• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Interesting pump design

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

arhines

Member
Joined
Jan 8, 2002
Location
Seattle, WA
Heres something for you guys to ponder:
119num4.gif



A fan drives this copper pipe cap which has impeller blades soldered on, made of more copper pipe. About 60gph, entirely silent, and costs $0.20 plus an old fan to make. I'll post the entire article (from my website) here:
This is by Rild, who writes at ZZZ. Just thought You guys might find it interesting.

A curiosity bump the size of Texas, that's what it is, and sure as sh@t it will get me into trouble every time. Recently I was told that one would have to be really "out there" (as opposed to being 'in' somewhere I suppose) to want to open a 65k E-mail attachment labeled "nude photo's of Bill Clinton". I mean it is almost certainly some horrible virus or worm, but just like poor Stimpy (of Ren and... fame) I am the last man on Earth that should be put in charge of guarding the big red button (the oh so shiny button, the bright wonderful candy like button....). Of course what this person didn't know is that a) I have only two cards left until a have a complete nude Clintons "the white water years" set from Topps (in case youre wondering it's the farm action scene with Bill entitled " piggly-wiggly " and the very rare "Vince Frosty" which has Hillary (wearing the panties with Satan on the crotch) cuddling a strangely pale and slightly limp Vince Foster) , and b) that the logic and safety conscience portions of my brain (you know the parts that try to warn you that you are locking your keys in the car or that maybe never is the best time to tell your significant other that that maybe she should lay off dessert) is constantly being beating and ridiculed, much like the school yard runt, by the parts of my brain that figure the best tool to probe the dark mysteries of a light socket is with a screwdriver, or that words like 'high voltage' and 'danger' are clearly intended to warn someone else (although if you stick a resistor in a light socket it "shoots" out a cool carbon flare ( don't blame me if you are actually dumb enough to try this )).

Now combine this with Andrews news update about the guy over at afrotechmods.com who turned a bunch of old hard drives into speakers, and you get my kind of insanity. To this end I endeavored to build a better mouse trap (except in my opinion the best mouse 'trap' money can buy is a loaded shotgun and a half a gallon of bourbon, plus maybe a chair and a radio). Or as the case may be, a better cooling system for my computer. Now I am not sure this is exactly ZZZ material, but besides the new DVD "blue" standard I really couldn't get behind anything this week so, maestro a little narcissism please...

I am apologizing in advance for the photos, they are simply terrible and do not do the subject matter justice at all (if you guys think this project has merit I may get some real photo's and generate a web page with more 'how to' type info). My digital camera works great outside, but indoors it has about the same cinematic sense as the guys filming "The Blair Witch Project". O.k enough whining on with the show .

A while back I purchased an AMD 1.4Ghz/ Abit combo with a Geforce2mx (which is now a geforce4mx440) and a 400 watt power supply . Including the power supply (it had two fans) I think the entire system had eight fans, with the CPU fan cranking out a tremendous shriek. Even though I tried a number of different air only solutions, my ears were still ringing hours after I turned my machine off, so I decided to take the plunge (ha, ha) and go to a water cooler. Figuring that I could at least eliminate the noisy CPU fan, and maybe even indulge in a little overclocking. The first thing I noticed was that all the (reasonably priced) pumps that I could find were AC and that they lacked a certain aesthetic value. Also, I noticed that the water block designs seemed really inefficient (I work with solar, water heat exchangers in particular). So I was not pleased to shell out in the neighborhood of $150-$200 and still need an ugly radiator to boot. The result is that I constructed a water cooler entirely from stuff readily (and fairly cheaply I might add) available from the hardware and computer store.





The heart of this system is designed around being as quiet as possible while still cooling with about the same efficiency as the air only solutions, to this end I have here a system that has no radiator, runs off a twelve volt pump (that can pull off the power supply) has a homemade water block, contains only two case fans (located bottom and just under the power supply both facing up), the chipset fan and the video card fan ( but more on that later ),and all the parts ( not including the orange "glow" lexan). Including the pipe and fittings it cost me less than $50 bucks. All told it produces less noise that my old celeron 433 system.

The pump itself is mostly clear, it measures about 3.3 x 6 cm produces better than about 60gph @ a hieght of eight inches (open pumping), while this may seem underwhelming it does this while pulling only about 1watt @12 volts dc . The nearest commercial pump available pulls 110v ac power which puts it at around 11 watts (eleven times more power) and only produces about 10% more flow than my unit . Plus this unit is completely self contained and does not rely on a plastic or ceramic type seal to protect the motor . Which makes it small, efficient, almost completely silent, virtually vibration free (In fact if I would have balanced it a bit better it would have zero vibration), and cheap - the proto has cost me about fifteen bucks to make (that includes the Mark I which got cannibalized due to a leak (crack) in the outer casing (working with acrylics can be a massive pain in the butt). I constructed it from and old Intel Celeron 433 stock fan, by simply encasing the bottom of the fan electronics in epoxy and then spray painting a couple of layers of paint on the coil windings, posts and 'magnets' , and then building this nifty impellor and attaching it to the face of the fan . There is a trick to building the impellor that makes it a piece of cake.






Here is a shot of the water block, pretty straight forward design:





All copper pipe, and assembled with a great deal less trouble than you might imagine (it actually took me longer to cut the pipe than it did to solder it all together, that little trick again). If you look closely at the top you can see where I dropped it , but dauntless I carried on... cough , cough .. the bottom or mating face is a very thin sheet of copper , the flexibility of the non mating areas allows a tighter fit with AMD's little spongy feet and my nifty hold down , you just need to make sure that your small mating surface is as flat as possible - not all that hard really





The largest pipe is 1-1/2" and the smallest is 1/4" ,this method was a test to get maximum balance between efficient use of flow turbulence and contact surface area. Before the outer pipe was fitted a hole was drilled ( not visible) directly above where the CPU slug would sit and the 1/4 " pipe. Actually two holes comprising a "V" shape as seen from the side so that the water is only separated from the CPU slug by the wall thickness of the 1-1/2" pipe combined with the thickness of the mating plate. It was added with the idea that under extreme load the heat transfer times could be decreased .
From experimental evidence I would guess that one would only need to go as high as 1" pipe (instead of going all the way to 1-1/2") as I believe that anything over twice the size of your feeder tubes is probably just a heat reservoir and cavitation trap. It is interesting to note that bolting the thing to the CPU just as it is pictured above and running a fan across it achieved better cooling numbers than my $30 aluminum/copper AMD approved store bought sink.
I had tons of trouble figuring out how to hold this baby down so I made this little wonder :





By simply squeezing some 1/4" tubing and soldering some Romex wire and a brass nut to it . To make sure I didn't crush my slug by overtightening the screw I cut a small section of 1/2" copper pipe (thick walled) then cut it in half so that seen endwise the pipe instead of forming an "o" formed two almost "c" shapes. You then place the half pipe between the waterblock and the screw (with the "c" facing block). This bit of pipe acts as a very, very rigid spring and makes it almost impossible to overtighten the clamp (unless you smash it completely flat ).

here is a close up of the heart of this gem:





The little rings are to keep the hose from crimping , the rest of the "case" is simply 1/2" copper pipe , and the only tough bit to find might be the piece of copper I used for the backside ( where the pump is bolted ). The whole thing holds about a quart (946ml) of fluid.


Ok so now the important bit: How well does it work?

First off, it is incredibly quiet, In fact I can just barely hear it running at a distance of about four feet. As for cooling power. Here are some numbers based on six weeks of the pumps operation. At ambients of 60F, the reservoir (once everything warms up - about an hour of CPU load) runs at 70F and the CPU runs with an idle of 85F and a load temp of 90F. At ambients of 70F the reservoir hovers between 79-80F and the CPU idles at 89-90 with load temps of 95F. With ambients of 75F (as High as I have been able to test so far) the reservoir has been about 82F , with the CPU idle : about 94 and the load : 99-100F.
these have been very consistent. If you are wondering about overclocking, I have jumped the system up to 1533 @1.82 V the numbers shift up the scale about 8-10 degrees F with a more noticeable separation between idle and load. One thing to add is that the air directed in from the bottom blows across the "waterblock" and lowers the temps by about 4-5 F (in fact the next model I make is going to be a combo air / water block cooler using some material donated from Solahart (Australian based solar company - special thanks go out to Jeff Brown for donating) the heat pipe which is a 1/2"copper pipe wrapped in aluminium fins , which , a four and a half foot section can dissipate about about a kilowatt an hour - passively!).
All In all I am fairly pleased with this set-up and it has now become my favorite gaming rig. Also, I believe that this type of system could be incorporated into existing case designs using the body panels to help the copper pipe radiate off the heat.

You like? If anyone wants a link to the story (with pictures, etc...) let me know (i'm under the impression that random linking isn't cool)

-arhines
 
interesting, topical links are perfectly fine! Plz share it with us, it looks like a really cool project! :D

The idea is, they just don't want people spamming links around or plugging their own site.
 
linky: http://www.zzz.com.ru/119.html . Has some pictures of the case, waterblock, etc...
a few comments at the bottom made by rild, are of interest:

just for the curious :

I was suspicious of my numbers so I calibrated all my thermostats (three letro dial gauges , three mercury thermos ,a digital meter, and a digital thermometer (like you use to see if you have a fever )) with the on MB sensors , so I am reasonably sure I have some acurrate measurements and ran a bunch of temp tests again. the only thing I changed was to add a shield to help direct more air over the "waterblock" .(temps after running the cpu under heavy load for about an hour)
here is what came out :
ambient room : 65.5F(18.6)
inside case : 78.0F(25.6)
water resavoir:84.5F(29.2)
cpu(Load): 92.6F(33.7)
using the hardware calculator from the aforementioned site :
using 72watts (assuming 100% load)
Ambient to cpu(load) C/W = .209
Resavoir to cpu(load) c/W = .063
using 63.36 watts (assuming 88% load)
A to cpuL: C/W = .238
R to cpuL: C/W = .071 -not too bad for something I made in the kitchen sink eh?

yes the cd-rom is transparent ;-)

, and no radiator - for several reasons
, to show that water cooling and case integration could be done without the need for a rad ( if you had aluminium sides on this thing it woudl have excellent numbers)
2. that most watercooling "blocks" are very poorly designed requiring huge flows and radiators
3. to show that with the correct design structure a simple low flow twelve volt system you could outperform any number of more expensive and elaborite systems (elegance and efficincy )-
4. I was showing off a bit :)
 
Bumping it up...just to let more people read this (more than one other person has to be interested...)
 
thats cool!
I LOVE THE CASE! and do you have a article or guide on how to make that pump?
 
If I read correctly, the fan's vitals were waterproofed, then the fan was used to pump water, instead of air. I believe the "impeller" is there to push water out from the center, into the fan's blades. Good way to kill the dead spot :D

Can you imagine how easy this would be with one of the new TMD fans? Since all the vitals and circuitry are on the outer ring, you would only need to adapt it for water use. Of course, depending on how much the fans cost, it might be better to just buy a pump.
 
I was hoping someone would ask that :)
Rild (the writer) is thinking of making a page with a howto on the whole deal...if he gets enough "please do"s he'll probably do it.
You can reach him by email with any questions/demands for a howto at: [email protected]

I've been trying for a while to get him a tip magnetic drive fan for this project, but YS Tech doesn't seem to have anyone who speaks english on their staff..hmm.
 
Yep they sure do. The thing that makes me skeptical about using a fan to pump water would be the fact that fans are designed to move air instead of water, and a pump may be more cost effective for this sort of application. the other thing that may worry me is how the life of the fan would last. Water is much more dence than air and thus be harder on the fan, therfore reducing the fans life.
 
Those were also initially my thoughts - but apparently, the thing spins very fast, fast enough that it isnt straining it on each cycle (ok, so it's obvious I don't really understand how they work/decay...but you get the point). Also, at that price, you could opt for quadruple redundancy :)
 
Back