TEC/Peltier CPU Chilled Water Cooling

[oroboros86]

I'm just determined to get everything to fit into a mid tower case.

WELLLL, back to the drawing board, I'm trying to figure out all the calculations for both hot and cold side, figuring the case out, voltage/temperature regulating the TEC voltage, and radiator setup for the case.
I'll be back with pictures and schematics!

 

[oroboros86]

I'm a retard, reading on further on the other thread, I'm now understanding why the radiator cooling the hotside isn't going to work. Ugh, Stubbornness! Back to the drawing board again...

 

[4Ryan6]

I'm just determined to get everything to fit into a mid tower case.

WELLLL, back to the drawing board, I'm trying to figure out all the calculations for both hot and cold side, figuring the case out, voltage/temperature regulating the TEC voltage, and radiator setup for the case.
I'll be back with pictures and schematics!

Calculations may not be as satisfying and accurate regarding peltiers because in some respects they defy logic.

Calculation wise I had many at overclock.net informing me that a heat pipe air cooler could not cool the peltier hot side according to their calculations.

It boiled down to, "What about 100% operational do you not understand!", for them to finally back off the, it won't work because of our calculations.

I did do some experimentation using smaller more compact air coolers like the solid copper server style coolers but they retain too much heat and that is bad, and use high speed 60mm cooling fans and that is bad as well, there are very good reasons servers are usually located in sealed rooms as they're extremely LOUD and high pitched irritating!

My next test was going to be low profile heat pipe air coolers like the AMD variants but I did not get to that testing yet, so they may work and save vertical mounting space?

Something like this maybe?

Or one similar? IDK?

I'll eventually test in that area, but I ran short of testing funds as other things were of higher priority, and since the present solution works just fine there wasn't much need for myself.

Just some possibilities for your consideration.

Edit Note: Peltier clamping pressure is very important and also needs to be in your planning, you will not be spring pressure mounting a peltier, as most clamping setups allow for, they will have to be direct screw mounted with about 100psi clamping pressure, something else for you to think about.

Look at the pictures in the beginning of this thread to see how mine are clamped together, clamping pressure is peltier performance critical!

 

[oroboros86]

Reading through the "exploring below ambient" thread, the statement that made realize that "radiators won't work (besides the fact that you directly told me) was that the pelt would have to remain hot, in order for it to stay cold. Looks like I'll have air cool it in the case with a lot more fan power pushing ask that heat out.

What would your opinion be in regards to voltage controlling the TEC?

I'm thinking it would be voltage/temperature controlled using a relay and a fan controller with a temperature sensor it.

https://docs.google.com/file/d/0B02MVpxYDgIOTFpwZXh0ZGNwTXc/edit?usp=docslist_api

Something like that.

 

[4Ryan6]

I'm using voltage straight from a PC single 12v rail power supply,and controlling the peltier with the speed of the cooling fan on the heatsink, no voltage regulation is needed.

The peltier is a 15.6v DC peltier so it is receiving less voltage with 12v being supplied to it, and that means the peltier hot side can be controoled with the cooling fan speed on the heat pipe air cooler.

It really couldn't get much simpler.

I am presently running a 10.4c water temperature in my insulated reservoir and my ambient room temperature is 24.4c, so I'm running a water temperature with zero condensation 14c below ambient.

My cpu idle temperature is 15.5c at a 5ghz 3770K Not Delidded CPU overclock, and just for the heck of it ran Intel burn test for a load temperature of 66c.

What's your idle and load temperatures at a 5ghz overclock?

 

[oroboros86]

Funny thing really, I'd have to say my current overclock is 0 Ghz because I currently don't have a PC, but I got my brother to send me his old amd x6 1050t, so I'll be doing all my testing on that before I upgrade the cpu, gpu and mobo.

I like said in my first post, it's all theory until I actually get it together.
Once the PC and peltiers arrive, I'll start to do what I can, and start planning on how I'm going to get everything to fit in the case (and running well.)

 

[4ntikythera]

I found this thread and your other while researching if refrigerated PC cooling was possible. I read first and last several pages of the other thread and this thread in its entirety. The work you have done is great. I'm not even at the level of water cooling myself but have begun to overclock some of my old PCs with their existing air cooling. So I was very interested in what is possible.

Some questions came to mind;

Did you consider using waterblocks not specifically made for CPUs? Some of the GPU or even HDD waterblock pictures look like they may allow the greater water flow you need without the modifications your CPU blocks required.

Does having the waterblocks in series not make it more likely for the second one to freeze? The water entering it already having been cooled by the first.

You wrote you tested several size reservoirs, different fan speeds, waterblocks and heatsinks. Was it a matter of tuning (for lack of a better term) the system until you reached desired performance? Aiming for just above dew point and experimentally testing things until you reached it. How much difference would running the reservoir half full make for example? Would it be more likely to freeze? How different would the idle temps be?

@threefingeredjack - Not sure if you'll ever come back to read this but... In the diagram for your twin reservior system you had the coolant loop running out of the radiator and into the cool tank closest to the TEC. Assuming everything else worked as intended I beleive you should consider exiting the radiator and entering the aluminum coil at the bottom and exiting closest to the TEC (http://en.wikipedia.org/wiki/Countercurrent_exchange).

edit: The pictures of the waterblock before you altered it show a rubber grommet where the water is closest to the CPU die. I don't understand the design principle for that. Any idea why? Surely more flow is better.

 

[4Ryan6]

Thanks for the interest and questions and Welcome to the thread!

Did you consider using waterblocks not specifically made for CPUs? Some of the GPU or even HDD waterblock pictures look like they may allow the greater water flow you need without the modifications your CPU blocks required.

As for myself No. Some suggested using GPU and Memory coolers but I got no feedback as to whether they actually tried them and were successful or not, I did use a dedicated channeled copper peltier water block at one time, but it was designed to cool the peltier not collect cold from it.

Logically it would seem it wouldn't make any difference if the block was designed to cool it should be able to collect the cool as well, but the problems arise in the thickness of the copper and it's ability to retain heat from the thickness.

The modified CPU water block base plate is only close to 1/8th of an inch thick so it transfers the cold to the water fast, that's why they work so well, the water is in and out quick, it picks up the cold and transfers it to the insulated reservoir.

After the fact I have considered blocks like GPU, however there is a problem of coming up with an adequate clamping pressure, as the clamping together of the CPU water block and the heat pipe air cooler is a perfect match, it is very important to get a good amount of clamping pressure close to 100psi (estimated), to get the full use from the peltier.

Does having the waterblocks in series not make it more likely for the second one to freeze? The water entering it already having been cooled by the first.

It would logically seem that way, but the modifications made to the flow so far have not froze up, keep in mind they are countering the CPUs constantly produced heat, so if the CPU was not involved they would freeze solid, so only the fact they produce more cold than the CPUs produced heat, allows for the cold produced past the CPUs heat temperature to be stored in the insulated reservoir.

You wrote you tested several size reservoirs, different fan speeds, waterblocks and heatsinks. Was it a matter of tuning (for lack of a better term) the system until you reached desired performance? Aiming for just above dew point and experimentally testing things until you reached it. How much difference would running the reservoir half full make for example? Would it be more likely to freeze? How different would the idle temps be?

Tuning is the right word! It all came down to a balance of reservoir volume, cooling fan speed, and pump flow rate, but a lot of testing took place experimenting seeking the right balance to be able to use it all together to reach the results achieved.

Even though the end of the other thread covers a lot of discoveries, it does not include the fine tuning and testing that I did to end up with the recommended cooling results this thread details, and basically guarantees, that if this setup is duplicated you'll get the exact same results as ArthurH has proven with his own setup.

Though his setup loop configuration is different than mine the components of the loop are the same and his results are almost identical to mine, so that proves no matter how you align or mount the individual units, the end result cooling performance is the same results.

edit: The pictures of the waterblock before you altered it show a rubber grommet where the water is closest to the CPU die. I don't understand the design principle for that. Any idea why? Surely more flow is better.

That design is to pinpoint and force the water flow directly over the CPU core area, it is basically a way to force the cold pickup from the center even with high or low flow rate and pressure, it is following in their endeavor to end up with a one size fits all solution.

Earlier water block designs either required a low flow rate or high flow rate, and the internal design of the water block accounted for the needs, but today you don't really see high and low flow rate water blocks anymore because the CPU heat transfer point is now accomplished with super fine copper pins and pin point flow concentration.

The old flow rate water blocks were great in that they would hardly ever get clogged or gunked up, however too high a water flow on a low flow water block seriously lost it's cooling efficiency, the water was not in the block long enough to transfer the heat to the coolant and they were very inefficient.

The new pin point design of water blocks today work great with low or high flow rate, but tend to get fouled with almost any type of trash, dust, floating in the air during the build, or not fully flushed from the radiator, that got into the water loop.

So that's the idea behind the micro O ring, it forces the water to compress into the pickup pins, effectively better transferring the CPU heat to the water.

The new pin point water block design is very effective in a traditional radiator loop, but has to be flow modified to pickup cold as I am using, or it would freeze up. (Unless of course you used some type of anti-freeze coolant additive and most don't like acrylic, and are caustic to computer components if a leak occurs, that's why I use Steam distilled water and stay above the condensation problems.)

 

[crisptofuring]

OK, complete change of tone here. I love the idea of increasing overclocking capability with below ambient water temperatures, been wondering for a while how to take the next "extreme" step in watercooling because radiators can only take you so far. However, aesthetics are really important to me. is it possible at all to avoid using an insulated res and tubing? I realize that's it would make the peltier assembly more inefficient but it would certainly look much better. Is it possible to fit the entire setup in a large case such as the obsidian 900d?

 

[4Ryan6]

OK, complete change of tone here. I love the idea of increasing overclocking capability with below ambient water temperatures, been wondering for a while how to take the next "extreme" step in watercooling because radiators can only take you so far. However, aesthetics are really important to me. is it possible at all to avoid using an insulated res and tubing? I realize that's it would make the peltier assembly more inefficient but it would certainly look much better. Is it possible to fit the entire setup in a large case such as the obsidian 900d?

No, this cooling approach will not work without an insulated reservoir!

Why?

Because the peltier has to have time to reach it's cooling potential or it gets stalled out by the CPUs heat output.

The insulated reservoir serves two purposes, it is a heat buffer to allow cold buildup, and it is insulated storage for the chilled water.

If it was not insulated at all a major amount of cold would be lost to ambient room temperature.

The insulation holds in the cold allowing the water temperatures to go much lower than without the insulation.

And No, it will not all fit in an Corsair Obsidian 900D, it will however all fit in a Mountain Mods Extended Ascension CYO!

As that case is on my wish list!

 

[4ntikythera]

After seeing what you had done I did some more reading and came across this http://www.aliexpress.com/store/product/KS112-the-semiconductor-refrigeration-water-cooled-air-conditioning-Movement-water-cooled-air-conditioners-360W/518335_834068613.html

it seems to be designed to blow cold air out the centre and cool the hot sides with water. however I'm thinking if you reversed the current to the peltier then you would be blowing hot air and be able to cool the water flowing through both sides.

the specs say its capable of 360w so I'm wondering if it might be worth experimenting with. I'm interested in what you think.

 

[oroboros86]

To my understanding, it could work, mind you that's 4 TEC's in there. the heatsink won't be able to cool the TEC enough to maintain it at a respectable level to keep the hotside stalling the cold side.
Someone correct me if I'm wrong.

 

[4Ryan6]

After seeing what you had done I did some more reading and came across this http://www.aliexpress.com/store/product/KS112-the-semiconductor-refrigeration-water-cooled-air-conditioning-Movement-water-cooled-air-conditioners-360W/518335_834068613.html

it seems to be designed to blow cold air out the centre and cool the hot sides with water. however I'm thinking if you reversed the current to the peltier then you would be blowing hot air and be able to cool the water flowing through both sides.

the specs say its capable of 360w so I'm wondering if it might be worth experimenting with. I'm interested in what you think.

Interesting little contraption, I guarantee that fan will be a screamer!

As to whether you could reverse engineer it?

Those are channeled aluminum water blocks and if you mix aluminum and copper in a loop you'll be forced to add some serious additives to stop the reactions of dissimilar metals, but that may be your intentions anyway?

I've seen a lot of tempting items out of that website, but never had the guts to buy from them?

If you do?

Then let us know how it goes?

 

[4Ryan6]

To my understanding, it could work, mind you that's 4 TEC's in there. the heatsink won't be able to cool the TEC enough to maintain it at a respectable level to keep the hotside stalling the cold side.
Someone correct me if I'm wrong.

They are claiming 360w so it must be 4 90w peltiers I tried something similar using a 240w peltier with a solid copper finned server heat sink and the copper heat sink retained heat too long and the high speed cooling fan was intolerably loud!







I just don't know how the aluminum heat sink would work, it may work better than the copper?

To keep the hot or cold side from stalling will take trial and error experimentation, there's really no Yes or No?

But more like I don't know, he'll just have to take a chance but with S&H it will be an over $100.00 lost if it doesn't work, but the individual parts themselves may be worth it? IDK

 

[4Ryan6]

Timeline Update:

First began Peltier/TEC water chilling experimentation July 13th 2012 and not all experimentation was successful, which included a lot of learned the hard way times and various rebuilds and project dead ends leading to.

The present water chilling cooling concept that was adopted on December 1st 2012 (Meaning 2 Peltier/TEC assemblies chilling the water cooling the CPU and the GPU solely cooled by traditional radiator cooling)

The various Peltier/TEC assembly elements have been rehoused in their same setup configurations, but the original setup configuration adopted December 1st 2012 has now been in operation over 16 months.

The GPU radiator cooling has been changed from a RadBox with multiple radiator configuration, to one massive side mounted Watercool MO-RA3, but the Peltier/TEC assemblies have just been relocated into a new case in their same setup configuration.

July 13th this year will be the 2 year point reached since the Peltier/TEC water chilling fist was assembled the first time and actually went into testing operation.

After over 16 months of dependable water chilling produced by these Peltier/TEC assemblies I can honestly say it is successful and continuing to be so.

I run a 5.0ghz 3770K overclock 24/7 and the CPU has not been delidded, my present water temperature cooling the CPU is 10.4c, my CPU present idle temperature 4 core average is 15.75c.

 

[faye__kane]

First of all, congrats on inventing a coo-well gizmo! [curtsies]

But I'm wondering why you chose to cool the water rather than the CPU directly when it is thermodynamically far less efficient. Also, the way you've done it, condensation is still a scary monster waiting to jump out at you and kill your MB. It's just a matter of your daughter or someone turning off your A/C and opening the window in the summer when you're not there. The humidity soars, and you killed your CPU and MB. I killed them myself doing it your way (trying to balance humidity, temperature, and condensation. That's why I decided to make condensation a non-issue.

I find it's better to make your board waterproof (as an emerg backup) and then insulate the CPU area so it doesn't get cold enough to condense the air no matter what.

This can be done with a surprisingly thin layer of clear silicone sealant:

Around ⅓ inch of silicone is all you need for it to not feel cold to the touch on the outside surface:

I don't really use that thick heat-sink paste; it's just there during the silicone injection so it doesn't stick to the top of the TEC.

Don't ignore the bottom side! (also good advice in bed)

Then put a normal watercool block on top:

That pic is not of a normal watercool block. it's a pre-built water cooler system with the plastic pump in the water block. I use a 420 Watt TEC (which requires a second PS slaved to the first), and it killed the cooling system you see above when the pump melted. Now I use a 3 x 120mm radiator array to cool the water, a plastic-free waterblock, half-inch I.D. hoses, and I put the radiator BEFORE the pump in the circuit so I don't melt another pump. If I want to make the water colder, I'll just add another pump and radiator array in series.

Also, very hi-pwr TEC devices do not come in 40 mm, so you have to use staggered aluminum plates on the cold side to reduce the width so it fits on the CPU. They must be insulated with silicone too. And my 420-watt TEC takes 24 V. I have a 24V transformer PS, but I don't use it (I use 12v) because I'm scared it will fail and my sys will overheat. But using PC pwr supplies, the sys will shut down if the "pwr good" (orange) wire goes low.

I did get a 120v DC-triggered AC relay to kill the sys pwr if the 24v pwr goes down, but I never hooked it up because it seems too Rube Goldberg. So far, I've only run my 420 W TEC at 12 volts.

Yours looks real cool, though! I don't put my stuff in a case because I'm always [strike] fu [/strike] hacking with it.

If anyone is interested, I'll take pix of v. 2.0 of my cooling system and post them. The pix above are v 1.0, which had only one 120mm fan, a 40mm TEC, and a pump that melted. I also killed 2 CPUs and one MB before getting it right. Per ardua ad astra!

☺,

-faye kane

 

[4Ryan6]

First of all, congrats on inventing a coo-well gizmo! [curtsies]

But I'm wondering why you chose to cool the water rather than the CPU directly when it is thermodynamically far less efficient. Also, the way you've done it, condensation is still a scary monster waiting to jump out at you and kill your MB. It's just a matter of your daughter or someone turning off your A/C and opening the window in the summer when you're not there. The humidity soars, and you killed your CPU and MB. I killed them myself doing it your way (trying to balance humidity, temperature, and condensation. That's why I decided to make condensation a non-issue.

I find it's better to make your board waterproof (as an emerg backup) and then insulate the CPU area so it doesn't get cold enough to condense the air no matter what.

This can be done with a surprisingly thin layer of clear silicone sealant:
That pic is not of a normal watercool block. it's a pre-built water cooler system with the plastic pump in the water block. I use a 420 Watt TEC (which requires a second PS slaved to the first), and it killed the cooling system you see above when the pump melted. Now I use a 3 x 120mm radiator array to cool the water, a plastic-free waterblock, half-inch I.D. hoses, and I put the radiator BEFORE the pump in the circuit so I don't melt another pump. If I want to make the water colder, I'll just add another pump and radiator array in series.

Also, very hi-pwr TEC devices do not come in 40 mm, so you have to use staggered aluminum plates on the cold side to reduce the width so it fits on the CPU. They must be insulated with silicone too. And my 420-watt TEC takes 24 V. I have a 24V transformer PS, but I don't use it (I use 12v) because I'm scared it will fail and my sys will overheat. But using PC pwr supplies, the sys will shut down if the "pwr good" (orange) wire goes low.

I did get a 120v DC-triggered AC relay to kill the sys pwr if the 24v pwr goes down, but I never hooked it up because it seems too Rube Goldberg. So far, I've only run my 420 W TEC at 12 volts.

Yours looks real cool, though! I don't put my stuff in a case because I'm always [strike] fu [/strike] hacking with it.

-faye kane

Hi Faye, Thank You for the compliments.

I chose to cool the water as a replacement for the ice cooling I was already doing, as I had already been playing around with below ambient temperatures and knew where I could go down to, (temperature wise), without condensation being a problem at all.

I was seeking a cooling solution to replace the ice I was regularly having to swap out, and had considered many other options the most expensive being aquarium type water chillers, because my experimentation was below ambient not below zero.

There's a lot of overclocking headroom below ambient but still above the condensation forming point as both camps AMD and Intel base their stock cooling on ambient room temperatures being the effectual cooling points the CPU protections are based on, so just a little manipulation in that territory can yield some great results.

It may seem extremely strange but it doesn't matter if my AC gets turned off, because if it does the peltier exhaust heats the room and as the peltiers input cooling air gets warmer the water temperature rises over time, that's how well balanced the heat pipe coolers and fan speed output do their job.

What I am saying is my setup is balanced, at a 23c ambient room temperature my peltiers deliver 10c water temperature, if that 23c ambient increases so does the water temperature, the AC is to maintain a 23c ambient for my personal comfort and in the process also controls my water temperature, so when the room temperature rises to say 25c the water temperature rises to around 12c as well and my condensation does not start until the water temperature begins going below 8c.

One thing I can say for my setup as it has come through a lot of experimentation, I have only lost one thing and that was a XSPC pump inside a reservoir attempting to water cool the hot side of the peltier!

It looks like you've gone through a lot of sacrificial components to arrive where you are now, and I commend you for having the determination to get there, many just give up after destroying something to reach their goals, but you have the tenacity to continue to reach your goal!

So what overclocks and temperatures has your cooling setup allowed you to reach? Ry

 

[DiabloDoom]

Great looking thread, I remember reading into peltier and seebeck effect for create electricity out int he wilderness.

I have a question, can't I just put the peltier straight onto the board between the heat sink and CPU or will it short the thing that's trying to power it?

 

[4Ryan6]

I have a question, can't I just put the peltier straight onto the board between the heat sink and CPU or will it short the thing that's trying to power it?

Of course you can, many do that at Overclock.net and that was what Faye Kane was just making reference to above.

The cooling described in this thread is another alternative to peltier use.

 

[DiabloDoom]

Ok, thanks. Another Q, to my understanding, a peltier will generate electricity if heated at one end and cooled at the other and attached to a power bank So what stops it generating electricity on a mobo? The current you add? or is the surface of the peltiers flipped in regards to whats being cooled and heated?

Sorry, I'm just having a hard time grasping the concept.

 

[4Ryan6]

Ok, thanks. Another Q, to my understanding, a peltier will generate electricity if heated at one end and cooled at the other and attached to a power bank So what stops it generating electricity on a mobo? The current you add? or is the surface of the peltiers flipped in regards to whats being cooled and heated?

Sorry, I'm just having a hard time grasping the concept.

Check this out for starters.

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

And yes if you can supply the hot and cold to the peltier itself it will produce electricity, to what degree remains to be tested, but for myself, that is irrelevant to the way I use the peltiers, might be a nice experiment for someone to conduct though? hint, hint.

 

[4Ryan6]

Today is a reference point I need to document May 12, 2014 changed out coolant in both the CPU loop and GPU loop.

Reason being I am only running steam distilled water period, and I've seen many claims that running 100% pure steam distilled water was bad mostly by reviewers and those that think the reviewers are 100% trustworthy.

So the only way for me to know 100% for sure myself, is to change out what I'm running by flushing the system with fresh steam distilled water, inspect for problems, and refill with fresh!

I used about 3 gallons of water to flush both reservoirs, by siphoning, refilling, running, shutting down, and continuing the process until I had run 3 gallons through the systems.

Everything was crystal clear, no bugs, no critters, no deterioration or buildup!

Note: 100% pure steam distilled water with zero additional additives is getting harder to find, if you buy steam distilled water make sure you read the label and see if it contains additives, you do not want any additives.

 

[4Ryan6]

This cooling solution is absolutely amazing, the difference between TEC/Peltier controlled water cooling and traditional Radiator water cooling, is like night and day!

They both start out energized at ambient room temperature, the difference is that with Radiator water cooling the water temperature inside the system starts to increase from the CPUs heat output, with Peltier/TEC chilled water cooling water temperature inside the system starts decreasing.

The CPUs heat output continues to increase the internal water temperature with radiator cooling until it reaches an equilibrium point at some temperature above ambient, and that equilibrium point is determined by the amount of radiator cooling field the system has available to it, to use.

The more radiator cooling field in the radiator water cooling system, the closer to actual ambient room temperature the water temperature gets when it reaches it's equilibrium point.

TEC cooling starts to drop the system water temperature shortly after the Peltiers are energized and constantly drops the water temperature below ambient down to it's adjusted control run point, so my system water temperature never reaches an above ambient state, at all.

My controlled run point could also be classed as a below ambient equilibrium point, which for my setup is an actual 10c water temperature, which is usually in relation to my actual ambient room temperature, around 15c below my ambient.

100% Fact:
AMD and Intel both design their systems to run in an above ambient state, all their motherboard design technology requirements controlling the supplied stock air coolers, are designed to operate and boost fan speeds at certain CPU operational temperatures reached, but what if you can operate and not reach those activation point temperatures because the CPU is operating below those points, have any of you even thought about those possibilities?

It sometimes surprises me that more of you are not experimenting with this type of cooling, but trust me if you do, you're going to discover things, very few are even aware of!

Ryan

 

[cmptrgydv]

Hi Ryan,

I have been working on an idea for a while now, like you I started thinking about lowering water temp, but my idea is a different angle then yours, I was looking at cooling the water first then pumping it through the loop, I found two GPU coolers with a flat surface, I attached them together, and put the peltiers on both sides of block, with heat sink and fan on the out side, I had read that a CPU block only raises the water temp by maybe 3 or 4 degrees, so I figured if you lower the temp by maybe 6 to 8 degrees, then the water would get very cold as it circulates around the loop, thus no need for radiator, so less water to cool, like you also thought of the tubing and res sweating, I thought just put res out side of case, problem solved, insulate tubing, but then also thought well just keep water above 50 degrees and below room temp, and turn up peltier as you overclock to keep water at an ideal temp,

I have never used a peltier before so was not to sure on their properties, but I knew I did not need one that gets -50 so I picked up two 60 watt peltiers, I figure I will make mistakes, and learn as I go, this is the first time I had read of anybody trying anything like what I have in mind, so I could use some insight if you don't mind,

first do you think the two 60 watt peltiers will cool the water enough, or will I need bigger ones,

 

[4Ryan6]

Hi Ryan,

I have been working on an idea for a while now, like you I started thinking about lowering water temp, but my idea is a different angle then yours, I was looking at cooling the water first then pumping it through the loop, I found two GPU coolers with a flat surface, I attached them together, and put the peltiers on both sides of block, with heat sink and fan on the out side, I had read that a CPU block only raises the water temp by maybe 3 or 4 degrees, so I figured if you lower the temp by maybe 6 to 8 degrees, then the water would get very cold as it circulates around the loop, thus no need for radiator, so less water to cool, like you also thought of the tubing and res sweating, I thought just put res out side of case, problem solved, insulate tubing, but then also thought well just keep water above 50 degrees and below room temp, and turn up peltier as you overclock to keep water at an ideal temp,

I have never used a peltier before so was not to sure on their properties, but I knew I did not need one that gets -50 so I picked up two 60 watt peltiers, I figure I will make mistakes, and learn as I go, this is the first time I had read of anybody trying anything like what I have in mind, so I could use some insight if you don't mind,

first do you think the two 60 watt peltiers will cool the water enough, or will I need bigger ones,

Hi cmptrgydv,

Since you already have the 2 60w peltiers and the GPU blocks, you may as well experiment with those to start out?

Spec wise the 60w peltiers have to be supplied the power they need and the most important thing is the amperage draw they pull, your power supply has to be able to handle the combined amperage load, because it is a constant load!

If you use a PC type power supply it needs to be a single 12v rail, not a multi 12v rail, and do not get your power from the 18g Molex 4 Pin or SATA cabling, use either the 4 or 8 pin motherboard connector, or 6 or 8 pin PCI-E, those are 14g wires.

You can snag some cable extenders for those wires, and not have to actually cut the power supply wires.

You want a 2nd power supply dedicated to powering the peltiers, do not attempt to run the same power supply you power the computer with!

The power supply powering peltiers is very important, this is not the situation of just barely getting by, you want at least 1/3rd power supply load handling capability over the actual amperage draw of the combined peltier load, to keep the power supply running cool.

I'll be honored to help you as much as I can, this thread covers a 100% operational cooling solution but there are other possibilities, you've already taken your first step, so lets see what you learn from your experimentation, so please share what you discover, and provide some pictures as well.