TEC/Peltier CPU Chilled Water Cooling

[4Ryan6]

' Cooling the Hot Side:

Cooling the hot side of the TEC is a challenge in itself, seeing as how it's best performance comes from allowing the hot side to get hot '


Er.. maximal performance of TEC comes when junction temperature is 0.

Even if that is the operational range goal, you still have to cool the hot side!

Thank You for your input!

I'm just saying you seem to imply in your article that letting the hot side get hot (or hotter) is a good thing but maximum COP / Qw is achieved when the difference between hot and cold approaches zero ( unless you are talking about how it is necessary to turn it on..). The more aggressively you cool the hot side the better it performs (as you probably know).

I am not implying anything, keep in mind this thread covers a 100% operational chilled water cooling solution that in 5 days has been operation for 2 years and 5 months.

It is designed to operate below ambient room temperature, but above the temperature range where condensation begins to form.

Thus it does not comply to a lot of other ideas and discoveries of other uses of the peltier and it's possibilities and capabilities.

The more aggressively you cool the hot side the better it performs (as you probably know).

From my own past experimentation I thought the same as what you're presenting, because I got that information from other websites and threads.

My first original thinking along those lines was to use chilled ice water to cool the hot side, I thought surely that would be an outstanding way to cool the hot side, because I could get much more lower end, but it wasn't.

It was however my very first encounters of causing the peltier to stall out.

If the hot side gets too hot or cold the peltier will stall, the same effect happens with the cold side.

There is a hot/cold balance needed for the peltier to constantly output the useable cold to store in an insulated reservoir, so the useable cold can build up, or for a better term build down to the coolant temperature needed to operate and cool an overclocked system.

There was much experimentation before this ever became a viable cooling solution that this thread covers.

Is that enough of an explanation for your understanding or do you need more?

 

[4Ryan6]

@ apelord

You might want to read this, eweast @ Xtreme Systems has his peltier explanations a lot closer to reality than many do.

http://www.xtremesystems.org/forums/showthread.php?38367-The-Ultimate-Guide-to-TECs

Even some of his explanations do not apply to this cooling solution covered in this thread, as I am using air to cool the hot side of the peltier that many told me was impossible, I'm glad I did not listen to them!

 

[apelord]

' Cooling the Hot Side:

Cooling the hot side of the TEC is a challenge in itself, seeing as how it's best performance comes from allowing the hot side to get hot '


Er.. maximal performance of TEC comes when junction temperature is 0.

Even if that is the operational range goal, you still have to cool the hot side!

Thank You for your input!

I'm just saying you seem to imply in your article that letting the hot side get hot (or hotter) is a good thing but maximum COP / Qw is achieved when the difference between hot and cold approaches zero ( unless you are talking about how it is necessary to turn it on..). The more aggressively you cool the hot side the better it performs (as you probably know).

I am not implying anything, keep in mind this thread covers a 100% operational chilled water cooling solution that in 5 days has been operation for 2 years and 5 months.

It is designed to operate below ambient room temperature, but above the temperature range where condensation begins to form.

Thus it does not comply to a lot of other ideas and discoveries of other uses of the peltier and it's possibilities and capabilities.

The more aggressively you cool the hot side the better it performs (as you probably know).

From my own past experimentation I thought the same as what you're presenting, because I got that information from other websites and threads.

My first original thinking along those lines was to use chilled ice water to cool the hot side, I thought surely that would be an outstanding way to cool the hot side, because I could get much more lower end, but it wasn't.

It was however my very first encounters of causing the peltier to stall out.

If the hot side gets too hot or cold the peltier will stall, the same effect happens with the cold side.

There is a hot/cold balance needed for the peltier to constantly output the useable cold to store in an insulated reservoir, so the useable cold can build up, or for a better term build down to the coolant temperature needed to operate and cool an overclocked system.

There was much experimentation before this ever became a viable cooling solution that this thread covers.

Is that enough of an explanation for your understanding or do you need more?

OK dude, if that is your experience, don't want to get into a drawn out exchange. It's possible we are talking at cross-purposes any way, not sure what is going on in your system and don't have time, knowledge or experience to dissect it in detail. I've a little practical experience with peltiers, not especially scientific and have been sketching out another peltier based system (trying to get a handle on the science first - hence this was in my mind). It's a tricky area though, with many inter-related variables which I must admit I don't wholly understand. My focus has been more on optimising for efficiency not absolute cooling power. In any case, thanks for the reply.

 

[4Ryan6]

OK dude, if that is your experience, don't want to get into a drawn out exchange. It's possible we are talking at cross-purposes any way, not sure what is going on in your system and don't have time, knowledge or experience to dissect it in detail. I've a little practical experience with peltiers, not especially scientific and have been sketching out another peltier based system (trying to get a handle on the science first - hence this was in my mind). It's a tricky area though, with many inter-related variables which I must admit I don't wholly understand. My focus has been more on optimising for efficiency not absolute cooling power. In any case, thanks for the reply.

UR Welcome!

 

[Gibbzy1991]

After reading all of a previous thread, and this one, there seems to be a few basic things you guys are overlooking. Energy is not lost or gained, only transferred. There are no free lunches, everything you put in has to go somewhere! You are getting too caught up analysing "results" without understanding the basics of thermodynamics and how a thermoelectric effect works.

A 1000w heater or kettle or whatever heat generating electrical device you can best understand uses around 1000w of electricity to generate around 1000w of heat.
When it comes to peltiers, they can be used to heat or cool, depending on what your aim is.

Simply put, a peltier uses energy to transfer energy. Its not very efficient at doing that though, so for a peltier to transfer 500w of (heat) energy it will need around 1000w of (electrical) energy. Therefore you have a total energy input of 1500w to dissipate, or "cool" if that makes more sense. Take that up a notch to chilling the water that watercools your cpu and gpu and you might have 800w to transfer, costing 1600w to transfer and totalling 2400w. That is 2.4kw of energy that has to go somewhere.

Another thing with peltiers is the relationship between the hot and cold side. Without fancy graphs, you are basically limited to a 70 degree difference between them, eg. if the hot side is over 70 degrees, the cold side cant get below 0 degrees.

Which isnt too drastic considering the condensation problems of running that cold, but if you want to go anywhere under 25 odd degrees you will need to understand how dew point works, ie the relationship between relative humidity and air temperature. What feels like "normal room temperature" may actually be very close to dew point, so by the time you notice it feels a bit muggy and warm condensation has already set in.

Just keep it simple and be realistic about what you are tying to achieve. You cant cheat physics!

 

[apelord]

After reading all of a previous thread, and this one, there seems to be a few basic things you guys are overlooking. Energy is not lost or gained, only transferred. There are no free lunches, everything you put in has to go somewhere! You are getting too caught up analysing "results" without understanding the basics of thermodynamics and how a thermoelectric effect works.

A 1000w heater or kettle or whatever heat generating electrical device you can best understand uses around 1000w of electricity to generate around 1000w of heat.
When it comes to peltiers, they can be used to heat or cool, depending on what your aim is.

Simply put, a peltier uses energy to transfer energy. Its not very efficient at doing that though, so for a peltier to transfer 500w of (heat) energy it will need around 1000w of (electrical) energy. Therefore you have a total energy input of 1500w to dissipate, or "cool" if that makes more sense. Take that up a notch to chilling the water that watercools your cpu and gpu and you might have 800w to transfer, costing 1600w to transfer and totalling 2400w. That is 2.4kw of energy that has to go somewhere.

Another thing with peltiers is the relationship between the hot and cold side. Without fancy graphs, you are basically limited to a 70 degree difference between them, eg. if the hot side is over 70 degrees, the cold side cant get below 0 degrees.

Which isnt too drastic considering the condensation problems of running that cold, but if you want to go anywhere under 25 odd degrees you will need to understand how dew point works, ie the relationship between relative humidity and air temperature. What feels like "normal room temperature" may actually be very close to dew point, so by the time you notice it feels a bit muggy and warm condensation has already set in.

Just keep it simple and be realistic about what you are tying to achieve. You cant cheat physics!

OK, thanks dude. I'd hope most would have a handle on these facts, but no harm in mentioning it.

What I would be interested in further understanding is the mythical 'Stalling' that appears to have occurred when the peltiers where over-cooled (the hot-stall is less of a concern to me). Any ideas what might be occurring here...?? Anyone??

 

[Gibbzy1991]

As a side note, the ultimate guide to tecs thread you mentioned is just plain confusing. A heap of contradictory nonsense about amps and watts and current draw, without really addressing the basic fact that by using a peltier or tec you are USING energy to TRANSFER energy. Take away all the numbers and think of energy in a more simple way and it will make sense. It is a very simple concept, that is technically quite complicated but at the end of the day doesnt need to be.

Work out how much energy (heat) you create, work out how much energy (electricity) you need to use to transfer that heat energy, pick a peltier that can transfer that energy and then deal with the massive amounts of energy (heat) the whole system will now have to dissipate.

 

[Gibbzy1991]

After reading all of a previous thread, and this one, there seems to be a few basic things you guys are overlooking. Energy is not lost or gained, only transferred. There are no free lunches, everything you put in has to go somewhere! You are getting too caught up analysing "results" without understanding the basics of thermodynamics and how a thermoelectric effect works.

A 1000w heater or kettle or whatever heat generating electrical device you can best understand uses around 1000w of electricity to generate around 1000w of heat.
When it comes to peltiers, they can be used to heat or cool, depending on what your aim is.

Simply put, a peltier uses energy to transfer energy. Its not very efficient at doing that though, so for a peltier to transfer 500w of (heat) energy it will need around 1000w of (electrical) energy. Therefore you have a total energy input of 1500w to dissipate, or "cool" if that makes more sense. Take that up a notch to chilling the water that watercools your cpu and gpu and you might have 800w to transfer, costing 1600w to transfer and totalling 2400w. That is 2.4kw of energy that has to go somewhere.

Another thing with peltiers is the relationship between the hot and cold side. Without fancy graphs, you are basically limited to a 70 degree difference between them, eg. if the hot side is over 70 degrees, the cold side cant get below 0 degrees.

Which isnt too drastic considering the condensation problems of running that cold, but if you want to go anywhere under 25 odd degrees you will need to understand how dew point works, ie the relationship between relative humidity and air temperature. What feels like "normal room temperature" may actually be very close to dew point, so by the time you notice it feels a bit muggy and warm condensation has already set in.

Just keep it simple and be realistic about what you are tying to achieve. You cant cheat physics!

OK, thanks dude. I'd hope most would have a handle on these facts, but no harm in mentioning it.

What I would be interested in further understanding is the mythical 'Stalling' that appears to have occurred when the peltiers where over-cooled (the hot-stall is less of a concern to me). Any ideas what might be occurring here...?? Anyone??

The mythical stalling is just that, mythical. The fundamental dynamics of peltiers, or any thermoelectric device, is that the cooler the hot side is, the cooler the cold side will be.

 

[apelord]

'The mythical stalling is just that, mythical. The fundamental dynamics of peltiers, or any thermoelectric device, is that the cooler the hot side is, the cooler the cold side will be.'

Er yeah, this was my understanding but hard to argue with a man's experience. I'll need to set up something roughly analogous to see if I can replicate.

 

[Gibbzy1991]

'The mythical stalling is just that, mythical. The fundamental dynamics of peltiers, or any thermoelectric device, is that the cooler the hot side is, the cooler the cold side will be.'

Er yeah, this was my understanding but hard to argue with a man's experience. I'll need to set up something roughly analogous to see if I can replicate.

True, but experience also tells me people will believe what they want to. And if they believe a system is working in a certain way, they will conjure up data to prove it.
You cant chase results around in circles forever, sometimes its best to step back and focus on the basics and the real physics of how it all works

 

[Ruwed]

As a side note, the ultimate guide to tecs thread you mentioned is just plain confusing. A heap of contradictory nonsense about amps and watts and current draw, without really addressing the basic fact that by using a peltier or tec you are USING energy to TRANSFER energy. Take away all the numbers and think of energy in a more simple way and it will make sense. It is a very simple concept, that is technically quite complicated but at the end of the day doesnt need to be.

Work out how much energy (heat) you create, work out how much energy (electricity) you need to use to transfer that heat energy, pick a peltier that can transfer that energy and then deal with the massive amounts of energy (heat) the whole system will now have to dissipate.

I don't know what thread you read, but it must have been a different guide than I read. It's pretty simple and specifically addresses the fact that you are using energy to cool....

"Prerequisites: Whatt is a Watt? -

Watt (in terms of electricity) = Current * Energy (Amps * Volts = Watt)
Watt (in terms of heat) = Energy/Time (Joule per second); 4.18Joules = 1 Calorie; 1 Calorie is the energy required to raise 1gram (or 1cc) of water by 1C."

"The larger the wattage of the TEC in proportion to the watts of heat generated by the heat load (CPU) the closer delta T approaches the theoretical limit of 69C. But, the closer you get to the theoretical limit (in this case 69C) the more and more watts it takes for the TEC to achieve it. Thus the less efficient it becomes. Fortunately it's not like TECs were the kings of efficiency to begin with."

It just explains HOW to determine which TEC you should use. If someone can't understand the basics in that guide, then they have no business attempting to build a cooler on their own and need to stick with buying pre-built coolers until they read and learn more.

Maybe I missed the contradictory information, can you point it out? Because I didn't see anything in there that contradicted or was flat out wrong.

Nobody ever said that this was the most efficient way to cool a CPU, this is an active cooling solution, it costs money to run it, but it does work. And as to everyone comparing results, results are what matter!

 

[Gibbzy1991]

Its not about reading guides written on someone elses take on how something works. I dont really want to go through that guide all over again and nit pick details, im not trying to do that.

You aren't using energy to "cool", you are using energy to transfer energy.

Read up on thermo electrics on wikipedia, real information on the real effects involved in what you want to achieve.

 

[Gibbzy1991]

I think that is the fundamental misunderstanding, when you think of it as "cooling" and "heating" it gets confusing.

At the end of the day, you are transferring energy. And when you use energy to transfer energy, you end up with a whole heap of energy that has to go somewhere. That energy has to be transferred to the surrounding air when its all said and done.

Its the cost of cooling in an inefficient way, and the cost of chasing that few degrees cooler by using a peltier is massive.

 

[Ruwed]

Its not about reading guides written on someone elses take on how something works. I dont really want to go through that guide all over again and nit pick details, im not trying to do that.

You aren't using energy to "cool", you are using energy to transfer energy.

Read up on thermo electrics on wikipedia, real information on the real effects involved in what you want to achieve.

You have the same problem as a bunch of engineers I've worked with.

Engineers Terms: You are transferring thermal energy from one location to another by utilizing the thermodynamic properties of two dissimilar metals (the peltier effect) when electrical energy is passed through them.

Laymen's terms: You are actively cooling your CPU.

They are both correct, one is all technical and spelling out the types of energy, one is a generalized statement that is still true.

I understand what I am attempting to achieve. While I appreciate advice, I don't think that coming in and bashing guides and saying we are thinking about it wrong is the way to give advice. Everyone has a different way of thinking about things. I don't need to break it down to "USING energy to TRANSFER energy". That leaves you in the exact same boat. How much energy? What type? How do you calculate it, how is it measured? What are the conversions required to convert electrical energy to thermal energy? Those are all critical questions to answer before you start purchasing components. Otherwise you end up with power supplies that can't handle the load, TEC's that can't keep the CPU cool, ect ect.

All that was spelled out in the guide..... Yup. You can't break it down much simpler and actually accomplish anything. Even in terms of "USING energy to TRANSFER energy" you still have to convert the energy into matching types to make accurate comparisons / estimates. Reading the wiki is even more confusing for most people, it tends to go WAY more in depth than is required for a home project. I'm not designing the next cooling system for the Lunar Lander, I'm cooling my CPU......

Just the way I think about it. Feel free to think about it however it makes sense to you.

 

[Gibbzy1991]

The other thing that hasn't been mentioned much is a thermostat. Buying a pre-built cooler because "I don't understand", as you are suggesting, would be bad news because even if I bought a cooler perfectly sized to keep my cpu at 22 degrees max load, its just going to freeze up when its not under load right?

Shouldn't a system running so close to dew point be regulated to stay at a safe temperature?

Or even better, a perfect system would measure the temperature and humidity and dynamically change the minimum temp to stay above dew point and avoid any chance of condensation while staying as cool as possible?

 

[Ruwed]

If you bought a pre-built cooler it would be "pre-built" which to me implies purchasing from a company who did their own research and incorporated a control scheme to prevent damaging the system it is installed in.

A control scheme has been mentioned in this thread BTW Read up a few pages, I think I linked my thread where I am doing exactly that. If you monitor the temps you can do it manually by adjusting voltage or turning TEC's on / off as needed. I'm lazy and want it automated though.

 

[Gibbzy1991]

Its not about reading guides written on someone elses take on how something works. I dont really want to go through that guide all over again and nit pick details, im not trying to do that.

You aren't using energy to "cool", you are using energy to transfer energy.

Read up on thermo electrics on wikipedia, real information on the real effects involved in what you want to achieve.

You have the same problem as a bunch of engineers I've worked with.

Engineers Terms: You are transferring thermal energy from one location to another by utilizing the thermodynamic properties of two dissimilar metals (the peltier effect) when electrical energy is passed through them.

Laymen's terms: You are actively cooling your CPU.

They are both correct, one is all technical and spelling out the types of energy, one is a generalized statement that is still true.

I understand what I am attempting to achieve. While I appreciate advice, I don't think that coming in and bashing guides and saying we are thinking about it wrong is the way to give advice. Everyone has a different way of thinking about things. I don't need to break it down to "USING energy to TRANSFER energy". That leaves you in the exact same boat. How much energy? What type? How do you calculate it, how is it measured? What are the conversions required to convert electrical energy to thermal energy? Those are all critical questions to answer before you start purchasing components. Otherwise you end up with power supplies that can't handle the load, TEC's that can't keep the CPU cool, ect ect.

All that was spelled out in the guide..... Yup. You can't break it down much simpler and actually accomplish anything. Even in terms of "USING energy to TRANSFER energy" you still have to convert the energy into matching types to make accurate comparisons / estimates. Reading the wiki is even more confusing for most people, it tends to go WAY more in depth than is required for a home project. I'm not designing the next cooling system for the Lunar Lander, I'm cooling my CPU......

Just the way I think about it. Feel free to think about it however it makes sense to you.

The thing is that over the few years the threads have been running, the same problems keep coming up and no real solutions have been found.

I'm just seeing it from another point of view and sharing that viewpoint because the people who have their heads down trying to work it out and get a system running are still having problems.

Unless the basics are kept in the back of your mind when you work out how much energy you are transferring and where it is going, you end up trying to cool the 500 odd watt a decent cpu and gpu will generate with a peltier. Which "costs" 1000 odd watt of power, which combined is like running a decent sized heater flat out. Which isn't all that great of a "cooling" solution.

Its like you took my posts as a personal attack because I tried to break it down to keep it all in perspective.

 

[Gibbzy1991]

If you bought a pre-built cooler it would be "pre-built" which to me implies purchasing from a company who did their own research and incorporated a control scheme to prevent damaging the system it is installed in.

A control scheme has been mentioned in this thread BTW Read up a few pages, I think I linked my thread where I am doing exactly that. If you monitor the temps you can do it manually by adjusting voltage or turning TEC's on / off as needed. I'm lazy and want it automated though.

Like a company in china that just wants to sell parts for hype value? I'm sure they put plenty of thought and research into it, and I'm sure they offer insurance if/when you fry something from using their cooler.

There is a reason why this type of cooling isn't used very often, the main one is because it isn't simple and easy.

But I'm out anyways, no point trying to share information with someone who knows it all to begin with!

 

[Ruwed]

If you bought a pre-built cooler it would be "pre-built" which to me implies purchasing from a company who did their own research and incorporated a control scheme to prevent damaging the system it is installed in.

A control scheme has been mentioned in this thread BTW Read up a few pages, I think I linked my thread where I am doing exactly that. If you monitor the temps you can do it manually by adjusting voltage or turning TEC's on / off as needed. I'm lazy and want it automated though.

Like a company in china that just wants to sell parts for hype value? I'm sure they put plenty of thought and research into it, and I'm sure they offer insurance if/when you fry something from using their cooler.

There is a reason why this type of cooling isn't used very often, the main one is because it isn't simple and easy.

But I'm out anyways, no point trying to share information with someone who knows it all to begin with!

I wasn't posting as a personal attack. There isn't really anyway that I could think of to respond that didn't sound like a dick, so I just answered Nothing personal about it man, I was just trying to point out that saying that we are all doing it wrong isn't the way to give constructive advice.

I'm not sure what the same problems that people keep running into are? I may have just missed it. There are going to be problems when building a custom cooler. That's why Ryan made the thread, so that people have a resource to look at and ask questions.

If you buy from China you get what you get.... I was thinking more of a water cooling system and AIO coolers, which do have warranties for your system if something goes catastrophic.
You are absolutely right, this type of cooing isn't used often because it is complicated and it's not cheap, and for the cost you can go sub-zero way easier. I'd bet I could have purchased a dual phase shift setup for what I'll have into this by the end. I'm doing it because I think it's cool and I enjoy projects.

I'm not sure where you got "someone who knows it all to begin with" from, but that is a personal attack since you want to try to point them out. I'm merely pointing out the flaws in your arguments and then backing them up with relevant facts.
If you have any information to add that I missed, feel free to point it out, but so far you've come in, said that everyone here is thinking about it wrong. That the guides that were posted are incoherent and contradictory. That none of us realize how inefficient TEC cooling is. When you are questioned or a rebuttal is posted, you take it as a personal attack. I can't speak to Ryan since this is his thread, but

 

[4Ryan6]

Are you two done?

I have a couple of comments on the discussion not points to argue so let me be clear that my comments are on the facts of this operational chilled water cooling system covered in this thread.

This type of cooling is active cooling using electricity, personally I thought that was painfully obvious since the peltiers themselves are powered by their own 1,000w power supply, is efficiency lost in the translation, of course it is.

Then why I am running this inefficient type of cooling is simply because it works.

It allows below ambient cooling, that allows higher 24/7 stable CPU overclocks than what can be achieved with any type of ambient cooling, and that's a fact that this inefficient cooling brings to the table.

Is this cooling for everyone?

No!

It cost money to build it, it cost money to run it, it is not for everyone, because not everyone even has the skills required to put it all together.

But for myself, and possibly some others, it is exactly the type of cooling I was after, and the extra inefficient electrical costs are mine to cover and I don't have a problem with that, however that is an operational fact and a prerogative of the end user.

Regarding controlling the electrical that seems to be a concern, I do not have to do that as previously stated way up somewhere in this thread, the hot side of the peltier is cooled by heat pipe air coolers and the cooling fan speeds on those air coolers still use ambient air to cool them.

I can control the actual water coolant temperature in the reservoir down to the levels I want to run the system, which is below ambient but above condensation building simply by controlling the ambient room temperature.

It takes more electricity to do that because the peltiers are always powered, which is my responsibility to pay for, which comes out of my pocket, but is very convenient when it all boils down to push a button and you're in business, everything is working great!

I did tinker with some electrical controlling which caused the myth you refer to as stalling out the peltier, which anyone actually running a peltier will eventually discover peltier stalling is no myth.

The stalling out of the peltier can happen and did happen to my setup, the main reason I posted the link I did is he references peltier stalling, because he has actually used peltiers himself and has first hand application information to share, not some personal theory.

So once you do your own actual peltier experimentation and run into peltier stalling and it becomes a reality you are forced to deal with, then you will know, it is no myth!

Peltier stalling is simply when either the hot side or cold side has reached a temperature high enough or low enough that it affects the temperature of the other side, you have to keep in mind the peltier is only about 1/8th of an inch thick, you have to keep the peltier in a hot/cold operating balance, if you intend to have the cold side constantly producing cold.

For my needs of storing the cold produced in an insulated reservoir, I need the cold side constantly producing cold and in balance with the hot side with zero stalling.

I hope that helps at least some of your understanding, but if you're just determined to argue and make your points, please try to keep in mind this thread covers a fully operational system, that has been running for 2 years and 5 months.

This thread is here so anyone interested that has the modding skills can duplicate this cooling and achieve the exact same results, it has been duplicated by one other forum member ArthurH who is also a Toms Hardware moderator and he has verified it works as claimed, somewhere further up in this thread.

 

[4Ryan6]

For the record this cooling was a successful experimentation to replace the prior Ice cooling I was running, and when it comes to the electrical loss because of the active inefficient peltier cooling, it equaled out to $50.00 a month less than what was being spent freezing ice in 2 different freezers.

So electrical cost wise, to me, it is a win, win, situation!

 

[MrTeal]

This type of cooling is active cooling using electricity, personally I thought that was painfully obvious since the peltiers themselves are powered by their own 1,000w power supply, is efficiency lost in the translation, of course it is.

Then why I am running this inefficient type of cooling is simply because it works.

It allows below ambient cooling, that allows higher 24/7 stable CPU overclocks than what can be achieved with any type of ambient cooling, and that's a fact that this inefficient cooling brings to the table.

Is this cooling for everyone?

No!

It cost money to build it, it cost money to run it, it is not for everyone, because not everyone even has the skills required to put it all together.

But for myself, and possibly some others, it is exactly the type of cooling I was after, and the extra inefficient electrical costs are mine to cover and I don't have a problem with that, however that is an operational fact and a prerogative of the end user.

Regarding controlling the electrical that seems to be a concern, I do not have to do that as previously stated way up somewhere in this thread, the hot side of the peltier is cooled by heat pipe air coolers and the cooling fan speeds on those air coolers still use ambient air to cool them.

I can control the actual water coolant temperature in the reservoir down to the levels I want to run the system, which is below ambient but above condensation building simply by controlling the ambient room temperature.

It takes more electricity to do that because the peltiers are always powered, which is my responsibility to pay for, which comes out of my pocket, but is very convenient when it all boils down to push a button and you're in business, everything is working great!

I did tinker with some electrical controlling which caused the myth you refer to as stalling out the peltier, which anyone actually running a peltier will eventually discover peltier stalling is no myth.

The stalling out of the peltier can happen and did happen to my setup, the main reason I posted the link I did is he references peltier stalling, because he has actually used peltiers himself and has first hand application information to share, not some personal theory.

So once you do your own actual peltier experimentation and run into peltier stalling and it becomes a reality you are forced to deal with, then you will know, it is no myth!

Peltier stalling is simply when either the hot side or cold side has reached a temperature high enough or low enough that it affects the temperature of the other side, you have to keep in mind the peltier is only about 1/8th of an inch thick, you have to keep the peltier in a hot/cold operating balance, if you intend to have the cold side constantly producing cold.

For my needs of storing the cold produced in an insulated reservoir, I need the cold side constantly producing cold and in balance with the hot side with zero stalling.

I hope that helps at least some of your understanding, but if you're just determined to argue and make your points, please try to keep in mind this thread covers a fully operational system, that has been running for 2 years and 5 months.

This thread is here so anyone interested that has the modding skills can duplicate this cooling and achieve the exact same results, it has been duplicated by one other forum member ArthurH who is also a Toms Hardware moderator and he has verified it works as claimed, somewhere further up in this thread.

Thanks for this thread. I'm going to read it through a little more, but something like this was actually very similar to an idea I've had to use some (ie, two dozen) Antec Kuhler Shelf coolers I have sitting around unused.
My basic idea was to remove/bypass the the radiator/pump in my desktop with a couple disconnects that mate to their counterpart on the wall. The rest of the loop I envisioned is what you can see in the drawing below.

I already have one of the pumps and a garage full of copper fittings/valves, though for the manifold I'll probably just buy one of the premade PEX ones. The two pumps would pump water through several (let's say 5) cooling modules. Each module would be one of the cheap 40x40mm AliExpress water blocks, with a 12076 TEC and Kuhler Shelf attached to it.

Some aspects might be still to come as I experiment a little, but is there anything that's a major cause of concern that you'd see before I get started? My main lines from the computer to the cooling setup would probably be a couple dozen feet of 1/2" copper pipe which will give a bit more restriction that I'd like, but if I like the results I can always substitute a different pump.

 

[4Ryan6]

@ MrTeal

I really don't think flow restriction will be a problem through the manifolds with 1/2" copper pipe but that's a lot of copper so make sure you do not have too many other metals in the loop especially not any aluminum!

What will the ball valves be made of?

How are you going to power the peltiers?

PC power supply or stand alone like the Meanwell types?

If you do not have a reservoir to store the cold built up by the TEC assemblies, you will loose a massive amount of cold to the uninsulated pipes and ambient air, have you thought about that?

Even my return lines are insulated.

 

[MrTeal]

@ MrTeal

I really don't think flow restriction will be a problem through the manifolds with 1/2" copper pipe but that's a lot of copper so make sure you do not have too many other metals in the loop especially not any aluminum!

What will the ball valves be made of?

How are you going to power the peltiers?

PC power supply or stand alone like the Meanwell types?

If you do not have a reservoir to store the cold built up by the TEC assemblies, you will loose a massive amount of cold to the uninsulated pipes and ambient air, have you thought about that?

Even my return lines are insulated.

The manifolds shouldn't be bad, it's more just all the length of copper pipe. 40ft (20 each way) of new 1/2 pipe at 1.5gpm gives almost a meter of head pressure drop. Sizing up to 3/4" would drop that by 90%, but most of my existing fittings are 1/2"

The ball valves are just standard copper sweat valves. I'd imagine the actual ball itself is stainless. TBH for a mockup and test I'm actually just looking at using a couple brass 1/2" copper to garden hose connectors, and then using the mating end on my 1/2" computer tubing.

I have a bunch of server PSUs laying around, so power for the peltiers would probably be one of those. I'll probably just do low-side PWM using a big FET to control the peltiers. That way they can also control the fans, though the fan assembly might be a big ducted box using a big AC fan. (IE, http://www.lowes.ca/portable-fans/utilitech-sfc1-500b-20-in-3-speed-high-velocity-fan_g1716654.html)

I was planning on insulating the cold feed line but leaving the insulation off the hot line until I see how the system works. This will obviously be dependent on flow rates and heat load, but I wasn't sure if the return water would be sub-ambient or not. How large a reservoir do you use relative to your system power? Some reservoir is nice, but from controls perspective too much can make tuning the PID loops a bit more of a PITA.

 

[4Ryan6]

My reservoir holds slightly less than 1/2 gallon of coolant.