TEC/Peltier CPU Chilled Water Cooling

[rubix_1011]

an old acer veriton case that i have stripped, prepped and trying to get painted and put back together before his broadband goes on... oh yeah, i have that to sort out as well...

Please do not take offense, I need this thread to stay on topic, because I do not want the moderators to feel it needs to be closed because it has not stayed on topic.

Thank You! Ry

I keep a firm watch on this section, so you should be OK unless someone reports this to someone else, although I'd think they would ultimately come track me down before closing the thread...either way, I don't see it happening without a kind nudge to 'stay on target'.

[video="https://www.youtube.com/watch?v=NnP5iDKwuwk"][/video]

 

[4Ryan6]

Nice!

 

[rubix_1011]

Small tangent - got the H110i GTX in the mail yesterday for testing. It's large, but very lightweight...so the radiator is aluminum and it's very, very thin...and doesn't seem to hold much liquid. The tubing is nice - nylon braided sleeving, but it's very stiff and difficult to move.

I haven't cracked it open yet for flow rate and delta testing (since it arrived yesterday afternoon). Got some photos, need to do some mechanical measurements and then....commencement of testing. Should be a good time. Once it's done, either the Swiftech H220-X or the EK Predator will be next on the bench.

 

[4Ryan6]

Small tangent - got the H110i GTX in the mail yesterday for testing. It's large, but very lightweight...so the radiator is aluminum and it's very, very thin...and doesn't seem to hold much liquid. The tubing is nice - nylon braided sleeving, but it's very stiff and difficult to move.

I haven't cracked it open yet for flow rate and delta testing (since it arrived yesterday afternoon). Got some photos, need to do some mechanical measurements and then....commencement of testing. Should be a good time. Once it's done, either the Swiftech H220-X or the EK Predator will be next on the bench.

That's a sealed system right?

If the radiator is aluminum you know the coolant inside is a mixed metal coolant, when you open it up don't let that coolant get on anything important.

It will ruin finishes on things like your desk, or floor, it's great for mixed metals but sucks for other things it may get on, it is also probably toxic to animals.

 

[rubix_1011]

Yes, the H110i GT/GTX are sealed units, much like the H100/H110 and any other Corsair/CoolIt/Asetek offering.

Yep, well aware of the coolant concerns - I've worked on cars my entire life and a lot now in my spare time, and glycol coolant of just about any color and composition has been on my garage floor. My brother is a mechanic and has $13,000 worth of Snap On tools in my garage.

But - I want to answer a lot of question we've all asked over the course of several years, such as actual flow rate on pumps like these, actual cooling delta, etc. We don't really get too many concise answers because, let's face it, no one wants to void a warranty in order to find these out. This is what I'm going to do, though. I want a concise, apples to apples comparison and testing suite in order to have a dataset that you can point at to evaluate scientific data points of one cooler to another. The radiator is wide and covers a lot of length x width....but is very, very thin...just a quick measure was around 27mm thick overall, but that was overall side to side...the core is only around 20-21mm thick, and the end tank is really quite small.

 

[4Ryan6]

Yes, the H110i GT/GTX are sealed units, much like the H100/H110 and any other Corsair/CoolIt/Asetek offering.

Yep, well aware of the coolant concerns - I've worked on cars my entire life and a lot now in my spare time, and glycol coolant of just about any color and composition has been on my garage floor. My brother is a mechanic and has $13,000 worth of Snap On tools in my garage.

But - I want to answer a lot of question we've all asked over the course of several years, such as actual flow rate on pumps like these, actual cooling delta, etc. We don't really get too many concise answers because, let's face it, no one wants to void a warranty in order to find these out. This is what I'm going to do, though. I want a concise, apples to apples comparison and testing suite in order to have a dataset that you can point at to evaluate scientific data points of one cooler to another. The radiator is wide and covers a lot of length x width....but is very, very thin...just a quick measure was around 27mm thick overall, but that was overall side to side...the core is only around 20-21mm thick, and the end tank is really quite small.

Just threw it out as a reminder of caution, I had no doubt you are fully aware of the caustic properties of automotive style coolants.

*********************

We cannot honestly expect the manufacturer to put anymore quality into these AIO coolers that they have to, marketing the product, or else they would be loosing money selling them right out of the factory door.

We look for price per performance but they sell on price per manufacturing cost, performance is really irrelevant when high production costs come into the picture, they look for profitable common ground.

Expense wise custom cooling is the greatest customer cost with a high proven manufacturing base of individual fully tested a time based (MTBF) components, and even though AIO coolers are more and more frequenting the water cooling scene, they still have to be cost effective to produce.

Aluminum radiators are cheaper than copper radiators, so of course they will get used to cheapen the manufacturing cost, that will be evident in your testing end results.

 

[4Ryan6]

I have to admit I am getting a little excited as my setup is 46 days away from 3 years in operation, it has been seen in operation and verified by 2 of Toms Hardware members, their verification statements are in this thread, one of them actually duplicated the setup, and achieved the same cooling results.

The below ambient cooling this setup provides allows the CPU to be stably overclocked higher than ambient cooling allows which was the initial goal of the experiments that led to this cooling solution.

So to me 3 years in operation is a milestone that has far exceeded my own expectations.

 

[oroboros86]

I'm not on here often, but I'd like to say a big thank you for such a great build. I wouldn't know anyone else on the internet, let alone the world with your knowledge at this point.
Your help and information has been great!

B

 

[4Ryan6]

I'm not on here often, but I'd like to say a big thank you for such a great build. I wouldn't know anyone else on the internet, let alone the world with your knowledge at this point.
Your help and information has been great!

B

Thank You For Your Kind Words!

 

[computers_are_freaky]

Hi guys, after hours reading through every post on this thread, searching through my physics book to get an idea of what the flip was going on and asking my physics teacher some wierd questions (i.e. best way to store cold, how a peltier works, ecc...) I think I've finally got my head around this awesome concept. I want to congratulate ryan on 3 years of operation with few failures, as well as anyone else who set up this cooling solution. Now, on to my question:
I am currently building a pc from used parts as a school project in the back of the IT lab, and want to male things more interesting for my final presentation. At the moment there is no need, but this Xmas, a friend of mine will upgrade and donate his current mobo and cpu to the cause. They are nothing fancy, just an fx 4300 and a resonably good overclocking board, all in working order. Would it be possible to set up a peltier cooling system for less than 100 euro (I live in Italy). It would have to be able to disperse the wattage from an overclocked cpu, probably more than 4.4 Ghz (what t is currently running at on the stock cooler). Any advice on how to do this would be more than welcome.

Oh yeah, I have access to loads of low quality psus from old school pcs, could use the best cable on there for powering the peltier. It would not need to run at full power 24/7, just an hour a week or so. Thanks again, and well done everyone for thinking of and perfecting this awesome concept.

---Update---
A friend of mine, his dad really, he has a CNC machine and says he might be able to make me some blocks if I give him a 3d model to give him some idea of what I want. I know my way around the basics of google sketchup, but not anywhere near the skill level to do this, so can someone please send me a rudimentary plan of a water block for the cold side of the peltier and for the cpu?
thanks a billion. Also, will I be able to cool the cpu on a 150W peltier with a 212 EVO on the hot side? Are there any better coolers for this out there? (this TRUE cooler I saw refferenced a few times, any links please?)

What thickness of insulation do you recommend for the reservoir and the tubing?

I plan to screw 2 cases together, like someone did a few pages ago, becasuse I have access to about 12 cases and the same number of PSUs, most of which are working.

Sorry for bad grammar, this was typed quickly during an IT lesson, thanks for the help and I need to say this again: THIS IS SO FLIPPIN' COOL!!!! (pun INTENDED!!!)

 

[4Ryan6]

I am currently building a pc from used parts as a school project in the back of the IT lab, and want to male things more interesting for my final presentation. At the moment there is no need, but this Xmas, a friend of mine will upgrade and donate his current mobo and cpu to the cause. They are nothing fancy, just an fx 4300 and a resonably good overclocking board, all in working order. Would it be possible to set up a peltier cooling system for less than 100 euro (I live in Italy). It would have to be able to disperse the wattage from an overclocked cpu, probably more than 4.4 Ghz (what t is currently running at on the stock cooler). Any advice on how to do this would be more than welcome.

Anything is possible but that seems to be a very small budget for even considering peltier cooling, the very things that make this cooling work are the peltiers themselves, taking into regards the FX 4300 is a 95w TDP and that is stock settings.

To take advantage of the peltier you have to keep it cool and have an adequate way to collect the cold from the cold side, and then store it in an insulated reservoir, that is of course to use the cooling this thread describes.

With that low of an experimentation budget you may want to consider a different approach, because the peltier alone needs to be at least 3 times the TDP just to hold even ground of a 95w CPU.

My 3770K is a 77w TDP and my peltiers total 400w actual, so you see why I say at least 3 times the TDP, but 4 times TDP is more like it to support a high overclock.

It is really the same principal of not having enough radiator cooling field to get the job done with traditional water cooling, it only gets you so far, and then they are disappointed that they cannot overclock further, then you hear well I guess I got a bad CPU when it comes to overclocking.

NO! That CPU will overclock as far as your cooling will allow, PERIOD!

That's why many that have undertaken experimental projects in this thread have fallen short of their goals, they failed to take into account how much wattage past TDP I am running for a successful high overclock setup, of a 77w TDP CPU.

I am not discouraging you just laying out the facts, which you cannot get around with less than 100 euro.

EDIT: Peltier cooling changes the ambient rules, it allows you to operate in a temperature range that is below the activation point of the CPUs automatic protection, meaning when other overclocks of the same level in the ambient realm have already activated the CPUs internal protection the same overclock under chilled water cooling has not reached those activation points yet.

Meaning the CPU overclock is still outputting performance at 100%, because it is operating below it's protective temperature throttling points, which is built into every CPU whether Intel or AMD.

That sir is why I am able to run a rock solid 5ghz overclock of a 77TDP 3770K Ivy Bridge that has not been delidded.

I am presently running a 12.5c negative delta and it is still dropping the coolant temperature and should reach an equilibrium of a 15c negative delta or 10c coolant temperature with a 25c ambient, in approximately 5 minutes.

 

[computers_are_freaky]

I am currently building a pc from used parts as a school project in the back of the IT lab, and want to male things more interesting for my final presentation. At the moment there is no need, but this Xmas, a friend of mine will upgrade and donate his current mobo and cpu to the cause. They are nothing fancy, just an fx 4300 and a resonably good overclocking board, all in working order. Would it be possible to set up a peltier cooling system for less than 100 euro (I live in Italy). It would have to be able to disperse the wattage from an overclocked cpu, probably more than 4.4 Ghz (what t is currently running at on the stock cooler). Any advice on how to do this would be more than welcome.

Anything is possible but that seems to be a very small budget for even considering peltier cooling, the very things that make this cooling work are the peltiers themselves, taking into regards the FX 4300 is a 95w TDP and that is stock settings.

To take advantage of the peltier you have to keep it cool and have an adequate way to collect the cold from the cold side, and then store it in an insulated reservoir, that is of course to use the cooling this thread describes.

With that low of an experimentation budget you may want to consider a different approach, because the peltier alone needs to be at least 3 times the TDP just to hold even ground of a 95w CPU.

My 3770K is a 77w TDP and my peltiers total 400w actual, so you see why I say at least 3 times the TDP, but 4 times TDP is more like it to support a high overclock.

It is really the same principal of not having enough radiator cooling field to get the job done with traditional water cooling, it only gets you so far, and then they are disappointed that they cannot overclock further, then you hear well I guess I got a bad CPU when it comes to overclocking.

NO! That CPU will overclock as far as your cooling will allow, PERIOD!

That's why many that have undertaken experimental projects in this thread have fallen short of their goals, they failed to take into account how much wattage past TDP I am running for a successful high overclock setup, of a 77w TDP CPU.

I am not discouraging you just laying out the facts, which you cannot get around with less than 100 euro.

So, this is a hypothesis, just brainstorming here:
get 2 150W peltiers and run each off of a separate PSU, even a terrible psu should be able to handle 1 peltier, right (they are about 350W each). It's only a few hours a week, so I can live with the ineficiency.
Have my friend make me 2 peltier blocks and a cpu block for a very low price (he said maybe 10 euro total). get a new pump, or ask around if anyone will give me an old aquarium pump. Get soem cheap tubing, this isn't ideal for long run, but will work for initial experimentation. Use a piece of pipe for res. Use a 212 EVO on each hot side of the peltiers.
If I can get the blocks off my friend and a free pump from somewhere (hopefully good quality), then I should be good on my budget, right? Thanks for the help, and I know your just giving me tips from your vast experience on this sunject, so thank yo very much for all the help

 

[computers_are_freaky]

don't worry about the science part, I've read every post on this thread (took me ages), so I get it, don't worry. thanks for the help.

 

[computers_are_freaky]

this seams susspiciously cheap, would it work though?
http://www.amazon.it/SODIAL-TEC1-12710-CELLA-PELTIER-TERMOELETTRICO/dp/B00AW65PDO/ref=sr_1_fkmr0_3?ie=UTF8&qid=1447758507&sr=8-3-fkmr0&keywords=150w+peltier
thanks for the help.

---Update---
found another peltier unit, same wattage, bit more expensive, same size. Does it look less sketchy? I'm not sure about how these things should look on paper, so I await your verdict. Thanks.
http://www.amazon.it/COLEMETER-TEC1-12710-CELLA-PELTIER-TERMOELETTRICO/dp/B00EE6FRMW/ref=pd_sim_421_7?ie=UTF8&dpID=41rYbzn9DqL&dpSrc=sims&preST=_AC_UL160_SR160%2C160_&refRID=0ZJCAGHMXYAS4G5BWQ43

 

[4Ryan6]

So, this is a hypothesis, just brainstorming here:
get 2 150W peltiers and run each off of a separate PSU, even a terrible psu should be able to handle 1 peltier, right (they are about 350W each). It's only a few hours a week, so I can live with the ineficiency.
Have my friend make me 2 peltier blocks and a cpu block for a very low price (he said maybe 10 euro total). get a new pump, or ask around if anyone will give me an old aquarium pump. Get soem cheap tubing, this isn't ideal for long run, but will work for initial experimentation. Use a piece of pipe for res. Use a 212 EVO on each hot side of the peltiers.
If I can get the blocks off my friend and a free pump from somewhere (hopefully good quality), then I should be good on my budget, right? Thanks for the help, and I know your just giving me tips from your vast experience on this sunject, so thank yo very much for all the help

I admire your determination, I wish you did not live so far away!

An EVO should be able to handle 150w, but regarding the power supplies they have to be a single 12v rail model power supply and it's 12v rail load amperage has to be at least 25a to be safe, your 150w peltier will probably draw 15a so you need to be well above that, or you'll burn out the power supply.

 

[4Ryan6]

this seams susspiciously cheap, would it work though?
http://www.amazon.it/SODIAL-TEC1-12710-CELLA-PELTIER-TERMOELETTRICO/dp/B00AW65PDO/ref=sr_1_fkmr0_3?ie=UTF8&qid=1447758507&sr=8-3-fkmr0&keywords=150w+peltier
thanks for the help.

Looks like a 40mm x 40mm Potted Peltier its specs are a little sketchy, but should work for your needs, it should draw 15a at 12v? But I am not sure regarding the written specs?

 

[computers_are_freaky]

dammit, I'll post a pic of the psu spec sheat on the side, maybe luck will shine through and they will just manage it (I doubt it though) Does it have to be modular? how rare are they (I might be allowed into a recycling centre to loot some gear, like cases and ram and things, so If they are fairly mainstream I maybe will be able to get my hands on one for free). How much are they new, and what would my electricity bill look like if I intrall this on my main system (about 1 hr a day).

Future idea (really far future, when I get some real cash):
I was thinking of maybe designing some sort of shared loop, so have a flow junction that I switch between what I want, standard custom loop on one side, TEC loop on the other for extremely cpu intensive games or compiling my code. That way I would still have the performance on hand when I want it but not wasting $$$ when I don't.

 

[computers_are_freaky]

another idea: my schools physics lab should have a power bank, maybe I could ask if they will let me use it to set up my loop and test different voltages and such. Need to ask permission, but could be an option. Are there any Power supplies that are not designed for PCs that could power 2 TECs and would be a bit cheaper?
Thanks for the help.

 

[4Ryan6]

dammit, I'll post a pic of the psu spec sheat on the side, maybe luck will shine through and they will just manage it (I doubt it though) Does it have to be modular? how rare are they (I might be allowed into a recycling centre to loot some gear, like cases and ram and things, so If they are fairly mainstream I maybe will be able to get my hands on one for free). How much are they new, and what would my electricity bill look like if I intrall this on my main system (about 1 hr a day).

Future idea (really far future, when I get some real cash):
I was thinking of maybe designing some sort of shared loop, so have a flow junction that I switch between what I want, standard custom loop on one side, TEC loop on the other for extremely cpu intensive games or compiling my code. That way I would still have the performance on hand when I want it but not wasting $$$ when I don't.

A lot of old power supplies were single 12v rail, look on the spec label on the power supply and see how many +12v are listed more than one and it is not a single rail, also look for the 12v amperage.

No it does not have to be a modular Power Supply.

Your electricity bill depends on what your utility supplier charges per kilowatt hour?

For just one hour of use, I don't think you'll be in jeopardy of the electrical Police knocking on your door.

 

[4Ryan6]

another idea: my schools physics lab should have a power bank, maybe I could ask if they will let me use it to set up my loop and test different voltages and such. Need to ask permission, but could be an option. Are there any Power supplies that are not designed for PCs that could power 2 TECs and would be a bit cheaper?
Thanks for the help.

Yes and No!

Powering 2 TECs Yes

Cheaper, Probably not.

Mean Well Power Supplies

 

[tech-wreck]

or, for about twenty monies, a server power supply is easily modded to deliver lots of nice smooth amperes which can be used to power high-end battery chargers...
http://www.rcgroups.com/forums/showthread.php?t=1170784
or if radio controlled flying isn't your thing, use it to power the cooling system.

 

[computers_are_freaky]

Ok, first: no way am I doing any psu modding, way too dangerous for my taste.
Ok, here is what is written on the side of the psu I have on hand from the pcs, they are not all the same, but simillar:

Vac ~ VOLTAGE. CURRENT. FREQUENCY
INPUT 230V. 5A. 47~63Hz

Vdc. +3.3V +5V +12V. -12V. -5V. +5VSB. BLK. GRN. GRY
MAXDC
OUTPUT. 14A 30A 15A. 0.5A. 0.5A. 1.5A. COM. P-ON. PG

PEAK. TOTAL
OUTPUT. 200W. 15A. 0.5A. 0.5A. 1.5A. RETURN REMOTE. POK

Ignore the full stops at the end of some numbers, auto correct is so annoying but I can't be bothered to turn it off. Is this enough to power even 2 weak peltiers? I spoke with the physics guys, and they said that 12V is a lot for a peltier, and so is 30amps. They suggested I get weaker units, but I'm not sure they gpt the gist of what I'm trying to do. Thanks for the help.

 

[computers_are_freaky]

Dammit, it removed all my careful spacing. Sorry guys.

 

[4Ryan6]

Ok, first: no way am I doing any psu modding, way too dangerous for my taste.
Ok, here is what is written on the side of the psu I have on hand from the pcs, they are not all the same, but simillar:

Vac ~ VOLTAGE. CURRENT. FREQUENCY
INPUT 230V. 5A. 47~63Hz

Vdc. +3.3V +5V +12V. -12V. -5V. +5VSB. BLK. GRN. GRY
MAXDC
OUTPUT. 14A 30A 15A. 0.5A. 0.5A. 1.5A. COM. P-ON. PG

PEAK. TOTAL
OUTPUT. 200W. 15A. 0.5A. 0.5A. 1.5A. RETURN REMOTE. POK

Ignore the full stops at the end of some numbers, auto correct is so annoying but I can't be bothered to turn it off. Is this enough to power even 2 weak peltiers? I spoke with the physics guys, and they said that 12V is a lot for a peltier, and so is 30amps. They suggested I get weaker units, but I'm not sure they gpt the gist of what I'm trying to do. Thanks for the help.

Looks like a 15a single 12v rail power supply.

I spoke with the physics guys, and they said that 12V is a lot for a peltier, and so is 30amps. They suggested I get weaker units,

:lol: Some peltiers require 34.5 volts

http://customthermoelectric.com/tecs_imax.html

 

[computers_are_freaky]

So, what sort of peltiers could I run on it?
(Sorry, not studied electricity yet, so don't know if you add the measurements or if they need to be that high or what)
Thanks for the help.