbit of playing with cords under my desk. And would unplug my pump alot. But I never lost the machine. I would notice temps getting really high and investagate. See where I am going with this. There is a buffer zone with more copper.

OCWC made a block like that. It performed pretty well, and he just epoxied it to the CPU.
-Kevin

Hi and thanks for the welcome bigben2k :)
I'm not the person from the OCAU, I just registered there yesterday when "unregistered" linked that post.. I can see the similarities though ;)

I've been thinking for a while about how to be able to make a simple microchannel block without the use of advanced mills or expensive tools. This block won't be a good idea for resellers though, since it'd be easier for them to mill a block, but the "hobbyist" I think this would be a good idea.

I said that you only needed regular tools.. You need a drill (preferrably drillpress) saw, file, sandpaper and some silversolder. The material in my pic is 1,5 mm copper sheet which should be very easy to get. Instructions to build this block would be sort of like this:

Cut a number of pieces like those seen in the picture.
Stack them as shown in picture.
Tighten the pieces together with bolts placed through the 4 holes.
Silver-solder the edges.
Drill holes for in- and outlet.
Sand and polish.
Fasten the hose-barbs.

Ofcourse you need a whole piece on each side as well, and preferrably with a bit thicker copper that will help holding the block together during soldering.

Re: Why a large waterblock?
 

Heat spreading.
I would expect that for an 80sq mm die a 50x50mm wb with a 4mm bp to give lower temps than a 15x17mm wb (of any thickness) for Heat Tranfer Coeffs(water to bp) below ~80,000W/m*m*c .
http://www.jr001b4751.pwp.blueyonder.../Catharbpa.jpg
The problem is determining the Heat Transfer Coeff from the water to the the bp.Some of my guesses(using Kryotherm):
http://www.jr001b4751.pwp.blueyonder.co.uk/hvva.jpg

I was thinking kinda the same thing the other day, but mine was why do people bother putting fin area far away from the core?, if you had it right it'd only have to overlap the core by about 5mm each side, carry every iota of heat away before the Cu 'stores' it?. It's working out the optimum though is'nt it?, how much fin area is needed to 'pass over' the heat(at what watt) to the water(and at what rate/velocity), working out an optimisation proggie for that would get my 'brownie point' star of approval :D ...
Sadly it's all above and beyond me:shrug: , I can only look and guess...

I think people need to work together and take each step as it comes, Cathar's worked out how to define optimum fin~H2o area. how to define optimum fin area for heatload(or range?100-200watt?) is next maybe?. find the best pump (how much for an Iwaki 20rt, or rzt?) use that as baseline, so you know what flow to work fin area to.

Sorry I'm wandering :p ...

PS, I think the thermal mass argument is 'bunk'(new word! W00T!! ;)), in this day we should have thermal protection (wiring the pump to the PC plug at least!) to protect our investments. A relay to power on pump(&TEC) are de~riguer to shut them down if the CPU overheats, as a electronic flow metre(proximity type) should be if we have a 'CPU fan fail shutdown' option that works, [cough]Mr Abit!, KX7![cough], or someones bright enough to make a curcuit breaker that works off the 12Volt switch...

Here is a nice little waterblock. I think it is about a year old

it definately looks funny against their northbridge block!

apparently it cools about as well as a stock heatsink

and if we continue the trend of ever decreasing water contact area we get to this:D
http://users.bigpond.net.au/volenti/DD_going.JPG

well if you're mad like me anyway:p

[edit] I use 2 pumps in series with this, I may be crazy, but I'm not stupid, heh.

This is, in fact, exactly the way that CaMMo does his implementation of sciencewhiz's "Lamiflow" design.

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

but with the barbs coming in at a tilted angle.

did you by any chance cut off the heatspreader on your P4?

nice design :)

Whoa! Finally :)

I have never seen anyone do this before, and to be honest I've been wondering why noone have tried it. (started doubting my ideas a bit as i thought that atleast someone had tried it)

Anyway, I have a number of plastic models of different laminar flow and microchannel blocks, but I haven't felt very interested in posting stuff about them on other forums as I probably would have been trolled or flamed for no reason, and I don't really feel like getting dumped on for trying new ideas. I have one design VERY similar to CaMMo's (the basic difference is that it's square, and it's not soldered on the end but a whole CU-plate.) that I intended for Peltier use, and several smaller block ideas for GPU's, CPU's and some for nothing special, but rater just trying out ideas :)

Thank you for posting that link Cathar, it gave me a sort of confirmation that I'm not all wrong here. Is there anywhere I can find any more facts about this? Flowrate, efficiency?

Sadly there's little to no real comparitive information on how it performs.

A little thought about this concept.. How good is silver-solder when it comes to heat-transfer? I mean it's not pure silver (sadly) which would be great, but some alloy.

About no data about the block.. well, then it's only one way to find out right? ;) But i'd rather stick to a smaller channel for the water like in my little mspaint masterpieces..

Well, there's also my block, which is milled (not what you want).

I can say that if (and that's a big if) you match up the "base-plate thickness" correctly with the plate thicknesses and the corresponding turbulence in the block, you could probably expect to see something slightly better than a Maze 3. My concern with the stacked plates is the possible lack of lateral heat spread from plate to plate.

The design is sound, but getting it to perform requires a little more effort than simply slapping some plates together.

http://www.employees.org/~slf/concept/prod1.jpg

Ahh, well that's maybe a little bit better than I thought ;)

The same concern I have.. I'm especially worried about the heat transfer capability of silver-solder as I'm not sure what it's made of (apart from a small part of silver) If the heat transfer is good, i think I have a solution for that problem.

Hehe, oh yes.. I'm well aware of that, and that's much of the reason I tried the idea in theory here first.

About your nozzle, if you could vacuum-mold (or whatever it's called) the top with plastic resin you could probably achieve a better flow by narrowing the inlet gradually. To see what i mean, take a hose, squeeze it flat about an inch from one end and voila, you have a circle -> slit adapter.. On the other hand you get excellent results from your block as it is right now :)

Another idea is to machine a "pocket" for a restrictor insert....

Good luck with your waterblock!

Right on Axly.

There are ASME standards for nozzles, and that's pretty much whay you're describing. Picture a funnel that's got a curved wall.

Oh heck, more l33t Ms Paint skill:

Who says I'm not already doing this nozzle design and just haven't showed the other side of the plate? ;) It's not a perfect nozzle inlet in the ASME sense, but it's not a straight cut either.

The reason why the solid metal plate is there is to provide a strong bracing pressure against the tops of the fins. With the very thin base-plates I'm using this is an essential part of the design. I investigated the inlaid nozzle plate idea, but for the low volumes of blocks I'm making, the tooling costs required to setup and stamp out some plates or make up a plastic mount, it was cheaper to just get the copper plate made up as is.

But not nearly as "l337" looking :cool: :D

I've been experimenting some with molding tops for blocks from plastic resin. Sadly I have just tried with the stuff I had at hand, which looks like muddy water, as well as being quite brittle. It wouldn't work very good for supporting thin baseplates, but that could be fixed quite easily with an extra copper plate between the baseplate and the lid. It would be a good way to get smooth flow down to the cooling spot..

Now if only my father could hurry up with building his new place (building new house and a 2 car garage + lots of workspace + 2 car carport) I use his place for my experiments as my apartment isn't very well suited for this kind of work ;)