Convection? Evaportaion?

[Crashman]

I'm looking at a couple concepts I haven't seen tried yet on a PC.

The convection method requires the bottom of the water blcok to be at or slightly above the bottom of the radiator. The top of the the water block must be below the top of the radiator. As water is heated, it rises from the water block and returns to the top of the radiator. As is cools, it falls to the bottom of the radiator. The hot water rising draws cooler water from the bottom of the radiator into the bottom of the water block. The advantange of this system is that it requires no pump. If you don't think such a concept could work, find out how the Model T was cooled!
The other method requires me to use a coolant that boils at the temperature I want the CPU to operate at. It's the same physical arrangement as the convection system, but the coolant boils at a certain temp (say 40C), meaning that the CPU stays at that temp. The radiator is instead used as a condensor, where the evaporated coolant liquifies and returns to the bottom of the coolant block.

The main advantage of these systems is that they do not require pumps. Anyone with experience in this area is welcome to add comments or caveats to aid in my decision.

Back to you Tom...

 

[Boondock_Saint]

Extremely ingenious and theoretically possible ideas. The only problem I could forsee with the convection system would be the size or layout of the radiator, and the speed at which water is moved. Concerning the size and layout, you would either need a fairly large radiator to dissipate the relatively slow moving water with any speed, or very thin and tightly packed thins for good operation, driving price up. As the water would not be being pumped either, concern would arise around too much static water over the waterblock. You do not have good thoroughput. This possibly might be remedied by the usage of larger diameter tubing, which would let more hot water exit at once. So, either a bigger radiator than normal, a lot more fins, or larger tubing might make this feasable. When weighed against current designs, though, I would rather go with a smaller radiator with cooler temperatures and a little noise than a <b>completely</b> silent system.

The evaporation setup is what I find extremely interesting though. <b>Excellent</b> idea, but I am unsure of what liquid would be used. None of the standard coolants I can think of could be used, and most likely the ones that could be used would be toxic or otherwise banned for consumer use due to environmental/human damage. The only substance I know right offhand that would be anything like this would be Bromine, boiling at around 60oC, I believe. Nasty [-peep-], you definitely do not want that in your system if you were to have an accident.

:tongue: Have you ever tried cooking an egg on your HSF? Tasty. :tongue:

 

[Crashman]

Don't forget pressurized gases. Certain gases become liquid at relatively low pressures.
Many of the most ideal coolants are extrememly flamable. What's the boiling point of Xylene? If it were ideal, the problem would be possible leakage!
At any rate, the system would need an expansion chanber for pressurized gases to maintain that boiling point. But that wouldn't be such a problem.

Back to you Tom...

 

[phsstpok]

Your second suggestion sounds like a heatpipe.

MaximumPC put out an article and used a heatpipe filled with water to cool a P3-600. The single pipe was drawn to the top back of the case where a large heatsink and fan sat. The key to the system was getting the correct amount of water in the pipe.

The article was in the Jan 2001 issue. No reprint at the MaximumPC homepage.

<b>Update:</b> I was just looking at the article again. The project used two pipes but in an editorial reply to a letter in a later issue it was mentioned that one pipe will work as well.

I still can't figure out how it works. Water boils at 212 degrees at one standard atmosphere. In a closed system, like the heatpipe, it would boil at a higher temperature. It just doesn't sound like it would work for a CPU and yet it did.

<b>We are all beta testers!</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 11/10/01 01:19 AM.</EM></FONT></P>

 

[Boondock_Saint]

<A HREF="http://physchem.ox.ac.uk/MSDS/XY/m-xylene.html" target="_new"> Click. </A>

Boiling point: 138 - 139 C

I don't think that would work.

Yeah, an expansion chamber would be needed.
You spoke of the contents being under pressure, but then you might possibly have to deal with condensation.




:tongue: Have you ever tried cooking an egg on your HSF? Tasty. :tongue:

 

[Crashman]

No condensation if the fluid boils above room temperature and cools down only to room temperature, condensation would require it to be below room temp.

Back to you Tom...

 

[Crashman]

The heatpipe they used with water relied on evaporation instead of boiling. Evaporation is a variable process, boiling is a fixed temperature process. As such I can use boiling to maintain a constant temp regardless of CPU output-increased heat results is increased rate of boiling at the same temperature. Also, boiling is a more efficient cooling process as it takes a lot of energy to change the state of a liquid to gas.

Back to you Tom...

 

[Crashman]

Acetone boils at only 59C! We're getting closer!

Back to you Tom...

 

[Crashman]

BUTANE! It boils at .5C! That means that under the apropriate amount of pressure it would boil at 40C! And imagine the explosion if it leaked! Maybe I should be looking at comercial refrigerants?

Back to you Tom...

 

[Boondock_Saint]

Risky, but feasible. No great solution ever comes without risk, at first.

Commercial refrigerants, do you mean Freon and such?

:tongue: Have you ever tried cooking an egg on your HSF? Tasty. :tongue:

 

[Crashman]

Freon and such. They have a very low boiling temp, which requires higher pressures to work. Maybe you could suggest a refrigerant with a higher boiling point, so I could drop the pressure?

Back to you Tom...

 

[phsstpok]

Boiling temperature is only fixed at fixed pressure. As the pressure rises so does the boiling temperature. (That's how a pressure kettle works. It relies on the pressure to increase cooking temperatures). How are you going keep the system at equilibrium?

<b>We are all beta testers!</b>

 

[Boondock_Saint]

Once you start dealing with Freon, you have to worry about condensation and such. Also, they already have CPU refrigeration systems that use Freon.

:tongue: Have you ever tried cooking an egg on your HSF? Tasty. :tongue:

 

[Crashman]

Expansion chamber. Maybe a piston type.

Back to you Tom...

 

[Crashman]

NO condensation as long as the COLD side is at room temp! The only reason you get condensation with an air conditioner is because it's operating at a temp below room temp. Using the proper pressure to cause the boiling to happen at a fixed point above abient temp (say, 40C), the cold side is at room temp and the hot side is always at 40C.

I would have to use a compressor to get it colder than room temp. I'm trying to devise a system that uses no pump. Compare a moonshine still to a refrigerator.

Back to you Tom...

 

[turing]

We are doing back of the envelope approxes for Na,Ga or Hg cooled systems. There are sys integ probs with each which may just rule them all out! Sodium burns in air,Mercury is pizen & Gallium might be too expensive as well as nasty!

 

[phsstpok]

How about a small peltier? Cools the CPU and adds more heat on the backside for boiling whatever liquid you are going to boil. You can use variable voltage to control the peltier and to some degree the temperature.

<b>We are all beta testers!</b>

 

[Crashman]

Peltiers have a relatively high failure rate. I was originally trying to eliminate a pump for similar reasons! But I'me reconsidering WATER, because if I apply a SMALL AMOUNT OF VACUUM to it, I can get it to boil at a lower temp.

Back to you Tom...

 

[Guest]

Crashman's on it!

I think it would be much easier to control the boiling point of water with a vacuum than trying to pressurize a more volatile liquid. (at least if your pump fails, you don't loose your coolant)

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

<b>Do YOU deserve Da Bird?</b>

 

[Crashman]

Remember, I'm using the change of state to move the coolant, instead of a pump. Now I need to figure out the expansion chanber that will maintain the lower pressure. Any ideas?

Back to you Tom...

 

[Guest]

I was thinking check-valves on the inlet and outlet. The expansion of the boiling at the heat source causes water to be forced out of that area and it must go to the exit because reverse flow is blocked. The check valve on the outlet prevents condensate from equalizing the pressure. If life was good, you would have a saturated system in the area between the two check valves (pressure reduction of the valves allowing a slightly higher pressure area to be maintained in the boiling section to prevent drying it out. (ie: reverse flow/back pressure prevented by inlet valve, and head losses of flow within the loop combined with an elevation change form a pressure for the outlet flow to move against.) The trick would be to have a vacuum maintained that is just enough to allow boiling only in the very near vicinity of the heat source. The rest of the system is solid (all liquid).

Please bear in mind that while this would be very novel (and do-able), your vacuum pump will be a lot noisier than any circulation pump. To eliminate the pump, you are back to exotic coolant with an acceptable boiling point/specific pressure.

well, enough of my ramblings, time for a beer

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

<b>Do YOU deserve Da Bird?</b>

 

[Crashman]

You don't really need a vacuum pump! A simple wieght and piston aperatus can creat a contant vacuum. The cylinder simply needs to be long enough to take up the expansion gases.

Back to you Tom...

 

[Guest]

I know the vacuum pump wouldn't have to run all the time, but it seems like evaporation in a reduced system will always be working against you. Sooner or later the piston will reach the extent of it's travel and you will need to evacuate the system somehow to "reset" it (if you will).

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

<b>Do YOU deserve Da Bird?</b>

 

[peteb]

Crash - I still don't buy the convection/natural flow model for this. I really don't think you will get a natural circlulation model to work effectively at this scale.

The principal you need will revolve around collecting condensed water at a higher point than the base of the waterblock, and effectively having water 'boil' out of the block itself. To maintain a boiling action in the block and not end up with big temperature spikes the block must always have water in it, but also need 'room' to boil. The block is, I suspect, going to need to be big. It works in a car because, well, the block <b>is</b> big.

In princpial this may work, however how are you going to maintain a partial vaccum in a system (to lower the boiling point) when the water vapour is expanding - the very mechanism you are trying to exploit.... <i>unless</i> the condensing water gives you the pressure drop you are needing to maintain the vaccuum. Whoa, this is going to be a bitch to tune and bench.

Okay, so you are still going to need a good sized water block, since that is the point of heating (actually did you consider the noise of a boiling system bubbling away?) to boil the water. As you've said you can provide a constant force to ensure vaccuum is maintained. The cooling water contracting (condensing) will be exactly counterbalancing the pressure increase of the boil. Now presumeably you need a gas in there as well. I don't know why, but might there be any reason to use something other than air?

What about the boiling point? You need to get down to ~ 40 degrees right. That sounds like a lot of depressurisation for water. Don't forget you do not need pure substances. Mixtures of water and alchohol can boil at substantially lower temps, although the result may still be flammable....

Hmm - hell of a project

-* <font color=red> Under Offer </font color=red> *-
email for application details

 

[Crashman]

This just keeps getting more complicated. OK, the best method for a boiling system is using a tall open heat pipe with the water black as a resevior and the heat pipe tall enough that no steam escapes before it's re-condensed. But I really don't feel like having a 10-foot tall pipe sticking out of the top of my PC! And water boils at 100C! So to continue using water in an open system requires a switch from bioling to evaporation, which requires a higher volume of water. UGH!
I was trying to turn the concept "on it's side" like they do with a moonshine still. The "whisky bottle" becomes the resevior at the bottom of the radiator (assuming a downflow radiator). To keep the system sealed requires a coolant that boils at the proper temp, and a way to regulate pressure. The pressure factor is the worst part. The moonshine still concept requires a large enough radiator that the output is completely condensed, the radiator itself becoming the expansion chamber. The excess air in the lines is expelled when the system is started, the vacuum draws air in after the process is shut down. To integrate such a large system into a case would be close to impossible. Further, finding a safe liquid to use in such a system with a 40C boiling point is quite difficult. I'm really digging for more information here!

Back to you Tom...