fluid immersion (boil the whole computer)

[plzdontbetaken]

I am planning a mini-itx build soon and remembered seeing this (http://www.corvalent.com/martinscorner/index.shtml). I am thinking I may give it a go and build the system in a sealed enclosure and use something like Novec 7000 because it is pretty cool looking at a boiling cpu.

I don't think I want to use the peltier though. How would you guys go about cooling the vapor back down to a liquid state?

(possibly the wrong place to post this but it didn't seem to fit anywhere else either)

[t800]

Evaporator sinked same pool as computer, lot efficient than turning vapor back to liquid from the air. But of this kind setup:

-Even if computer is sinked to this liquid, i would still use heatsink, since copper surely beats this liquid in heat conductivity.
-Add some strong fan to move the liquid inside pool, for making heat transfer process more efficient.

[plzdontbetaken]

Evaporator sinked same pool as computer, lot efficient than turning vapor back to liquid from the air. But of this kind setup:

-Even if computer is sinked to this liquid, i would still use heatsink, since copper surely beats this liquid in heat conductivity.
-Add some strong fan to move the liquid inside pool, for making heat transfer process more efficient.

Thanks for the reply t800.

Also, I don't quite understand what you are proposing? Are you saying to use another evap system to cool the bath fluid? The fluid boils at 34C (@ 1atm) so any component that gets hot in the fluid bath will cause the fluid around it to boil, but the fluid is in a sealed (air tight) enclosure so there is no where for the vapor to go. Somehow the fluid has to be brought back to the liquid state.

In the video I linked they used a peltier to move the heat across the system boundary so condensation would form on the cold heatsink. I would want to avoid using a peltier for this purpose.

I was just thinking about keeping the same concept they used in the corvalent demonstration but changing the peltier cooling to using a pump and radiator type design to cool the vapor back down. I guess since this section of the forums is about very very cold I didn't explain my first post very well and probably should have posted somewhere else to begin with. I don't intend to go below ambient with the system. I would just be making it for the novelty of it.

Any cool ideas would be appreciated

[mytekcontrols]

I would just use something similar to the evaporator from a window AC unit as your cooling coil, and either suspend this directly above the liquid coolant (allowing it to condense the vapors, and drip back down), or simply submerge it in the liquid coolant along with the mother board and CPU.

Depending on the temperature you wish to have the liquid coolant be at, you could run it straight from the window AC condensing unit (simply modify the AC unit to allow the evaporator to be put in close proximity to the liquid coolant), and even use the stock refrigerant charge. Or maybe do like many of the people do on this forum, and use a colder refrigerant with a resized cap tube.

[plzdontbetaken]

I would just use something similar to the evaporator from a window AC unit as your cooling coil, and either suspend this directly above the liquid coolant (allowing it to condense the vapors, and drip back down), or simply submerge it in the liquid coolant along with the mother board and CPU.

Depending on the temperature you wish to have the liquid coolant be at, you could run it straight from the window AC condensing unit (simply modify the AC unit to allow the evaporator to be put in close proximity to the liquid coolant), and even use the stock refrigerant charge. Or maybe do like many of the people do on this forum, and use a colder refrigerant with a resized cap tube.

That would be pretty slick.

[wdrzal]

Thanks for the reply t800.

The fluid boils at 34C (@ 1atm) so any component that gets hot in the fluid bath will cause the fluid around it to boil, but the fluid is in a sealed (air tight) enclosure so there is no where for the vapor to go. Somehow the fluid has to be brought back to the liquid state.


Any cool ideas would be appreciated

Just be aware adding heat/boiling this fluid will add/increase the pressure in a sealed container. The container must be strong enough to deal with the pressure or some means of allowing for expansion must be used.

[Callsign_Vega]

I could not find the price of Novec 7000 anywhere. It's not one of those $200+/gallon fluids is it? I know they say its non-toxic, but is it long-term non corrosive to sensitive materials like rubber/plastics used on motherboards?

[mytekcontrols]

The Novec HFE-7000 fluids go for about $250/gallon, and I believe that is the minimum amount sold.

However there is another kid on the block called Vertrel XF which is an HFC, having a boiling point of +55 C, and from what I gather, it is a dielectric, which is a must have for the proposed application. And it presently goes for 1/2 the price as compared to the HFE-7000 when bought in 1 gallon quantities (also sold in as little as 1 quart).

Buy Dupont Vertrel XF (R-43-10mee)

The application brochure states:

Brine/Coolant
DuPont Vertrel is a very effective brine/
coolant to prevent a part or device from
overheating. In a brine / coolant application,
DuPont Vertrel is enclosed in a heat
exchanger system. When DuPont Vertrel comes in contact
with parts at high temperatures, it evaporates, cooling the
parts. As DuPont Vertrel moves to the cooler side of the
heat exchanger system, it condenses back into liquid state.
DuPont Vertrel is used to continuously cool down the parts
that could become damaged due to excessive heat.

Heat Transfer
DuPont Vertrel specialty fluids offer a
convenient and affordable solution to
meet the complex challenges of heat
transfer applications. Their low boiling point
(55 ̊C/131 ̊F) and thermal stability guarantee continued
effectiveness over several boiling cycles.

DuPont Vertrel specialty fluids are typically used as a heat
transfer fluid for semiconductors, semiconductor tools
and pharmaceutical manufacturing.

High Voltage
DuPont Vertrel is widely used as a dielectric
or testing fluid in many high voltage
applications. The dielectric properties
and breakdown voltage of DuPont Vertrel
make it an ideal choice for this application. Typical uses
include electrical testing of insulation sleeves, commercial
glove testing, and sleeve testing equipment.

Depending upon what temperature you want to keep the outside surface temperature of the CPU at, the best you could do with the evaporative cooling option would be +55 C. However if you were to treat this the same as done for water cooling, with the exception of the coolant being in direct contact with the CPU, then the limitation is only in the effectiveness of the heat exchange with the evaporator being used to remove the heat from the liquid. So this gets us back to what I suggested earlier, that being to submerge a hermetically sealed phase change evaporator and the motherboard in a common reservoir of liquid coolant. In this instance there would be no pressure build-up what-so-ever (noting Walt's warranted concern), since the coolant would never be allowed to change state from a liquid to a gas. It simple becomes a liquid heat sink.

The main and only advantages of such a system versus direct contact of a CPU with the evaporator (as done by most of the phase change systems on this forum), is no condensation problems, and possibly more effective heat transfer when you factor in the maximum contact area with the heat producing parts (although as noted earlier, the heat transfer of copper is still superior to the liquid coolant from a "heat sinking" standpoint, so best to leave a heat sink still attached to the CPU).

So in a nutshell we would be talking about using the Vertrel XF as a secondary coolant in exchange with the primary phase change coolant (via some kind of heat exchanger/evaporator). However on the plus side, no liquid pump should be required, since the difference in temperatures (CPU versus Evaporator) should keep things moving around quite fine.

BTW; Dupont Vertrel XF freezes at -80 C, so for those Cascade and AutoC people out there, don't get too carried away

[t800]

So this gets us back to what I suggested earlier, that being to submerge a hermetically sealed phase change evaporator and the motherboard in a common reservoir of liquid coolant. In this instance there would be no pressure build-up what-so-ever, since the coolant would never be allowed to change state from a liquid to a gas. It simple becomes a liquid heat sink.

The main and only advantages of such a system versus direct contact of a CPU with the evaporator (as done by most of the phase change systems on this forum), is no condensation problems, and possibly more effective heat transfer when you factor in the maximum contact area with the heat producing parts (although as noted earlier, the heat transfer of copper is still superior).

There raises problem that you cannot cool computer which fully sinked in liquid lot lower than 0c, since motherboards don't work very well in minus temps.

[mytekcontrols]

There raises problem that you cannot cool computer which fully sinked in liquid lot lower than 0c, since motherboards don't work very well in minus temps.

Good point (thanks t800), but to clarify, only certain components create this limitation. So unless my information is incorrect, it is the Hard Drive that is most prone to low temperature problems, with the manufacturer's recommendations to not go below +10C.

Next in line would be the capacitors, which can go considerably colder (Aluminum Electrolytics -25C to as low as -40C for higher quality variants, Ceramic and Tantalums typically as low as -55C).

Other components:
Lithium Battery: typically -55C
Carbon Film Resistors: typically -55C
Silicon Diodes: typically -65C
misc support IC's: typically -55C

So based on this data, we can assume safe operation down to -25C (colder if the electrolytic capacitors being used are rated for it). Obviously the hard drive will be kept outside of liquid coolant's influence.

Now for the real problem:
Are all the materials compatible with the Vertrel XF? This I would not even want to hazard a guess at. So before going down this road, it would be best to talk with someone at Dupont about what the intended application is, and what effects the Vertrel might have on the various materials be used (of special concern, would be the electrolytic capacitors).

[[XC] gomeler]

Very interesting find on that fluid. Bookmarked for future reference.

[mytekcontrols]

Very interesting find on that fluid. Bookmarked for future reference.

Yes very interesting indeed

I happen to have a gallon sitting right next to me, and have plans to conduct tests on using it for an AutoC. Tentatively I'm thinking about a blended HFC charge using the following refrigerants:

R-43-10mee
R-134a (or R-125)
R-23
R-14
Argon

Sorry for getting

[Callsign_Vega]

Since only the real advantage like you said is eliminating the condensation problem, I think sealing your MB with something like Dragonskin is still the safer/cheaper route while going sub-ambient. Especially with these expensive liquids having unknown long-term corrosive effects on MB components.

[t800]

Since only the real advantage like you said is eliminating the condensation problem.

Well if looking this objectively there is more advantages eg normal phase-change system:

+ You can cool _everything_ at same time, cpu, gpu(s), memory, chipset, power regulators..etc. Normal way you can really cool just cpu and gpu and those require seperate evaps which makes system more bigger and complex.

+ Compressor don't need run on all the time...well depening which kinds temp you are reaching.

+ No condensation

- Messy? (when need to maintenance system, change parts)
- Liquid price (The Novec HFE-7000 fluids go for about $250/gallon)?

[[XC] gomeler]

Since only the real advantage like you said is eliminating the condensation problem, I think sealing your MB with something like Dragonskin is still the safer/cheaper route while going sub-ambient. Especially with these expensive liquids having unknown long-term corrosive effects on MB components.

I already hate having to wait 10 minutes to wrap the socket and evap enclosure in toilet paper. This would be even easier with zero prep work for insulation. How long does it take for dragonskin to cure? 6 hours? 24 hours? No thanks, I want to bench when I want to bench

Yes very interesting indeed

I happen to have a gallon sitting right next to me, and have plans to conduct tests on using it for an AutoC. Tentatively I'm thinking about a blended HFC charge using the following refrigerants:

R-43-10mee
R-134a (or R-125)
R-23
R-14
Argon

Sorry for getting

No risks of hydrolock due to the high evaporation temperature? Just thinking if it evaporates at 55C at 0psig, at 15-20psig which is a decent suction pressure you are talking an even higher evaporation temperature. Unless of course I'm missing something

[plzdontbetaken]

I could not find the price of Novec 7000 anywhere. It's not one of those $200+/gallon fluids is it? I know they say its non-toxic, but is it long-term non corrosive to sensitive materials like rubber/plastics used on motherboards?

I was quoted £163.30 for 4.54kg bottle. I am going to make a mini-ITX enclosure that will use roughly that much to submerge the motherboard plus a dual gpu graphics card.

[plzdontbetaken]

The Novec HFE-7000 fluids go for about $250/gallon, and I believe that is the minimum amount sold.

However there is another kid on the block called Vertrel XF which is an HFC, having a boiling point of +55 C, and from what I gather, it is a dielectric, which is a must have for the proposed application. And it presently goes for 1/2 the price as compared to the HFE-7000 when bought in 1 gallon quantities (also sold in as little as 1 quart).

Buy Dupont Vertrel XF (R-43-10mee)

The application brochure states:


Depending upon what temperature you want to keep the outside surface temperature of the CPU at, the best you could do with the evaporative cooling option would be +55 C. However if you were to treat this the same as done for water cooling, with the exception of the coolant being in direct contact with the CPU, then the limitation is only in the effectiveness of the heat exchange with the evaporator being used to remove the heat from the liquid. So this gets us back to what I suggested earlier, that being to submerge a hermetically sealed phase change evaporator and the motherboard in a common reservoir of liquid coolant. In this instance there would be no pressure build-up what-so-ever (noting Walt's warranted concern), since the coolant would never be allowed to change state from a liquid to a gas. It simple becomes a liquid heat sink.

The main and only advantages of such a system versus direct contact of a CPU with the evaporator (as done by most of the phase change systems on this forum), is no condensation problems, and possibly more effective heat transfer when you factor in the maximum contact area with the heat producing parts (although as noted earlier, the heat transfer of copper is still superior to the liquid coolant from a "heat sinking" standpoint, so best to leave a heat sink still attached to the CPU).

So in a nutshell we would be talking about using the Vertrel XF as a secondary coolant in exchange with the primary phase change coolant (via some kind of heat exchanger/evaporator). However on the plus side, no liquid pump should be required, since the difference in temperatures (CPU versus Evaporator) should keep things moving around quite fine.

BTW; Dupont Vertrel XF freezes at -80 C, so for those Cascade and AutoC people out there, don't get too carried away

That is an interesting fluid. Might be a better choice than the Novec 7000 to use on another project depending on how it interacts with plastics. This first build I am not going to shoot for subzero temps. If this one goes well I imagine I will tweak my design and build another for subzero. I bookmarked your link to test in the future. Thanks .

[Callsign_Vega]

I already hate having to wait 10 minutes to wrap the socket and evap enclosure in toilet paper. This would be even easier with zero prep work for insulation. How long does it take for dragonskin to cure? 6 hours? 24 hours? No thanks, I want to bench when I want to bench



No risks of hydrolock due to the high evaporation temperature? Just thinking if it evaporates at 55C at 0psig, at 15-20psig which is a decent suction pressure you are talking an even higher evaporation temperature. Unless of course I'm missing something

Ah, for benching and switching out components, ya you don't want to wait for Dragonskin to cure. I meant for a permanent 24/7 setup.

[Callsign_Vega]

Well if looking this objectively there is more advantages eg normal phase-change system:

+ You can cool _everything_ at same time, cpu, gpu(s), memory, chipset, power regulators..etc. Normal way you can really cool just cpu and gpu and those require seperate evaps which makes system more bigger and complex.

+ Compressor don't need run on all the time...well depening which kinds temp you are reaching.

+ No condensation

- Messy? (when need to maintenance system, change parts)
- Liquid price (The Novec HFE-7000 fluids go for about $250/gallon)?

Ya, $250/gallon is pretty steep. Springing a leak is a dramatic event!

I will be going chilled water. Have a nice 15k BTU window unit. Going to chill cpu plus 4 GPU's to hopefully -20C at least. Compressor shouldn't have to run all the time but at least I can dump the heat outside with the window setup. But ya, condensation will be a big problem as always. Going the Dragonskin route.

[pythagoras]

No no no no no and just incase you didnt get that no again. I would never recommend a sealed enclosure for a diy build. To many areas for failure of the cooling system, which will result in an explosion.

regards

john.

[n00b 0f l337]

Although the components will survive, what of the changes in resistance with temperature? This can already be an issue with unregulated parts of the circuitry.

[mytekcontrols]

No risks of hydrolock due to the high evaporation temperature? Just thinking if it evaporates at 55C at 0psig, at 15-20psig which is a decent suction pressure you are talking an even higher evaporation temperature. Unless of course I'm missing something

Good point Chris, and something I am tuned into, but keep in mind that HFE-7000 has a boiling point of 34C, and it has been proven to work fine in a water-cooled AutoC (liquid line temperature 18C). Although excessive flood back would not be recommended, so the amount used would be somewhat critical. But remember that in an AutoC, refrigerants such as these can act as carriers (for both oil and other refrigerants -- creating blends with boiling points in-between) and conductivity improvers.

We'll have to take this up in a separate thread, because once again I'm going off topic.

Although the components will survive, what of the changes in resistance with temperature? This can already be an issue with unregulated parts of the circuitry.

Nol another good point of concern.

Quoting from Wikipedia: "Temperature coefficients of thick film resistors are high, typically ±200 or ±250 ppm/K; a 40 kelvin (70 °F) temperature change can change the resistance by 1%."

And thin film resistors are even more stable. So unless I'm not seeing something, it would appear that the resistance changes would be rather small. And I can't conceive of there being any circuits on a MB that would be so critical that they would be bothered by such a small change in resistance. After all we are talking digital and not analog circuitry. But this is certainly something to investigate.