Sub-Zero Cooling Project

This is my first post here (Hi.), and it seems to be the only resource I can turn to to get the discussion and questions I need answered. While I'm not looking for handouts, support would be appreciated in the sense of pointing me in the right direction or to the right resources. I've attempted to use Reddit for support, and while I did get a good conversation with one person about my project, most of my response was to the effect of "You're stupid for thinking this and are just going to waste a lot of money and time.". Granted, this may be true, but nevertheless, it's a project that's fascinating me and one I want to see to completion. I've been doing a lot of research lately, so I'll describe my goal, and the phases of my project. I don't have the time or funds to make this happen immediately, but it will be steadily worked on.

• Goal: Provide entire sub-zero cooling for an enclosed computer.
• - Sub-Goal: Reduce sound-print at terminal to under 20db, or negate if possible.
• - Sub-Goal: Learn basic and advanced theory on the way of the heating/cooling loop. (Especially since this relates to my job, though on a tiny scale by comparison.)
• - Sub-Goal: Eliminate or minimize the maintenance required for general computer usage.
• - Sub-Goal: Provide an environment that a computer can be overclocked while still maintaining entire computer sub-zero temperatures.

Project Phases:

1. Research cooling medium and target temperature. (In Progress) - Originally this was two separate phases, one where I attempted to locate the ideal cooling medium, and a separate one to pinpoint the ideal target temperature. Since they go hand in hand, I've decided to combine them. This is the part where I'm devoting most of my time to right now, researching the ideal liquid.
   
3. Research cooling method - Assuming I get this project off the ground, and find a medium, I'll have to figure out how to cool it to subzero temperatures. Since I want to maintain my goal of minimizing the sound, my thoughts are going to end up being where I have a cooling station located separately away from the computer, and coolant will be piped into the unit. I have a few ideas for this, but I haven't heavily researched it yet.
   
5. Containment design - When I get all the technical aspects planned out, I'll start having to come up with a modular design for both the unit and the cooling system.
   
7. Warm testing - Out of all the research I've done, none of the fluids I've come across will make it work. Once phase one is completed, phase three will commence when I start testing the fluid at warm temperatures to determine the ability to maintain and function as a cooling medium. This will probably be with spare computer parts I have laying around collecting dust, or some other custom built heat load electronics to test the heat transfer capacity under load.
   
9. Containment design and construction - This is more of a sub-phase where I will be combining phases three and four together using an actual containment that was designed, and with actual computer components.
   
11. Cooling system - This phase will commence once phase two is complete, where the design, build, and testing of the cooling system will occur. The goal of this phase is to obtain and maintain sub-zero temperatures.
    
13. Cold testing - This is where it'll all start to come together as I'll be able to cold test an actual computer system, build a temperature profile for that system, and adjust and bring temperatures to the targeted goals under various load conditions.
    
15. Final testing - At this point I should have an operating system, able to maintain the targeted temperatures at any load conditions that meets all design goals.
    

Overtime, my project phases will be adjusted accordingly, as well as my project goals if they need to be based on what I learn and experience. This will also serve as a database for everything that I research and learn over the course of this project.

Since I have enough information to talk about this now, I can go a little more in detail for my goals and what I've learned so far.

Target Temperature:

While most people look for cooling on their computer system, it's mostly limited to the CPU and a few other key components. There are also pros and cons for every choice, depending on how much time, or effort you put into them. What people don't seem to consider is entire system cooling, at least to sub-zero temperatures. I've read about submersion builds before, such as the one that Puget Sound computer shop did, and the success and learning they had while doing it; along with a few other people who have done this type of cooling with mixed results. The one thing I wasn't able to find was a minimum temperature for all the different components in a computer system. (This is where that one guy on Reddit came in, and he pointed out some key things that he knew about minimum temperatures.) Along with that, capacitors seem to be the most limiting component, depending on the type they are, failing once they start to get under minus 20C.

Other temperatures that I'm at least partly aware of, depending on the type of RAM, it starts getting weird under 0C. Intels start getting weird around -100C depending on the chip. NVidia GPUs, -130C and VRAM -80C, along with various other components starting to fail around -50C. That being said, I'm pretty sure a target temperature of -20C system wide should be reasonable, at least from a temperature consideration type of thing. Of course since there's a lack of material on this subject, it's one of the key points that I would need assistance with researching, or various other experiences of people getting specific component temperatures too low, etc.

Cooling Medium:

Novec would be ideal, as it offers good temperature control, people have done real world testing, and ease of maintenance. The big issue is there is a huge cost, especially since most losses would be unavoidable and would quickly drain funds. Mineral oil combats this, but at a cost. It basically wouldn't allow good temperature control as system response of the temperature would react sluggish, and it gels at lower temperatures.

So why isn't there a better option? I recently discovered a fluid that interests me, that I want to get some feedback on. Therminol. The link to specific data about it is below. Key points: - Max use temp is -115C - Pour temp is -135C - Very low acidity and corrosion - Dielectric constant is 1.99 (which is lower than mineral oil) - It also has a high safety margin, max temp of 175C and almost twice that for auto ignition.

Things I don't know specifically yet: - What the rate of heat transfer is. - Costs. (Though sources I did manage to find hinted it could be inexpensive.) - Long term component effects. If these are comparable to mineral oil or better, I think I might have just found an ideal immersion fluid. The comparable con to it would be that it's still an oil, so cleaning a component would be difficult for upgrading. Now if mineral oil itself is a suitable fluid for immersion, then why wouldn't something else that's equivalent be just as good, if it was able to handle lower temperatures and had good heat transfer qualities?

https://www.therminol.com/products/Therminol-VLT

This effort is... well, a lot for what the gains would be. So I most certainly understand where people are coming from when they say its a waste of time, effort, and money.

Typically, sub zero cooling isn't a 24/7 thing. As you found already, part of the problem is the varying temperatures of cold boot bugs and cold bugs of the various parts. For example, some Intel chips are as low as -100C or so while some AMD chips don't seem to have one (good to -192C or LN2 temps). Another part is condensation... it will constantly build while the unit is on so long as there is humidity in the air. And if you turn it off, that melts and turns into water.

But sitting around -20C that will buy you a couple/few hundred MHz versus custom water. Is it worth it? Its your project, time, and money, so that isn't up to me. If it was me, I wouldn't touch this with a 10' pole...

... that said, I am interested in seeing this happen (with someone else's time/money/effort, LOL) and will be happy to assist with knowledge transfer!

Let me ask you this... what benefits are you expecting to see with this type of cooling?

I'm doing something similar to you. Fortunately we are not the first to try this. My search has led me to utilizing a "chiller box". An air tight(ish) insulated enclosure that is cooled by an A/C evaporator. Air is forced thru the evaporator and then through a closed loop liquid cooler radiator. This both cools the air in the box and sub-zero cools the loop. No condensation forms. The condenser and compressor can be installed remotely for noise control and the sealed insulated box reduces the fan and component noise to nearly nothing. There is a great number of conversations and test results here http://www.overclock.net/f/60/specialized-cooling

EarthDog said:

This effort is... well, a lot for what the gains would be. So I most certainly understand where people are coming from when they say its a waste of time, effort, and money.

Typically, sub zero cooling isn't a 24/7 thing. As you found already, part of the problem is the varying temperatures of cold boot bugs and cold bugs of the various parts. For example, some Intel chips are as low as -100C or so while some AMD chips don't seem to have one (good to -192C or LN2 temps). Another part is condensation... it will constantly build while the unit is on so long as there is humidity in the air. And if you turn it off, that melts and turns into water.

But sitting around -20C that will buy you a couple/few hundred MHz versus custom water. Is it worth it? Its your project, time, and money, so that isn't up to me. If it was me, I wouldn't touch this with a 10' pole...

... that said, I am interested in seeing this happen (with someone else's time/money/effort, LOL) and will be happy to assist with knowledge transfer!

Let me ask you this... what benefits are you expecting to see with this type of cooling?

Click to expand...

The effort is part of the whole experience. There are so many aspects to consider, that I am going to take my time with this instead of just rushing into it, dumping oil on a computer and throwing it in the freezer.

So my question then is, why isn't sub zero cooling a 24/7 thing? Putting aside the overclocking aspects, what makes it detrimental for computers to operate at those temperatures for extended periods of time. (This is assuming that the temperature isn't out of the operating range of individual components, such as the capacitors I mentioned before.)

Another question is, temperature wise, what does custom water give you? Also, I'm still attempting to come up with the concept design of a variable software controlled cooling deployment system. My theory is, that if there is a shroud with an inlet/outlet over a component, such as the CPU, it could receive different cooling levels than the rest of the components. I think the biggest issue with lower levels of cooling is condensation, which this would end up completely neglecting. (Hell, even if for some reason condensation did form in the unit, it would move to the bottom, at which point there could be a system that would detect water, open a valve and then drain a little.)

- - - Updated - - -

Thick8 said:

I'm doing something similar to you. Fortunately we are not the first to try this. My search has led me to utilizing a "chiller box". An air tight(ish) insulated enclosure that is cooled by an A/C evaporator. Air is forced thru the evaporator and then through a closed loop liquid cooler radiator. This both cools the air in the box and sub-zero cools the loop. No condensation forms. The condenser and compressor can be installed remotely for noise control and the sealed insulated box reduces the fan and component noise to nearly nothing. There is a great number of conversations and test results here http://www.overclock.net/f/60/specialized-cooling

Click to expand...

I'll definitely be checking out that, and the whole forum in general since it seems to have additional useful information.

The problem isn't running at those temperatures...As described, one of the main problems is condensation. Anything below the dew point, and you get condensation. If you are below freezing, that isn't a problem as it will freeze (ice doesn't conduct electricity well). But when it melts, then you have a problem. For short durations (think a couple to several hours), we use sub zero cooling to benchmark/race our CPUs and GPUs. We typically put something on the component such as artist eraser, a silicon conformal coating, or liquid electrical tape. This helps prevent moisture from coming in contact with the part. When we are completed with the session, the system is then broken down and left to dry for a couple of days. So the problem is not with how to drain the water.

That said, the point to running that cold is to allow, for example, the CPU to run a lot faster than stock. To do so, you need to increase voltage. Increasing voltage, regardless of the temperatures, will degrade your CPU. How fast? Who knows... it depends on the voltage and the silicon.

As far as what custom water gives you... more thermal headroom than air. Depending on the setup is how much it cools better than air.


You listed goals (my concerns):

*I'm not sure how it will be quieter than air or water when you are going sub zero as well as adding all kinds of components for temperature control (fans, etc). Though I suppose you did say you will have parts in another room...

*What you are doing is actually going to INCREASE the amount of maintenance required! You are adding an incredibly complex cooling system (which needs maintained) to a PC which is already a very low maintenance unit.

If you are just overclocking, do what Thick8 did and go Geothermal (unless he has a new project that is sub zero???)! I am just not sure how you will reach your listed goals (particularly on the less maintenance front) with this type of project.

Sorry, not trying to hate, I think this is a neat idea, but it doesn't seem to align with your listed goals (as far as I can see). How far do you plan on overclocking your CPU when you are sub zero?

I agree Joe, Projects like this are rarely based upon making sense. or even real world performance gains. More so simply because we can.
This will be a lot of fun as long as the OP has the money and time to spend.
I will be following along. There are a few guys who have done cool boxes over at EOCF
May want to pop in and look at some of there trials and tribulations.


Good luck with the project

EarthDog said:

*I'm not sure how it will be quieter than air or water when you are going sub zero as well as adding all kinds of components for temperature control (fans, etc). Though I suppose you did say you will have parts in another room...

Click to expand...

So the main cooling portion will indeed need to make the use of compressors, radiators and fans. I'm finishing up the design details, but taking a lot of inspiration from people who have effectively gutted out old AC units, put the coils in a cooler, and put the liquid in the cooler there. The primary cooling unit will be located farther away from the actual computer and piped in. Logistically I still need to account for materials and costs, but that will be eventually.

EarthDog said:

*What you are doing is actually going to INCREASE the amount of maintenance required! You are adding an incredibly complex cooling system (which needs maintained) to a PC which is already a very low maintenance unit.

Click to expand...

The more complex the system is, then yes. Then again, I only clean out my actual computer (vacuum/air) about once a year, so this would be the goal, of just a single day once a year to do whatever needs to be done. Something like this I won't know until I either have to do it or until I plan it out more.

EarthDog said:

If you are just overclocking, do what Thick8 did and go Geothermal (unless he has a new project that is sub zero???)! I am just not sure how you will reach your listed goals (particularly on the less maintenance front) with this type of project.

Click to expand...

I do really enjoy looking at all of the other projects, sub zero or not, as I can draw inspiration from them, and learn the tricks or failings that have occurred.

EarthDog said:

Sorry, not trying to hate, I think this is a neat idea, but it doesn't seem to align with your listed goals (as far as I can see). How far do you plan on overclocking your CPU when you are sub zero?

Click to expand...

The first computer I plan on doing this with will be more of a server type computer with extra hardware I have just laying around collecting dust. I might as well put it to use with a project, and then go from there. I've only dabbled in overclocking, so this is something that I'll have to learn when I get to that point. A good balance would be between a 24/7 stable system and something that can handle the gaming load that I put on it. But these are just ideas that are floating around in the distance right now.

- - - Updated - - -

Witchdoctor said:

I agree Joe, Projects like this are rarely based upon making sense. or even real world performance gains. More so simply because we can.
This will be a lot of fun as long as the OP has the money and time to spend.
I will be following along. There are a few guys who have done cool boxes over at EOCF
May want to pop in and look at some of there trials and tribulations.


Good luck with the project

Click to expand...

Thank you much, and I appreciate the support and interest.

You were right, it's more than just making sense, since my primary computer is an i7-3820/16GB/Dual 670GTX SLI and my wife's computer is just a slight step down in performance, but more so about learning and about actually putting a use to all this computer equipment that I collected over the years.

Phase One - Update

So, I know it's been a couple of days since I was on here. (I should really change my email notifications or check my email to make sure notifications are coming through.) I'm basically dual posting between this forum and overclock.net, mostly to make sure I get a broad range of input, but also because everyone at one point has considered this sort of thing, or various aspects of it. EDIT: Oh, and the biggest thing that I didn't notice about the Therminol fluid? -6C flash point. Enough said.

But not to digress, the upset.

After a lot of research, bouncing ideas back and forth, and coming up with real world vs theory, I think I've potentially settled on a cooling fluid.

Meet, Dynalene FC series: http://www.dynalene.com/v/vspfiles/templates/210/datasheets/Dynalene_FC_Technical_Data_Sheet.pdf

Here are some key points that I've been putting together:

• Thermal Conductivity of .325378 W/m*K (Water is 0.6)
• Specific Heat of 3.546 kJ/k*K (Water is 4.187)
• Viscosity of 0.6815 Pa*S (Water is 0.653 @ 40C)
• Pour Point of -40C (Usable range would be more around -30C+)
• Conductivity of <10 uS/Cm (DI Water is 0.1)
• No flash point or autoignition temperature, non-toxic, non-flammable, corrosion inhibitor

So on paper, it looks that it's better than anything else I could find, almost twice as good in fact. And pretty close to water, without the whole conductivity issue that water has. I actually contacted Dynalene and they scheduled a technician who can answer my questions to give me a call on Monday. These are the initial questions I wrote to them about, but if you have anything else that would be good to ask, let me know:

• It lists the recommended temperature range from -30C to 80C. Now, while I am guessing the lowest possible point is -40C since that's when it starts to freeze, what are the effects between the -30 to -40 range? Also, what starts to happen to the fluid when it approaches the 80C range, and/or is there a time associated with it at these higher temperatures? (Would it be possible to come into contact with a point that was higher than 80C, is there a time associated with it being above 80C that would cause it to fail?
• It says that it's electrical conductivity is less than 10 uS/cm, and then says that it can maintain low conductivity for at least two years. Is it the fluid's property that it will still be under 10uS/cm at the two year point? Is it possible to know what the effects are of a fluid at the 5 year point, such as any change in the conductivity?
• One of the features says that it eliminates the need for deionizing filters. Do they have any effect on the fluid over time, as such are they able to maintain the electrical conductivity properties if used?
• Concerning closed system use: What the the design characteristics of the fluid when used in an open system? If it is still used in a closed system, yet exposed to an air layer, are there any negative effects to the fluid? If so, what would be the ideal way to completely close the system, using either a CO2 blanket, or other gas, or just filling completely? Similar fluids I've researched, such as the Novec 7100 engineered fluid will evaporate in an open system requiring replacement or an enclosed system, and it sounds similar to this?
• It mentions a non-ionic corrosion inhibitor. What specific type of corrosion is it protecting against? Are there any other corrosion concerns with using this fluid with general electronics such as resistors, capacitors, rubber, etc?
• There is a chart that mentions FC-BG, FC-PG, and FC-EG, but I can't find any specific information on the variants of this fluid. I am guessing that BG is referencing to a BioGlycol base fluid. Is there any difference between the BG/PG/EG fluids?
• This is more of a general question about Dynalene fluids, instead of the FC specifically. You guys have a lot of fluids, and while I've gone through the technical specs for most of them, I wanted to make sure there wasn't any that I missed. Are there any other fluids that exceed this one in terms of thermal conductivity and specific heat, while maintaining the similar properties of electrical conductivity and non-flammable, etc?

The other benefit to this fluid is the cost is a lot cheaper. Not to mention the Novec fluids, but this should run the cost of about the same as mineral oil.

After talking to the Dynalene representative and discussing the future goals and qualities of the fluid, it turns out that even though the fluid has a good dielectric strength, the breakdown voltage isn't so good. Basically, it's perfect for closed loop water cooling systems when you want the peace of mind that leaks aren't going to cause any sort of problems, but past incidental contact, it won't work.

However, as a company they're working very well with me to get me exactly what I need or what would work best. They have their own R&D department that the guy I talked to said he'd ask them, though if I needed to get a fluid designed from scratch that wouldn't be free. Also I should be getting some 1 gal samples of some of their other fluids that should work better, for material compatibility, temperature, etc, that I can test out here. The next guy I need to talk to that deals with these specific fluids that they'll be hopefully sending me is on vacation right now, so I'm going to have to push that back a little bit while I work on other design aspects.

You have me interested.

Keep an eye on the viscosity of your fluids at different different temps. That Dynalene stuff got really thick at the temps you are hoping to run.

Pumps will become an issue, especially if the thermal load keeps changing. Pumps for low viscosity fluids do not pump high viscosity fluids and vice versa.

It would be amazing if that Novec fluid would come down to usable pricing (or my salary increased substantially ).

For those of you that don't know about Novec...... it's amazing.

Soulcatcher668 said:

You have me interested.

Keep an eye on the viscosity of your fluids at different different temps. That Dynalene stuff got really thick at the temps you are hoping to run.

Pumps will become an issue, especially if the thermal load keeps changing. Pumps for low viscosity fluids do not pump high viscosity fluids and vice versa.

It would be amazing if that Novec fluid would come down to usable pricing (or my salary increased substantially ).

For those of you that don't know about Novec...... it's amazing.

Click to expand...

After I talked to Dynalene, that was one of the things that they mentioned, the Novec fluid. Apparently, at least for real world major industry applications, it isn't all that's cracked up to be, and one of the companies they were working with was previously using that fluid and they helped them clean their system out and get a different fluid. I'm still actively talking to them though, so I am going to get some more parts of the project going. (I need to pick up some tools and setup a little workshop so I can start building the enclosures and tank containments.) Hopefully by about that time I have a small test bench setup for some fluid samples that I should be getting here so I can start testing real world applications of the fluids their recommending.

Tellik said:

After I talked to Dynalene, that was one of the things that they mentioned, the Novec fluid. Apparently, at least for real world major industry applications, it isn't all that's cracked up to be, and one of the companies they were working with was previously using that fluid and they helped them clean their system out and get a different fluid. I'm still actively talking to them though, so I am going to get some more parts of the project going. (I need to pick up some tools and setup a little workshop so I can start building the enclosures and tank containments.) Hopefully by about that time I have a small test bench setup for some fluid samples that I should be getting here so I can start testing real world applications of the fluids their recommending.

Click to expand...

Before I say anything: if you do find a good, inexpensive and safe sub-zero emersion fluid, please let me know. I would be thrilled to be part of a project making that available.

That said: I think you might be much better off in continuous cooling using a normal phase change system, cascade if you want to go fairly cold (ie -80C to -120C), and a cold trap. I don't see full immersion in a sub-zero liquid as something that could be sustainable or feasible on anyone's budget outside of the corporate world. I can say this because I have been in the ULT (ultra low temperature) cooling industry for a little while and have been involved in quite a few projects using everything from CO2/LN2 to multi stage refrigeration.

Someone mentioned before something to the effect of a cold trap, or freezing the moisture out of the air. So in theory, you could use one machine to remove moisture (a cold-trap) continuously, while using another to chill components (phase-change) like CPU/GPU/RAM. This would work, but for an innovator such as yourself, might be both overly expensive and way-long as far as an R&D project goes. Also, the machines aren't tiny (at least the affordable ones). I would suggest looking through what is available... Taking what is there and then build upon that - given that you don't infringe. I have thought of developing a forced air chiller, but chilling ambient temps and individual components with a FAC doesn't add up. You need direct contact to chill individual components with any semblance of accuracy. I think for your application a phase change and cold trap would be the way to go if you could afford it and have the patience for R&D. To my knowledge, there aren't any PC size cold traps out there. One thing I didn't mention though is that Cold-Traps must be defrosted every so-often - and how often depends on the humidity of the outside environment and amount of ice on the trap. If you do find an immersion fluid that is affordable and safe, you could use a fluid chiller to maintain temps - but that too is ridiculously expensive - even at low load.

Another problem you will face is control. You must be able to only chill down to a certain temperature, which you and others in part have eluded to (cold bugs, cold boots). While Tech companies are only beginning to build hardware around ultra low temperatures, which is certainly exciting, you will presently need to be able to control your cooling system pretty precisely at ULT. This is why it takes a Pro (like folks on these and other forums) to manually maintain ULT with LN2/Dice and control the computer into extreme modes at the same time.

As far as sub-zero cooling of all of the components (ie the whole case) we do that now with the OCC purge case. People have said that components (theoretically) will fail at -0C etc., but in practice we have not had that happen. The OCC and purge case is a no-ice solution using one machine to cool both CPU and maintain a completely dry environment. We have operated with the purge case at MOBO temperatures down to -50C for hours at a time. It uses LN2 feed from a Dewar, so in an application of continuous cooling for days at a time, unless you had a large LN2 supply and could stand by and swap out dewars every couple days, it would not be feasible. Here is a vid of the OCC running at -180C and Purge Case running at -16C with no ice anywhere and everything running smooth as butta.

http://www.overclockers.com/forums/...ltier-TEC-Assembly-Now-17-Months-in-Operation

You can adapt the CPU Chilled Water Cooling part of the thread link above to do exactly what you want to do, and not have to run it from another room either.

All you would need to modify is adding some type of anti-freeze to the coolant and fully seal and fully insulate around your motherboard socket both sides, and nearby slots and hardware such as your graphics card and system memory, especially if the main computer stands vertically.

My cooling has been in operation for over 3 years now, and I run it below ambient and above dew point, but I have allowed it to go down to 4c running high speed fans to evaporate the condensation.

I normally run it 15c below ambient or -15c DeltaT and that temperature range is completely condensation free, that temperature range below ambient has allowed a rock solid 5ghz OC of my i7-3770K and the CPU has not been delidded.

To run sub-zero all that would have to happen in my case is not stop it from generating cold as I am only running 100% steam distilled water and BioCide (Copper Sulfate), in 3 years of operation all I have had to replace is one of the D5 pumps.

You can continue to investigate all you want but this cooling can be adapted to your needs and the hardware will cost close to $1,000.00 US.

Best of Luck to You! Ryan

Might be worth reading as well: http://www.overclock.net/t/1202757/build-log-project-rainfall-finished
Project Rainfall. I was hoping to do a volenti cooler myself, but I can't find a good container that I can thread valves into.