Overclock.net banner

Rad?

  • Nah.

    Votes: 2 15.4%
  • Its "cool"

    Votes: 2 15.4%
  • Totally!

    Votes: 6 46.2%
  • I want one!!

    Votes: 3 23.1%

Low Temp Freezer PC testing

2713 48
Purple Plant Violet Pink Petal

Blue Purple Light Entertainment Violet
Blue Purple Entertainment Electricity Violet
Green Blue Light Automotive tire Automotive lighting
Font Technology Insect Gadget Automotive exterior


So, I did a thing...

I've heard a-lot of people saying that you cannot put a PC in a freezer, and have seen quite a few try and fail.

So, I got my torches out and went shopping for materials to build an insulated box. Somewhere around the 300$ mark i saw a little chest freezer that was just about the size I wanted for 115$. Sold.

Well, to anyone interested, here's a PC in a freezer. Albeit heavily modified freezer*

Gutted a freezer. Cut a hole in the top. Installed and sealed a viewport. Put a much bigger condenser on it. Put a much bigger evaporator in it. Put a bigger compressor in it (still quite a bit smaller than the ones I see people use from window AC's, about half the size in comparison).

Used cheap fans for the coils as I wasn't sure if they would have the static support to pull through the coils.
Turns out they are great. I will be buying some magnetic high static fans to replace these.

Tips and Tricks:
Fixed metering device is a no go and will only get you to one temp and will be inefficient to maintain.
Internally equalized TXV will be problematic at low temps (below freezing).
Use externally equalized TXV as this will balance the temps under various different loads. Get an adjustable one for the best refrigerant control.
For ultra-low temps, use a custom sub-cooler on your liquid line. Mine is ten 7/8" loops of 3/8" copper through 1-3/8" canister.
Use a suction line accumulator to protect the compressor from liquid refrigerant under various loads.
When balancing the charge in the system, bottles of hot water simulate a heat load very well.
PSU and HDD/SSD's go outside the unit.
Insulating foam and cork tape to seal penetrations.
Vacuum port to remove moisture.
Use Larger than 1/4" acrylic as the 1/4" will bend and buckle under 15" vacuum.
Use a solenoid in conjunction with a high pressure / low pressure cut out control. This will prevent warm refrigerant from migrating to the evaporator during off cycles and help the box retain low temps longer and fewer runs per hour.

**If you run the system with no cold storage, it will cycle constantly and fail to keep up under heavy load.
Currently 3 gallons of water in the bottom of the freezer. This is where the latent heat gets absorbed during normal operations and prevents the system from cycling constantly.

Testing with cold storage at -10c yesterday, the compressor would cycle 2 times an hour rather than the 6 I had allowed it. During CPU benchmarks (nzxt z73 surprised it's still pumping haha) the CPU will still get up into the 40's at 4.8ghz and into the low 60's at 5.4ghz. The GPU is still air cooled (custom low temp loop coming soon. sponsors?) and will get up to 68after repeated benchmarks and actually takes a bit to cool down where as the CPU returns to negative temps within seconds. With the CPU radiator in direct exchange with the evaporator (below the test bench) it takes a very long time to heat soak it. I am currently relying on a mobo probe and GPU probe that can read negative temps, so after the CPU or RAM are below 0c, its all a guess.

Pulling my cold storage down to -15c today my onboard SSD stopped reporting somewhere around -12c (I forgot it was still in there. It still works in the laptop, so only problematic at low temp.)

So there is is guys, a freezer PC. It is, the most expensive and impractical PC case I have ever built, but I dig it haha.
Charging and balancing the system to respond correctly under different loads had been, tedious, so if you're going to build one, pack patience.

Tons of pics and additional info for anyone interested. Hope at least a few of you get a kick out it, i got a kick out of building it.
Window Azure Purple Automotive lighting Rectangle
21 - 40 of 49 Posts

· Registered
Joined
·
33 Posts
Discussion Starter · #21 ·
This is the reason for the high and low pressure controls me the condensing temp will at times run a bit lower than the garages ambient temp (meaning the condenser has a cooling affect on the ambient air. If the ambient in the garage drops below 40f the system will run in very low pressure and likely ruin the compressor. The pressure controls are to help manage the pressures under various circumstances. Hot garage, cold garage, hot box, cold box etc this in conjunction with the externally equalized TXV is where all the magic happens that most others that attempted similar builds simply did not have an adequate understanding of refrigerant dynamics and control.
 

· Registered
Joined
·
188 Posts
so the box is in limp mode until my custom loop materials come in. I may put a fan on it in the interim to have a comparison of air over heat exchanger at low ambient temp vs water cooled at low ambient. The reason I consider this worth doing primarily is viscosity. Once your operating in lower temps, liquid all get thicker and less effective. I’m sure there is a point at very low temp where air would trump water as it has not viscosity and therefore won’t be restricted in its quest to remove heat.

while typing it, it feels necessary to point out that I understand that at conventional temps and I most applications, liquid is almost always more effective than air.

but if you start shopping for non corrosive fluid that retain low viscosity at -40°c, you’ll understand why I am considering that air may be more effective

"Once your operating in lower temps, liquid all get thicker and less effective"
This is very true. The higher the viscosity, the less effective. Even sub zero liquid cooling I have a 240mm fan blasting directly into CPU and VRMs. I was using distilled water with just enough glycol to keep it fluid. I wanted to try using winter windshield wash as coolant, but havent had the time. It is good for -40C to -50C. Lowest ambient I have benched at is -38C to -40C. And yes, the coolant get very sluggish at those temps.

"I’m sure there is a point at very low temp where air would trump water"
I agree as well. I have done a ton of sub zero clocking. And from what I have experienced, I have to agree with this point as well. I have clocked video cards further on air than with liquid cooling.

If you can look back (2012-2013) and find the score boards for Realbench at ROG Forums, you will see that 5.2hgz-5.3ghz on all six cores was very doable even back then. Albeit, benchmarking, not 24/7 operation. I have a few scores there with clocks at 5.2ghz. I actually held top score for a bit at one point. Usually in the top 5.
My point being is that it seems strange that it took this long for mainstream clocks of 5.0ghz and above. But still, not on all cores. Sad part is that unless the CPU is manually tuned, it will never reach it's full potential. But this has been the case for some time.
 

· Registered
Joined
·
33 Posts
Discussion Starter · #24 ·
"Once your operating in lower temps, liquid all get thicker and less effective"
This is very true. The higher the viscosity, the less effective. Even sub zero liquid cooling I have a 240mm fan blasting directly into CPU and VRMs. I was using distilled water with just enough glycol to keep it fluid. I wanted to try using winter windshield wash as coolant, but havent had the time. It is good for -40C to -50C. Lowest ambient I have benched at is -38C to -40C. And yes, the coolant get very sluggish at those temps.

"I’m sure there is a point at very low temp where air would trump water"
I agree as well. I have done a ton of sub zero clocking. And from what I have experienced, I have to agree with this point as well. I have clocked video cards further on air than with liquid cooling.

If you can look back (2012-2013) and find the score boards for Realbench at ROG Forums, you will see that 5.2hgz-5.3ghz on all six cores was very doable even back then. Albeit, benchmarking, not 24/7 operation. I have a few scores there with clocks at 5.2ghz. I actually held top score for a bit at one point. Usually in the top 5.
My point being is that it seems strange that it took this long for mainstream clocks of 5.0ghz and above. But still, not on all cores. Sad part is that unless the CPU is manually tuned, it will never reach it's full potential. But this has been the case for some time.
This is why The cpu is not really my concern. The system has plenty of processing power without breaching thermal. If I need more processing power, it’s as easy as upgrading to 12900k. I don’t have much desire to spend hours adjusting clocks that I’ll never use in regular application. The GPU on the other hand gets maxed quite often during training sessions and the goal here is to have a system run reliably, 24/7 in sub ambient temps low enough to stress the GPU for long periods without approaching thermal limiters.
CPU fan should be here in the AM and I’m still picking parts for custom loop so I will have time to thoroughly test air over applications at different temps to compare against the custom loop when it’s complete.

long term I will likely run the cpu at 4.8/5.0 and the GPU with a conservative Air cooled over clock. As stated above. The goal of this is to create an environment for my GPU to operate in subzero temps for long periods to avoid interruptions or inconsistencies in my training sessions.
 

· Registered
Joined
·
33 Posts
Discussion Starter · #26 ·
Rectangle Handwriting Slope Font Parallel


Basic mechanical design for anyone interested.
High pressure cut out set to 350 psi at point “A”
Low pressure cut out set to -15psi at point “E”
Subcooler must be installed to conjoin along point “B” and point “D”
Accumulator must be installed at point “E”
Solenoid installed between point “B” and point “F”
Metering device between point “F” and point “C”
Sensing Bulb and equalizing tube at point “D”
 

· Registered
Joined
·
33 Posts
Discussion Starter · #27 ·
Hood Automotive lighting Automotive tire Blue Purple

Automotive tire Product Bicycle part Motor vehicle Wood

Bicycle tire Sports equipment Bicycle fork Automotive tire Bicycles--Equipment and supplies

A couple of pictures of the subcooler / heat exchanger in action. At the time of this picture the box was at 5°f. Suction line (blue) temp was -15°f. Liquid saturation temp 92°. Actual liquid temp after subcooler 40°f.
 

· Registered
Joined
·
33 Posts
Discussion Starter · #28 ·
Electrical wiring Gas Audio equipment Technology Electrical supply

Suction line accumulator after protecting the compressor for a 4 hour 80°f to 0°f stress test.

The dark spot under it is from the little puddle it leaves when it defrosts.

It is a looooot of mass to freeze that way and I’ve only seen it build frost like this on the initial run down to 0° And todays stress test. During the normal tests I ran over the last week it never ran long enough to build up any significant frost.

Needless to say, it has done its job in protecting the compressor under various loads.
 

· Registered
Joined
·
33 Posts
Discussion Starter · #29 ·
Watch Purple Light Green Automotive tire

Watch Clock Electrical wiring Safety glove Gadget

Some decibel readings from the condenser and the compressor. For comparison, when I went back to listening to my podcast at 70% through my phone, volume of the phone was between 70-78 decibels.
 

· Registered
Joined
·
33 Posts
Discussion Starter · #30 ·
Font Material property Screenshot Parallel Circle
Font Material property Parallel Screenshot Circle
Font Parallel Circle Screenshot Number
Product Font Material property Parallel Circle
Font Material property Parallel Circle Magenta

Font Parallel Circle Screenshot Diagram


Some refrigerant data from todays stress test.
Blue circle is the evaporator coil temp.
Red circle is the actual liquid temp.
Purple circle is the subcooling.
Yellow circle is the box temp.
All diagnostic temps are Fahrenheit.
Test was run from 80°f to 0°f and took just over 4 hours with around 3 gallons of 80° water in the cold storage. PC was running regular ops during this time. Web browsing, Netflix, and steam, no heavy loads.

As the low side temp drops below freezing and starts to freeze the subcooler you can see a drop in liquid temp and rise in subcooling. This additional subcooling is what makes sustaining very low temps a non issue regardless of the units outside ambient temp.
 

· Registered
Joined
·
188 Posts
Again, just my observations. Like I said, I love creativity. Prove me wrong. I am in no way trying to discourage you.

Would the test show more integrity if there was some sort of a constant controlled source of heat inside during the stress test? This would simulate having an active system inside. Would give you a better idea of how it would operate under load. Would have to be a static temp for accuracy and calculations.

Your test only demonstrated it's ability to lower the temperature of a mass. The function of any freezer. But in your case, a super freezer. You didnt monitor and maintain any kind of inside temperature with an active heat source inside. That should be your next test. Maintain a constant (+/- 5C) inside acceptable temp with a controlled heat source. 5C is a big variable when sub zero overclocking, so should I would shoot for something in that range to start. If you can achieve that, you got something.

I think it would take an exceptionally powerful cooler to achieve what you want to do, if even possible. Even the cost alone of the electricity to run it if it worked. I believe your biggest downfall is volume. There just isnt enough volume inside that freezer to dissipate the heat generated while maintaining a worthwhile temp, without overloading the cooler. Like I said in the other post, equilibrium. I believe physics says the temperature will reach an equilibrium. That harmonious temperature might be lower than what is of any usefulness.
It might be fine for a few hours benchmarking, but I dont think it would serve for an extended period of time.

Just food for thought. This might give you an idea of what you would want to achieve. I will use my i7-3930K as an example. Say at -35C outside ambient and PC outside. Pumping close to 1.6V into it and running it at 5.2ghz, it maintained a CPU core temp of about -10C*. That is with distilled water/glycol for fluid, and a 240mm fan blowing on it as well. There is no way any mainstream or abstract cooling system short of Ln2 or helium would achieve anything lower. Anything practical for that matter.
That was pretty much an absolute constant -35C to -40C total environment. So unless you can maintain temps close to that, I dont see much usefulness. If you can, you got a money maker. Otherwise, far more economic and practical solutions are available.

So if all this works.... great.

If it dont...... mod it further and you got yourself one bad-ass phase change unit. Pump that **** thru a cold plate.

Go Hard or Go Home!

EDIT: The CPU core temp of -10C I mentioned above was reported by my Asus Rampage IV Formula motherboard. The Rampage mobo series will report sub zero temps.
 

· Invalid Media
Joined
·
11,327 Posts
View attachment 2556706 View attachment 2556707 View attachment 2556708 View attachment 2556709 View attachment 2556710
View attachment 2556711

Some refrigerant data from todays stress test.
Blue circle is the evaporator coil temp.
Red circle is the actual liquid temp.
Purple circle is the subcooling.
Yellow circle is the box temp.
All diagnostic temps are Fahrenheit.
Test was run from 80°f to 0°f and took just over 4 hours with around 3 gallons of 80° water in the cold storage. PC was running regular ops during this time. Web browsing, Netflix, and steam, no heavy loads.

As the low side temp drops below freezing and starts to freeze the subcooler you can see a drop in liquid temp and rise in subcooling. This additional subcooling is what makes sustaining very low temps a non issue regardless of the units outside ambient temp.

This is a really 'cool' project you're doing.

Something to consider down the line :D :

 

· Registered
Joined
·
33 Posts
Discussion Starter · #33 ·
Again, just my observations. Like I said, I love creativity. Prove me wrong. I am in no way trying to discourage you.

Would the test show more integrity if there was some sort of a constant controlled source of heat inside during the stress test? This would simulate having an active system inside. Would give you a better idea of how it would operate under load. Would have to be a static temp for accuracy and calculations.

Your test only demonstrated it's ability to lower the temperature of a mass. The function of any freezer. But in your case, a super freezer. You didnt monitor and maintain any kind of inside temperature with an active heat source inside. That should be your next test. Maintain a constant (+/- 5C) inside acceptable temp with a controlled heat source. 5C is a big variable when sub zero overclocking, so should I would shoot for something in that range to start. If you can achieve that, you got something.

I think it would take an exceptionally powerful cooler to achieve what you want to do, if even possible. Even the cost alone of the electricity to run it if it worked. I believe your biggest downfall is volume. There just isnt enough volume inside that freezer to dissipate the heat generated while maintaining a worthwhile temp, without overloading the cooler. Like I said in the other post, equilibrium. I believe physics says the temperature will reach an equilibrium. That harmonious temperature might be lower than what is of any usefulness.
It might be fine for a few hours benchmarking, but I dont think it would serve for an extended period of time.

Just food for thought. This might give you an idea of what you would want to achieve. I will use my i7-3930K as an example. Say at -35C outside ambient and PC outside. Pumping close to 1.6V into it and running it at 5.2ghz, it maintained a CPU core temp of about -10C*. That is with distilled water/glycol for fluid, and a 240mm fan blowing on it as well. There is no way any mainstream or abstract cooling system short of Ln2 or helium would achieve anything lower. Anything practical for that matter.
That was pretty much an absolute constant -35C to -40C total environment. So unless you can maintain temps close to that, I dont see much usefulness. If you can, you got a money maker. Otherwise, far more economic and practical solutions are available.

So if all this works.... great.

If it dont...... mod it further and you got yourself one bad-ass phase change unit. Pump that **** thru a cold plate.

Go Hard or Go Home!

EDIT: The CPU core temp of -10C I mentioned above was reported by my Asus Rampage IV Formula motherboard. The Rampage mobo series will report sub zero temps.
I have been stress testing the system with benchmarks all week. Yesterdays stress test was to see how long cold storage would take to pull down as it is a significantly greater load on the system than a benchmark.

benchmark and heavy load stress tests are a non issue. Even using the 3090 for mining over a 4 hour period only raised my cold storage temp by 3°. Although my cold storage was not under -20f at this time, I expect it would have held up better if it had hardened.

Again when you speak to not having enough “volume” I believe this is because you don’t understand the dynamics of refrigerant adequately, or perhaps have not considered the capacity of the evaporator coil.

-5c under heavy load was achieved day 1 lol I’ve been shooting for -20 to -40 c as water takes a hardened state at this point and will absorb much more heat before changing state. The balancing of the system has been figuring out how to get enough low temp refrigerant into the coil at -30c to handle the heat off put by the GPU and CPU. This is about the point it starts to get difficult. At -5°c there is plenty of refrigerant in the coil to absorb this heat and maintain box temp. As the temp lowers, the metering device restricts the flow of refrigerant to lower the temp more. My “problem” threshold has been -25c. There is simply too little refrigerant in the coil at this point to maintain and after a long run will bring the box back up to the -15c range. During the build, this is when I introduced the cold storage, as even if I lose control of the box temp, if I have a “battery” of cold mass to absorb excess heat, the compressor will have less work to get the box back to -25.

I can tell from your last two openers that you probably get a lot of kick back from your posts on forums.

Some advice:

It’s much easier to ask questions or ask how something works as opposed to making assumptions with incomplete data. You’ll get better results. 👍

If I didn’t know better, I would almost accuse you of posturing to know MORE about the design of a refrigerant circuit than I, so please, if you have improvement suggestions, please make them. I love new ideas.

Note* this isn’t a money maker. Anyone silly enough to spend this kind of money on a pc case should have their head examined. I do HVAC for a living and intend to use the boxboth for practical training on refrigerant theory and as a VR training rig. The box is probably only valuable to me and a handful of HVAC nerds or trainers. If anyone here asked me to build them one, they would laugh me out of the forums when I told them the cost haha

I still am not sure if you even have a question or just here to flex brain cells. If you do have a question, I would be happy to answer it. If your just here to flex, I’ll be back after my coffee 👍 🍻
 

· Registered
Joined
·
33 Posts
Discussion Starter · #34 ·
This is a really 'cool' project you're doing.

Something to consider down the line :D :

This is super rad!! I am trying to stay away from condensation inside the box though and I believe even with the box at -20, I will still get condensation issues if I use LN2. I think there would be a balance point where the loop would actually help cool the box and once the two met a fair equilibrium, the concern would go away.

But, I’d have to hurt someone if they ever opened the box, it would ruin everything.

Currently, I can be have one of the kids play horizon 5, open the box at -10, grab a few otter pops and close the box. The differential is not great enough currently to worry about short term exposure.

The temps with LN2 would make that a scary situation lol I could see myself diving at one of the kids as I see them reaching for the freezer lol

“NOOOOOOOOOOOOO”
 

· Registered
Joined
·
33 Posts
Discussion Starter · #35 · (Edited)
Automotive lighting Light Electrical wiring Cable Motor vehicle


The evaporator coil. About 140x140x360mm as a water radiator, it would be awesome! For refrigerant, in this circuit, it is perfect 👍 The depth of it is really what makes it shine in this application. If you can see, I had to add plexiglass panels on top and bottom to make sure the air would be drawn across the entire coil. Because the Evap is so deep, the air spends a lot of time in contact with very low temp refrigerant making it very effective at absorbing heat.

Single circuit evaporator coil (this has proven to be semi problematic, and if I had it do do again I would have used a dual or quad circuit coil)
This was taken from a failed unit and repurposed for this project. It’s intended for use in a low temp box around 10x larger than my freezer. This is why I chose to use low airflow PC fans instead of a high rpm refrigeration fan. High airflow would have led to short cycles and it would operate more like a freezer and less like a pc cooler.

This entire coil will be the temperature indicated in the blue circle in each of the pics. When this coil is at -15f, I cannot generate an amount of heat inside the box that the refrigerant cannot absorb. As I add heat to the box, the TXV increases the flow of refrigerant to account for the additional heat and keeps the box at temp.

with cold storage at -20f run cycles are even fewer that I had expected.

there were no heavy load stress tests yesterday as it’s still running with the failed kraken cooler. Air over fan should arrive today and I can get back to testing.

It’s worth noting that in a low temp design like this with a solenoid and pressure cut outs, the compressor does not need to run for the evaporator to absorb heat.

When the air in the box reaches the set temp, the solenoid closes and pumps all of the refrigerant into the condensing coil to cool it down before the next run cycle. This is where a lot of the efficiency comes from. When it does it, it prevents the warm refrigerant from migrating back into the evaporator and equalizing the box temp. Instead, the evaporator coil remains at a lower temp than when it was running, allowing it to continue to absorb heat. The cold storage acts like a battery during this period and helps the box maintain it low temp in the off cycle for much longer. Once the box temp comes up, the solenoid opens and once the system reaches operational pressures the compressor re-engages and pumps the evaporator coil back down to the temp it needs to be to accommodate the box temp.
 

· Registered
Joined
·
33 Posts
Discussion Starter · #36 ·
The dynamics of 404a are really what make this system functional beyond those designed by others. It’s capacity to absorb heat at the temps that one would want to to cool a PC make it perfect. Most other refrigerants have much higher target temps in mind, making them more effective at high temps but much less effective at low temps like the ones I want to achieve. Outside of exotic refrigerant blends, 404a is perfect.

Here’s a spreadsheet on the dynamics of 404a. Anyone with a pocket calculator can run some numbers and tell you that it does not take much 404a to absorb a massive amount of heat at the temps I am aiming for.

it does, like I stated, leave a bit to be desired below the -25c mark and I do wish I was knowledgeable enough to create a consistent blend myself (not off the table) to maintain higher pressure at lower temp. But, to be honest, I knew going in that I would have a threshold around the -20 to -30c mark that I wasn’t exactly sure I would be able to overcome. I do know now that from testing, I would have much more capacity at the -25c mark if I had used a dual or quad circuit coil for the Evap like i did for the condenser. And is not really necessary to overcome. The box can currently maintain below freezing under all different workloads and operations. And that really is what I wanted to achieve with it.

 

· Registered
Joined
·
33 Posts
Discussion Starter · #37 ·
The capacity curve of the system is worth noting as well due to the use of 404a refrigerant.

If you use the data in the spreadsheet to run your numbers for capacity, you will notice that there is a “sweet” spot for 404a and it’s right below freezing. Perfect 👍

the problem here is, what happens if the box is warm already? The refrigerant does NOT have the capacity to absorb the heat from an 80°f box as well as the system under full load.

once you get the box temp down into the 20°f range, it starts to meet the capacity requirements of the box and the system under full load.

with the box at 0°f it runs optimally and has a significant greater capacity than is necessary to maintain the system under full load.

then again as you near the -20 mark, the refrigerant is capable of absorbing heat, there is simply much less of the refrigerant present at this temp to absorb heat, this is where you start going backwards.

introduce, custom made subcooler, suction line accumulator, and cold storage, these additions are there to combat the issues I knew I would have at certain temps.

will I be able to achieve a -40 box under load? Not likely with 404a. But it can idle at -40 and I’ll be much happier after a long training session to see the pc sitting at -15c rather than the 60’s and 70’s it would constantly run at previously.
 

· Registered
Joined
·
94 Posts
View attachment 2556026
View attachment 2556027 View attachment 2556028 View attachment 2556029 View attachment 2556030

So, I did a thing...

I've heard a-lot of people saying that you cannot put a PC in a freezer, and have seen quite a few try and fail.

So, I got my torches out and went shopping for materials to build an insulated box. Somewhere around the 300$ mark i saw a little chest freezer that was just about the size I wanted for 115$. Sold.

Well, to anyone interested, here's a PC in a freezer. Albeit heavily modified freezer*

Gutted a freezer. Cut a hole in the top. Installed and sealed a viewport. Put a much bigger condenser on it. Put a much bigger evaporator in it. Put a bigger compressor in it (still quite a bit smaller than the ones I see people use from window AC's, about half the size in comparison).

Used cheap fans for the coils as I wasn't sure if they would have the static support to pull through the coils.
Turns out they are great. I will be buying some magnetic high static fans to replace these.

Tips and Tricks:
Fixed metering device is a no go and will only get you to one temp and will be inefficient to maintain.
Internally equalized TXV will be problematic at low temps (below freezing).
Use externally equalized TXV as this will balance the temps under various different loads. Get an adjustable one for the best refrigerant control.
For ultra-low temps, use a custom sub-cooler on your liquid line. Mine is ten 7/8" loops of 3/8" copper through 1-3/8" canister.
Use a suction line accumulator to protect the compressor from liquid refrigerant under various loads.
When balancing the charge in the system, bottles of hot water simulate a heat load very well.
PSU and HDD/SSD's go outside the unit.
Insulating foam and cork tape to seal penetrations.
Vacuum port to remove moisture.
Use Larger than 1/4" acrylic as the 1/4" will bend and buckle under 15" vacuum.
Use a solenoid in conjunction with a high pressure / low pressure cut out control. This will prevent warm refrigerant from migrating to the evaporator during off cycles and help the box retain low temps longer and fewer runs per hour.

**If you run the system with no cold storage, it will cycle constantly and fail to keep up under heavy load.
Currently 3 gallons of water in the bottom of the freezer. This is where the latent heat gets absorbed during normal operations and prevents the system from cycling constantly.

Testing with cold storage at -10c yesterday, the compressor would cycle 2 times an hour rather than the 6 I had allowed it. During CPU benchmarks (nzxt z73 surprised it's still pumping haha) the CPU will still get up into the 40's at 4.8ghz and into the low 60's at 5.4ghz. The GPU is still air cooled (custom low temp loop coming soon. sponsors?) and will get up to 68after repeated benchmarks and actually takes a bit to cool down where as the CPU returns to negative temps within seconds. With the CPU radiator in direct exchange with the evaporator (below the test bench) it takes a very long time to heat soak it. I am currently relying on a mobo probe and GPU probe that can read negative temps, so after the CPU or RAM are below 0c, its all a guess.

Pulling my cold storage down to -15c today my onboard SSD stopped reporting somewhere around -12c (I forgot it was still in there. It still works in the laptop, so only problematic at low temp.)

So there is is guys, a freezer PC. It is, the most expensive and impractical PC case I have ever built, but I dig it haha.
Charging and balancing the system to respond correctly under different loads had been, tedious, so if you're going to build one, pack patience.

Tons of pics and additional info for anyone interested. Hope at least a few of you get a kick out it, i got a kick out of building it. View attachment 2556036
You sir... are awesome... I will hold your beer any day!
 

· Registered
Joined
·
33 Posts
Discussion Starter · #39 ·
The principal and practice is the same as if someone were to call me and tell me they were converting a 20,000 sq ft warehouse to a call center. Give me a floor plan and get an idea of how many people, computers and lights there will be. Calculate capacity.

Or if they were turning an old 1800 sq ft insurance office into a pizza kitchen. Account for the extra capacity from the ovens and walk in and done.

If it’s a grow facility that needs 100,000 sq ft conditioned and humidified and dehumidified simultaneously….And they want a chiller that will cool large masses of material from 220°F+ to -80f quickly, you get out your pad and paper and you run the numbers carefully.

This is easy. When you know exactly the potential heat of each component in the system. Math math math = capacity.

If someone can use a r410a/134a window ac to cool a bunch of glycol and cool their PC, Its no doubt one would obtain much better performance much more efficiently with a 404a circuit.

Another important note for comparisons, when comparing to water or glycol cooling (cooling glycol first then pumping it to the PC) with that method the glycol is simply a carrier to absorb and transfer heat to your cooler. The longer it has to travel and the energy used needs to be accounted for. And neither glycol, nor water are bad at absorbing heat, but in comparison with direct refrigerant contact, they simply to not compare and cannot remove heat quickly enough to sustain high loads.

In Hvac, your typical water loop or glycol system will show you about 10° change on an efficient loop. The reason they are used is because they are not expected to remove heat quickly and they can simply run on a pump continuously rather than an expensive compressor.

When you work with refrigerant loops you can expect temperature changes up to 60° In different applications.

All that matters is capacity. How much heat do you need to remove? Put that much refrigerant through your system. Know the rate at which you will gain this heat? Match the airflow to your heat load.

In all practicality, this is a phase change system, just much more efficient, no worry for condensation, where it lacks is the direct refrigerant contact with the cpu so it will not achieve sustained low core temps. The core temps throughout testing have made it up to the 60°c range regularly with little to no effect on the box temp. The highest liquid temp I recorded during testing while the cracken was still running at -10°f was 1°c. So the liquid would be a bit warmer than the box with no real effect on the box, I blame this primarily on whatever fluid they use in the kracken. The GPU tests had a bigger impact on the box than the CPU did actually bringing my cold storage temp up over long tests
 

· Registered
Joined
·
188 Posts
I have been stress testing the system with benchmarks all week. Yesterdays stress test was to see how long cold storage would take to pull down as it is a significantly greater load on the system than a benchmark.

benchmark and heavy load stress tests are a non issue. Even using the 3090 for mining over a 4 hour period only raised my cold storage temp by 3°. Although my cold storage was not under -20f at this time, I expect it would have held up better if it had hardened.

Again when you speak to not having enough “volume” I believe this is because you don’t understand the dynamics of refrigerant adequately, or perhaps have not considered the capacity of the evaporator coil.

-5c under heavy load was achieved day 1 lol I’ve been shooting for -20 to -40 c as water takes a hardened state at this point and will absorb much more heat before changing state. The balancing of the system has been figuring out how to get enough low temp refrigerant into the coil at -30c to handle the heat off put by the GPU and CPU. This is about the point it starts to get difficult. At -5°c there is plenty of refrigerant in the coil to absorb this heat and maintain box temp. As the temp lowers, the metering device restricts the flow of refrigerant to lower the temp more. My “problem” threshold has been -25c. There is simply too little refrigerant in the coil at this point to maintain and after a long run will bring the box back up to the -15c range. During the build, this is when I introduced the cold storage, as even if I lose control of the box temp, if I have a “battery” of cold mass to absorb excess heat, the compressor will have less work to get the box back to -25.

I can tell from your last two openers that you probably get a lot of kick back from your posts on forums.

Some advice:

It’s much easier to ask questions or ask how something works as opposed to making assumptions with incomplete data. You’ll get better results. 👍

If I didn’t know better, I would almost accuse you of posturing to know MORE about the design of a refrigerant circuit than I, so please, if you have improvement suggestions, please make them. I love new ideas.

Note* this isn’t a money maker. Anyone silly enough to spend this kind of money on a pc case should have their head examined. I do HVAC for a living and intend to use the boxboth for practical training on refrigerant theory and as a VR training rig. The box is probably only valuable to me and a handful of HVAC nerds or trainers. If anyone here asked me to build them one, they would laugh me out of the forums when I told them the cost haha

I still am not sure if you even have a question or just here to flex brain cells. If you do have a question, I would be happy to answer it. If your just here to flex, I’ll be back after my coffee 👍 🍻
Dont get you panties in a knot. Constructive criticism from somebody that has been around this industry for 25+ years. Nothing more. I have seen a lot of different ideas on cooling PCs. A PC inside a freezer is nothing new. One that is user friendly, doesnt exceed the cost of the system it is cooling, efficient, not a colossal, and cost effective, is new.

"I have been stress testing the system with benchmarks all week."
You never mentioned this testing before, or I missed it. That is why I suggested a static controlled heat source inside during testing. If the inside heat source isnt static and controlled, how could you possibley have accurate data if no control numbers? There has to be a constant somewhere.

"because you don’t understand the dynamics of refrigerant adequately"
I wont lie. No, I dont have the refrigerant dynamics understanding you have. I do have a basic understanding though. And a little knowledge of physics. Also, I have been around this industry long enough to see and know what works and what doesnt. You arent the first to try this nor will you be the last. I am sure some of those who have attempted this, have your level of knowledge and some greater.

"-5c under heavy load was achieved day 1 lol"
I was suggesting for integrity, the inside temp cant vary more than +/- 5C. Not maintain -5C. A variation of 5 degrees at sub zero is huge when overclocking. This I know for fact. Binderdundat.

"I can tell from your last two openers that you probably get a lot of kick back from your posts on forums."
History will tell. Yes, I have been somewhat of an ******* in some forums in the past. But in this case, I am simply offering my observations. Is your ego so big that you beleive you are on a level above all and nobody can offer any insight?


"It’s much easier to ask questions or ask how something works as opposed to making assumptions with incomplete data."
If I had a question, I would ask. As I said, I simply offered observations from a different view. Take it as you wish. Dont get too high on yourself.

See, historically I have been an *******. People change. I managed to get through this reply civilly. I am kinda proud of myself. LOL

No more comments from here. Good luck.
 
21 - 40 of 49 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top