Water cooling system with Peltier thermoelectric support

[Lochekey]

Sweet ice effect - when I set t-stat below -4 eg -5,-6 this is what happened

View attachment 195940

Easiest way to deal with that is to run some windshield washer fluid instead of water. Try to find one that is methanol based instead of glycol based, also try to avoid one that has a lot of additives like rain-x or bug cleaner.

Another alternative if you have access to straight methanol is to mix a 25/75 mixture with water and that should do you down to about -18c.

 

[EarthDog]

I plan to overclock Ryzen 7 to 4Ghz+ with that system.

You can do that on air.

 

[Deous]

You can do that on air.

4Ghz+ - meaning also 5GHz and above to whatever max limits )

 

[EarthDog]

4Ghz+ - meaning also 5GHz and above to whatever max limits )

huh? Not sure im understanding you...

Wait.. think i get it...you wont reach 5ghz... you will be lucky to reach 4.5ghz. That is done on cascade which is colder than your setup.

 

[Mr.Scott]

He's thinking he's gonna get to 5+ on a chiller.
Ain't happening.
Might see 4.3-4.4 if lucky.
5 gig needs LN2

IMO, not worth the hassle.

 

[Deous]

Update
I experimented with only one TEC - 10A/12V (actual is 9.5A) plus water cooled hot surface
Result is almost -35C on tec side



I wonder if I could go lower for higher delta...

 

[Mr.Scott]

Yeah....unloaded.
Put a 150w CPU on it and it'll be +20c in a heartbeat. Higher than that even, because your hot side will now heatsoak your loop over time.

 

[Deous]

Yeah....unloaded.
Put a 150w CPU on it and it'll be +20c in a heartbeat. Higher than that even, because your hot side will now heatsoak your loop over time.

Not necessarily if I add one more 120W TEC. It will transport 2X more heat. Taking it higher - adding 3 TECs will cool CPU to freezing temperatures.
I was right at the beggining of this thread - TECs are great for transporting heat and not exchanging it.

I also noticed some interesting effect - when you stack two tecs same way and connect bottom one till cools then switch it off and switch the top one on then the last hot surface is like 2 times hotter then it should - it may mean than all the heat from the bottom was moved to top - that could indicate that bottom surface can go twice colder.
Unfortunately you must build a switching circuit to control the currents. I wonder if any people did experiments like that.

 

[Mr.Scott]

Not necessarily if I add one more 120W TEC. It will transport 2X more heat. Taking it higher - adding 3 TECs will cool CPU to freezing temperatures.

LOL.
Not at all.
You still have to dispose of the added heat from the TEC itself, thus heatsoaking your loop even faster, thus producing even worse temps.

 

[Deous]

TECs are used wrong - they shouldn't be used to cool water under any circumstances. It is simply useless.
They transport/remove heat from the object - and still that process is not understood by people.
'Delta' doesn't mean that much in terms of heat power.

Go here: https://www.calculatorsoup.com/calcu...ions/power.php

Assuming you have 150W CPU - it generates 511.821 BTUs per hour
Take 100W TEC with 70% efficiency - it removes 238.85 BTUs per hour (341.21*70%)
Two TECs remove more than 470 BTUs/h - so you need 3 Peltiers like that removing more than 700BTUs per hour from the object
Result is 200BTUs spend for the freezing
Conclusion - five 100W TECs are able to completely super-freeze the CPU - it will literally turn into superconductor

 

[Lochekey]

I do not understand your insistence that a TEC cannot be used to cool a liquid filled loop. Using a liquid filled loop between the CPU and the TEC is a good way to maximise the potential surface area of multiple waterblocks mounted to multiple TECs. Conversely the same can then be done on a "hot" side loop to dissipate the heat transfeered and produced by the TEC array.

If you disagree with this concept please explain why and provide evidence to the contrary.

 

[Mr.Scott]

TECs are used wrong - they shouldn't be used to cool water under any circumstances. It is simply useless.
They transport/remove heat from the object - and still that process is not understood by people.
'Delta' doesn't mean that much in terms of heat power.

Go here: https://www.calculatorsoup.com/calcu...ions/power.php

Assuming you have 150W CPU - it generates 511.821 BTUs per hour
Take 100W TEC with 70% efficiency - it removes 238.85 BTUs per hour (341.21*70%)
Two TECs remove more than 470 BTUs/h - so you need 3 Peltiers like that removing more than 700BTUs per hour from the object
Result is 200BTUs spend for the freezing
Conclusion - five 100W TECs are able to completely super-freeze the CPU - it will literally turn into superconductor

The heat they remove doesn't magically disappear. It has to go somewhere. It goes into your loop that you're using to cool your TECs. If you can't keep the loop cool, your TECs remove less heat and your temps rise.

5 100w TECs + an overclocked CPU would require at least 6 120 rads just to stay even.
This is why TECs were deemed inefficient for cooling.
There are a few exceptions. Silver Surfer for example.

 

[EarthDog]

Conclusion - five 100W TECs are able to completely super-freeze the CPU - it will literally turn into superconductor

Whaaaat? This thread is about to get closed as its getting plain silly now.

 

[Deous]

I do not understand your insistence that a TEC cannot be used to cool a liquid filled loop. Using a liquid filled loop between the CPU and the TEC is a good way to maximise the potential surface area of multiple waterblocks mounted to multiple TECs. Conversely the same can then be done on a "hot" side loop to dissipate the heat transfeered and produced by the TEC array.

If you disagree with this concept please explain why and provide evidence to the contrary.

Loop is already cold and chilled by 5K-BTU AC. 100W tec is freezing in seconds without even noticeable change on loop which kicks into chill mode after maybe 25min.
Water shouldn't be used to cool CPU - I told you that million times - water should be used to transport heat from tecs to ac-chilled tank.
Tecs should be surfacing CPU directly with cold side and on heat side the chilled water block should be attached.
I tested 3 tecs together - all works fine

 

[Deous]

Whaaaat? This thread is about to get closed as its getting plain silly now.

Well maybe not a superconductor but will be in deep LOWs like -20C ~ -30C

 

[Mr.Scott]

Not for nothing but, my chiller alone does -20c without wasting my time on TEC's at all.
You're just over complicating the wheel that has already existed for more than a decade.

 

[Deous]

Not for nothing but, my chiller alone does -20c without wasting my time on TEC's at all.
You're just over complicating the wheel that has already existed for more than a decade.

I was referring to tec cooling. I assume ""your chiller"" is aerosol refrigerant based with compressor in the middle.
Stuff like that is hard to build as I mentioned before but you can easily chill water to cool with tecs below ambient.
I don't know how low temperature tecs can deliver but if it is comparable to phase transition cooling - I think it is worth it.

 

[Silver Surfer]

TECs are used wrong - they shouldn't be used to cool water under any circumstances. It is simply useless.
They transport/remove heat from the object - and still that process is not understood by people.
'Delta' doesn't mean that much in terms of heat power.

Go here: https://www.calculatorsoup.com/calcu...ions/power.php

Assuming you have 150W CPU - it generates 511.821 BTUs per hour
Take 100W TEC with 70% efficiency - it removes 238.85 BTUs per hour (341.21*70%)
Two TECs remove more than 470 BTUs/h - so you need 3 Peltiers like that removing more than 700BTUs per hour from the object
Result is 200BTUs spend for the freezing
Conclusion - five 100W TECs are able to completely super-freeze the CPU - it will literally turn into superconductor

Using TECs to cool water is exactly how my chilled water cooling works which gives me the ability to operate in a temperature range that is below ambient. No offense but from your experiments they appear primitive and seems you are trying to make a justification and sometime peltiers just don't do exactly what you think they will. Which means you have to experiment past the math to see what they'll actually do under various applied voltages. Some of your assumptions are ridiculous like

Conclusion - five 100W TECs are able to completely super-freeze the CPU - it will literally turn into superconductor

, that's completely absurd!

I have invited you to look at my Chilled Water Cooling thread to see a fully operational TEC/Water cooling system in operation, today, cooling an 8700K 6 core Intel CPU overclocked to 5ghz all cores with hyper threading. Unfortunately you're happy to stay in your world of negativity but when you make statements like

TECs are used wrong - they shouldn't be used to cool water under any circumstances. It is simply useless.

, you are completely clueless! SS

 

[Deous]

Using TECs to cool water is exactly how my chilled water cooling works which gives me the ability to operate in a temperature range that is below ambient. No offense but from your experiments they appear primitive and seems you are trying to make a justification and sometime peltiers just don't do exactly what you think they will. Which means you have to experiment past the math to see what they'll actually do under various applied voltages. Some of your assumptions are ridiculous like , that's completely absurd!

I have invited you to look at my Chilled Water Cooling thread to see a fully operational TEC/Water cooling system in operation, today, cooling an 8700K 6 core Intel CPU overclocked to 5ghz all cores with hyper threading. Unfortunately you're happy to stay in your world of negativity but when you make statements like , you are completely clueless! SS

It is cooling water with peltiers that is absurd. Ok you didn't get joke about semiconductor, cpu will just cold rock freeze.
I have to look at your setup because removing all that huge heat from your peltiers requires really lot of loud crazy fans.
What is so hard to understand about transported BTUs? Voltage is electric pressure and when pressure drops the temperature does too.
You should study gas laws in more details. How much power your system takes anyway?
And what temperature has your CPU?? You cannot cool 5ghz cpu with water that easy - that's crazy.
You would have to chill a lot lot of water and keep decent pressure on it so CPU gets enough BTUs.

 

[Soulcatcher668]

I think that you need to start with understanding the basics of normal water cooling first. It really sound like you are trying to run before you can walk.

I would love to see some new TEC stuff working, but telling people that have working systems that they do not understand and that it can not possibly work is not helping your credibility.