TEC/Peltier CPU Chilled Water Cooling

Page 12 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
If you're even considering using a peltier to utilize the cold you can get from it, you need to harness as much as possible and the wire idea will take way too long to reach a useable cold, IMO. Using a water block to collect the cold from the peltier is a very good cold transfer to the water or coolant, using the same chilled water to actually cool the CPU in the same loop. Even though the water block may not be the most efficient use of the peltiers cold collection, it will be far more efficient than using wire for cold transfer, water flowing inside the tubing transferring cold vs solid wire transferring cold? It's kinda a no brainer as to which will yield the best useable results, if you really think about it.

If you take the operation of the traditional radiator water cooling, vs, the active chilled water cooling explained in this thread as a direct example of peltier use.

When a traditional radiator loop is powered up after a time period of being off, the coolant temperature in the loop equalizes to ambient room temperature.
Once powered the coolant starts to flow collecting the CPUs heat and the coolant temperature begins to rise and continues to rise until it reaches a temperature point of equilibrium, above ambient room temperature.

When the chilled water system is powered after a period of being off, it's coolant temperature has also returned to ambient room temperature, the coolant starts to flow collecting the CPUs heat, but unlike the radiator the peltier assemblies are outputting cold, so the chilled water cooling never ever goes above ambient, but steadily drops below ambient until it's cold equilibrium is reached, which normally takes about 30 minutes of operation and which is usually around 15c below ambient room temperature.

Now if you use solid copper wire to transfer the cold to the reservoir, even if you insulate all the wire and reservoir the cold transfer will take much longer to occur and be also IMO be limited, which will allow the CPU to have dumped a significant amount of heat that the impeded cold transfer cannot recover.

Keep in mind the water flowing is constantly transferring heat, (from the CPU) ~ to the coolant ~ cold to the coolant (from the peltier assemblies) ~ cold stored inside the insulated reservoir, ~ pumped back to the CPU to collect it's outputted heat, and the cycle continues.

I would think if you intended to expend the electrical power to run the peltiers in the first place, you would want as much from them as possible.
 


Thanks for that confirmation ArthurH, I really appreciate you taking the time to add your input!
 

Davil

Distinguished
Feb 2, 2012
310
0
18,960
Well in my estimation what you're really doing is taking a cold side of the peltier putting it through a medium, i.e. the waterblock, and going from the waterblock to the water. What I'm suggesting is actually doing the exact same thing only slightly differently. It goes from the peltier, to the copper wire, and then to the water. And actually I may have more surface area contacting the water in this way because I want to coil the copper wire inside the reservoir in direct contact with the water. I believe another interesting idea I had with this was to use hollow copper tubing and fill it with either water, or a better option is I would think maybe the gel they use in cold packs for lunch boxes since it won't actually freeze. But that should help to maintain the temperature and capture it more effectively even while the peltier is off. Which could lead to potentially not leaving the peltier on the entire time and only cycling it on and off when necessary.

Do you see what I'm saying? What I want to do is make a much smaller version of what you have. I envision a much smaller power supply, possibly just something I could put in a project box and plug into a wall with an on/off switch. I'm actually going to run to home depot today to get some insulating foam and I'm going to do a quick experiment. If it goes according to plan I'll put up a quick video to at least prove the theory.
 


Good luck to you, let us know what happens?

 

Davil

Distinguished
Feb 2, 2012
310
0
18,960
Yea sorry I got a bit busy. The experiment did seem to work but I decided it's not a very good demonstration since it's not really representative of what would actually be happening. Plus I picked up a small peltier powered refrigerator for $30 and I'm working on tearing it down to get at the 25w peltier inside. It's actually perfect for what I was thinking of because it's already got the electronics I need plus a working heat sink on the back. I really only need to work on my coil design.
 
A 25w peltier might cool a couple of static drinks while you're working all this out, but you can forget it being capable of cooling a computer producing it's own heat.

You have to be capable of overriding the wattage of the CPU.
 


That is my goal as well.

What you need to understand is there are ways to shrink it down to a smaller size, BUT, you still have to have the raw capabilities to accomplish handling the heat load produced by the CPU, and that load is increased when you overclock the CPU.

If you do not have the wattage capabilities to handle the CPU without it even being overclocked, what will happen when it is overclocked.

Starting where you are with a 25w peltier, you're going to have to have an extreme love of failure and starting over and a lot of tenacity, or else after a few failed experiments you'll just quit and give up.

I'm not trying to discourage you!

You need to understand simple logic, you are not going to cool a 90w CPU with a 25w peltier.

Now if you can I'll gladly eat my words and start experimenting in that direction myself, but all my past experimentation has led me to this setup I'm running, and I'm running 2 peltiers producing a combined 400w, that is how I can cool the CPU and store cold in the reservoir, I am simply producing more cold than the CPU is producing hot.

But it ain't going to happen with 25w.

 

Davil

Distinguished
Feb 2, 2012
310
0
18,960
I see your point, I thought I read something about 25w peltiers at the start of the articles but I was probably reading the amps. Guess I'll just have to keep waiting for the 245w one to get here. At least in the mean time I do have something to cool my soda :) while I wait.
 


That is an excellent attitude Davil !!!!!!!!!!!!!

I look forward to your discoveries! :) Ryan

While you are waiting go back up to the opening thread posts and read up on the power supply capability and where you actually get the power to power the peltier, you do not use the 18g 4 pin Molex and SATA cabling as it will not handle the amperage load.



 

Davil

Distinguished
Feb 2, 2012
310
0
18,960
I'm considering building my own power supply or seeing what radio shack has available since I'm still in an experiment phase. I just want to power the peltier and a fan to cool it. I'm basically going to just see if I can freeze a small quantity of water in a cup by attaching the cold side to a copper coil and putting the other end in the water. Should that work I think I have a fairly decent method of doing things without all the hassle of using all that other stuff.
 

CustomBuilt

Distinguished
Jul 23, 2006
17
0
18,510
Sorry, I couldn't resist.

Hello Ryan, I first off want to say congrats to you on your success with this endeavor! I have been following this thread for several months now while slowly buying parts for my very first water cooled build. I have been a system builder for about 15 years now but have never taken the plunge to water cooling till now. I am still in the process of ordering parts and fabricating parts for my first water/chilled system, which leads me to some questions I have for you about the TEG I am considering buying.

First off I have a 1000watt PS with 70amps on the single 12v rail which will be dedicated to the TEG only.
Secondly, this is the TEG I am considering buying.
http://www.ebay.com/itm/280580259103?_trksid=p2055119.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

My question to you is: will these two work together in harmony or will one destroy the other? (burnout)

knowing that the theoretical max of the 8pin EPS line is 28amps@336watts, will the TEG rated @ 32amps and 550watts try to draw to much from the power supply and burn it out? or itself?
http://www.playtool.com/pages/psuconnectors/connectors.html#atx12v4

TEG spec @16 volt is 32amp @ 550watt max.
I will be running 12v to the TEG so I should expect a max of 24amp @ 412watt. Is this correct?
412watt being max dissipation?

If you can enlighten me on this, It would be much appreciated.
Thank You.
 


Your 12v amperage estimation of 24a looks right to me and the 412w wattage is very close as well.

To power it the power supply you have is plenty capable very close to mine.

Wiring wise to keep from overloading the cabling I used the 8 pin M/B connector which is 14g wire, consisting of 4 12v yellow positive and 4 black 12v ground, I combined 2 of the yellow for the positive lead wire to TEC #1 and 2 of the black for the negative.

I used 12g stranded wire to run from the power supply connections to the peltier connections.

Since I have 2 TECs I used the other 2 yellow and the other 2 black from the 8 pin connector for the positive and negative to the 2nd TEC, they don't even get warm, they have zero problems handling the TECs.

My question to you is, that peltier is huge 62mm x 62mm, how do you plan cooling the hot side and collecting the useable cold from the cold side?

 

CustomBuilt

Distinguished
Jul 23, 2006
17
0
18,510


This will be the water block used to pickup cold.
http://www.shop.customthermoelectric.com/Water-Block-Assembly-30-x-30-x-085-In-outlet-Opposite-Sides-WBA-30-085-AL-R2.htm;jsessionid=A549EEF2EFE01D7D5E4140D692668E1B.m1plqscsfapp06

This will be the cold/hot plate used.
http://www.shop.customthermoelectric.com/Cold-Plate-w-Mounting-holes-30-x-30-x-025-CPT-30-30-025-AL.htm;jsessionid=A549EEF2EFE01D7D5E4140D692668E1B.m1plqscsfapp06

Then I have a Noctua NH-D14 that will sit atop the cold/hot plate.

For the reservoir, I am using dual PVC pipe. one 6 inch and one 4 inch. The 4" will be capped at each end and then wrapped with owens corning insulation and then put inside a painted 6" tube for aesthetic purposes. The 6" tube will then be mounted to the outside of a steel reinforced Lian-LI case. I am planning on this build to be a single case unit which is why I am mounting it to the outside of the case along with the water block assembly. This Lian-Li case also has room for two power supply units. One at the top and one at the bottom. I am a PC builder who is concerned with aesthetics and therefore will not be covering up my acrylic water lines with insulation. Only one line will be insulated and that will be the one from water block to insulated reservoir. Please keep in mind that my goals differ from yours in that, I am only wanting to achieve "better than water alone" temps, Somewhere between ambient watercooling and sub-ambient and I will be happy. If I achieve sub ambient with this setup, then im just that much happier. I believe that with this large of a TEC and this style of "zig zag water flow" water block, my goals are within reason. Your thoughts?

BTW: My GPU's will be on a separate conventional RAD and fan loop.

In progress photo's below.

image_preview.php

image_preview.php

image_preview.php

image_preview.php

image_preview.php

image_preview.php


 

CustomBuilt

Distinguished
Jul 23, 2006
17
0
18,510
Notice the 6 rivets which are holding a steel plate for reservoir mounting reinforcement.


image_preview.php
[/url][/img]

image_preview.php
[/url][/img]

image_preview.php
[/url][/img]

image_preview.php
[/url][/img]

image_preview.php
[/url][/img]


 
For the hot side you may need to consider adding a 2" x 2" cold plate extender to the hot side to step the 3" x 3" size closer to the Noctua Base dimension.

The Noctua base is going to get lost in the middle of the 3" x 3" plate, you need something to kinda bridge the outer area that won't be covered to transfer the heat to the Noctua base.

http://www.shop.customthermoelectric.com/Cold-Plate-Extender-Block-20-x-20-x-025-CP-20-20-AL-02.htm

Nice work so far, I'm getting excited for you! :)

This Lian-Li case also has room for two power supply units. One at the top and one at the bottom. I am a PC builder who is concerned with aesthetics and therefore will not be covering up my acrylic water lines with insulation. Only one line will be insulated and that will be the one from water block to insulated reservoir. Please keep in mind that my goals differ from yours in that, I am only wanting to achieve "better than water alone" temps, Somewhere between ambient watercooling and sub-ambient and I will be happy. If I achieve sub ambient with this setup, then im just that much happier. I believe that with this large of a TEC and this style of "zig zag water flow" water block, my goals are within reason. Your thoughts?

My next build will be all about aesthetics and getting everything inside one case, including the Watercool MO-RA3 radiator cooling my graphics, but that's not going to happen until after Christmas this year as after that point the system will be crossing the 2nd year of successful operation. Mine may not be the most eye pleasing setup, I know that, but it all started as an experiment and has gone through numerous changes to get to the point it is at now. I think your goals are definitely reachable for the sub ambient water and I am looking forward to seeing your finished results and what it does for you!
 

CustomBuilt

Distinguished
Jul 23, 2006
17
0
18,510
Thank You for the extender block advice. I have ordered it.

"Nice work so far, I'm getting excited for you! :)"
Me too! I'm like a 10 year old kid on Christmas eve, yet I'm much older than that : )
Speaking of Christmas, this build should be done shortly after that. Once completed, I will post temps, benchmarks and more pics. The complete system specs will be as follows.

CASE: Lian-Li PC-A70F
PSU: Corsair AX1200i
CPU: 3970x
MB: ASUS Sabertooth X79
MEM: 16gb G-Skill Ripjaws 1866 @ 8-9-9-24
SSD: 2x Samsung 256gb 850 pro's in raid 0
HDD: WD Black SATAIII 1TB
SOUND: Creative Sound Blaster Z
GPU's: 2x GTX 690's
DVD-RW: Plextor
Blu-Ray: LG
MONITOR: BenQ XL2420T 120hz.
KEYBOARD: Logitech G510
MOUSE: Logitech G700
SPEAKERS: Logitech Z906 5.1 500 watt
OS: Win7 Pro x64

CPU WB: EK Supremacy EVO plexi
GPU WB's: XSPC Razor
MOSFET WB: Alphacool
RAD: Alphacool NexXxoS ST30 Full Copper Dual 140mm Radiator
RES: Alphacool Lighttower
PUMPS: 2x D5's
FANS: Cougar CFD14HBB
TUBING: PETG 3/8x1/2
FITTINGS: Monsoon Economy
 
I'm not sure you've planned enough radiator for 2 GTX690s those are dual GPU cards, you might better rethink your radiator cooling field for your GPUs?

When I was running 2 580GTX I was running 2 radiators a Black Ice 240 and a Black Ice 360 and they were sufficient, for the 2 single core cards, but with you running 2 dual core cards?
 

CustomBuilt

Distinguished
Jul 23, 2006
17
0
18,510
I currently run a stock 690 in my rig. I bought it soon after it released for 999$. I had no plans on getting another because that 4th GPU would be of no use 90% of the time and I can't justify the price/performance ratio. However, I just bought a second 690 from a friend who gave me a deal I COULD NOT RESIST. His 690 was the same model I already have and it is already fitted with a XSPC water block. He offered the card, water block and alphacool rad to me for 350$ (That's not a typo) I told him I would take it but I just need to see it in action first. He launched Furmark which is very heat intensive and ran it for about 15 minutes. Temps never exceeded 40° so I figure I will atleast try both cards on that rad and see how it goes before I go spending more cash.
Keep in mind the lithography of a 580 and 690 are quite different. Fermi 40nm vs. Kepler 28nm. If I remember right, the 580 was known to run quite hot?
 


350 bucks! YES! :) What a deal! :)

I still think you'll be short on rad space.

 
For those keeping up with this thread I'm fast approaching 2 full years in operation and I've been seriously contemplating adding a third peltier assembly and going sub zero.

It will require a major revamp of the entire system, with greater reservoir insulation, and additionally insulating the motherboard and running an anti-freeze type of coolant to go sub zero and not freeze the coolant.

My greatest concern is that when overclockers that cool with LN2 insulate their motherboards, they are setup for the motherboard laying flat and the LN2 pot vertical, but for my water block setup and the machine standing vertical, I'll have to come up with some way of keeping the melt condensation from dripping down on my graphics card. (I do not want that happening, EVER! :pfff: )

Presently condensation is no issue at all, unless I go below an actual 8c water temperature, but going below 8c condensation is a problem!

I just ran these tests at an 8c water temperature which is 15c below ambient.



I have not exactly resolved how I'm going to address that issue?, so any ideas you guys may have, I would appreciate you sharing?

Thanks Everyone! Ryan
 

Memhorder

Distinguished
This may sound Absurd but perhaps you could seal the case and have Silica Dioxide Filters. Silica absorbs 40% of it's weight in moisture and you can simply dry them off. There's got to be something that will control humidity in the air for electronics. Industrial Technology.
 
Status
Not open for further replies.