Bushman-Tech WB design - comments appreciated

[DX2]

OK first off, i dont have any formal engeenering experience, just physics you learn in school. So a basic knowledge of physics and thermalphysics.
ive put alot of thought into the design, ive been watching procooling for a couple years and have looked at other designs too.

As for the name "Bushman-Tech", it is a term i coined up while at school, basicaly what others might call "ghetto" . But having grownup in Zimbabwe, and being South african, Bushman is alot closer to home. I might even name the individual concept and designs after variious southern african things. might be cool

I have this post in HTML format on my little webspace. The content will be close to identical to this post.

http://members.fortunecity.com/keimo...t/Cooling2.htm

Ive been tweaking the design on and off for about 2 years now, so be gentle before you tear it apart. Further, if you want to build it, be my guest. just give credit. And please for the love of god dont sell it. ive been tinkering with it for a long time , i reserve all the rights for production (if ever) and i guess you could call it free for personal use but no commercial sales. Call me paranoid. But people are evil.

Design goals:
High water flow

Use of dimples

Use none restictive open channels

High surface area

Use of dimples

Use of ribs in fins

Use of interchannel miniribs

Use of hole in vertical and horizontal fins and ribs respectively

Use of segmentation in ribs

Low thermal resistance (heat moves easily through block away from heatsource and reaches coolant quickly)

Use of basethickness for lateral heat movement

Use of uninterupted fins

Use of segmentation in ribs

Use of dimples

Use of holes in fins and ribs

home buildable with basic tools

Use of upside down design

Use of straight lines

keeping tools in mind

Please see Attached .rar with video of how i imagine making the WB. drill and dremel tools, little cutting discs

easily applicable to gpu / cpu / chipset / anything

easily adaptable in all dimensions

At present i'm sitting on a 2.4ghz celeron that hardly needs WC but im pretty enthusiastic none the less . Even if the WB doesnt see the light of day for several months too come.

Right lets get to the meat. (yes all images were made by hand in Microsoft Paintbrush)



I did a few quick surface area calculations and decided apon a ribbed fin design.
the tweaks that followed were to optimise this design.

this is the basic idea. a ribbed fin. ill explain the individual features shortly.


A profile view through the Waterblock.


Vertical segmentation:




The fins are not only ribbed to increase surface area but also have horizontal segments. these increase surface area but also produce some turbulance to help scrub the heat form the copper surface. the segments dont cut right through the fins no maintain good flow along the channels and to allow for horizontal spreading of heat


Holes in vertical ribs

To further increase surface area for heat dissipation but also to bring the water closer to the heat source holes are drilled throught the horizontal fins. the holes are small and few as to not negativly affect the flow of water along the channels

Later on really liking the idea of bringing the water ringht into the ribs, without cutting the ribs completely and thus allowing the heat to move along the length of the rib, i basically went wild with holes. which you will see in the latest build at the end of the thread

miniribs with Dimples

In centre of the channels we have a row of mini ribs. these are to increase the surface area. To further increasse the surface area we have, at the same intervals, as the holes and the vertical segmentation, segments between the miniribs. Thus keeping the intervals simplifies the manufacturing. To further increase surface area and reduce the distance from heat source to water we drill dimpels into the miniribs and between segments, as in the surface of a golfball these might serve to improove flow speed but reduced surface friction.

.
------------------------------------------------------------------------------------------------
OK so now we have the basic design concepts. the block is essencially an upside down design for build ease. Hence the channeles are not cut into the copper, but the fins sit on a flat base with a cupped lid which wil make the channels. ive included the channels in the top down view though that one sees the flow. this is flowConcept 1, i have another slightly different flow concept which ill list after this one.



water flow v1

Removing the fins and concertration on on the flow of the water we have the diagram above. The thickness of the blue arrows and lines indicates flow rate. in this flow design the speed of the water remains constant throught the block. having the connections on opposing sides.

block with lid


-------------------------------------------------------------------------------------------------

Flow concept 2

The fin and channel layout V2



The water flow V2


Again the thickness of the blue arrows and lines indicates flow rate.in short. 2 cannel in the centre, and 4 on the outsides thuis flow in the centre is double that on the outsides. In long: The water enters and leaves on the same side, making this conept better for graphics cards. The water enters and flows into 2 channels. the water speed is the same as when it entres, each channel then turns and splits into 2 more channels making a total of 4 in the other direction. The water flows at half the speed here. Thus we dont have a constant flow rate in the channels. we have the highest flow rate in the centre directly over the die.

block with lid


------------------------------------------------------------------------------------------------
DIY construction:

Materials ive considered, are ofcourse copper, which id use for the final thing. but my first attempt will probably be with alluminium. much easier to work with . for copper work id have to find a place to machine this. My dad might have access to a robotic CNC at work.

the tools would be basic hardware store stuff, dremil bits and drill bits. the way ive done the design it should all be do able with just those shown tool (see video attached) a cutting disc and drill bit.

the actual dimensions are not set yet. i was thinking like 1mm fins and ribs. allowing for about 0,5mm - 0,8mm holes

Attached:
fin+ tools.
Most interesting bit. a fin with hwo i imagine cutting the block with drill bits, and basically those little dremel tools. little cutting discs and such. Think very much DIY. Think Low budget.

[DX2]

Updates to design and imagery

Since then ive added more holes to the design and gotten a little application called sketch up (which i saw someone use here). Since using mspaint is leet, but time consumign and not very practicle. My full on CAD skills are weak. But all in due time.


fins + holes are the latest concept of how a fin would look.



groove + channels is just the 3d look of the block

[bigben2k]

Not impossible to make, except maybe for the finer details.

Interesting design. Ribbed fins are indeed an improvement.

The weak point in this design (if you are open to it) is in the flow geometry. As it stands, I would expect a bit more performance than early generation DangerDen Maze blocks.

There's potential here, if you can improve the flow. Maybe cut the top rib, and look into putting in a jet inlet; the backwash can flow upwards through the ribbed fins.

[jaydee]

No comment on the design but a post to think about quoted below:

Originally Posted by Roscal

A minichannel WB from my friend Eve, a similar "tree fin" design I never realized because of lack of time. Channel width=~0.2 mm , using wire EDM as usual (like my mini-fins/pins protos if you know). It's just for fun and single model because cost is prohibitive ( >> $800 ) but very good performer. Full gold anodization to finish and protect it (not shown here) :

[Spot]

I remember that.

The process it takes to make......that's a killer. Was it EDM?

[bigben2k]

Yep.

[DX2]

Originally Posted by jaydee

No comment on the design but a post to think about quoted below:

its a pretty block indeed. but beyond my scope.
While they look similar. there are a couple differences.

1. I have a winding channel design. (maze) / Eve has a single pass mircochannel
2. i plan to make mine with a dremel and have free spaces between the fins as a result
3. less flow restrictive

no way am i going to get features that small using a dremel, infact the design Eve did for his/hers its down right impossible to achieve with a dremel.

i don't know what wire EDM is, but i guessing they pulled a rough wire through the block.

i remember this design, but only saw it well over a year after id done my initial scribble for my water block. but indeed there are similarities,
*top down lid
*and ribbed fins (tree fins)

[DX2]

Originally Posted by bigben2k

Not impossible to make, except maybe for the finer details.

The bottom of the fins, next to the dimpled miniribs. yeah im guessing ill have to stand the block on its side and drill through its width. it wouldn't look as neat. but when the time comes ill do a thread and/or update this one.

Originally Posted by bigben2k

The weak point in this design (if you are open to it <-- Thats why i posted here ) is in the flow geometry . As it stands, I would expect a bit more performance than early generation DangerDen Maze blocks.

There's potential here, if you can improve the flow. Maybe cut the top rib, and look into putting in a jet inlet; the backwash can flow upwards through the ribbed fins.

i'm not i have understood "flow geometry" and Cut top rib (tree branch), properly
but i did understand inlet jet. In my initial design ideas a passed over inlet jet to reduce pump size and maintain high flow rate.

the heat scrubbing as a result would be a favourable one. Also too kepp the design simple from a DIY point of view i thought i might add litle metal plates to the main inlet nozzel, like washers which can be swapped out.

[bigben2k]

Much better.

Flow geometry -=> path of water.

Make it. Test it. refine it. You're off to a good start.

[bobo5195]

The idea of using extra fins is a good one and I especially like the long detailed write up as it makes my brain hurt so much less.

Given your unlimited machining budget I would say it is a fairly good design. I don’t expect performance to be right up there as the fins are very thick (due to the horizontal fins). This means that the vertical fins have to be thicker and overall heat transfer area is less. Could be an interesting exercise to compare thermal transfer per unit area to something like the apogee. It may be slightly lower as the horizontal fins stop hotter water from displacing from the surface. I know this will be a small effect but it could account for a lot. There was a paper I read which tried to determine if heat transfer rate could be increased by machining 1mm wide grooves to increase surface roughness. That is another approach that could be useful.

The few little things I would change without going to something completely different

I disagree with the flow splitting approach as it half’s your mass flow rate for no reasons.
Increase the number of fins if possible.
Remove the small vertical fin at the base as it kills machinability.
Forget drilling holes in the top of the small vertical fins. It reduces performance as the water in those holes is going nowhere (just getting hot and naturally convecting) and adds extra pressure drop across the device.
The holes in the side reduce the performance as the water goes through these instead of around a nice long path in the block.

Happy building. Extended fins like these are one good approach and with tweaking maybe able to generate a quite high level of performance.

I was born in that area of Africa so I to have heard of bushman tech quite a lot .

[DX2]

I would to thank everyone for the feed back so far.

I have thought the machining through alot. and yes the bottom rib in the form presented makes machining very difficult. but keeping with the DIY theme i'd already figoured somthing out. the nice edges look prettier though.

I admit my paintbrush or sketch up drawing arnt 100% of what it will end up looking like.

so here is my solution to the bottom of channels.


place the block on its side and drill straight down.

[bigben2k]

Interesting...

I don't know what budget you have for this, but you might start with a simplified version of it. Then have one made as intended, to compare results.

[Dukemurmur]

well i will tell you right now that a dremel wont have that turn out looking ANYTHING like u want it to...even to use a mill (hand mill of course) would be a rather time consuming process but MUCH MUCH eaiser and will look like what you want....on the other hand a CNC could bang that out witht the right cutter...pretty much the same as the hand mill would need...and it would look EXACTLY like whatever the 3d model looked like down to a MAX deviation of about 5/1000...that would cost you 300+ to have that machined...unless you had the 3d file and it coded to there CNC then you can take about 100 bucks off the price....i think that design would work ALOT better for a TEC then for a CPU even with the IHS on...

Duke

[joshuass]

How tall are those fins ? It would be interesting to see if you could find the efficiency of them. I know for the block we just made ours ended up at about 70% which isnt all that great. We could have easily shaved 2mm from the block pin heights. (A pic of it is in the showoff thread)

Just somethin to think about.

[Alexio]

I don't think it will be really affective. The pin surface that touches the base is only very small. And I think it's just that connection that is the limiting factor these days. As you can see many blocks only have 6mm high fins or pins or whatever. That's because it's the optimal spot for the lowest possible deltaT with certain flowrates, that's also why a certain base thickness is optimal.

I think that with your design it will not be a very good block with a medium to high flow through it as explained above. I do think that with very low flows (say without a pump) tree fin blocks can be pretty efficient, but a maze design can never be efficient in a passive loop.

And you really have to think about the holes you want to make in the fins. I think there won't be any flow trough them and they will only be trapping the heat.

Sorry for this negative comment