Slashdot is powered by your submissions, so send in your scoop

 


Forgot your password?
Close
typodupeerror
×
Hardware

CPU Convective Water Cooling 267

Posted by chrisd from the ugly-computers dept.
biso writes "The possibility of cooling a CPU with gravitational convective flow of water is here analyzed and experimented with positive results. Many liquid cooling systems have been experimented by overclockers to better dissipate the heat from CPUs. The major part of these coolers is characterized by a relatively complex system requiring pumps or other active devices. Sometimes even liquid nitrogen is used. I built a simpler cooler, able to dissipate the same heat flux of a normal heatsink."
This discussion has been archived. No new comments can be posted.

CPU Convective Water Cooling More

CPU Convective Water Cooling

Comments Filter:

  • /. effect (Score:5, Funny)

    by Exiler ( 589908 ) on Tuesday February 04, 2003 @11:11PM (#5228400)
    He's going to be boiling noodles in a minute

    • Re:/. effect (Score:5, Funny)

      by killthiskid ( 197397 ) on Tuesday February 04, 2003 @11:20PM (#5228463) Homepage Journal

      I'm just impressed with a big bowl of water balanced on top of a open tower case. Now THAT takes balls.

      The guy obviously doesn't have cats or any other pets.

      • liquid and open cases (Score:4, Funny)

        by trmj ( 579410 ) on Tuesday February 04, 2003 @11:52PM (#5228604) Journal
        I leave the sides off of my computer for better air flow and less dust collection, but mainly because I'm too lazy to put in case fans.

        Since my case has to sit on top of my desk, this also gives me a little more space to put stuff (inside the case, out of the way of boards / wires etc).

        What do I keep in there? First is the reciever for my wireless keyboard / mouse [gyration.com], the syncing cradle for my handspring visor edge, and a few important papers. Those are the things that don't really move.

        The fun things in there are the mountain dew beer glass, the caffeine shot glass, and usually 2 - 3 open cans of dew. All have soda / caffeine in them and are actively used.

        No, nothing has spilled yet, and yes, I do have cats. 3 of them. They know that if they go near the sacred tower, they get stuff thrown at them (usually a slipper, but on occasion a t-shirt if I find that first).

        On a side note: I in no way have the balls to water-cool my case, other than replacing the cold soda every 30 minutes.

      • Re:/. effect (Score:2, Informative)

        by Lasuuco Tulkas ( 598141 )
        Actually, the water he is using is most likely de-ionized water as this prevents corrosion in the pipes and water block - although sometimes it is best to mix in some detergent to lessen the water tension in the lines to work out the bubbles.
      • or even children :)

    • Transformer Oil - Electrical & Thermal Propert (Score:5, Informative)

      by BigBlockMopar ( 191202 ) on Wednesday February 05, 2003 @02:10AM (#5229400) Homepage

      Nah... I'd use transformer oil, and I don't think a Lipton Cup-a-Soup would taste quite the same.

      Transformer oil, however, is probably quite suitable for use in a CPU cooling system.

      It has a higher breakdown voltage than air and is almost infinitely less conductive than real-world (ie. impure) water. Transformer oils are specifically designed for use as an insulating material in large power distribution transformers. Electric utility transformers at power substations, operating in the range of hundreds of thousands of volts, would arc between windings if the oil leaked out of them and air - with its lower breakdown voltage - seeped in. (Air breaks down at about 3kV per millimeter.) You can feel pretty confident that leaked oil won't short out IC pins on your motherboard. Hell, you could also ditch your power supply fan and fill that full of oil, too - just beware of relays and other mechanical components.

      Heat transfer is a big reason for oil, too. In a car engine, much of the heat is generated by friction in the bearings, and motor oil pumped through the bearings takes that heat away. Transformer oil doesn't have to lubricate, nor does it have to carry away huge amounts of impurities or combustion by-products as in a car engine - the biggest requirements are heat carrying capability and high breakdown voltage. Large pole pigs (pole-mounted power transformers) are usually oil-filled and often have pipes coming from the bottom and going to the top - they serve as radiators. Oil flow is not by pump, the reliability would be too low - they're convective, too.

      Finally, viscosity. Yes, this might be difficult, but transformer oils are available in a variety of thicknesses. You want a viscosity corresponding to SAE 0, which is the same as water. Even less might be available, though I've personally never seen it.

      Density changes with temperature rise will have to be considered, since the lower density of hot liquids causes them to rise in the system (and is also the physics behind lava lamps). The system that guy designed is based on the density changes of water. Transformer oil won't behave the same way; accordingly, you'll have to whip out the old slide-rule and do some math. Calculus is your friend. Fortunately, the data on transformer oil should be readily available, it's an important design criteria.

      Voltesso and Diala are good trade names which I've personally used in transformers loaded to hundreds of kilowatts at over 250,000V, at RF frequencies. (FAA obstruction lights on large VLF radio transmitting towers.) They're ALL PCB-free, and while you don't want to drink it, they're no more toxic than motor oil. And it takes a hell of a lot of work to make them catch fire.

      In short, transformer oils are available in a variety of viscosities, are specifically engineered for their thermal transfer capabilities, are not electrically conductive, not dangerous, and are suitable for almost all of your electronic cooling needs.

      The only problem I forsee is that you're gonna have a hard time buying them in quantities less than 45-gallon drums... though the drum would make a great passive radiator. Seriously, talk to a couple of linesmen with your local power utility, maybe you'll be able to talk your way into a couple of gallons of it.

      And once that's done across all the machines in your compile farm, you can get to work tackling the big problems of why Linux isn't ready for the desktop yet [glowingplate.com].

      • That was a very informative post, does anyone know if motor oil is conductive, I would think that a quart of 5W20 or another light oil would flow pretty well through a small submersable pump. My favorite transformer oil story was from a local Utility guy who came to our school years ago. He asked us if we ever wanted to shoot the transformers with our pellet guns, and then told us that he used to want to as a kid, but that it would be a bad idea since the oil would leak and the transformer would fail possibly explosivly.
        What I found funny was that I don't think a single person had ever thought about shooting a transformer until he mentioned it, and the fact that they could explode might be something that would encourage young boys to shoot one with their pellet gun.
        • I dunno, and even if someone did the motor oil they used would probably be different from yours.

          So the best way is to check it yourself - stick a multimeter and measure the resistance. Given that most stuff on a motherboard is low voltage I wouldn't be overbothered by resistance-voltage properties of the oil.

          • Re:Transformer Oil - Electrical & Thermal Prop (Score:4, Insightful)

            by gl4ss ( 559668 ) on Wednesday February 05, 2003 @07:16AM (#5230226) Homepage Journal
            simple multimeter tests don't just cut it, unless it can measure the 'hit through'(sorry, my english sucks and i don't know the proper word) capabilities of the substance.

            water would make a fine insulator unless the 'hit through'(amount of voltage differential needed for the electricity to jump/hit like lighting through something) was small(and no, i'm not an electricity-engineer).

            anyways, this kind of cooling has been done(submerging the whole mobo in something), couple of times. one guy used some biograde mineral oil succesfully.

            theres at least one no-pump commercial solution too for liquid cooling, but it has a fan on the radiator.

            though, in my opinion, getting rid of the pump doesn't bring you anything 'extra' since the pump makes next to zero noise, and noise damping the pump from environment is easy too. the real problem lies in how to get the water to keep cool without having extra fans(heatload on it gets quite big if you have cpu, gfx-card, chipset, psu, hd's and etc watercooled for silence). the bowl of that size that's in the article won't cut it.

            and really, k6-2 could be cooled enough with just about any lump of metal compared to the 76w+ modern cpu's.
      • Why not take in one step further and install a CPU-powered lava lamp? It's the next logical step from AOpen's valve amplifier motherboards.

  • odd (Score:5, Insightful)

    by The Other White Boy ( 626206 ) <theotherwhiteboy.gmail@com> on Tuesday February 04, 2003 @11:12PM (#5228408)
    ok, i can see the extra bit of silence from not having a pump would be nice. but ... isnt a bowl of water on top of your computer just asking for trouble? something tells me this guy doesnt have cats.

  • not that complex... (Score:5, Insightful)

    by AmigaAvenger ( 210519 ) on Tuesday February 04, 2003 @11:14PM (#5228422) Journal
    I've built numerous different water cooled systems, and the $20 pump from the aquarium supply place is NOT the most complex piece. A good waterjacket for the cpu is by far the most complex and generally most expensive single piece, and also the one that is most critical for good performance. Still need the waterjacket in this design, so it isn't really saving anything...

    • Re:not that complex... (Score:4, Informative)

      by jasonkohles ( 546421 ) <slashdot@jasonkohles.com> on Tuesday February 04, 2003 @11:19PM (#5228462) Homepage
      Sure it is, he says right out he didn't eliminate the pump because it was the most expensive part, he wanted to eliminate it because it was the loudest part.
    • I'm sorry but I fail to see how a water jacket could be the most complex part. The jacket itself if made properly (which is to say, as simply as possible) will consist of two pieces. I would personally then weld/solder/braze/WTFever you call it the two pieces together. The whole thing could be made with a 2.5dimensional mill in a matter of seconds with only one or two tool changes and with repositioning each "half" of the block once.

      The pump, on the other hand, is going to involve either a motor and a turbine and something like it (dramatically more complex, perhaps 10-20 parts there alone if you break them down) or something similar. At the very least you'll have an impeller driven by magnetic induction which implies a certain parts count on its own.

      A water jacket is just one flat piece (a lid) and one channeled piece, and a couple of fittings.

      • for 2 reasons water jackets are not produced in bulk like pumps and a good one like the one he uses, is more like a mini car radiator ( look at the pic) something like that is going to cost a lot to produce
      • also, the jacket has to fit on the cpu and attachment points without cracking the cpu (athlon) or breaking the tabs. submersible aquarium pumps are several magnitudes quieter than the quietest hard drive, you will NOT hear them in a normal environment. Flow rate makes very little difference, even a very small rate is enough to keep liquid circulating through the jacket. (watch the output temp as an indicator of the needed flowrate.)
    • ...the motherboard manufacturers just don't put the processors on the back side of the motherboards.

      Think about it. If the processor and other heat generating chips were on the reverse side you could mount the motherboard in such a way as to press against the large metal backside of your case cabinet. The case cabinet could be designed in such a way as to have indentations that force the processor flat against the case with heat sink gel. In fact the entire case back could be a water-cooled heat sink. This would keep the inside of the case "water-free".

      Sure, this would take a radical new case design and motherboard (another industry standard), but that seems to be where we are headed right? I mean processors are getting hotter and Intel and AMD are trying to figure out what to do.

      Why not?

      BTW, I think it has something to do with the fact that the motherboard components are all wave soldered on one side. This would melt most plastic components on that side. I'm not sure about this.
      • force the processor flat against the case with heat sink gel


        A nice idea in principle, but cases are there to protect the components inside. If you go physically attaching the components to the case, all you have to do is knock the case and end up stressing the CPU die.


        That said however, I agree some sort of more "holistic" approach could be wise.

      • I'm glad you're not too sure about this. Since wave soldering requires getting the metal bits (pins, copper traces, etc) up to soldering temperature for the soldering process to actually work it doesn't matter where the plastic is if it touches metal that needs to be soldered.

        Typicially medium temperature plastics can easily handle the 300-500 F temperatures needed for soldering. The problem with mounting the socket on the backside is that you'd have to solder the opposite side of the board, and the socket would have to be covered so the wave soldering process wouldn't force solder into the socket holes.

        Since the chipset is often BGA (Ball Grid Array), and many components are surface mount and require a seperate process anyway (usually infrared heating, sometimes oven heating) then the extra soldering isn't that much of an issue, nor is the protective cover over the socket. Rerouting the board backwards would be a pain though, and since it'd essentially require new case designs and could cause burns if not properly vented...

        Well, there really isn't a *good* reason to go to that amount of trouble for what appears to be a very small market of people who would trade $200 - $300 for a passive heatsink on the outside case. (define very small as < 5 million units sold worldwide)

        -Adam

  • Nitrogen is a GREAT coolant! (Score:5, Funny)

    by Tsar ( 536185 ) on Tuesday February 04, 2003 @11:14PM (#5228425) Homepage Journal
    It's cheap, reliable, odorless and environmentally friendly, and refills are available everywhere. I myself have been using a 78% nitrogen gas mixture at 14.7 PSI to cool all my computer components for years now. Where did I get the idea? I guess I just pulled it out of thin *SMACK*

  • too hot (Score:5, Insightful)

    by Anonymous Coward on Tuesday February 04, 2003 @11:14PM (#5228426)
    I really think that phase change cooling systems are the future of the PC. Only with phase change cooling systems do you get high quality cooling able to remove the utmost heat away from a CPU and cool it to below freezing.

    I saw a presentation by Intel last year in which it pointed out that modern CPU's emit more heat per area than molten lava, and they expect that within a few years they will emit more heat per area than the sun.

    With these considerations passive water cooling is only a good first step and bound to be insufficient, even over the short term.

    • Re:too hot (Score:5, Funny)

      by havardi ( 122062 ) on Tuesday February 04, 2003 @11:35PM (#5228528)
      I was just thinking thinking about the same system he has setup-- but where the water is boiled off the chip into steam... If it could be totally sealed that would hella cool. Of course, it would have to be large. But, you could add plants and simple animals to the ecosystem and it would be an interesting conversational piece

    • Re:too hot (Score:4, Insightful)

      by PsychoKiller ( 20824 ) on Tuesday February 04, 2003 @11:41PM (#5228556) Homepage
      I saw a presentation by Intel last year in which it pointed out that modern CPU's emit more heat per area than molten lava, and they expect that within a few years they will emit more heat per area than the sun.

      Not quite true. I saw the same graph, and it was an extrapolation of current trends.

      Obviously we aren't going to emit more heat/area than a nuclear reaction, something is going to have to change in the design of Intel CPUs.
      • Remind me to shitcan my furnace and buy a couple dozen Pentium VI's.

      • Nuclear reaction? (Score:3, Interesting)

        by dirtyhank ( 135909 )
        I guess Intel meant the surface of the Sun. This outer layer (photosphere) has a temperature of 6000C aprox. Yeah, it's damn hot but quite below from the 15000000C of the core (where the nuclear reactions take place).

        Molten lava temperature is ~1000C so only 5000C to go :-)
    • I saw a presentation by Intel last year in which it pointed out that modern CPU's emit more heat per area than molten lava, and they expect that within a few years they will emit more heat per area than the sun.

      Hrmm...I was going to buy a space heater, but I guess I'll just wait for the P5's to come out instead.

      All kidding aside, I suspect that we're running into some physical barriers with respect to clock synchronized computing. Seems to me that we really need a paradigm shift such as asynchronous computing [sciam.com] to take us to the next level.

  • Content of the page (Score:3, Informative)

    by Anonymous Coward on Tuesday February 04, 2003 @11:15PM (#5228429)
    Abstract

    The possibility of cooling a CPU with gravitational convective flow of water is analyzed and experimented with positive results.
    Introduction

    Many liquid cooling systems have been experimented by overclockers to better dissipate the heat from CPUs. The major part of these coolers is characterized by a relatively complex system requiring pumps or other active devices. Sometimes even liquid nitrogen is used.

    My intent was instead to build a cooler able to dissipate the same heat flux of a normal heatsink, but without the annoying noise of the fan.

    A first prototype was built out of a regular heatsink. Holes were drilled in the aluminium finning, and copper tubes passed through them. An aquarium pump provided the necessary pressure for circulation.

    Figure 1. First Prototype--Front View
    (picture)
    Figure 2. First Prototype--Side View
    (picture)

    The system was silent and reliable. But with bigger pipes and a lower pressure drop would it have been possible to take away the pump? Simple calculations showed that it would have been perhaps feasible and a prototype was built.
    Temperature on Heatsink Surface

    Roughly:

    Power to be dissipated: powd = 80 W

    If the heatsink is a little copper box to put over the CPU, a reasonable value for the surface available at copper-water interface can be: surfc = 0.01 m2

    The heat transfer coefficient on the water-copper boundary layer can vary from a few watt per square meter per kelvin if the flow is slow and laminar to more than 1 kW K-1m-2 when the flow is very fast and turbulent. If the coefficient is supposed to be: texc = 100 W K-1m -2

    The difference of temperature on surface will be: delt = powd / (texc surfc) = 80 K

    It appears that the water should boil on the surface of such a little heatsink, but radiation wasn't taken into account and the geometry of the box is complex, so it's not clear if there could be turbulence and with which effect. If necessary the surface could be enhanced with fins or by increasing the dimension of the equipment.
    Convection

    Supposing that the heatsink could be able to exchange the heat between the CPU and the water, would it flow through the pipes?

    Power to convey: powd = 80 W

    Length of the circuit branches between the CPU and the radiator on the top of the computer case: heigh = 0.8 m

    Equivalent length of the circuit (we take into account the bends too): len = 2 m

    Radius of the pipe: rdp = 9 10-3 m

    Rate of change of water density against temperature: dct = 0.55 kg m-3 K -1

    Water density: rho = 103 kg m-3

    Water viscosity: eta = 10-3 decapoise

    Specific heat of water: wsh = 4180 J kg-1 K -1

    Gravitational acceleration: grav = 9.8 m s-2

    Pi: pi = 3.14

    Difference of temperature between ascending and descending branch: deltat

    Difference of density of the water in the two branches: deltarho = deltat dct

    Difference of pressure due to the difference of density: deltap = deltarho grav heigh

    Volume of water conveyed per unit time: vot

    Pressure drop in the pipe: deltap = vot 8 eta len / (pi rdp 4)

    Power conveyed: powd = wsh rho deltat vot

    Putting it all together: deltat2 = 8 powd eta len / (wsh rho pi rdp 4 dct grav heigh) = 3.4 K2

    Everything should work with a temperature difference of less than 2 kelvin. Consequently the radiator isn't required to be very efficient.
    SIRPAL-1 Prototype

    The SIRPAL-1 prototype was made using a 5 mm thick copper sheet for the base, and 2 mm thick copper sheets for the walls. The edge of the square base is 55 mm long. Inside there are two plates 25 mm wide. One is vertically aligned, soldered to the base, to increase the exchange surface near the CPU, the other is horizontal, soldered between the input-output pipe fittings, to guide the fluid in the right direction.

    A test was performed on a K6-2 450MHz which dissipates a power of about 25 watt. The ambient temperature was 18 celsius degrees. After a few hours the CPU temperature, measured by the PC board sensor, was at least 1 kelvin lower than when the fan is used. External surface temperatures: 19 celsius degrees on the pipes; 24 celsius degrees on the copper box.

    A drop of ink in the water revealed a slow flow as expected. It worked so well that I think a more powerful CPU would be efficiently cooled too.

    Figure 3. SIRPAL-1
    (picture)
    Figure 4. SIRPAL-1--Testing
    (picture)

  • Allow me to suggest... (Score:5, Funny)

    by Ikoma Andy ( 41693 ) on Tuesday February 04, 2003 @11:16PM (#5228434)
    ...using sponges as sound dampening material inside the case. Who knows, it just might save your computer when somebody bumps the frickin' table and dumps that bowl full of water over everything!!

  • Welcome to the party, PAL! (Score:3, Funny)

    by freeze128 ( 544774 ) on Tuesday February 04, 2003 @11:17PM (#5228444)
    Didn't I see that figure 3 photo in Die Hard with a vengance?

  • Why not use transformer oil? (Score:3, Interesting)

    by xtronics ( 259660 ) on Tuesday February 04, 2003 @11:19PM (#5228460) Homepage
    Trasformer oil is insulating and would not short-out anything. It could also cool the powersupply at the same time, without the inherint saftey risk water subjects us to.
    • thats the problem its insulating i.e. not tranfering the heat outside of the case, people have tried it and found it just just transfer enough heat to matter
    • I'll tell you why. It is extremely hazardous stuff. In order to increase the boiling point of the oil, PCB's (ya know, polychlorinated biphenyls) and other toxic nonvolatiles are added to the oil. If you spill that crap, it's worse for you and the environment compared to Freon.
    • There are probably much less hazardous fluids than thransformer oil. Transformer oil probably wouldn't work for a convective system anyway, since my bet is it's way too viscous for the forces at hand to work efficiently.
      • Let's pump it and two read the following standards. The oil has to be thin to cool transformers.

        Polychlorinated Biphenyls (D 4059) Regulations prohibiting the commercial distribution of polychlorinated biphenyls (PCBs) mandate that insulating oils be examined for PCB contamination levels to assure that new products do not contain detectable amounts.

        Viscosity (D 445) Viscosity is the resistance of oil to flow under specified conditions. The viscosity of oil used as a coolant influences heat transfer rates and consequently the temperature rise of an apparatus. The viscosity of an oil also influences the speed of moving parts in tap changers and circuit breakers. High viscosity oils are less desirable, especially in cold climates. Standard viscosity curves can be generated using Method D 341 by measuring two or three data points and plotting the data on special chart paper. The resulting curve can be used to interpolate or extrapolate values at temperatures where the viscosity is not measured directly.

  • tesla turbine... (Score:2, Interesting)

    by Aine ( 647301 )
    ...powered by water (or even antifreeze) could conceivably do two jobs: provide cooling for the cpu and produce power. It wouldn't have to be all that big either. It might not produce that much power, but probably enough to light a small desk lamp.
    • I doubt you could come anywhere near that level of power. While tesla's disc turbine is impressive it is not going to generate that much energy unless your "small desk lamp" is an ultra low power LED. Look at the basic facts of the situation, your CPU is only dissipating 40-120 watts, assuming it's modern. Less if it isn't. What kind of efficiency do you really think you're going to get between the cpu and the fluid? How much heat energy will be lost to radiation before it reaches the turbine? How efficient is the turbine? The generator?

      In the end you might be able to utilize excellent design to get enough power to light a small bulb. I think you could get more power more cheaply by using a sizable heat pipe system which ran into a stirling engine made out of soda pop cans :P

  • PC or 400psi TCM? (Score:4, Interesting)

    by gelfling ( 6534 ) on Tuesday February 04, 2003 @11:22PM (#5228474) Homepage Journal
    It's kind of ironic istn't it that CMOS PCs will eventually need the same plumbing as the TCMs they were supposed to replace.

    Hey I had a 9021-721 MVS/ESA mainframe that used TCM's cooled by a 400psi cooling system. The great thing about the next gen CMOS mainframes was that even though one TCM was now replaced with 3-6 CMOS units, we didn't need a massive chiller system.

    Oh well, guess everything will have huge ass chiller pumps now.

  • Simpler, Cheaper Method... (Score:5, Interesting)

    by jaybird144 ( 558619 ) <jaybird144@@@gmail...com> on Tuesday February 04, 2003 @11:24PM (#5228485)
    While I admire all of those who are willing to actually pipe water into their computers for the sake of keeping them cool, I am still more than a little frightened of this idea. Knowing me, something would end up leaking (probably through fault of my own) and my precious tower would go up in a ball of flame or some such. Rather, I choose to take your regular old air cooling and make it more effective. Through proper cable management, good airflow paths, a set of nice quiet fans, and the ability to control them with respect to how much they are needed, I keep my fan almost as cool as a friend of mine who does have a liquid cooling system, and with very comparable noise levels. If you're willing to do some work and set up a decent air-based system (and be willing to clean out the dust), there's no need to balance a bucket of water precariously on top of one's computer... : |
    • Reading your post gave me an idea....

      The reason that generally only dedicated individuals go to the trouble of using a water-cooling system in their case is because water conducts electricity, right? And air doesn't, which is why air is much safer -- well, risk free. Spilling water on the innards of one's computer would short it out, or cause a surge, or some electrical-type thing; but no amount of air flowing through a case is going to cause any damage, unless we're talking hurricane-force winds.

      So why don't liquid water-cooling systems use some liquid with the useful thermal properties of water, but without its ability to conduct electricity? In other l33t cooling stories on /. I've read about liquid nitrogen being used, but I imagine it's so cold that the damage from a leak there would be largely from temperature. Surely there is some easily-obtainable liquid, oil or something, that wouldn't cause damage to computer components if it came into contact with them, but would be effective at cooling the system.

      • So why don't liquid water-cooling systems use some liquid with the useful thermal properties of water, but without its ability to conduct electricity?

        They do. It's known as 'distilled water'. Water is non-conductive (or close enough for most purposes) when it is free of salts.

        Some people have tried using Flourinert for cooling but it's very expensive and I don't know what the minimum quantity is but I think it's multiple gallons.

        In the end distilled water is probably your best bet. Putting filters on any intake fans or vents on your case will prevent distilled water mixing with conductive dust and shorting things out. If your case has multiple intake paths, either it's poorly designed, or you just need multiple filters. Most cases should have only one intake point, and air should more or less move throughout the entire machine before exiting, whether because you have multiple exhausts, or because air is forced to flow in such a manner.

        People have also used mineral oil but with any cooling medium other than distilled water (which also does not corrode things for obvious reasons) you have to ask yourself if the seals in your pump can handle it and so on. Water is the most commonly pumped fluid, so we're quite good with it these days.

        • Water is corrosive. A small leak might not immediately be noticed but water will destroy everything on its path over the mobo if it gets the time.
          • While in the long term even distilled water is the 'universal solvent' and will take apart almost anything, that is a long. ass. time. For instance, if a cat pisses in your keyboard (happened to a VT100-AA of mine once) you can soak it in distilled water overnight, and let it dry.

            I should think it would be instructive to have some way to measure the water level in your cooling system. I would probably use an old-school water glass such as those used on steam locomotives, and of course thermal monitoring with automated shutdown, and/or cpu idling. You DO want to know if you have sprung a leak.

        • Distilled water, when brought into contact with, well, just about anything (metals, for example) become conductive very, very quickly. Go ahead and soak your running computer in distilled water and see how long it remains non-conductive for.

    • Re:Simpler, Cheaper Method... (Score:5, Interesting)

      by Tidan ( 541596 ) <tidan_md@y[ ]o.com ['aho' in gap]> on Tuesday February 04, 2003 @11:50PM (#5228589)
      So much for innovation...

      This guy is trying to be creative and show us a unique way of cooling a computer. Obviously this type of water cooling is somewhat of a new idea, and has yet to be perfected. I mean, he even says that he tested it on a K6-2 computer (relatively low power dissipation).

      Give the guy some credit here! Instead of saying why fans are better and that water (or any liquid) spells disaster, why not try and understand why this is better than the old system?

      First - liquid cooled without a pump = no mechanical parts to fail. Those heat-pipe heatsinks that are becoming popular have liquid inside them, and you probably aren't worried about having them leak. It's possible that there might be a commercial product that does the same thing with gravitational convection of liquids.

      Secondly - eliminating (or reducing) fans not only eliminates noise, it also improves reliability. How many of us have had fans get noisier over time and eventually die on us?

      Anyway, I say more power to this guy. someday we might have the CPU mounted at the bottom of our case, and have a cool lava-lamp looking effect running up the side wall. Let's not be quick to fear new technologies - look at the possibilities first!

  • "Sometimes even liquid nitrogen is used. " (Score:4, Informative)

    by Chris_Stankowitz ( 612232 ) on Tuesday February 04, 2003 @11:30PM (#5228503)
    Yeah, and other times they go a little nutz and use Fluorinert [slashdot.org]
    • And then they don't bother to check the material data sheet for the type of Nert they spent a good chunk of cash on, immersed dry ice in it, and had it turn into jelly. Smart. (As I understood, there ARE varieties of the stuff that can be cooled to that level without becoming viscous).

  • I like my fan (Score:2, Funny)

    by DaHat ( 247651 )
    Heat sink and fan, such a nice and simple system, is best when you touch the heat sink wrong and can get a nice burn!
  • I know I have. I think eventually we'll see some commercially available sealed systems that use some liquid (probably not water because of its corrosive properties) and convection to move it around.

    Something idiot proof and self contained with no mechanical parts, save perhaps a slower turning large fan to exhaust air.

    So far existing heat pipes are not much more than a novelty. The wick style ones with alcohol as a fluid (mentioned here awhile back) look promising.

    The giant skivved aluminum/copper blocks with the big ass fans aren't going to cut it, not just because of noise, but practicality.

  • 5 -- You just inherited $700 from your great-grandfather that's just waiting to be thrown out the proverbial case window

    4 -- A loud smelly watercooling kit is the perfect complement to your neon light tubes

    3 -- Why spend $200 for a brand new P4 when you can pay $500 for supplies and crank your P2 up 30 MHz?

    2 -- That rock fountain you got for Christmas can be put to a much geekier use

    1 -- To impress your classmates at Chubb Institute

    Free blogging for geeks at *nix.org [starnix.org]

  • Newer processor would be more interesting (Score:3, Insightful)

    by eamber ( 121675 ) on Tuesday February 04, 2003 @11:34PM (#5228522) Homepage Journal
    Alright - dude did this on a K6-2 450. They're about HOW many years old now?

    I'd be much more interested to see him cooling an overclocked Athlon XP 2100+ with 1.9 vcore running at 2400 MHz - or something of the like - with the method in the article.

    Ah well, I'm happy with my good 'ol pumps and radiator, myself.
    • Yes that would be interesting. Infact, it might work even better especially if his water column was better designed with the larger processor. Larger processor = more heat = less dense water = faster rising water = more water moving to replace the warmer water. But I have nothing to back this up. But hey, Ford was able to cool the Model T with passive water cooling, so it could work. Key word is could. Don't quote me on it. Side Note - Nuclear subs now use convection cooling to end even the *quite* pump noise.
    • The question here is how much heat can be removed how quickly regardless of the processor involved. If the processor puts out x amount of heat and this setup can remove y amount of it, it should be able to do that for any processor that puts out x amount of heat. Besides, if the new processors ran any hotter than the old socket 4 waffle irons, overclocking would be far and away the prime cause of house fires.

      If you were doing this as an experiment to see if it's workable or not, wouldn't you try it out on old equipment before risking the expensive new stuff?

    • Alright, maybe he didn't want to spill water on a new comp? Obvious? I guess not.

  • You know...for the cost of a cooling system... (Score:4, Funny)

    by Quaoar ( 614366 ) on Tuesday February 04, 2003 @11:36PM (#5228534)
    You could just buy a *second* computer.

  • You wanna build a quieter computer? (Score:3, Interesting)

    by jsse ( 254124 ) on Tuesday February 04, 2003 @11:40PM (#5228547) Homepage Journal
    You may look here [7volts.com]

    Without the need to implement that bulky bowl of water. :)
  • How about recycling? A water, being heated, can become as a steam. It's energy can be used to produce a rotation to for fan propeller. Just kidding.

    Seriously, the steam can condensate in pipes outside of PC and come back as a liquid. Just add a compressor and we've got a refrigerator. Can I keep my soda cool inside my PC?

    • err, assuming the thing works, your not going to get any steam, my cpu a p4 2ghz, runs about 65C (with fan) way below 100 required for steam, even if it failed chances are the cpu would fail before it got hot enough for steam to be produced esp with that much water
      • All right. Just keep your motherboard working in a fridge. But warn your wife to avoid placing some chicken on a top.
        • thats couse a fridge keed things cold enough for water to form from the air, what this guy set up keeps the water at room temp if the guy used a pump (he does not) and the cump was fast enough the water would get colder then room temp and water would form around the pipes

  • I found it! (Score:3, Funny)

    by LongJohnStewartMill ( 645597 ) on Tuesday February 04, 2003 @11:42PM (#5228562)
    What? You guys are still using water? I found The Ultimate Cooling Device: Hilary Rosen (aptly named The Ice Princess).

    • Re:I found it! (Score:3, Funny)

      by Cyno01 ( 573917 )
      Yeah, but w/ her sitting on your processor, she'll just nag you about downloading stuff. A water cooling system would be *much* quieter.
    • Sorry, Ms. Rosen is going to be busy running for elective office. I'm not kidding, I saw her on a talk show the other day discussing a number of issues unrelated to the RIAA. She didn't announce any intention to run for anything, but she's obviously positioning herself for a run for some office and based on what I heard and saw she's going to be a very effective candidate.
  • everytime there is a new inovation in cpu disign that reduces heat insted of just remaking a p4 to run cool enough to run without a fan, they crank up the clock so that its putting out even more heat then before, making water cooling required for a quiet system
    • Actually, Intel uses the CPU technology they develop to make a CPU that runs cooler. It's called the strongarm; it runs at 200Mhz and is cool enough to keep in your pocket. They've got a faster CPU (the xscale), but it's got some problems in the current revision.

      The x86 market is still pushing for speed, but I suspect that the next generation of the cheap PC will run cool instead of being faster.
  • And will this water-cooling pump disentgrate pthe cpu pump and leak water all over that nice chip?

    Oh well. The chip was probably a DOritos..
  • Is that a 5 1/4 floppy drive in that machine? What's he running, a 486?
  • Well, gravity based home heating systems have been obsolete with the advent of the water pump? Why bother when it's well known that forced water flow will provide better thermal conductivity?
    • I think you meant to say 'better heat dissapation'? Thermal conductivity (of water) is fixed, based on the medium, as I recall.
      • You're right. The conductivity stays the same but it's the faster flow in a pump system that ensures more reliable heat transfer.
        • You're right. The conductivity stays the same but it's the faster flow in a pump system that ensures more reliable heat transfer.

          Only more reliable if the pump doesn't fail. This method has no pump to fail, making it more reliable.

        • If I remember my chemical engineering degree properly, forced circulation gives a higher Nusselt number (fluid engineers love their dimensionless groups) due to higher turbulence, which reduces the effect of the fluid insulating itself. More stuff here [wolfram.com].

          Water is a very bad conductor of heat without good convection - you can boil water in the top half of a test tube while ice is happily sitting at the bottom - wrapped in gauze to make it sink, obviously.

  • Excellent (Score:3, Funny)

    by worst_name_ever ( 633374 ) on Wednesday February 05, 2003 @12:08AM (#5228670)
    Always good to see continuing advances being made in the field of compu-bong technology.
  • I built a simpler cooler, able to dissipate the same heat flux of a normal heatsink."

    So it works as well as a regular heatsink? whats the purpose then?

  • So? (Score:2)

    by blair1q ( 305137 )

    When do we see the first Lava Lamp mod?

  • This is how liquid cooling should be done (Score:3, Insightful)

    by UniverseIsADoughnut ( 170909 ) on Wednesday February 05, 2003 @12:27AM (#5228786)
    The thing that should be pointed out is this is how liquid cooling should have been done from the start. The reason their not is because most if not all of the home pc case cooling systems are things slapped together by people without engineering background in such things. Good liquid cooling systems are designed like this. There should be no pump to cool your cpu. The problem is to do it right involves doing some calculations which most don't do so they just use a pump. It's not a very hard thing as you can see by the article.

    Now on a side note, if you like the idea of passive cooling loops Shuttle XPC cases now come with them. They use very similar ideas. You have liquid in a tube, when it vaporizes it rises to the upper radiator where it condenses and a fan carries the heat away and the water returns to the cpu side of the loop. Not sure if it vaporizes or just gets hot to move up the tube, really just a matter of effectiveness. It does a good job in lowering cpu fan noise. Unfortenly the noise of the PSU is very high in these things, or at least my SN41g2, though i belive the P4's have same PSU. The need the same thing on the PSU as the cpu. Also the "heat pipe" is a work of art.

  • efficiency (Score:3, Insightful)

    by whereiswaldo ( 459052 ) on Wednesday February 05, 2003 @12:49AM (#5228944) Journal

    Sure, heat convection flows, but your efficiency goes way up if you have an active flow. Same principle with hot water heating. They used to not have pumps to circulate the water - because it worked - but it's more efficient to have a pump.

  • I live in wisconsin (Score:3, Funny)

    by codepunk ( 167897 ) on Wednesday February 05, 2003 @02:57AM (#5229600)
    It is about -12 deg outside at the moment. Why do this when I can just put my machine outside my window. You guys in California have to resort to something like this? Time to move....

  • cool! (Score:2)

    by Sj0 ( 472011 )
    I hope this can be implemented for modern PCs soon. the fans on modern PCs are the single most dangerously unreliable parts of the PC -- when it goes, the PC dies a slow and painful death. Not everyone has the know-how to clean these fans sporatically, and even among computer people, the knowlege to lubricate fans manually is a rarity. I'd really like to see systems like this, which rely on natural forces, and are less prone to mechanical failure, used instead of high-RPM dust magnets used in heatsinks today.
  • If you ever go to a farm show you may see the people who restore old industrial engines. These things used to drive small farm generators, milk pumps, air pumps and the like. They were almost all water cooled by convection, and the smaller ones did not have radiators (so called) but just had a tank of water on top which was allowed to boil off slowly and be replaced from time to time. Exactly the same as this cooler method. The benefit was, in fact, generally increased longevity and reliability, provided that the ports in the jacket and the head were properly designed to prevent dead spaces where the slow convective flow did not circulate.

    I saw one guy once using one of these to cook his lunch, using a pan in the water tank as a water bath to heat soup, and they can also get water to the boiling point to make tea. However, and computer system which gets water to the boiling point is not likely to be terribly effective (unless we go back to tube technology, of course)

  • I built a simpler cooler, able to dissipate the same heat flux of a normal heatsink

    No, that's not what you did. The cooler you built was substantially more complex than a normal heatsink+fan. It was more expensive in materials, took more time for you to make, and required you to make more mods to your case.

    It might have been quieter, though :-)

Slashdot Top Deals

E = MC ** 2 +- 3db

Close