Just another crazy thought experiment... stop in if your curious.

[Dlaw]

Well in this case volatile is kind of a misnomer. Volatile not like gasoline but as in boils at low temp. I wont disagree with you, but it would seem that they dont use the more volatile liquids because they would evaporate at a higher rate and leave air in the system. A completely sealed loop would prevent any loss in this way. Once again id have to check with the physics professors, but I think if it boils at room temp, it would transfer heat better than water. The use of water very likely has something to do with availability and not needing special certifications to handle it as well as how reactive it is with the tubing. Im thinking isopropyl though. Iso would take in the heat and give it back faster.

Now its time for me to bow out though, because Im an air cooler after all.

Edit: I seem to remember reading that copper heat pipes on air cooling rigs were sometimes filled with inert gasses. Im just thinking of an expansion of this concept and having the gas move through the radiator.... Im ceertain there are thermal issues to be overcome, but the base concept is (IMO) solid. I do see the problems that could arise though. I really think cost would be the biggest issue though. Each system would have to be custom built, or there would have to be a wide array available for popular case designs.

I understood that you didn't mean volatile as in explosive, but think of this: When a liquid reaches the boiling point, the molecules in the liquid are moving extremely fast, bumping into each other and the surfaces of what contains them. Now think of a pot full of boiling water, the bubbles that are moving from the surface of the pot to the air are not full of air, but rather water vapor. They do this because the water vapor is less dense than the liquid water. Now, if you put a tight-sealing lid on that same pot, what happens? Momentarily, the water stops boiling. Why? Because the water vapor quickly equalizes the pressure above the liquid water. But then what happens? If left long enough, the pressure will build to the point of "boiling over", meaning that the pressure inside the pot will exceed the seals ability to hold.

Now, if you were ever young, you may have at some point taken the cap off of the radiator in your car after driving it for a few minutes. If you did, you may have found that the liquid inside the radiator boiled out of it. This is for the same reason as the pot in the paragraph above, but what makes it different? The radiator cap. It's designed specifically to maintain a low positive pressure (somewhere between 9 and 15psi, generally), and if the pressure becomes greater than that, the fluid is blown off into the reservoir. Why don't they just allow the pressure to continue to build in the radiator? Because, as it is with computer water cooling loops, there are seals within all of the parts being cooled, that will fail at various pressures. Even if they don't fail immediately, the higher the pressure in that loop, the better the chance for an explosion once something does fail.

Now back to the more volatile liquids, the molecules in them move faster than that of water at the same temperature. This means that at any given temperature, the pressure in a closed loop will be greater for those liquids than water, thus increasing the risk of failure to the components of that loop, and the possibility of an explosion.

As far as the inert gasses, yes, a lot of them are. But you have to figure, these heatpipes are solid, with no seals and no moving parts. They're designed to handle much more pressure than the gas can produce at the temperatures of current computer equipment.

 

[EarthDog]

So, for the third time, LOL, how does anything boil inside a sealed loop with no room for it to boil? What basic principle am I missing?

 

[Dlaw]

So, for the third time, LOL, how does anything boil inside a sealed loop with no room for it to boil? What basic principle am I missing?

Check this out, 4th post, 3rd point. What I took away from it, is in a sealed system (loop), the ratio of liquid to gas will remain constant. What that means, is that so long as the system remains sealed (and thus, the positive pressure is kept), a liquid will not boil, nor evaporate.

 

[TuKr]

Check this out, 4th post, 3rd point. What I took away from it, is in a sealed system (loop), the ratio of liquid to gas will remain constant. What that means, is that so long as the system remains sealed (and thus, the positive pressure is kept), a liquid will not boil, nor evaporate.

Correct but it will still try to expand. Which is where volitile comes into play. It will boil up until the point where the pressure is great enough to make the boiling stop. If the pressure rises too much it can destroy the pressure vessel. Air in this kind of sealed loop is ok. We want it to build its own pressure if possible so no compressor, We wont have to move the fluid, properties of convection do that for us.

 

[Mr.Scott]

Edit: I seem to remember reading that copper heat pipes on air cooling rigs were sometimes filled with inert gasses. Im just thinking of an expansion of this concept and having the gas move through the radiator.... Im ceertain there are thermal issues to be overcome, but the base concept is (IMO) solid. I do see the problems that could arise though. I really think cost would be the biggest issue though. Each system would have to be custom built, or there would have to be a wide array available for popular case designs.

Heat pipes have water and Glycerin and wicking all under vacuum. No inert gasses.
There were a couple that tried Nitrogen filled, like Silverstone, but it wasn't any more effective.

 

[Conumdrum]

Dis is a funny thread.

 

[knoober]

Just out curiosity...

Is there any reason (beyond frying your mobo and other components) not to use gallium in the loop instead of water? Easily available at online retailers and not horribly expensive. I dont know what volume of liquid 20g makes (the amounts it is sold in) but would that even be able to be considered ? I like to think I might build a water cooling loop someday. Ive got dreams of running the loop to the sub zero (winter) temps outside and running it through a motor cycle oil radiator I am sympathetic to the problems of condensation as well, but I hope that insulation and sleeving of the loop might overcome that.

 

[EarthDog]

Considering gallium doesn't melt until like 28C... how are you going to manage that???????

You would have to have a heater keeping it a liquid, then you would have to have a pump strong enough to pump a very viscous liquid metal through the loop.

So, that is just a plain old full stop no...

 

[Otter]

So, for the third time, LOL, how does anything boil inside a sealed loop with no room for it to boil? What basic principle am I missing?

There is room for it to boil because some the gas cools back to a liquid at the other end of the system, reducing pressure.

Maybe.... 9 years ago, there was company prototyping a sealed cooling system that worked similar to a heat pipe but was modular and used a large radiator. I can't remember the name, and I'm not sure if they ever put it on the market, but it was and remains a very interesting idea. I'd been thinking along those lines myself when I heard about it, and I was eager to give it a try. But I'm not sure what became of the project.

It had evaporator blocks for each component connected by metal tubing to a big condenser with fins. Everything was sealed and designed to withstand the necessary pressures.

The fan blades were the only moving parts.

I think the liquid may have returned by a different line so that it would be sure to reach all of the blocks, but it's been years so I may have some of the details wrong.

IIRC, they were having trouble keeping the seal, and that may have been what killed the project. Or perhaps they just didn't have the capital to keep going until they attracted enough customers to make it viable.

 

[knoober]

Considering gallium doesn't melt until like 28C... how are you going to manage that???????

Well, a delayed start on the pump might do the trick, as it is highly heat conductive and will liquefy the whole shebang pretty quick. My temps jump 10 degrees in seconds on air and all that would be needed is a bare 2-3 degrees to liquid (depending on ambient, I'm assuming 26-27C average) Also I can't imagine it to be much more viscous than mineral oil. CLU and CLP have a gallium (or maybe indium) backbone and seem to flow pretty well at room temp. I'll grant that they are alloys, but the alloyed metals are usually things that melt at higher temps (silver, strontium, etc)

Edit: it's also nice to know that it might be impractical to do a sealed loop, but that I'm not the only one who's had the idea

 

[EarthDog]

That is the core of the CPU hitting 10c higher in seconds. What about the rest of the loop? How would the heat get there? How long of a delay to luquify the whole loop? Have you ever seen gallium melt at 98.6F/37C?

26C ambient? Negative. That is 79F+. Most people are somewhere around 72F/22C. But again, how are you going to get the entire loop to liquify and not overheat the GPU in the process? How are you going to pump a viscous(read: Liquid metal) fluid? You are well into the realm of the ends not justifying the means to me.

There are ideas... there are good ideas, and then there are idea's that just are not possible (or worth it) due to Physics. You can guess which one the pressure thing and gallium falls into... its not the the first three.

 

[knoober]

That is the core of the CPU hitting 10c higher in seconds. What about the rest of the loop? How would the heat get there? How long of a delay to luquify the whole loop? Have you ever seen gallium melt at 98.6F/37C?

)

Well Ill admit to having trouble interpreting the search results i got and not really having the education needed to really understand but.... it appears that liquid gallium has a viscosity between water and milk. Milk being 3 cp, water being 1cp ( http://www.apacks.com/resources/choose-best-fit-filling-needs/viscosity-chart ) and liquid gallium coming in at a nice round 2. blah blah blah cp (http://www.answers.com/Q/What_is_the_viscosity_of_liquid_Gallium_at_30_degrees_Celsius -- hey be carefull with this link guys its the only one I could understand but it made my AV go nuts "Malicous website blocked" x a ton of times. I got a clean scan afterwards but we all know how much that really means) The point is if the pump can take oil, it could deal with the liquid metal....once it liquifies anyway

I honestly don't know how long it would take to liquify and if that amount of time would be harmfull to the cpu. I assume that it would be rather fast and the sealed loop itself might draw off enough heat until liquid temps were reached. I can understand that most might be fairly nervous with that, and I would be as well. A better liquid metal might be found then. Mercury would be ideal, except for its toxic nature. Mercury ranks much higher on the w/mk charts I have seen. This one lists mercury at over 8w/mk while water is @ .58 w/mk http://www.engineeringtoolbox.com/thermal-conductivity-d_429.html Others I have looked at were similar.

The only real huge stumbling block here (and what makes this real useless spitballing) is the fact that there are pro engineers who are all over this field and they surely must have thought of this and there surely must be a reason why it isnt done already. But it looks good on paper

 

[EarthDog]

Viscosity isn't an issue it seems... noted!

I don't agree however that it would turn to liquid pretty fast. I suppose that depends on its convective properties... But you see in the video that it took a couple of minutes in his hand. I can't imagine it will get through a liter of it ( i use liter because that is about how much my loop hold and no idea how many grams of a solid material it will hd) in time to not make sure it does not overheat. There isn't any flow and would rely solely on convection to reach at least 28C throughout. A loop.

I just don't see it as viable, thought it is possible.

 

[knoober]

Im always just trying to build a better mousetrap. Less stress that way. As you can see most of those plans end up in the round file. Ill stop cluttering this thread up now. Thanks for reading guys!

 

[Mr.Scott]

At least you're entertaining.

 

[TuKr]

Nice loop to fit a litre worth of liquid in earthdog... And i too think gallium, and the original idea of this thread are unworthy. I wonder if gallium carries the same properties as other metals along the lines of tempering, where it gets harder to melt each time its heated/melted again.

Would dry ice stay cold if its sealed inside of a container with no way to escape? Now calm down and dont tell me this is a bomb... i know. Just a thought because lets say we did build a small cpu block, with a thin strong heat plate, and a thicker top. inside, a small piece of dry ice. i don't know any of the properties of dry ice and the pressures released as it melts, but if the pressure kept it from melting(boiling) then it would never get warm. adding a heat source would drastically change pressures though. a cylindrical container with a screw on top that fits on top of your CPU, strong enough to hold pressure. we may not be able to keep it sub zero, but if it always stayed around 30C, because of the pressure increase when the CPU goes under load and heats up forcing more to be condensed, that would be pretty awesome. But like I stated earlier, I know its a bomb... could we find something else to not be so volatile at room temp but maybe work along the same principle. Im enjoying this no matter how impossible it sounds.

 

[knoober]

I had been thinking that the liquid was the "weak" point here but perhaps it is the radiator. The radiator is what is responsible for the most difficult part of the heat exchange. Bringing the heat out of the loop and into the air. There are college phsysics studens somewwhere studying heat transfer that are gonna make great strides someday

 

[EarthDog]

Before the radiator is the block... the more heat that comes from it, the more goes into the loop/rad!! Chicken... egg... which is first?

 

[knoober]

Before the radiator is the block... the more heat that comes from it, the more goes into the loop/rad!! Chicken... egg... which is first?

Ihs is copper, I assume the cold plate us copper as well ( only thing better is silver and gold for thermal conduction). I'd love to start there but I don't see anything to add to a copper/tim /copper sandwich. Except of course for directly cooling the plate with dry ice I think some folks may already be doing that.

I think the only places for improvement would be in the liquid and the radiator. If we could get the radiator to exchange heat with the air better, but don't have any clue how that would be done except by changing the material. Aluminum is used mainly because of cost and it's general non reactive nature. Let me know if you guys get anything figured out

 

[Otter]

The weak point is at the block. First, all the heat has to go through the heat spreader and a very small area of TIM to get to the block, then it has to get picked up by the water from the still fairly small surface area inside the block. The radiator is easy. If it's not cooling enough, just make it bigger or give it more powerful fans.