passive watercooler pump?

C

category_five

Limp Gawd
Joined
Sep 10, 2004
Messages
193
Just a thought as I am not likely to go and watercool my system, but you never know:
The reason I am not inclined to watercool is the addition of multiple points of failure into my system. The second reason I am hesitant to watercool is the additional power draw of a pump. That said there is a way to kill both these birds with one stone.

If it was possible to have a passive system to cycle the water it would eliminate a major point of failure for watercooled systems, the pump. Also it would negate the additional power draw of a pump. Seeing as how heat rises it might possible to cycle the water using the heat differential between the cooled water out of the radiator and the hot water from a processor.

Anyways I will try to think of a way to passively cycle water using the heat difference. I will try to come up with something and bounce it off you guys here You never know, it could happen! :)

Edit: Another thought- Ideally this passive pump would physically move the hot water, not just transfer the heat without moving the water (what I don't want is a heat pipe).
 
I don't think it would break the laws of thermodynamics as it is not perpetual motion we are talking about. Heat would propel the system and that heat comes into the system as waste from the processor.
 
quick sketch don't laugh:

hot water goes up tube (heat rises)
--------------------------------------->
[P] Processor, Radiator [R]
<---------------------------------------
Cold water goes in lower tube (cold sinks)

Please feel free to tell me this is naive but also if at all possible try to post a way to get it to work :)



Edit: another thought - Well, at rest the hottest the water would get is around the heat of the processor, 120-160 degrees F. I think that is hot enough to start some kind of sustained siphon...

Edit: another thought - If something like this were able to work it would actually surpass the reliability of a conventional fan cooler solution. While this wouldn't be enough to mainstream watercooling due to the cost involved, it would take it out of the realm of less reliable solutions.

Edit: a quick google turns up nothing on passive watercooling. A few things on passive solar heaters using the tendancy for hot water to rise... Maybe a small scale model to test...
 
Actually, I think it is perpetual engine (might be wrong tho) - u want to build a self-contained system where energy of the movement is used to generate the movement itself - or so it seems to me :D

Let me know when u come up with something, I'll patent it and get filthy rich :D
 
No no no, look at it like this. You hook up this system and leave it. Don't turn on the computer, nothing moves. The water stays stationary. Once you apply energy in the form of heat, stuff happens. The water will start moving. Heat it up enough, and the water will boil and move very rapidly. What I am trying to do here is channel the natural movement of the water when heated in such a way that the water will go to a radiator where it will leech out.

To put it a crude way, imagine your processor was a little hotter. Now imagine you put a bowl of water on your processor (bear with me here). The water will eventually boil, and steam will come up. Add a turbine spun by the steam and you have electricity to power whatever you want. This is obviously different than what we want but I'm just showing you the energy IS there.

Heat energy, mechanical energy, electrical energy, they are all interchangable with the right devices. Of course there is loss due to the laws of thermodynamics but what I am proposing here is not perpetual motion. We have a (relatively) huge amount of heat coming off the processor. We just need to find something constructive to do with it :) .

Ok and as far as proof of concept I present to you the work shown here I did some more google mining and found it :) It is nowhere near as efficient as I am invisioning. I think he might have made his tubes too large. I imagine it like sucking through a straw, the small straw is MUCH easier to suck through for long distances than a huge one. This guy went with the HUGEST tubes he could find. Perhaps I'll make a proof of concept with tubes 1/4 the size.

BTW my furniture refinishing project took me over 6 months so don't expect to see anything soon, if ever.
 
this sounds a little like the stirling engine-- ran on pure heat. dean kamen (think segway ht) has reportedly managed to get one to work.

but actually, if you design the motor for the pump so that it is very hardy indeed, then it will last you a very, very long time.
 
A long time yes, but if this can be gotten to work it will last basically forever as it has no moving parts. Yes, yes, barring corrosion, thermal expansion, mechanical stress failure. But as least nothing is spinning. I would expect the best pump to have about the same mean time between failure as the best fan; likely they will both run a long time but will eventually fail. What I am talking about has no moving parts that can fail. If you make it right as long as there is heat the water will be cooled. I find passive technologies to be fascinating.
 
Here are his temperature readings. We don't neccesarily have to assume absolute accuracy as the temerature shift is more telling for our purposes than the temperature itself.

CPU temp. = 55deg. Celsius
Waterblock temp. = 43.7deg. C (measured at the base)
Radiator inlet temp. = 33.3deg. C. (measured at the top)
Radiator Outlet temp. = 31.8deg. C. (measured at the base)
Ambient temp. = 25deg. C. (monitored by LCD thermometer)

So he had a 10 deg C loss from the tube going up and only 3 deg C loss going through the radiator. His inlet tube was doing more cooling than his radiator. What does this mean? Well, I assume it means smaller tubes are worth checking into ;)
 
I need to find out the wattage of heat/ temperature heat my 64 3500 puts out before I make a model.

Edit: The guy on that website estimates he gets 300 litres per hour waterflow, which translates to roughly 79 gallons per hour. Commercial waterpumps used in OCing range from 75 to 300+ GPH. So far so good!

BTW trying to edit posts as much as possible to avoid bumping my own thread ;)

edit: more thinking out loud: IF convection alone provides 79 GPH and IF your system is set up vertically it becomes very important to cycle your pump WITH the current that convection provides, especially if your pump is a weaker one.

edit: I'm thinking copper barbs from a hardware store with sheets of copper 1/8-1/16th inch thick cut with a hacksaw, filed, sanded and then soldered together with plumbers solder stuff. Assuming it can be soldered with a standard soldering iron. No way I'm going to learn / purchase welding supplies for this, nor will I buy a lathe. Drilling holes might pose a problem though...
 
There was a company that have made pumpless watercooling (kinda) Heatsink with waterblock and 80 mili radiator/fan. Though it was like 4 maybe even 5 years ago. Somone, if not even Kyle himself, (maybe some other good hardware page though) have tested it and it was preforming poor. It was using effect that hot water tends to rise.
Water was circulating but slow and preformance was bad.

Pumps aren't so critical if You use good pump specialy. If You use flow indicator then there is no risk at all.

Personaly I think there is a greater risk of burning system with pumpless system then in normal setup. Why? Simple pumpless system generates very small preasure and flow is minimal. Some algees or what ever and system stops.

BTW great dude makes convenction system and hangs it on radiator to warm up water additionaly :)

MD
 
I remember that radiator as well. Here is a link to it. I also recall it did not perform well in the reviews. However, it has a fan on it, and what I am proposing will be fanless. Second, What the creater of that heatsink was aiming for was better performance than attainable with a standard fan, and missed. I am not aiming for lower temperatures than you can get with a fan. I am aiming for equivelent performance without a fan. Furthermore, what I am proposing would remove the heat from the case altogether. The radiator in the link simply removes the heat from the processor and puts it into the case.

In conclusion this example does not put off my goal as my goal is modest, and even though it is modest it would have very tangible advantages to currently available solutions.
 
no, energy is introduced into the water through the cpu.

TehQuick said:
Actually, I think it is perpetual engine (might be wrong tho) - u want to build a self-contained system where energy of the movement is used to generate the movement itself - or so it seems to me :D

Let me know when u come up with something, I'll patent it and get filthy rich :D
Click to expand...
 
This is basically the same principle that drives heatpumps, minus the phase change. In either case, you are driving fluid motion by a difference in densities (bouyancy - the cause of natural convection).
 
well, if you don't allow phase change to aid in your movement (the nice thing about boiling systems is that the density difference between liquid and gas is so huge and aids nicely in getting your system moving) I suggest a large copper block with a 1" copper pipe loop. What kills you on these systems is flow resistance, so you need to avoid tight corners or anything else that restricts flow. Plan on a a 2 ft radius for your loop. No radiator per se, just cooling fins on the outside of your loop on the "down" site. At the flow rates that system will achieve cooling the water on it's way will not be an issue.
I'd still be worried of local overheating at the CPU so; heat flow is heat capacity x mass flow, so no flow - no cooling.
 
Just in case nobody else has seen this yet, I just want to point out that you all are talking about the very foundation of heatpipe technology. Further investigation into this field might generate more results for you (and show you the limits of what you are intending to do). For instance, water alone might not be your best media. Saltwater conducts better. Also, pressures can be lowered to lower the boiling point/evaporation point of water (or any liquid) so that say...the water near the processor boils at 90degrees, and that phase change energy traps heat until it travels to the top of the 'pipe' until it dissipates enough energy and condenses...turning cool and falling back down to the processor until enough energy is absorbed and the water evaporates. If you intend to use water passively, or a liquid medium, a phase change...even if not induced like a prometia...is one of the only ways to make an effective heat transfer medium out of water. Otherwise, there are many other better elements for passive cooling...metals in particular. Mercury is one element that we are doing alot of work with for some of our more demanding clients.

Why do passive with water anyways? Water cooling, unless involving a phase change, needs to be active to really do anything to a constant energy source. Im sure an eheim pump would work fine. My first eheim to fail me ever just quit on me last year after 15 years of constant service without a trouble or a peep.
 
What I was talking about has nothing to doo with heat pipes. In water cooling Your goal is that water doesn't evaporate/boil .

Heat pipes on other hand are also kinda phase change coolers. Much more then water coolers.

To remove heat heatpipes are better way then pumpless watercooling IMHO.

That wasn't cooler that I was thinking of although It's good example that I've allready forgot :) That one came even later then one I was talking about. I might need to find that article it eaches Me now :)



MD
 
MD said:
What I was talking about has nothing to doo with heat pipes. In water cooling Your goal is that water doesn't evaporate/boil .
Click to expand...
Sure it does, in each situation you create fluid flow by a difference in density. Of course you don't want the water to boil in a watercooling loop, but you could very well build a heatpipe using water as the phase change fluid.

Heat pipes on other hand are also kinda phase change coolers. Much more then water coolers.

To remove heat heatpipes are better way then pumpless watercooling IMHO.
Click to expand...

Naturally, because the phase change absorbs more heat than the passively flowing single phase water.
 
I bought some 1/4 inch acrylic tubes and copper fittings from a hardware store today. No radiator yet, I'm going to be bending some copper refridgerator tube for a cheap radiator initially. I plan to play with it and measure temps and water flow with different setups. I will eventually get some 3/4 inch tubes and compare the two.
Taking into account the comments in this thread on heatpipes I am going to also investigate closed loop evaporative cooling.
 
category_five said:
I bought some 1/4 inch acrylic tubes and copper fittings from a hardware store today. No radiator yet, I'm going to be bending some copper refridgerator tube for a cheap radiator initially. I plan to play with it and measure temps and water flow with different setups. I will eventually get some 3/4 inch tubes and compare the two.
Taking into account the comments in this thread on heatpipes I am going to also investigate closed loop evaporative cooling.
Click to expand...


Home heatpipe construction will be a little more involved (quite a bit actually). I won't discourage you from learning/trying though :D
 
not necessarily. The goal is to cool your processor. If you could do that by causing the water to boil (which soaks up a huge amount of energy) and cool it back down with an effective flow rate, it would be done all over the place.

MD said:
What I was talking about has nothing to doo with heat pipes. In water cooling Your goal is that water doesn't evaporate/boil .
Click to expand...
 
Yes, reading about heat pipes they are very involved. Filling a pipe with Freon and then welding it shut /shudder. However evaporative cooling of water might be worth looking into, especially in a closed loop. It seems that many people are water cooling with a large open bucket reservoir, which is essentially bong cooling. I would suspect they get almost as much or more cooling from the reservoir evaporation as they get from the radiator.
 
category_five said:
Yes, reading about heat pipes they are very involved. Filling a pipe with Freon and then welding it shut /shudder. However evaporative cooling of water might be worth looking into, especially in a closed loop. It seems that many people are water cooling with a large open bucket reservoir, which is essentially bong cooling. I would suspect they get almost as much or more cooling from the reservoir evaporation as they get from the radiator.
Click to expand...


Actually, the system has to be charged with refrigerant (probably not freon these days) AFTER it is sealed shut. You'd have a tough time holding any kind of pressure in there with the ends open :D
 
My point had more to do with this...
Water, like any material, absorbs energy at a certain rate in the form of heat to increase its temperature. The amount of energy that it absorbs per degree of change in temperature varies with each phase it is in. It takes more energy to raise ice one degree then it does to raise liquid one degree, and even less for vapor. And, there is a certain threshold between phases where there is a constant amount of energy absorbed to go from that one phase to another. If you can make water change phases by adding enough heat to it so it evaporates, it will be much harder for it to release that energy any time before it gets to a place cool enough to release it all. In heat pipes it is part of the design to account for how much energy will be absorbed, how much of a given liquid this energy can phase change, and hold that energy until it reaches the other end, while trapping the energy in a vapor phase until it cools enough at the other end to return to liquid so as to not release the heat on the way because it is 'locked' into the vapor and can not be released until the vapor cools enough. This process is what makes the liquid the material of choice (unless you use a solid in a a strong enough vacuum that it would change from solid to gas without a liquid stage...whoah...now that is heat transfer!). This phase change is what is necessary to make water your material of choice. Without this phase change (or an active solution like a pump), water is easily outperformed because it is not a naturally conductive substance like say....ANY METAL. So, what I am trying to say is, unless you plan to use an active solution like a pump...or plan to use a carefully controlled phase change like an evaporative cooler, prometia, or heatpipe...water is not the material of choice. It would actually be considered an insulator. You would be better off just putting a solid copper block on your proc...because without a way to move that water any metal will carry heat away faster anyways. And you are pretty much just putting a solid block of something less conductive than metal on your CPU. Pointless. Water's attractiveness is the fact that it can be moved...take that away and it looses its meaning. Other than the fact that its a liquid, water is no comparison to metal with regards to heat transfer.
 
zer0signal667 said:
Actually, the system has to be charged with refrigerant (probably not freon these days) AFTER it is sealed shut. You'd have a tough time holding any kind of pressure in there with the ends open :D
Click to expand...
And how do you get the refrigerant in there after it's sealed? :confused: :D
 
You breach the tube with needle. Ok, I'm sure theres more to it than that but it's basically what it would amount to.

As for making a heat pipe - I'm not sure if this is really the way to go as the tech and technique for this is far beyond my abilities. Furthermore, the reliability of a home made solution using materials that will evaporate in seconds on contact to air leaves much to be desired. The process of creating one at home can be found here.
 
category_five said:
You breach the tube with needle.
Click to expand...
At which point it's not sealed... :D

I was more teasing zer0signal than being serious. Anyways, with some heatpipes you can see where they were crimped or capped or plugged to seal them after they're charged with refrigerant.

Making a decent heatpipe is beyond most DIY people; too much specialized equipment and knowledge needed.

Besides which, isn't the thread about passive water?
 
Yes it is about passive water movement, but an enclosed evaporative passive water system would amount to a heat pipe. The two subjects overlap in this regard.
 
Big power transformers that you see inside substations cool exactly how you are describing. The core and windings heat up the oil that is inside, hot oil rises and moves over to the big radiator fins, "falls" down the fins as it cools and starts at the bottom again. No moving parts. However, I do not think this process would work (at least not well) mainly because you have such a large heat source from such a small surface area. The movement of heating water or whatever liquid will not move fast enough to cool down the proc before it burns itself up.
 
category_five said:
If it was possible to have a passive system to cycle the water it would eliminate a major point of failure for watercooled systems, the pump.
Click to expand...

If you had your shut off temp set in the bios, wouldnt the bios shut off the computer before the system got over that temp provided the pump fail?

Edit: grammer.
 
Quote:
Originally Posted by category_five
If it was possible to have a passive system to cycle the water it would eliminate a major point of failure for watercooled systems, the pump.


If you had your shut off temp set in the bios, wouldnt the bios shut off the computer before the system got over that temp provided the pump fail?

Edit: grammer.
Click to expand...

Yes, but reliability isn't about fixing things when they break, it's about things not breaking in the first place.
 
SparksNelec said:
Big power transformers that you see inside substations cool exactly how you are describing. The core and windings heat up the oil that is inside, hot oil rises and moves over to the big radiator fins, "falls" down the fins as it cools and starts at the bottom again. No moving parts. However, I do not think this process would work (at least not well) mainly because you have such a large heat source from such a small surface area. The movement of heating water or whatever liquid will not move fast enough to cool down the proc before it burns itself up.
Click to expand...


Yah, the power density is one difference there... Also, I think transformers can handle temps a little hotter than what we would like for our CPUs. That factor gives a larger delta-T right off the bat for transformer cooling, given the same ambient air temps. Having a chiller inline with the water loop could greatly help, although it's not really in the same mindset as passive water-cooling.
 
yeah, the passive oil cooling works in that case because the amount of heat being moved is so much greater than the ambient air used for cooling, and the heat source is so much greater. with a cpu, we are talking about a relative narrow margin...we are talking about keeping a cpu sometimes within just a few degrees of room temperature (thinking summer here)...not alot of room for slow moving passive systems. Not a difference of hundreds of degrees like a huge generator or engine...
 
not only will the water not move quickly enough in a passive system to keep it cool enough for most of our likings, as was pointed out in that review link, you need to have poor temperatures in order to keep the coolant moving.

This idea is novel, but not feasible in the sense that you will get any gains from it...it is like trading in your sports car for a golf cart because you think that the sports car engine will fail 10 years or 100,000 miles from now.
 
Erasmus354 said:
not only will the water not move quickly enough in a passive system to keep it cool enough for most of our likings, as was pointed out in that review link, you need to have poor temperatures in order to keep the coolant moving.
Click to expand...

Not necessarily poor temps, but a large temperature difference. If you were to put the "radiator" into some chilled water, or out your window in the winter, it might be feasible.
 
This natural circulation system is exactly how many early automobile cooling systems worked (Ford Model T for instance) but they are inefficient compared to pump driven systems and stopped being used in the 1920's
 
This idea is novel, but not feasible in the sense that you will get any gains from it...it is like trading in your sports car for a golf cart because you think that the sports car engine will fail 10 years or 100,000 miles from now
Click to expand...

The reliability is the only reason I want to do this. There are many applications where the "golf cart" is a superior solution over the "sports car". Sports cars certainly have their uses but not everyone wants the maintenance associated with one.

Currently if you wanted to implement a computer in a remote location and want zero maintenance time you are stuck with a passive heat sink and a processor several generations old.

Couple a passive cooling solution that works reliably on a higher heat output processor with solid state memory mass storage (USB flash drive type) and you have a computer that you could set up running in a storage closet and potentially never have to physically maintain.

It is becoming increasingly obvious that one of the bottlenecks to higher processor speed is efficient thermal solutions. Unless there is a massive shift in the method by which computer processors run more efficient cooling is needed. As it stands water cooling is certainly efficient enough but is too unreliable for the mainstream and this reliability needs to be addressed before we will see water cooling taken seriously by the industry.
 
You must log in or register to reply here.
Back
Top