Sign in with
Sign up | Sign in
Your question
sticky

TEC/Peltier CPU Chilled Water Cooling - Page 2

Tags:
  • Overclocking
  • Water Cooling
  • Product
  • Below Ambient Cooling
  • CPUs
  • Cooling
Last response: in Overclocking
Share
a c 249 K Overclocking
a c 110 à CPUs
June 30, 2013 10:31:42 AM

06yfz450ridr said:
I did have my res on the back mounted but the brackets were crappy and broke. I dont have anymore room on my desk due to my three monitors for eyefinity so I removed that case as well as the peltier.

I've always wanted to mount my res on its side up top but I'm always worried it won't flow right


It will flow right as long as you set it up right, the tubing would need to be setup in a constant slope to the pumps intake, and if the present holes in your case don't work out to do that, that's what drills and drillbits are for.

Sometimes you have to make your ideas work, that's what this entire thread is about.

As one of my friends at THGF says; "There's rarely an out of the box solution that fits perfectly, so the water cooler user is actually automatically inducted into case modding."

Or stated as something similar! Credit to Moto.



a b K Overclocking
a b à CPUs
July 3, 2013 8:37:23 PM

Textfield said:
Thank you for going through this 4Ryan6, and for posting all your findings. I wanted to ask your advice on something.

Most sane people would stick with the full or super towers for this kind of thing, and stray away from even mid towers. But I'm not sane, so I've been hypothesizing a build that I might actually attempt: a mini-ITX below-ambient machine.


Okay, before you say I'm absolutely insane, which I am, the case I would use is actually a bit large for mini-ITX. It's the Bitfenix Prodigy. I'd use it mainly because it has space for two 240mm radiators (1 intake, 1 exhuast) and one 140mm exhaust radiator. It also appears you have to rip out most of the drive bays to make room for everything, but there's always the option for a bottom-mounted SSD.

Oh, and I wanted one more thing out of this. I was thinking about using it as a LAN party machine. So it would have to be mobile.

This need for mobility immediately removes the large heatsink, so I was thinking of an alternate way to cool the Peltier.

I was thinking about using a lower-wattage Peltier and water cooling the hot side using the 240mm exhaust. I could put the pump and reservoir on the cooler side of the loop to prevent overheating. Then I was thinking I could have the CPU loop go through a 140mm exhaust before going to the Peltier, so that only a little bit of cooling was required to bring the water below ambient.

However, I wanted to combat one of the big issues with water cooling: low internal case airflow. So, I was thinking of using the cold side of the Peltier for a second purpose. If I had water running across the Peltier and then across an intake 240mm and then back to the Peltier, I could blow below-ambient air into my case, which would help make up for the lack of airflow. The process would also feed on itself, because the below-ambient air would make each exhaust radiator more effective.

I know I'd still need some insulation for this, but I was thinking a lower-power Peltier and below-ambient internal air temperature would remove the need for giant insulation and make a mini-ITX build possible.

I was planning on using one of AMD's APUs for this. They're easily overclocked and would remove the complexities of a video card, at least for a little while.

What do you think? Would this even remotely work?


The biggest problem I see is using a small peltier. A decent apu is about the same wattage as my CPU of about 125w I believe and over clocking both will cause more heat. You really need a 200w peltier to run it nice and cool or even a decent water cooling loop would be nice if you can modify it to fit
July 3, 2013 10:45:47 PM

Wow. I'm sorry if this is a redundant post and I'm not contributing anything here, but I had to post just to say this is absolutely amazing. I'm familiar with peltier/TEC devices and for years I have dreamed of cooling with one, and the way you implemented TEC here is just stellar. Everything looks incredible, and the way you were able to break 5GHz...I'm speechless. Sure, record holders are able to reach 8 with liquid-helium and liquid-nitrogen and all that stuff, but like you said, since this doesn't require cryogenics, you can run this on a daily basis and it stays colder than underclocked computers. Also, those record holders probably don't run stability tests. Their only goal is to just get there. You actually use 5+ GHz speeds on a daily basis with this incredible piece of equipment you built yourself. It is out of this world.

I'm actually kind of hoping you sell out to some big manufacturer that will take everything you've done and wrap it up into a piece of consumer level equipment that anyone can just strap to their CPU and be done with it. You've clearly already done all the homework.
Related resources
a c 249 K Overclocking
a c 110 à CPUs
July 4, 2013 3:49:57 AM

Textfield said:
Thank you for going through this 4Ryan6, and for posting all your findings. I wanted to ask your advice on something.

Most sane people would stick with the full or super towers for this kind of thing, and stray away from even mid towers. But I'm not sane, so I've been hypothesizing a build that I might actually attempt: a mini-ITX below-ambient machine.

Okay, before you say I'm absolutely insane, which I am, the case I would use is actually a bit large for mini-ITX. It's the Bitfenix Prodigy. I'd use it mainly because it has space for two 240mm radiators (1 intake, 1 exhuast) and one 140mm exhaust radiator. It also appears you have to rip out most of the drive bays to make room for everything, but there's always the option for a bottom-mounted SSD.

Oh, and I wanted one more thing out of this. I was thinking about using it as a LAN party machine. So it would have to be mobile.

This need for mobility immediately removes the large heatsink, so I was thinking of an alternate way to cool the Peltier.

I was thinking about using a lower-wattage Peltier and water cooling the hot side using the 240mm exhaust. I could put the pump and reservoir on the cooler side of the loop to prevent overheating. Then I was thinking I could have the CPU loop go through a 140mm exhaust before going to the Peltier, so that only a little bit of cooling was required to bring the water below ambient.

However, I wanted to combat one of the big issues with water cooling: low internal case airflow. So, I was thinking of using the cold side of the Peltier for a second purpose. If I had water running across the Peltier and then across an intake 240mm and then back to the Peltier, I could blow below-ambient air into my case, which would help make up for the lack of airflow. The process would also feed on itself, because the below-ambient air would make each exhaust radiator more effective.

I know I'd still need some insulation for this, but I was thinking a lower-power Peltier and below-ambient internal air temperature would remove the need for giant insulation and make a mini-ITX build possible.

I was planning on using one of AMD's APUs for this. They're easily overclocked and would remove the complexities of a video card, at least for a little while.

What do you think? Would this even remotely work?


Hi welcome to the TEC cooling world once the bug bites you and you find a successful solution that actually works you need to use those principals to guide your experiments.

This works for various reasons that all have to work together or it doesn't work at all.

I've already done major testing regarding what works and what doesn't from the first idea of attempting to use this type of cooling, I suggest taking the time to read all the trial and error which starts approximately 2/3rds the way though the Exploring Below Ambient Cooling thread stickied above.

I've to some extent tested using different heat sinks like the copper server style to shrink down the size of it but they did not work to the results I needed to continue down that path, each testing takes money and time to accomplish, and those resources are limited for me.

This is what you need for this cooling to work, you need a heat pipe style of air cooler capable of passive cooling to cool the hot side of the peltier, you need a flow modified water block to pick up the cold from the peltiers cold side, you need an insulated reservoir that will hold enough coolant to use a thermal mass cooling principle to store and build the cold, and a pump to push it all.

Do not tie radiators into a peltier chilling loop, the radiators will work against the cooling you could achieve.

If you have ideas that are within all that's been discovered here that does work then they'll probably work too, so if you undertake attempting them please post your results here successful or not.

a c 249 K Overclocking
a c 110 à CPUs
July 4, 2013 4:04:11 AM

jlan86 said:
Wow. I'm sorry if this is a redundant post and I'm not contributing anything here, but I had to post just to say this is absolutely amazing. I'm familiar with peltier/TEC devices and for years I have dreamed of cooling with one, and the way you implemented TEC here is just stellar. Everything looks incredible, and the way you were able to break 5GHz...I'm speechless. Sure, record holders are able to reach 8 with liquid-helium and liquid-nitrogen and all that stuff, but like you said, since this doesn't require cryogenics, you can run this on a daily basis and it stays colder than underclocked computers. Also, those record holders probably don't run stability tests. Their only goal is to just get there. You actually use 5+ GHz speeds on a daily basis with this incredible piece of equipment you built yourself. It is out of this world.

I'm actually kind of hoping you sell out to some big manufacturer that will take everything you've done and wrap it up into a piece of consumer level equipment that anyone can just strap to their CPU and be done with it. You've clearly already done all the homework.


It's nice to run across someone that gets it, and fully sees the potential of this type of cooling, I wish I had a laboratory with unlimited resources, unfortunately I don't.

There's a big difference between running a shoot for the moon overclock that usually ends up sacrificing a CPU to reach it, vs a high useable 24/7 stable overclock running cool enough for the CPU to last a long time.

I do hope this testing will open the eyes of manufacturers to realize what's possible utilizing a simple little invention called a peltier, and milking what can be gleaned from it's use.

I've had so many tell me why this is not possible and go into detail of why they say so, and the one thing they don't seem to understand is, "What about fully operational do you not understand?".

So for me it's a breath of fresh air for someone like you that gets it!

Thank You! Ryan



July 18, 2013 7:24:09 AM

Ryan I'm getting started on some TEC cooling building configurations. I had an idea while reading your posts and would be nice to know if you think it would work. It consists of two separate circuits: one for the cold-size and other for the hot-side. The main idea is to maintain a constant defined temp on each circuit even if dealing with water temp from 70-90F on the hot circuit.

Lets say the water goes like this on a single loop circuit, based on one of your first replies (illustrative temperatures):
- From the reservoir, cooled by the TEC;
- Through the CPU waterblock, raising water temp;
- Through the TEC hot side's waterblock, raising again the water temp (to 160-180F);
- Through the radiator, lowering it's temp (but raising the radiator temp like hell);
- Back to the reservoir, where it heatens the reserved water reducing the efficiency of the TEC cooling.
- The system reaches a balanced, higher than room, temperature.

The delta temp the water goes through the system is too wide. The proposal of setup is having two shorter delta temp systems, something like this:

* Cold side circuit
- From the cold-reservoir, cooled by the TEC;
- Through the CPU waterblock, raising water temp;
- Through the air cooled radiator, lowering water temp;
- Back to the reservoir at a lower temp than the first setup;

* Hot side circuit
- From the hot-reservoir to the hot-side waterblock, raising water temp;
- Through an air cooling solution or a TEC (with lower power than the CPU cooling one- the goal is to reduce the water temp by a lower delta temp than the first setup);
- Back to the reservoir at a higher temp than the first setup;

This way you have two smaller delta temp systems, with well defined and different temperatures, where you can change even the water flow capacity. The difference between the hot circuit's operating-temp and the cold circuit's operating-temp has no longer to be dealt with.

By the way I'd appreciate if you know where I can get simple and cheap watercooling parts, just for testing.
a b K Overclocking
a b à CPUs
July 18, 2013 11:20:05 AM

kauedg said:
Ryan I'm getting started on some TEC cooling building configurations. I had an idea while reading your posts and would be nice to know if you think it would work. It consists of two separate circuits: one for the cold-size and other for the hot-side. The main idea is to maintain a constant defined temp on each circuit even if dealing with water temp from 70-90F on the hot circuit.

Lets say the water goes like this on a single loop circuit, based on one of your first replies (illustrative temperatures):
- From the reservoir, cooled by the TEC;
- Through the CPU waterblock, raising water temp;
- Through the TEC hot side's waterblock, raising again the water temp (to 160-180F);
- Through the radiator, lowering it's temp (but raising the radiator temp like hell);
- Back to the reservoir, where it heatens the reserved water reducing the efficiency of the TEC cooling.
- The system reaches a balanced, higher than room, temperature.

The delta temp the water goes through the system is too wide. The proposal of setup is having two shorter delta temp systems, something like this:

* Cold side circuit
- From the cold-reservoir, cooled by the TEC;
- Through the CPU waterblock, raising water temp;
- Through the air cooled radiator, lowering water temp;
- Back to the reservoir at a lower temp than the first setup;

* Hot side circuit
- From the hot-reservoir to the hot-side waterblock, raising water temp;
- Through an air cooling solution or a TEC (with lower power than the CPU cooling one- the goal is to reduce the water temp by a lower delta temp than the first setup);
- Back to the reservoir at a higher temp than the first setup;

This way you have two smaller delta temp systems, with well defined and different temperatures, where you can change even the water flow capacity. The difference between the hot circuit's operating-temp and the cold circuit's operating-temp has no longer to be dealt with.

By the way I'd appreciate if you know where I can get simple and cheap watercooling parts, just for testing.


like he explained to me using a tec with the radiator is pointless since all the cold is going to escape there since its colder then the surrounding air just essentially being re heated again by the air going through the rad.

im rather confused on why on the cold side your using the hot sides waterblock? you should be using the cold side with no rad and then the hot side with use the hot side waterblock with radiator to cool it. i think the way you described it will actually cause your system to over heat fairly quickly.
July 18, 2013 6:12:17 PM

06yfz450ridr said:
kauedg said:
Ryan I'm getting started on some TEC cooling building configurations. I had an idea while reading your posts and would be nice to know if you think it would work. It consists of two separate circuits: one for the cold-size and other for the hot-side. The main idea is to maintain a constant defined temp on each circuit even if dealing with water temp from 70-90F on the hot circuit.

Lets say the water goes like this on a single loop circuit, based on one of your first replies (illustrative temperatures):
- From the reservoir, cooled by the TEC;
- Through the CPU waterblock, raising water temp;
- Through the TEC hot side's waterblock, raising again the water temp (to 160-180F);
- Through the radiator, lowering it's temp (but raising the radiator temp like hell);
- Back to the reservoir, where it heatens the reserved water reducing the efficiency of the TEC cooling.
- The system reaches a balanced, higher than room, temperature.

The delta temp the water goes through the system is too wide. The proposal of setup is having two shorter delta temp systems, something like this:

* Cold side circuit
- From the cold-reservoir, cooled by the TEC;
- Through the CPU waterblock, raising water temp;
- Through the air cooled radiator, lowering water temp;
- Back to the reservoir at a lower temp than the first setup;

* Hot side circuit
- From the hot-reservoir to the hot-side waterblock, raising water temp;
- Through an air cooling solution or a TEC (with lower power than the CPU cooling one- the goal is to reduce the water temp by a lower delta temp than the first setup);
- Back to the reservoir at a higher temp than the first setup;

This way you have two smaller delta temp systems, with well defined and different temperatures, where you can change even the water flow capacity. The difference between the hot circuit's operating-temp and the cold circuit's operating-temp has no longer to be dealt with.

By the way I'd appreciate if you know where I can get simple and cheap watercooling parts, just for testing.


like he explained to me using a tec with the radiator is pointless since all the cold is going to escape there since its colder then the surrounding air just essentially being re heated again by the air going through the rad.

im rather confused on why on the cold side your using the hot sides waterblock? you should be using the cold side with no rad and then the hot side with use the hot side waterblock with radiator to cool it. i think the way you described it will actually cause your system to over heat fairly quickly.


I think I was not clear... Water comes from the cold-reservoir (does not mean the water is cold, just means it belongs to the cold circuit), gets chilled by the TEC, goes to cold side waterblock. The CPU die exchanges heat with the cold size waterblock, wich is cooled by the cooled water. This raises the water's temperature (above ambient temp) and then it goes to the radiator to cool a little before going to the cold reservoir again. Maybe the rad is useless in this setup, I still don't have material to make tests.

As the TEC's cold side raises it's temperature (due to the heat exchange with the waterblock), the hot side gets hotter. The goal here is that the hot-circuit maintains the TEC's hot-side at a regular temperature (going through rad, reservoir and so on).

I will give some (very) made up temperatures here, to explain:

- Single circuit setup water flow and water temperatures:
Water reservoir --- 86f ---> TEC's cold side --- 41f ---> Waterblock on CPU --- 100f ---> Waterblock on TEC's hot side --- 160f ---> rad --- 86f ---> water reservoir

- On a double circuit setup water flow and water temperatures:
*Cold side
Water reservoir --- 86f ---> TEC's cold side --- 41f ---> Waterblock on CPU --- 100f ---> rad --- 86f ---> water reservoir

*Hot side
Water reservoir --- 95f ---> Waterblock on TEC's hot side --- 120+f ---> rad --- 95f ---> water reservoir

So, despite working on higher water temperature, it's not needed to cool it to the cold side's desired operating temperature... you don't have to get rid of this temperature difference between the two systems.
a c 249 K Overclocking
a c 110 à CPUs
July 19, 2013 1:54:01 AM

kauedg said:
06yfz450ridr said:
kauedg said:
Ryan I'm getting started on some TEC cooling building configurations. I had an idea while reading your posts and would be nice to know if you think it would work. It consists of two separate circuits: one for the cold-size and other for the hot-side. The main idea is to maintain a constant defined temp on each circuit even if dealing with water temp from 70-90F on the hot circuit.

Lets say the water goes like this on a single loop circuit, based on one of your first replies (illustrative temperatures):
- From the reservoir, cooled by the TEC;
- Through the CPU waterblock, raising water temp;
- Through the TEC hot side's waterblock, raising again the water temp (to 160-180F);
- Through the radiator, lowering it's temp (but raising the radiator temp like hell);
- Back to the reservoir, where it heatens the reserved water reducing the efficiency of the TEC cooling.
- The system reaches a balanced, higher than room, temperature.

The delta temp the water goes through the system is too wide. The proposal of setup is having two shorter delta temp systems, something like this:

* Cold side circuit
- From the cold-reservoir, cooled by the TEC;
- Through the CPU waterblock, raising water temp;
- Through the air cooled radiator, lowering water temp;
- Back to the reservoir at a lower temp than the first setup;

* Hot side circuit
- From the hot-reservoir to the hot-side waterblock, raising water temp;
- Through an air cooling solution or a TEC (with lower power than the CPU cooling one- the goal is to reduce the water temp by a lower delta temp than the first setup);
- Back to the reservoir at a higher temp than the first setup;

This way you have two smaller delta temp systems, with well defined and different temperatures, where you can change even the water flow capacity. The difference between the hot circuit's operating-temp and the cold circuit's operating-temp has no longer to be dealt with.

By the way I'd appreciate if you know where I can get simple and cheap watercooling parts, just for testing.


like he explained to me using a tec with the radiator is pointless since all the cold is going to escape there since its colder then the surrounding air just essentially being re heated again by the air going through the rad.

im rather confused on why on the cold side your using the hot sides waterblock? you should be using the cold side with no rad and then the hot side with use the hot side waterblock with radiator to cool it. i think the way you described it will actually cause your system to over heat fairly quickly.


I think I was not clear... Water comes from the cold-reservoir (does not mean the water is cold, just means it belongs to the cold circuit), gets chilled by the TEC, goes to cold side waterblock. The CPU die exchanges heat with the cold size waterblock, wich is cooled by the cooled water. This raises the water's temperature (above ambient temp) and then it goes to the radiator to cool a little before going to the cold reservoir again. Maybe the rad is useless in this setup, I still don't have material to make tests.

As the TEC's cold side raises it's temperature (due to the heat exchange with the waterblock), the hot side gets hotter. The goal here is that the hot-circuit maintains the TEC's hot-side at a regular temperature (going through rad, reservoir and so on).

I will give some (very) made up temperatures here, to explain:

- Single circuit setup water flow and water temperatures:
Water reservoir --- 86f ---> TEC's cold side --- 41f ---> Waterblock on CPU --- 100f ---> Waterblock on TEC's hot side --- 160f ---> rad --- 86f ---> water reservoir

- On a double circuit setup water flow and water temperatures:
*Cold side
Water reservoir --- 86f ---> TEC's cold side --- 41f ---> Waterblock on CPU --- 100f ---> rad --- 86f ---> water reservoir

*Hot side
Water reservoir --- 95f ---> Waterblock on TEC's hot side --- 120+f ---> rad --- 95f ---> water reservoir

So, despite working on higher water temperature, it's not needed to cool it to the cold side's desired operating temperature... you don't have to get rid of this temperature difference between the two systems.


What works is covered in this thread, what it took to get to this point is covered in the Exploring Below Ambient Water Cooling Thread linked on the first page of this thread. You can save yourself some time and money and go straight to what I've proven that works in this thread and has been in operation now over 8 months, or do your own experimentation. In most all cases what seems mentally logical regarding peltier cooling may not work at all, so good luck to your endeavors.

July 19, 2013 4:42:02 AM

4Ryan6 said:


What works is covered in this thread, what it took to get to this point is covered in the Exploring Below Ambient Water Cooling Thread linked on the first page of this thread. You can save yourself some time and money and go straight to what I've proven that works in this thread and has been in operation now over 8 months, or do your own experimentation. In most all cases what seems mentally logical regarding peltier cooling may not work at all, so good luck to your endeavors.



Thank you, I'll go through the mentioned thread again. Do you know any online store wich sells simple waterblocks (not CPU specific)?
a c 249 K Overclocking
a c 110 à CPUs
July 19, 2013 6:17:54 AM

kauedg said:


Thanks. I live in Brazil, the shipment prices are high wherever I buy... Do you think these EK kits are reliable and could be used in some TEC cooling experiments?

http://www.frozencpu.com/cat/l2/g57/c607/list/p1/EK_Pro...


You don't need a radiator for the type TEC cooling I'm running, I suggest reading the thread to see what all is involved.

A kit will be giving you stuff you don't need and short you on what you do need.

a b K Overclocking
a b à CPUs
July 19, 2013 3:20:04 PM

Also ill point out that in your idea kauedg you assume that the water temp will rise that fast in reality at most its maybe one degree change if at all from the cpu die to the next component. that water is flowing way to fast even on low settings it takes my system at least 6-8 hours to raze water to room temp. depending on the start temps. (using ice) --- errm thats low usage ,netflix,wordexel-youknow work.
a c 249 K Overclocking
a c 110 à CPUs
July 20, 2013 5:24:49 AM

thequn said:
Also ill point out that in your idea kauedg you assume that the water temp will rise that fast in reality at most its maybe one degree change if at all from the cpu die to the next component. that water is flowing way to fast even on low settings it takes my system at least 6-8 hours to raze water to room temp. depending on the start temps. (using ice)


So what is your daily ice routine?

Are you freezing blocks?

How many, how long it lasts, etc.?

a b K Overclocking
a b à CPUs
July 20, 2013 3:25:18 PM

um if just normal everyday computer work no gameing just one load of two two litter bottles of ice.

If i am gaming all day ill add in two in the morning then let the watter chill up then replace them with two more, after about 30 min then replace them half way though the day, so it depends.

Even if I forget to replace them all day this system is still better then my noctua nh-d14 in performance. -lazy days.

I have not played a game for more then 4 hours at one time yet. since doing this. i been playing Uncharted 2,3 on my ps3 over the last few weeks in my free time. because i never got around to it.
a b K Overclocking
a b à CPUs
July 20, 2013 3:35:11 PM

i have the advantage of having an ivy bridge cpu its not pulling more then 10-20 watts on average in games maybe 45-50, i think the pump puts out more heat on average. LOLOLOLOLOL
a c 249 K Overclocking
a c 110 à CPUs
July 20, 2013 7:37:28 PM

So how much water volume are you running?
a b K Overclocking
a b à CPUs
July 21, 2013 1:40:34 AM

total water 3.8 to 4.1 ? once it gets low I add in once extra bottle per run, so 3 two liter bottles until I get more distilled water. I stopped using the default 1 gallon distilled water containers because they would rupture after a week or so where the Pepsi bottled only rupture while in the freezer after a month ish.

I not planning on starting a TEC system until after the winter holiday,

i wonder how effective it would be with a 5 gallon res and maybe a bottle in the morning to get the water cold faster. :) 

Ryan do you happen to have any photos of your modified TEC Rasa block?
a c 249 K Overclocking
a c 110 à CPUs
July 21, 2013 7:09:05 AM

thequn said:
Ryan do you happen to have any photos of your modified TEC Rasa block?


No I do not, and you are not the first that has asked me the exact same question.

You have no idea how many times I have kicked myself for not taking before and after modification pictures!

At the time I was modifying it, I was pretty much at a point of desperation, I did not actually know for sure it would work, it was a test that turned out to be 100% successful.

Sometimes pictures are worth a 1,000 words, because I also modified the Swiftech Apogee XT which having the solid brass top was much more difficult to do vs the Rasa acetal top.

In words, once taken apart the acetal top has a center injection stem, that protrudes inside to about a 64th of an inch from actually touching the pin grid on the upper side of the copper base plate.



Note: Some of the Rasa blocks are different there's been pictures posted here somewhere on their differences but mine had a fine rubber O-ring seated in a grove (shaped like <>), exactly as pictured above, cut into the acetal and that rubber O-ring actually touches the copper base pin grid.

Toss the fine O-ring, and cut (using a dremel tool or similar) approximately 3/32nds off the protrusion, you will be going a minimum of at least twice the depth the of the fine O-ring seat.



Take your time and work on it until it is as returned to flat as much as possible, when that step is completed, the groove <> that was for seating the fine O-ring will be completely gone.



Then cut small grooved channels on the new face you've created, twice the width of the fine groove O-ring seating you completely removed.

Cut the grooved channels oriented as a wagon wheel spoke design at 12 - 3 - 6 - and 9 oclock as points on a clock, then an additional set of grooves at 1:30 - 4:30 - 7:30 and 10:30, these channels help direct the water to the corners of the copper baseplate.

Smooth all the edges of your work, clean it up real good of all loose particles, and reassemble the water block.

Wallah! One modified flow Rasa Water Block!

As long as you have a Dremel to do it with, it is actually a very simple process, the acetal is very easy to cut, just take your time and don't rush it.

Snagged these Rasa pictures off the net and did a quick paint referencing of the cutting details, I hope it helps in your understanding of what to do.



a b K Overclocking
a b à CPUs
July 23, 2013 11:20:04 AM

Thanks Ryan looks good, thanks for that information. What Dremel tool insert did you use ?
a c 249 K Overclocking
a c 110 à CPUs
July 24, 2013 3:20:32 AM

thequn said:
Thanks Ryan looks good, thanks for that information. What Dremel tool insert did you use ?


I used the largest ball tip cutter head to slowly whittle away the face, then switched to the drum sander attachment to smooth it back out flat, then went to the proper sized ball cutter to cut the channeling grooves into the new face.

A very small desktop vise would help tremendously in holding it steady for you but it can be done without one.

Once you make the first cut you're pretty much committed so the end results are totally up to you, you may feel more confident using other cutter heads as the Dremel kits have many cutter options to take advantage of.

Edit: I reworked the pictures above.

July 24, 2013 6:13:04 AM

You can't (or don't need to) use a radiator in this loop, because it only makes the water get to ambient temperature. What if you insulate the radiator, cooled by a peltier?
a b K Overclocking
a b à CPUs
July 24, 2013 6:32:37 AM

kauedg said:
You can't (or don't need to) use a radiator in this loop, because it only makes the water get to ambient temperature. What if you insulate the radiator, cooled by a peltier?


that really defeats the purpose of the radiator since you would have to cover the entire radiator.

so your better off not using one at all.
a c 178 K Overclocking
a c 117 à CPUs
July 24, 2013 6:34:20 AM

Using a peltier to cool the water a radiator is submerged in I suspect is just a very roundabout way of achieving what Ryan already has, peltier cooled water.
Plus, you would need to buy a pump to move water past the fins of the rad.
July 24, 2013 8:10:30 AM

06yfz450ridr said:
kauedg said:
You can't (or don't need to) use a radiator in this loop, because it only makes the water get to ambient temperature. What if you insulate the radiator, cooled by a peltier?


that really defeats the purpose of the radiator since you would have to cover the entire radiator.

so your better off not using one at all.


Ok, I don't have much experience with watercooling but that does not defeat it's purpose. The main concept of a radiator is to make a hot liquid cool to about room temp, right? Basic thermodynamics: two or more bodies with different and constant temperatures tend to reach a balancing temperature. In the most common case a radiator exchanges heat with the liquid. At the same time it exchanges heat with the ambient.

It's pointless to add a peltier (cold side) to a radiator because the rad will cool BUT also will exchange heat with the ambient and the water at the same time, balancing the system's temperature.

But, if we take the ambient heat exchange from the system (or at least reduce it a lot), by insulating the rad, the balance will happen between the isolating material, the water and the rad itself. If you attach a peltier to a totally insulated rad (no water flowing), you agree with me that the rad will remain cold, right?

I don't mean to be hard headed at all, my intention is to learn. I'm still waiting for my WC parts to arrive and begin testing.


(manofchalk I didn't mean to submerge the radiator in water)
a c 178 K Overclocking
a c 117 à CPUs
July 24, 2013 8:20:30 AM

So no heat exchange is happening between the radiator and the air since its all insulated. The rad has the cold side of a peltier stuck to it (and the heat output of the peltier is a non-issue).
So you get a cold radiator that cools the water that flows through it. Why do that over just having a water-block stuck to the cold side?

July 24, 2013 10:11:25 AM

manofchalk said:
So no heat exchange is happening between the radiator and the air since its all insulated. The rad has the cold side of a peltier stuck to it (and the heat output of the peltier is a non-issue).
So you get a cold radiator that cools the water that flows through it. Why do that over just having a water-block stuck to the cold side?



The circuit in a waterblock is much smaller than the rad's circuit. Assuming they are in the same system, hence same waterflow, the water stays much longer exchanging heat with the rad's fins than exchanging heat with the waterblock. It's very hard to explain the overall idea because english is not my native language.

I think that there are two goals when setting up a system like this: chilling the die as much as possible or chilling it to a constant temperature.

As soon as I receive all my orders I will run extensive tests using an IR thermometer and lots of different setups. I will post the results here, then.
a c 249 K Overclocking
a c 110 à CPUs
July 24, 2013 11:59:17 AM

kauedg said:
You can't (or don't need to) use a radiator in this loop, because it only makes the water get to ambient temperature. What if you insulate the radiator, cooled by a peltier?


kauedg said:
Ok, I don't have much experience with watercooling but that does not defeat it's purpose. The main concept of a radiator is to make a hot liquid cool to about room temp, right? Basic thermodynamics: two or more bodies with different and constant temperatures tend to reach a balancing temperature. In the most common case a radiator exchanges heat with the liquid. At the same time it exchanges heat with the ambient.

It's pointless to add a peltier (cold side) to a radiator because the rad will cool BUT also will exchange heat with the ambient and the water at the same time, balancing the system's temperature.

But, if we take the ambient heat exchange from the system (or at least reduce it a lot), by insulating the rad, the balance will happen between the isolating material, the water and the rad itself. If you attach a peltier to a totally insulated rad (no water flowing), you agree with me that the rad will remain cold, right?

I don't mean to be hard headed at all, my intention is to learn. I'm still waiting for my WC parts to arrive and begin testing.


(manofchalk I didn't mean to submerge the radiator in water)


kauedg said:
The circuit in a waterblock is much smaller than the rad's circuit. Assuming they are in the same system, hence same waterflow, the water stays much longer exchanging heat with the rad's fins than exchanging heat with the waterblock. It's very hard to explain the overall idea because english is not my native language.

I think that there are two goals when setting up a system like this: chilling the die as much as possible or chilling it to a constant temperature.

As soon as I receive all my orders I will run extensive tests using an IR thermometer and lots of different setups. I will post the results here, then.


So your idea is to use an insulated radiator as the cold exchanger instead of a water block?

So you won't be pulling any ambient air through the radiator fins to neutralize the cold, just using the radiator with the fins insulated to capture the cold from the peltier.

Interesting Idea!

I do suggest you go take a look at overclock.net and see if anyone else has attempted that, if anyone has, you can expect the same results.

Most of their peltier/radiator attempts have failed miserably, but they were still running fans on the radiator, neutralizing the cold from the peltier, your idea may actually work, IDK?

But if it does you may be stumbling onto another peltier cooling solution?

The cooling solution using modified flow water blocks as cold exchangers that is covered in this thread works, 100% guaranteed, and has been in operation for over 8 months.

If you decide to proceed please keep us informed with your progress.

I hope it works for you!

This cooling that this thread covers started from an idea. :) 



a b K Overclocking
a b à CPUs
July 24, 2013 1:08:45 PM

I would imagine it would have the same effect as a slush box. another thing is how much would it take to freeze the lines in side a rad? sounds a bit easier to freeze it i would think? and then what size to use anything larger then 120mm rad would be pointless right? because the size of the pelter? or would the extra cold leakage help on larger ones.
July 24, 2013 1:26:25 PM

4Ryan6 said:
kauedg said:
You can't (or don't need to) use a radiator in this loop, because it only makes the water get to ambient temperature. What if you insulate the radiator, cooled by a peltier?


kauedg said:
Ok, I don't have much experience with watercooling but that does not defeat it's purpose. The main concept of a radiator is to make a hot liquid cool to about room temp, right? Basic thermodynamics: two or more bodies with different and constant temperatures tend to reach a balancing temperature. In the most common case a radiator exchanges heat with the liquid. At the same time it exchanges heat with the ambient.

It's pointless to add a peltier (cold side) to a radiator because the rad will cool BUT also will exchange heat with the ambient and the water at the same time, balancing the system's temperature.

But, if we take the ambient heat exchange from the system (or at least reduce it a lot), by insulating the rad, the balance will happen between the isolating material, the water and the rad itself. If you attach a peltier to a totally insulated rad (no water flowing), you agree with me that the rad will remain cold, right?

I don't mean to be hard headed at all, my intention is to learn. I'm still waiting for my WC parts to arrive and begin testing.


(manofchalk I didn't mean to submerge the radiator in water)


kauedg said:
The circuit in a waterblock is much smaller than the rad's circuit. Assuming they are in the same system, hence same waterflow, the water stays much longer exchanging heat with the rad's fins than exchanging heat with the waterblock. It's very hard to explain the overall idea because english is not my native language.

I think that there are two goals when setting up a system like this: chilling the die as much as possible or chilling it to a constant temperature.

As soon as I receive all my orders I will run extensive tests using an IR thermometer and lots of different setups. I will post the results here, then.


So your idea is to use an insulated radiator as the cold exchanger instead of a water block?

So you won't be pulling any ambient air through the radiator fins to neutralize the cold, just using the radiator with the fins insulated to capture the cold from the peltier.

Interesting Idea!

I do suggest you go take a look at overclock.net and see if anyone else has attempted that, if anyone has, you can expect the same results.

Most of their peltier/radiator attempts have failed miserably, but they were still running fans on the radiator, neutralizing the cold from the peltier, your idea may actually work, IDK?

But if it does you may be stumbling onto another peltier cooling solution?

The cooling solution using modified flow water blocks as cold exchangers that is covered in this thread works, 100% guaranteed, and has been in operation for over 8 months.

If you decide to proceed please keep us informed with your progress.

I hope it works for you!

This cooling that this thread covers started from an idea. :) 



Thanks for the forum link Ryan. At first I thought you were being sarcastic on the first lines ("So, you think you're that good?" haha). Even if I end up "reinventing the wheel" I will begin my project from scratch, taking notes on the separate part's performance and then putting the best ones together.

I will have in hand:
1 reservoir (0.8L with temp meter)
1 water pump 480lph/12v/950ma
2 120mm copper radiator
2 120mm Cooler Master Fan1
1 stock amd air cooler
3 simple copper waterblocks
IR thermomether
Diverse electronic parts

1 12V/6A TEC
2 12V/6.4A TEC
1 PSU 240v with 3x 12v/20A outs
a b K Overclocking
a b à CPUs
July 24, 2013 5:19:08 PM

kauedg said:


Thanks for the forum link Ryan. At first I thought you were being sarcastic on the first lines ("So, you think you're that good?" haha). Even if I end up "reinventing the wheel" I will begin my project from scratch, taking notes on the separate part's performance and then putting the best ones together.

I will have in hand:
1 reservoir (0.8L with temp meter)
1 water pump 480lph/12v/950ma
2 120mm copper radiator
2 120mm Cooler Master Fan1
1 stock amd air cooler
3 simple copper waterblocks
IR thermomether
Diverse electronic parts

1 12V/6A TEC
2 12V/6.4A TEC
1 PSU 240v with 3x 12v/20A outs


Good luck man please take photos and post them here so we can obverse and comment on how awesome it is.

@ 4ryan
wow I just read Your TEC post there on OCN and they kinda just took a big steaming pile on your topic. there loss
one more question you point out that your second tec block is runing a brass top? What effect dose it add to the line because brass often impure and nickle is often used to plate brass for less corrosive at room temp.


a c 249 K Overclocking
a c 110 à CPUs
July 25, 2013 2:42:40 AM

kauedg said:
Thanks for the forum link Ryan. At first I thought you were being sarcastic on the first lines ("So, you think you're that good?" haha). Even if I end up "reinventing the wheel" I will begin my project from scratch, taking notes on the separate part's performance and then putting the best ones together.

I will have in hand:
1 reservoir (0.8L with temp meter)
1 water pump 480lph/12v/950ma
2 120mm copper radiator
2 120mm Cooler Master Fan1
1 stock amd air cooler
3 simple copper waterblocks
IR thermomether
Diverse electronic parts

1 12V/6A TEC
2 12V/6.4A TEC
1 PSU 240v with 3x 12v/20A outs


I went through similar testing when I first entertained the idea of using a peltier to cool with, sometimes the only way to find out what works is to try it, if it works you've got something, if not, back to the drawing board and start over.



a c 249 K Overclocking
a c 110 à CPUs
July 25, 2013 2:52:04 AM

thequn said:
@ 4ryan
wow I just read Your TEC post there on OCN and they kinda just took a big steaming pile on your topic. there loss
one more question you point out that your second tec block is runing a brass top? What effect dose it add to the line because brass often impure and nickle is often used to plate brass for less corrosive at room temp.


The OCN guys are stuck on direct peltier mounting and to them anything less is irrelevant, they were always trying to explain to me why what I was doing wouldn't work.

I guess they missed the part that I was sharing a fully operational cooling system not an idea or concept.

The brass top water block has produced zero problems and it is not plated inside.

July 25, 2013 5:18:02 AM

4Ryan6 said:

The OCN guys are stuck on direct peltier mounting and to them anything less is irrelevant, they were always trying to explain to me why what I was doing wouldn't work.

I guess they missed the part that I was sharing a fully operational cooling system not an idea or concept.

The brass top water block has produced zero problems and it is not plated inside.



What's the difference between running an OC'ed CPU at -5C or 10C, at the same CPU freq? It seems to me that there's a point where it gets useless to go below 0C if you are not going to raise the freq.

TEC systems are by itself almost not worth setting up, as far as I've read. You can't expect it to work like a chiller, phase-change system, but it has to be somewhat better than water cooling. This is the goal I'll be persuing in my project: keeping cooler than WC, without going below 0C, but making the exceeding power and heating worth. Maybe then, later, try to go to lower temperatures giving room to more overclocking. This is how technology/science advance.
a c 249 K Overclocking
a c 110 à CPUs
July 25, 2013 5:46:55 AM

kauedg said:
4Ryan6 said:

The OCN guys are stuck on direct peltier mounting and to them anything less is irrelevant, they were always trying to explain to me why what I was doing wouldn't work.

I guess they missed the part that I was sharing a fully operational cooling system not an idea or concept.

The brass top water block has produced zero problems and it is not plated inside.



What's the difference between running an OC'ed CPU at -5C or 10C, at the same CPU freq? It seems to me that there's a point where it gets useless to go below 0C if you are not going to raise the freq.

TEC systems are by itself almost not worth setting up, as far as I've read. You can't expect it to work like a chiller, phase-change system, but it has to be somewhat better than water cooling. This is the goal I'll be persuing in my project: keeping cooler than WC, without going below 0C, but making the exceeding power and heating worth. Maybe then, later, try to go to lower temperatures giving room to more overclocking. This is how technology/science advance.


Your goal is exactly the same as mine!

I was simply after a cooling solution that was independent of ambient room temperature cooling, as is Air cooling, and Water cooling using the traditional radiator loop, (which is still at ambient mercy), and I was not after going down to condensation, or ice producing levels of cooling.

That's what I presently have, even though it is capable of going down further than I actually run it, as I choose to stay above the condensation point, which still allows me to run water temperatures at 13c below ambient without condensation.

July 25, 2013 6:07:00 AM

4Ryan6 said:


Your goal is exactly the same as mine!

I was simply after a cooling solution that was independent of ambient room temperature cooling, as is Air cooling, and Water cooling using the traditional radiator loop, (which is still at ambient mercy), and I was not after going down to condensation, or ice producing levels of cooling.

That's what I presently have, even though it is capable of going down further than I actually run it, as I choose to stay above the condensation point, which still allows me to run water temperatures at 13c below ambient without condensation.



Good to know I'm on the same path. I'm also focusing on building a peltier module that work as an "add-on" to a watercooler loop. Then it can be automatically turned on/off when a certain temperature is reached, to save electricity. But that comes after finding the optimal setup.
July 25, 2013 1:13:37 PM

The more I read, the more question pop in my head... How do I control a TEC's temperature, by voltage or current? How can I set a pot to control this?
a c 249 K Overclocking
a c 110 à CPUs
July 25, 2013 4:25:33 PM

This is only relating to my cooling setup!

Temperature is controlled by fan speed on the heat sink cooling the hot side of the peltier.

Voltage is constant feed on the primary TEC assembly, and the secondary TEC assembly is toggle switch controlled when needed.
a b K Overclocking
a b à CPUs
July 25, 2013 5:16:15 PM

kauedg said:


TEC systems are by itself almost not worth setting up, as far as I've read. You can't expect it to work like a chiller, phase-change system, but it has to be somewhat better than water cooling. This is the goal I'll be persuing in my project: keeping cooler than WC, without going below 0C, but making the exceeding power and heating worth. Maybe then, later, try to go to lower temperatures giving room to more overclocking. This is how technology/science advance.


This is a good attitude for the project. sadly not more people think like this, always better to start small then work your way up. then again most people consider custom water-cooling a large endeavor, LOL good luck man, ill be a few months behind you with a TEC, Currently i am loving my ice system, and i am not ready to destroy it just yet.
July 25, 2013 6:21:11 PM

thequn said:
kauedg said:


TEC systems are by itself almost not worth setting up, as far as I've read. You can't expect it to work like a chiller, phase-change system, but it has to be somewhat better than water cooling. This is the goal I'll be persuing in my project: keeping cooler than WC, without going below 0C, but making the exceeding power and heating worth. Maybe then, later, try to go to lower temperatures giving room to more overclocking. This is how technology/science advance.


This is a good attitude for the project. sadly not more people think like this, always better to start small then work your way up. then again most people consider custom water-cooling a large endeavor, LOL good luck man, ill be a few months behind you with a TEC, Currently i am loving my ice system, and i am not ready to destroy it just yet.


Thanks a lot. Going through the nitty-gritty of stuff makes us think of new ways to do it. Even failed theories may lead to successful ones. Right now I'm setting up a spreadsheet for calculations while my equips aren't delivered. But I will be documenting it all here as I progress.
July 27, 2013 6:29:53 AM

Hi Ryan,

I've been working on a system of my own regarding the Peltier TEC idea, just found yours on the net. Looks complicated! I was wondering if you'd take a look at what I'm proposing and tell me your thoughts? I've put a thread on overclockers, as linked here:

http://forums.overclockers.co.uk/showthread.php?p=24673...

My theory is that the high temperatures of the hot side of the TEC can be dispersed through the radiator externally by a fan, down to ambient temp, having collected the heat from the CPU as well. Meanwhile, the return is passed through a heat exchanger in the super chilled section of a sealed tank. Take a look.

I'm interested in how you've wired yours up, and how you control it all. These TEC plates seem simple enough, and I read your warnings on amp draw, so I'll bear that in mind when I install. My MOBO came with some clever programme which controls fan speeds, I was wondering if I could plug into this and convince it to regulate voltages of the TEC according to the CPU temps, as well as the pump flow rates. Theres probably some ratio of control, probably graphable, which can be used here, speeding up the pump while reducing the TEC voltage, and viceversa, so as to prevent overheating/cooling the tanks?
a b K Overclocking
a b à CPUs
July 27, 2013 11:56:29 AM

A. No your mother board cannot control the TEC fan controls are a max of 5 volts.
B. aluminum is not good for water-cooling loops due to corrosive nature of it. best if its all copper. unless you gonna use that anti freeze inside the loop as well and in the chamber.

The big issue is the one loops, I see two possibility ether its wont work at all or it will work so well that it wont work at all.
If it is gonna work that rad must be able to disperses the heat on one cycle on entering the radiator. otherwise its will be to hot once entering the cool Coil
how large are theses bulk heads? how large is the Rad? how long and with is the cool coil?


Also this is a really cool idea,
a b K Overclocking
a b à CPUs
July 27, 2013 1:58:38 PM

fallow up question how are you goina take the cold and add it to the cold bulkhead? and the same with the top??
a b K Overclocking
a b à CPUs
July 27, 2013 2:15:00 PM

Hey 4ryan6, can you help me get my peltier to run off my 12v rail? It seems my power supply has some safety feature causing it to shut down as soon as I try and power the peltier off the 12v rail. The peltier will run just fine off the 5v red wire but anytime I try and use 12v yellow wires it will come on for a second then power off and need to be reset.

I've tried the PCI-e 6+2pin wires, the 4+4pin motherboard wires, and the yellow ones from the 20+4pin all with the same results.

Here is the power supply which only runs the peltier and the two fans on the 212+:
http://www.newegg.com/Product/Product.aspx?Item=N82E168...

Here are the peltier specs:
http://www.ebay.com/itm/New-1-PCS-50mm-TEC1-12726-TEC-T...

Any ideas are appreciated!

Feel free to PM me if you don't want to muck up this thread.
a c 249 K Overclocking
a c 110 à CPUs
July 28, 2013 4:15:49 AM

mlcaouette said:
Hey 4ryan6, can you help me get my peltier to run off my 12v rail? It seems my power supply has some safety feature causing it to shut down as soon as I try and power the peltier off the 12v rail. The peltier will run just fine off the 5v red wire but anytime I try and use 12v yellow wires it will come on for a second then power off and need to be reset.

I've tried the PCI-e 6+2pin wires, the 4+4pin motherboard wires, and the yellow ones from the 20+4pin all with the same results.

Here is the power supply which only runs the peltier and the two fans on the 212+:
http://www.newegg.com/Product/Product.aspx?Item=N82E168...

Here are the peltier specs:
http://www.ebay.com/itm/New-1-PCS-50mm-TEC1-12726-TEC-T...

Any ideas are appreciated!

Feel free to PM me if you don't want to muck up this thread.


Do not run your cooling fans for the 212+ off the same power supply powering the peltier, that's bad, the power supply may sense it's an overload with the molex cables connected, run the power for the cooling fans from the main computer by extending the 4 pin molex.

The peltier should be powered by a stand alone power supply, period.

The peltier is a constant load according to your P/S and Peltier specs, the peltier is using half the P/Ss available amperage and 2/3rds the P/Ss available wattage, that is a 400w peltier you have.

Minus the mumbo jumbo talk from the sidelines that you're only supplying 12v, so you're not using the spec load, but unless you test the actual draw load you really don't know, the peltier is made in China and it may be exactly as rated or not.

I am getting my power from the 8 pin M/B connector, = (4 yellow 12v positive and 4 black 12v negative 14 gauge wires).

I'm using 2 of the yellow 14g 12v positive wires soldered to a single 12g lead wire from the P/S to the peltier, and 2 of the black 14g 12v negative soldered to the negative side of the 12g lead wire.

Using a 12g lead wire to handle the current between the P/S and Peltier keeps the wiring from overheating.

Hope this helps.










a c 249 K Overclocking
a c 110 à CPUs
July 28, 2013 4:23:54 AM

threefingeredjack said:
Hi Ryan,

I've been working on a system of my own regarding the Peltier TEC idea, just found yours on the net. Looks complicated! I was wondering if you'd take a look at what I'm proposing and tell me your thoughts? I've put a thread on overclockers, as linked here:

http://forums.overclockers.co.uk/showthread.php?p=24673...

My theory is that the high temperatures of the hot side of the TEC can be dispersed through the radiator externally by a fan, down to ambient temp, having collected the heat from the CPU as well. Meanwhile, the return is passed through a heat exchanger in the super chilled section of a sealed tank. Take a look.

I'm interested in how you've wired yours up, and how you control it all. These TEC plates seem simple enough, and I read your warnings on amp draw, so I'll bear that in mind when I install. My MOBO came with some clever programme which controls fan speeds, I was wondering if I could plug into this and convince it to regulate voltages of the TEC according to the CPU temps, as well as the pump flow rates. Theres probably some ratio of control, probably graphable, which can be used here, speeding up the pump while reducing the TEC voltage, and viceversa, so as to prevent overheating/cooling the tanks?


Thegun has provided some excellent answers and some good questions for you.

So where are we at now?

a b K Overclocking
a b à CPUs
July 28, 2013 11:59:25 AM

Thanks for the suggestions, as soon as I have some time to move some wiring around I will give that a try and I'll let you now if it is successful.
a c 249 K Overclocking
a c 110 à CPUs
July 29, 2013 3:06:18 AM

Your sig link to Coolimus Prime isn't working?
July 29, 2013 6:02:47 AM

thequn said:
fallow up question how are you goina take the cold and add it to the cold bulkhead? and the same with the top??



Not sure what you mean by "take the cold and add it to the cold bulkhead", maybe the image wasn't clear. Sorry, the hot and cold chambers are part of a singe theoretical container, which is separated into a top and bottom chamber by a single central bulkhead, this bulkhead will house the TEC plate and exchange heat between the top and bottom chambers. Heat from the lower chamber to be drawn out and moved upwards into the upper chamber. The bulkhead will allow contact between the fluid in the lower part and the cold side of the TEC, and fluid in the upper part and hot side of the TEC.

Yes I figured on not using water inside the loop as per corrosion and danger to internal components of the PC. I was thinking antifreeze throughout or possibly mineral oil, or some alternative, at this point it's a hypothetical liquid with good heat exchange properties. I'll find out whats it called later :p 

"It'll work so well that it won't work at all"? I don't follow, I'm afraid.

The idea with the single loop was to round off the issue Ryan seemed to refer to regarding the difficulty of controlling the temperatures on either side of the TEC. Ie, when the hot plate is cooled down too far the cold plate doesn't cool as much or something to that effect. In my head that looks like a sharp negative bell curve, which I imagine if you're also pumping varying degrees of hot through it from the CPU, the TEC is going to have to be all over the place trying to keep on top of it. SO! I figured, if you run the heat from the CPU and TEC together through the same radiator/air cooled heat sink, getting down to ambient as best as possible, (so not, no exact size of radiator, just the biggest I can afford, or a number in sequence) then the resultant liquid temperature leading back into the cold chamber will always be more or less the same temperature, whatever ambient is. Ergo, the TEC can be tuned to run at a more specific voltage, as heat from the CPU will be lost in the fluid of the hot chamber. As long as you can cool the hot chamber, the CPU temp doesn't matter.

Also, chamber volumes, not sure, there is likely some math can be applied here to give a figure, however, lets just ball park go with whatever Ryan used in his cold tank reservoir thing? Seemed to do the job :)  I'm not an expert hence why I am asking you guys, it seems like it would work in my head, provided the heat from the radiator can be dispersed. I have an old chimney hole in the wall I was going to put the radiator into. Big, big radiator. Lots of fans. Dedicated PSU all that jazz.

What else. Yes, right, pumps, in the wrong place. Should be between the CPU and hot tank. Ignore that. And... the shape of the hot and cold chambers is hypothetical too, the one I was debating building doesn't have a bulkhead but is instead two separate chambers in a sort of hourglass arrangement so there is no contact between the hot and cold sides, they also have internal insulation against the walls. But, that over complicates things and I don't have CAD access right now.

so that lot in mind, what do you think? Ryan is the expert it seems as he's actually built one of these devices which works, so, Ryan, do you think that your system could be enhanced using a single loop system such as this or is something I don't understand going to occur resulting in a resonant cascade scenario transporting me to Xen with no hope of return but to find the Gman?

    • Previous
    • 2 / 6
    • 3
    • 4
    • 5
    • 6
    • More pages
    • Next
    • Newest
      • 1
      • 2 / 6
      • 3
      • 4
      • 5
      • 6
!