• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

My Chilled Water Cooling Obsession Is Now Version 4.0.

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Silver Surfer

Member
Joined
May 8, 2011
Location
Darlington, South Carolina
A friend called my chilled water cooling a hobby but actually it is absolutely an obsession! In one way or another I have been Chilled Water Cooling my CPU for a higher overclock threshold since 2011, it all started with 10 gallons of water in a Coleman cooler. Using gallon jugs of distilled water that had been frozen and exchanged out daily to keep the water in the cooler around 10c. The 10c coolant range was a high level of curiosity for me to experiment with, which allowed experimentation into below ambient cooling. Many think that any temperature below ambient produces condensation but that's just not true, but there are affecting factors such as dew point, and relative humidity, elevation relative to sea level, Blah, Blah, Blah. However those factors do not have as much of an impact in a home that is air conditioned from the outside raw environment. So my experimentation was simply to go below ambient but above condensation forming and the 10 gallons of water in the cooler provided the first limitation. Which was no matter how much ice you add to that 10 gallons of water you cannot get much below 10c or around 50f. Because as long as you still have liquid water with the ice there will be an equalization point of ice vs water. Now for arguments sake you can force lower temperatures, by simply adding salt, producing the same results of freezing ice cream, but you're not going to pump salt water through your cooling system, that would be really bad.

So for me 10c coolant temperature range became an operational target window, WHY? Because all CPUs are designed to operate in an ambient world whether you live in Alaska or India which have drastic temperature differences CPUs are designed to operate in those environments successfully. The CPU manufacturers work with the Motherboard manufacturers sharing their design parameters so the motherboard which supports the processors cooling knows exactly from the CPUs signals when to increase cooling fan speeds to keep the CPU cool enough to operate efficiently, all the way up to the point that the cooling is running at it's maximum capability to attempt to keep the platform cool. These trigger signals of course happen at various internally set temperature levels in each CPU, and if the motherboards cooling solution can not handle keeping the CPU cool enough, the CPU itself protects itself by various means to get it's temperature down, down clocking itself to pitiful performance levels. Now what happens if you can actually operate below these ambient trigger levels, take a minute to think on that.

So back to my obsession, the 10 gallons of water in Coleman cooler with ice blocks, served a learning purpose to discover there was an operational level below ambient without condensation destroying my computer, but freezing the ice everyday, swapping out jugs, got old! There had to be a better way of reaching a 10c coolant temperature without freezing water and swapping out jugs, so after much research I decided to give peltiers a try, even though I got literally nothing but negativity from the idea. Especially when I divulged I was going to cool the hot side of the peltier with an air cooler, instead of a water block and radiator cooling loop. Those that were actually using peltier cooling at overclock.net, thought trying to use air coolers to cool the hot side was ridiculous and not even possible, but they were wrong!

Chilled water cooling vs #1 was just an experiment to see if a heat pipe air cooler could actually cool the hot side of a 12v 200w 15a peltier and it could, but back then 9 years ago heat pipe air coolers were pitiful compared to whats been available for the last 5 years. So today's heat pipe air coolers can definitely get the job done cooling the hot side of a peltier, but back when chilled water cooling vs #2 was put into operation the Thermalright Ultra-120 eXtreme Rev.3 was my choice to use, running 3 peltier assemblies. Version #2 ran 3 200w 12v 15a peltiers, requiring the cooler to have it's own power supply to handle the peltiers and when all 3 peltiers were powered it was 600w just for the peltiers power consumption. But it actually worked successfully and even though it was an electrical power sucking monster it was still using less power than freezing the water in the jugs using 2 freezers.

The main goal of Chilled Water Cooling vs #3 was to see if the job could be accomplished using less electrical power, and possibly be accomplished with just the computers power supply, so the end result of vs #3 was a maximum peltier draw of 300w, using only one power supply. The secondary goal was to house it all inside one case, which was accomplished, but turned out to be a bad idea, for various reasons, but since I am not writing a book here unless asked, I forego the details, lets just say I learned what not to do. The goal of vs #4 was to take the best of vs 2 and 3 and put it into a smaller case, with quick disconnects and have the cooling unit separate from the actual main computer but powered by the main computers power supply. The picture below is a look under the hood of Chilled Water Cooling Version #4.

20200501_100906.jpg

More to come, but I have to get this picture posting figured out. SS

20200501_100841.jpg

20200421_183205.jpg

20200425_180808.jpg

20200425_180819.jpg

20200425_180829.jpg

20200425_180837.jpg

20200427_145645.jpg

The Graphics Card is Independently Cooled Using a Watercool MO-RA3 radiator, Alphacool Reservoir, and D5 Pump.

20200428_121519.jpg

20200429_080954.jpg

All of the Chilled Water lines and Reservoir are insulated to retain the cold.

20200429_182841.jpg

My helper needed a break!

20200429_100453.jpg

20200429_081009.jpg

Just in case I added this fan over the CPU water block, if any condensation even begins to form it will dry it before it is a problem.

20200501_175604.jpg

10c is my coolant temperature it supports overclocking the 8700K to 5200mhz at 1.35v Bios setting.

I hope all of you are well during this time, Bless You All! Silver Surfer
 

Attachments

  • 20200501_175610.jpg
    20200501_175610.jpg
    2.4 MB · Views: 559
  • 20200501_174503.jpg
    20200501_174503.jpg
    1.6 MB · Views: 561
  • 8700K at 5200mhz.jpg
    8700K at 5200mhz.jpg
    386 KB · Views: 543
Last edited:
Do you have a PID controller for the TECs to maintain 10C? Otherwise I would think coolant temp would fluctuate during load. Otherwise a very cool setup. Have you pushed the CPU harder with good results?
 
Do you have a PID controller for the TECs to maintain 10C? Otherwise I would think coolant temp would fluctuate during load. Otherwise a very cool setup. Have you pushed the CPU harder with good results?

I have a 12vdc 30a Thermal Controller that I bought from ThermoMart, the exact model I have they do not carry anymore, but it is something similar to this.

https://www.thermomart.com/Temperature-Controller-Digital-Thermostat?product_id=179

Two of the three TECs run all the time and the 3rd TEC assembly is set to cut in at 11.0c and cut out at 9.9c which maintains the 10c range.

I am using 225w 15a 24v peltiers they are 40mm x 40mm, but they are only powered with 12v, which cuts their wattage output to just barely over 100w and their amperage draw is still at least 8a, which allows my main computers power supply to run itself and all three TECs at full power load problem free.

https://customthermoelectric.com/19911-5p31-15cq-wt-thermoelectric-peltier-module.html

The water blocks I am using have been modified they were covered in the version 3.0 link if you're interested.

As far as the pushing it harder than Intel Burn Test, No, but that is because I spent 9 days and close to 80 hours getting this thing back operational and I'm presently enjoying using it. I will be doing further testing later on as I am not done with this setup, I just modified an old Swiftech water block that was used in version 2.0 to test as well, that is one reason I bought the quick disconnects, to be able to add in an experimentation without having to drain the entire loop.
As far as benchmarks and heavy gaming like FarCry 4 it has no problems maintaining its set parameters, but the 8700K is overclocked to 5200mhz which has forced other necessary testing. Which is can the system provide a stable boot at the BIOS 1.350 voltage when the coolant is at ambient room temperature and so far it can. Since it does not start generating cold until powered, the insulated reservoir coolant temperature has risen to ambient and only increases by .02c before it starts to drop, and is down to 10c in 9 1/2 minutes.
 
Last edited:
Nice thanks for the reply. I guess by pushing things I was thinking more extreme OC stuff. Not that 5.2 isn't impressive in it's own right, but if the CPU is only peaking at 50 running IBT have you thought about some higher voltage runs?
 
After numerous runs of Intel Burn Test.

I have heard that the Intel Burn Test should only be used if you want to shorten the life of your CPU.

Question: Why does it show only 11 watts? Surely the wattage was far higher than that during the test.

Final note: I have yet to see a TEC that was more than 10% efficient. By that I meant, it took 200 watts of power to remove 20 watts of heat. I don't know anyone using TEC.

For my own data collection, I would love to see the following numbers:

1. Ambient temperature of the room wherein the chiller is being tested.
2. Average core temperature under full load.
3. Wattage being reported under full load.

That will give me enough information to rate your entire cooling solution.
 
I have heard that the Intel Burn Test should only be used if you want to shorten the life of your CPU.

Question: Why does it show only 11 watts? Surely the wattage was far higher than that during the test.

Final note: I have yet to see a TEC that was more than 10% efficient. By that I meant, it took 200 watts of power to remove 20 watts of heat. I don't know anyone using TEC.

For my own data collection, I would love to see the following numbers:

1. Ambient temperature of the room wherein the chiller is being tested.
2. Average core temperature under full load.
3. Wattage being reported under full load.

That will give me enough information to rate your entire cooling solution.

Yes and we would like to see the same data for YOUR cooling setup!
 
I have heard that the Intel Burn Test should only be used if you want to shorten the life of your CPU.

Question: Why does it show only 11 watts? Surely the wattage was far higher than that during the test.

Final note: I have yet to see a TEC that was more than 10% efficient. By that I meant, it took 200 watts of power to remove 20 watts of heat. I don't know anyone using TEC.

For my own data collection, I would love to see the following numbers:

1. Ambient temperature of the room wherein the chiller is being tested.
2. Average core temperature under full load.
3. Wattage being reported under full load.

That will give me enough information to rate your entire cooling solution.

As I have said before, efficiency and TECs should never be in the same sentence, however this setup is far more efficient than my original experiments were, so that is a solid step forward.

This is a possible cooling solution that actually works, but is not something I am trying to publicly sell that's why it is posted in Extreme Cooling Section, so take it as you will, I could care less what your final assumptions actually are, to me they are irrelevant. I do not need your approval for what I am presently doing or what is to come. :) SS
 
Let's see it!

When I get around to it ED, it is not a priority, presently I am modifying a test water block to improve cold transfer, the entire reason for using the quick connects was to be able to add in additional TEC assemblies for testing. The goal is still to accomplish it all off of one power supply carrying the load and I am presently running an Enermax 1200w Platimax D.F. to power it all. So this post update 4.0 is just to get it visible and I chose to put it here instead of overclock.net for hopefully positive input, but I do realize the popcorn crowd is in every forum. :) Hope you and your family are doing well through this time! SS

- - - Auto-Merged Double Post - - -

I have heard that the Intel Burn Test should only be used if you want to shorten the life of your CPU.

Question: Why does it show only 11 watts? Surely the wattage was far higher than that during the test.

It is not like I run Intel Burn Test all day, it is a fast way to establish a first step in stability, followed up by Prime 95 to ensure full stability.

Now your Question: Is also a curiosity to myself and a friend of mine because even under full load Core Temp is reporting 11.1 watts?
I noticed it doing that after my last Motherboard BIOS Flash and I am investigating into that, I am running the last available release of Core Temp?
 
...I have yet to see a TEC that was more than 10% efficient. By that I meant, it took 200 watts of power to remove 20 watts of heat. I don't know anyone using TEC.

...give me enough information to rate your entire cooling solution.

Doc, i think you are obsessed. Some people just make things for fun, for self learning, to occupy their free time etc... Most of us know that this "contraption" (no offense) makes very little sense in efficiency or cost, but we still appreciate the sharing and we learn from it. I'm sure we all strive to improve, as SS is on it's 4th generation of this cooling solution, but sometimes we dont care or want to rate our projects.

After all this is a community of enthusiasts...
 
Doc, i think you are obsessed. Some people just make things for fun, for self learning, to occupy their free time etc... Most of us know that this "contraption" (no offense) makes very little sense in efficiency or cost, but we still appreciate the sharing and we learn from it. I'm sure we all strive to improve, as SS is on it's 4th generation of this cooling solution, but sometimes we dont care or want to rate our projects.

After all this is a community of enthusiasts...

Well Said! :)

ThankYou! SS
 
Nice work as always SS, love it.

Question: How long do the TEC usually last?

Thank You! :)

Actually my original peltiers from 2011 testing are still 100% functional they were 15.6vdc and only got 12vdc so they were under powered as also was the entire version 2.0. Version 3.0 and 4.0 are 24vdc peltiers getting 12vdc so they are seriously under powered, I have no idea how long they will actually last like that. A friend and I were actually discussing that this morning, some have destroyed peltiers by overpowering them but I don't know of any that have failed from under powering them.

Note to all: I would like to commend the original Artic MX4 thermal compound, my Version 2.0 ran for over 4 years, and then was in my storage room for a year and a half. Artic MX4 was used on the peltier mounting both hot and cold sides. I disassembled one of those older TEC Assemblies to use in version 4.0. The Artic MX4 thermal compound was still soft and able to be wiped off with a paper towel. I was expecting to see it completely dried out and literally turned to dust, especially on the hot side of the peltier. I was shocked to see it still in a slightly moist state. To me that was absolutely amazing! :thup:
 
Last edited:
That's good to hear. I was expecting to hear they burn out every few years. Good to hear about the MX 4 as well. :thup:
 
Modified one of the Swiftech water blocks from version 2.0, cut out the center of the top that was the original jet plate support down to the copper baseplate, this block works great because the pin grid on the copper plate is the same size as the 40mm x 40mm peltier, so the cold produced is picked up all across the peltier. The little copper wall between the center inlet and the outlet forces the coolant back to the rear and sides and then around the wall and out the outlet, without the copper wall the coolant would follow the past of least resistance right out of the outlet. This block was clear epoxy poured to allow a consistent gap across the water block of approximately 3mm, the flow rate can control how long the coolant is in the block as the cold is transferred to the coolant.

The block was poured with the vinyl tubing in the inlet and outlet holes, the tubing removed and the holes tapered for a better flow, this was done before the epoxy had fully cured so the copper wall could be forced into place, the top of the copper wall actually is pressed into the copper base plates ultra fine pin grid array. So that copper wall is a solid part of the water block now, completely different than what Swiftech intended it for. Cooling a CPU processor and using the same water block with no modifications at all does not work, because the tight tolerances will allow the coolant to freeze, and then Partys Over. I do not run any type of anti-freeze because I do not run this cooling sub zero and never have, but I discovered with an unmodified water block, it would absolutely freeze blocking the flow and then all of them would freeze, eventually splitting the block open if you don't catch it in time.

20200503_073711.jpg

20200503_081418.jpg

20200503_081432.jpg

20200503_081444.jpg

20200503_081456.jpg

So now I have used that modified water block and another 40mm x 40mm Peltier with quick connects and an older Thermalright 120 Extreme Heat Pipe air cooler, to add in a 4th TEC assembly on the backside of the 4.0 case. This unit is switch controlled and when energized can hold my temperature in the 10c range during stress testing. The main computers 1200W power supply is powering it all, the total peltier load is about 400W maximum draw, the normal operation draw is always at least 200w. The thermal temperature controlled TEC adds and additional 100w, when it is automatically energized, bringing the load occasionally to 300w, the 4th TEC is only switched on when needed totaling 400w. The picture below is the 4th TEC assembly.

20200508_165617.jpg

I also modified the top where the Silverstone Air Penetrator AP182 180mm, because the fine mesh of the case top was restricting the fans air flow, cut that fine mesh out and used a 180mm wire fan grid so the fan could breathe.

20200508_165604.jpg

That's all for now folks, any other modifications I'll post when they are done to it, Thanks for reading! SS


Almost forgot: Just in case any of you are not old school and cannot figure out what a Swiftech Apogee XT CPU water block ever looked like inside, here it is.

Swiftech Apogee XT Inside.png
 
Last edited:
This is a possible cooling solution that actually works, but is not something I am trying to publicly sell that's why it is posted in Extreme Cooling Section, so take it as you will, I could care less what your final assumptions actually are, to me they are irrelevant. I do not need your approval for what I am presently doing or what is to come. :) SS

Way, way out of line.

I was just asking a question.

I am just collecting data. How could you possibly see an attack from a mere question?
 
Surely you must realize that offering an unsolicited "rating" of someone's cooling system could have easily been interpreted with the connotation of a review or passing judgement on the system. That's certainly how I interpreted it at the time it was written. If anything I found your comment out of line.

Of course in hindsight we all know what you were trying to accomplish, but at the time you had not proposed your formula yet. For a self proclaimed wordsmith, however, "rate" still seems to be an unusual choice of words for something that has little basis in providing apples to apples comparison between dissimilar systems. You could have easily actually described what you wanted to do by saying, "calculate the effectiveness of" vs "rate."

Anyway as far as forums go, this is one of the nicest and most friendly out there, so the fact that you seem to be having frequent disagreements with a number of long standing members should tell you something. Remember very few posters here actually know each other personally, and I'm in no position to say what is actually ad hominem vs what is misinterpreted as such. Just something to consider.
 
Back