Thermosyphons and similar craziness

[johan851]

Article Here: http://overclockers.com/articles1246

These look like they could actually be a pretty viable solution for CPU heat. A lot of ideas in the past could be quickly recognized as a total flop, but a commerical version of this shows a lot of potential.

It almost makes me sad that I just finished my elaborate watercooling box.

 

[boardboyd]

It looks like it will still be awhile before this comes out.

I thought that another company tried this approach before but had bad results:
http://overclockers.com/articles626/

 

[I.M.O.G.]

Interesting prospect, and even better link boardboyd, it does have a very similar appearance to that previously reviewed model.

Johan, please place this link in the OP - its the article this thread is in reference to:
http://overclockers.com/articles1246

 

[Skeen]

This was the conclusion Joe came to about the device you linked, boardboyd:

"These results are not overwhelming. I believe there is a fundamental problem with a system such as this:

In order for this system to work, there must be a pressure differential to drive the liquid through the system - this makes up for the lack of a pump. As such, the system will not work if it's too cool. As C-System states:

"The fan's capacity is 36 cfm, if you would like to change this to a higher cfm fan, it could happen that the cooler will not operate. This cooler is an automatic-circulation-type driven by the difference of pressure, so if a high cfm fan is used in cooling, trouble will be caused between the difference of the pressure, which will cause a disruption in the flow of refrigerant..." "

I wonder how he got around that. Maybe the use of water in a vacuum was the difference, as the other company used some fluid with a low boiling point.

A full load temp of 47C at 1.4v isn't all that great, I'd say. The other thing that bugged me was the stock and overclocked results. He went from 2.0 to 2.2GHz. Whoop dee doo, did he even have to raise the voltage for that?

 

[macklin01]

I don't think a high overclock is the target market. It could be great for a moderate overclock in a near-silent computing environment. It looks like it could be very good for the large market of people who crave some performance boost, but not at the cost of a lot of noise. Many, I'm sure, have toyed with the idea of water cooling but haven't been willing to take the plunge. It could also make an interesting stock cooling option for Prescotts, etc.

The thing I find myself asking is how a "thermosyphon" is any different than "big heat pipe". It seems to be the same thing to me. Also, the boiling point reduces with pressure? That (though the vacuum) would help enhance the circulation / performance.-- Paul

 

[Overlag]

i like the idea very much. it would be a good step up for OEM's, although im not too sure about extreme overclockers.

My vapochill micro works on the same basics, although it not really a cooling loop. It still uses liquds that boil to the top, then recondenses back to the evaporator.

There is very little temp change with the Vapochill micro too, but as the first link form 2002 shows, if its too cold, the system doesnt really run well. But it keeps my X2 4400 cooler than my silent boost did. 65c vs 42-46 depending on day....

 

[John G]

I was thinking the whole article, "This would probably pair up really well with TEC's..." Then, he had the small print comment at the end.

I agree that the target market for this isn't big overclocks, although a high capacity unit with a TEC could change that. This is more targeted to provide a solution for high heat CPU's, dual cores, etc... Heatsinks with fans eventually don't cut it or get too noisy. Standard heatpipes aren't that great. Various water cooling and phase change solutions out there do work, but are too elaborate to be feasible in general. There was nothing in the middle other than to pioneer new ideas. This looks like one that has been now pioneered and worked into something that works.

 

[Overlag]

I was thinking the whole article, "This would probably pair up really well with TEC's..." Then, he had the small print comment at the end.

I agree that the target market for this isn't big overclocks, although a high capacity unit with a TEC could change that. This is more targeted to provide a solution for high heat CPU's, dual cores, etc... Heatsinks with fans eventually don't cut it or get too noisy. Standard heatpipes aren't that great. Various water cooling and phase change solutions out there do work, but are too elaborate to be feasible in general. There was nothing in the middle other than to pioneer new ideas. This looks like one that has been now pioneered and worked into something that works.

It sounds to me like the fluid in this unit is slightly pressurized in order to adjust the boiling point closer to the desired temperature. Hence the "once you open it up, it's toast" comment.

i think a TEC version would probably need different pressures to make its fluid boil at the right temps. But i agree the design would probably work very well with TEC's.

 

[Elif Tymes]

Its in a vacuum, therefore it's anti pressurized....

sorta .


Anyways, that looks really cool, I bet it has some serious potential in the OEM/other worlds.

 

[John G]

Its in a vacuum, therefore it's anti pressurized....

sorta .


Anyways, that looks really cool, I bet it has some serious potential in the OEM/other worlds.

Yeah, as soon as I posted that I realized it was backwards and edited it to say vacuum....then, I realized that the comment I was referring to specifically said it was under a vacuum to begin with. Duh. So, I edited it again to take it out.

 

[John G]

i think a TEC version would probably need different pressures to make its fluid boil at the right temps. But i agree the design would probably work very well with TEC's.

Actually, it wouldn't need to be different. For example, if you take a pot of water and boil it on the stove. Pressure stays constant at atmospheric. If you turn up the heat (which a TEC would essentially do to the system) the water just boils faster and the temperature stays the same: 100 deg C. The boiling regulates the temperature. In order for the temperature of the water to rise above 100C, it has to ALL boil off into gas... So, basically the system would just boil faster with a TEC. As long as the radiator could keep up, things would be fine.

BUT....the TEC gives you the ability to adjust the system to boil at a higher temperature without running the CPU hotter. That would allow your radiator to run hotter, which increases the heat transfer rate to the environment. So, you could run a similar size radiator at a higher temperature and have lower CPU temperatures at the same time. The cost would be the extra component and the energy to run it. There was an article just a little while ago about pumpless water cooling with a TEC run at low voltages that weren't prohibitive. All that would have to be done to raise the boiling temperature would be to decrease the vacuum on the closed system.

 

[boardboyd]

Agreed with John_G, if you decrease the pressure you increase the boiling point of a liquid, in pipe flow if you decrease the pressure too much you get cavitation, where at room temp the water turns into a gas.

I think a TEC would increase efficiency of the thermosyphon, I remember some article I think on this site that talks about how heat radiation increases with temp. increase. When I have more time I will find the article.

 

[Elif Tymes]

Technically, in a perfect vacuum, water boils at room temperature. In all honesty, in a perfect vacuum, water changes from a water, to a gas due to the fact that there is no pressure to keep it together, while @ 1 AMU there is enough pressure to keep it liquid till about 100c, after which the pressure/chemical bonds have absorbed too much energy, so that the molecules seperate from each other(I.E., it changes into a gas)

Once the liquid reaches the radiator, and the energy is taken out of it(I.E. the heat is absorbed by the metal, which is then absorbed by the air, and then taken away via the fans. Depending on Airflow/surface area, the heat could be absorbed fairly rapidly.

What I don't see, is how it makes sure that the absorbed water doesn't go back down the same pipe, I.E. the tube on the left takes the evapped water up, and the tube on the right drops it down, or does it merely do whatever it feels like?

 

[Avg]

Here is what I'm thinking though, since the liquid in it has to reach a certain temp to transfer heat efficiently , the cpu temps wouldn't necessarily increase with overclocking since the temp at default frequency is only that high because that is the amount of energy that makes the system work. Do you get what I'm trying to say? So if you were to overclock temps might not increase very fast. Well I'm not as informed as some of you guys are but that is what I was thinking about while reading this thread. Anyway I would like to see how it does when voltage has to be increased and there is a considerable increase in heat.

 

[macklin01]

What I don't see, is how it makes sure that the absorbed water doesn't go back down the same pipe, I.E. the tube on the left takes the evapped water up, and the tube on the right drops it down, or does it merely do whatever it feels like?

That's an interesting point. I think it doesn't matter. I'd think of it more like rain: it can still rain while water vapor rises. (A little inaccurate, but you see where I'm going. ) Actually, I doubt that so much fluid is condensing per time that it would totally fill one tube or the other. Instead, I'd imagine it trickling down both pipes, while vaporized fluid rises in both tubes. I wouldn't think that circulation (in terms of a one-way loop) is so important as allowing the vapor to rise to the radiator and allowing the fluid to collect (via gravity) in the block. I think that this is the most important point of the orientation. (In fact, it may not need a full loop at all. Neither do heat pipes.) -- Paul

 

[I.M.O.G.]

I wondered that also actually... I was thinking along the lines of I wonder why two tubes are used, and why not one twice as large, or more smaller tubes... It would have been interesting to have done the testing on this.

 

[skou]

What I don't see, is how it makes sure that the absorbed water doesn't go back down the same pipe, I.E. the tube on the left takes the evapped water up, and the tube on the right drops it down, or does it merely do whatever it feels like?

The answer, is gravity, or the "sucking of the Earth."

They set it up, so the vapor is always on the high side of the circuit, and the liquid is on the low side of the loop. (Although it really won't matter. Heatpipes work the same way, and only use one tube.) I would imagine that these need to be properly oriented, when installed. (This end must face up.)

steve

Edit, here is the write-up on the testing.

http://www.hpl.hp.com/research/papers/2002/thermosyphon.pdf (#1)

/edit

 

[John G]

Here is what I'm thinking though, since the liquid in it has to reach a certain temp to transfer heat efficiently , the cpu temps wouldn't necessarily increase with overclocking since the temp at default frequency is only that high because that is the amount of energy that makes the system work. Do you get what I'm trying to say? So if you were to overclock temps might not increase very fast. Well I'm not as informed as some of you guys are but that is what I was thinking about while reading this thread. Anyway I would like to see how it does when voltage has to be increased and there is a considerable increase in heat.

You're absolutely right. The temperature will stay the same until you reach the point of overloading the system. It'll just boil faster until reaching that point.

There will be a marginal temperature increase along the way of a couple degrees, but it is just from the thermal resistance between the heat source and the boiler. Much like voltage drop across a wire or resistor as you increase current flow.

 

[I.M.O.G.]

That document which is linked shows a vapor line and a liquid line, which makes sense in their example due to the placement of the lines... the vapor goes up and gravity would keep water from condensing there, while the liquid line is lower and gravity would pull the fluid through. In Joe's pictured example, I'm not sure how to explain it though.

 

[Sjaak]

Thermal resistance of 0.15 C/W...thats reason for droooooooolage