TEC/Peltier CPU Chilled Water Cooling

Page 16 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.


But you do not want the end consequences of the jelly doughnut punishment!

So what are you referring to?

 
But you do not want the end consequences of the jelly doughnut punishment!

So what are you referring to?

You mentioned Private pile and the douhnut scene is one of the funniest imo, theyre paying for it you eat it!
if I had a few minutes I'd parody the riflemans creed tuned to your exploits...
this is an Intel 2500k quadcore processor, there are many like it but this one is mine...
and so forth,
And could always throw in Ff details for fun 🙂
compression fittings a nuthin part till you dont got one 😛
Moto
 


O, IC :)

The Gomer Pyle reference was not to the movie "Full Metal Jacket" it was to Gomer Pyle USMC, the TV show.

I realized you were commenting off the movie, thus my comment of the end consequences, (Meaning when Leonard put a 7.62mm full metal jacket bullet clean through the Drill Sergeant), but that wasn't the actual intention of the Gomer Pyle "quote". :)

 

7.16? I'd wager its 7.62mm hehe now drop and give me fifty!
🙂
Didnt know there was a series so will look that up when life allows
Moto

 
Update: Today I changed out the XSPC Raystorm water block cooling the CPU to the EK Supremacy EVO, I had already run tests with the XSPC Raystorm earlier to compare with the EK Supremacy EVO.

Ran Intel Burn Test at 10c actual water coolant temperature, ambient room temperature is totally irrelevant with my cooling setup, but it was 24c for those concerned.

Intel burn test: (Temperature readings in c)
XSPC Raystorm Max Core = 64, 75, 71, 69, Core Average = 69.75
EK Supremacy EVO Max Core = 61, 72, 70, 66, Core Average = 67.25

So the EK Supremacy EVO beat the XSPC Raystorm by 2.5c

That's as solid a performance comparison between the 2 water blocks as you can ask for, the only thing changed on the machine was the water blocks, both were tested at 10c water temperature, running the same pump flow rate, and as already stated, ambient room temperature is an irrelevant factor with chilled water cooling.
 
What CPU and speeds were those tests with? I've been holding back from buying the new Heatkiller 4.0 series.... I *just* built my system. My build log
And the temps are decent, but I think I can do better by adding a cooled water system similar to yours.

Sorry, I've been digging around reading posts for a couple of days and probably saw the update somewhere in this thread but can't remember...

Thanks for posting up all this info, I am contemplating building a chilled water system and for space vs cooling capacity and maintenance this seems to be the best solution out there! Evap cooling looks really interesting and maybe someday I'll play with it, but I don't have the desire to constantly add water or the space to build a really large system, I've been running some quick figures and I'm hoping that I can build a peltier cooler in a custom built cabinet that's around 24"l x 8"w x 8"h. We'll see, I'll have to get pieces and parts here before I can figure that out! The build will have to wait until I get a bit more money built up again (that system build was quite a bit more expensive than I originally planned!) but I'm keeping an eye out for used pieces and parts on the cheap to slowly acquire.

Did I see you mention in another forum that you've gone down to only one peltier? Or are you still running two for load intensive applications?

 
CPU-Z validation is in my sig and the voltage is what is set in my BIOS, and the CPU has not been delidded.

I run 2 TECs all the time, and have a 3rd that can be energized on demand for stress tests and such.
 



Thanks!

What is the rating on the TEC's you're running? (edit#2 - I found a direct link to what you are running in another forum) I saw a size of 50x50 but have not found the rating or part numbers listed in the main post?

Edit(#1): I went ahead and picked up some parts today to get started on the build of the cooler.
3 - TEC1-12715 (Roughly 137w Qmax each, probably around 110-120 actual at 12v), they were the largest 12v I could find in the 40mm to match the cooling blocks.
3 - Copper Blocks I liked these because they are already high flow with 6mm channels, they are cheap and hopefully only require lapping. At least if they don't work well I haven't wasted much on them. Looking over some of your conversations with others where you recommend the modded CPU cooler due to it being 3mm thickness, I'm hoping these will work ok, 10mm overall thickness with 6mm channels should leave 2mm of wall, now to see how much surface area the water actually covers... never thought I'd be wishing for a snake camera!
A temp display/controller with probe and a solid state relay.

That should give me enough to at least start playing around and seeing what to expect. I'm still on the hunt for some decent coolers, I have a couple cheapy stocks and H60 kicking around that I can use while I do some tests, but eventually I'll have to find better cooling for the hot side before I actually put the system into operation.

My thoughts are to run one, possibly two all the time, with the third on that solid state relay and have it kick in if the water temp goes over 15-16c, where ever I decide the limit is, only testing will tell! And then kick back off a degree or two before dew point. I'm not sure if 330w-ish Qmax will be enough to keep the CPU going for long periods under load, I'll have to give it a whirl, I'm going to try to design the case to allow room for a fourth if needed.

I've been tossing around idea's on how to tie the enable for the second PSU to my MB so that I don't have to flip a separate switch for the cooler. I'm very much a 'one button = on' kinda guy, we'll see if I actually do it or not...

Maybe I can pick someone's brain who's gone through this and get some ideas for temp differences ect. I've got a bit more reading to do before I can finalize my questions though!
 


Thanks again!

I picked up three of the 12715's, I'll just hope that they're enough.

I've been going over temperatures and dew points ect, I'm a bit OCD on occasion... so I've decided to build a custom controller that will measure the temp of the insulated reservoir and then the ambient temp and humidity in the case, calculate the dew point, set a temperature of 1 or 2c over that and then kick peltiers on and off by itself, no user input required. Here's to hoping I manage to figure it all out!

I've got most of the parts for that on the way now. I looked around but couldn't find anything that was within a reasonable price range that could do what I wanted, I'm hoping to not spend too much on the controller, it won't have a pretty housing, but should be functional and will have an LCD display to show you what's going on. Programming it is going to be a learning experience for sure!

 
You're going to have to discover at what temperature condensation begins to form where you live, I live in South Carolina, in the USA, and our state is a high humidity area, and for my situation condensation begins forming at 8c water temperature, so I normally run safely at 10c, which is normally 15c below ambient room temperature.
 



I live in a low humidity area (think nosebleeds in winter). Humidity today is 27% for example. 10c is a pretty safe temp for me :) This morning the dew point was -10c! Even if I take the average humidity, add 10% to it and then calculate the dew point I'm still under 10c. I just have to watch my wife..... she likes to turn the A/C off in the house, so I can see a jump from 72f to 80f in the summer.

But there are those rainy days, and with such a dry climate rainy days will really have a large impact on our dew point, so I figured rather than setting up a temp / humidity display in the office and then calculating the dew point every day when I turn on the computer, I'd just have something do it for me. Is it necessary? Nope, but neither is cooling my PC to below ambient 😀 It's just a neat side project that I can have fun with!

Man, looking at the efficiency and power curves of your peltier compared to the one's I purchased.... :ouch:
I hope I can get more efficiency out of mine than you have to be getting out of yours, otherwise I'm screwed! I'll be buying larger TECs and water blocks!!
On average around 12v you should be pushing 200w of Qc. I'll be pushing closer to 120w at 12v..... Heh, I'm building a hell of a insulated reservoir, but I don't think cold storage is going to be able to overcome that large of a deficiency!!! IF I can transfer 80-85% of that 120w on each peltier, I should be able to keep up with my CPU, maybe.
 


It is definitely not necessary cooling, and I never said it was!

What it allows is worth it to me!

Cooling below ambient allows stable overclocking higher than above ambient does, that's a simple fact and the main reason I experimented into this type of cooling.

It also allows the main PC to operate using less CPU voltage to reach the overclock and lowers the power consumption of the overclocked PC, a pleasant unexpected side effect. (Of course the peltiers still draw their power!)

Anyway you look at it the wattage sizes of the peltiers you choose have to overcome the wattage heat being outputted by the overclocked CPU, or you won't have any cold to store in the insulated reservoir.

Simply because the CPU will neutralize your cold gain.

 
I'm going to give it a shot using these lower rated TEC's and see if they can handle it. If not, I'll be upgrading the same TEC's you're using!

And I completely agree, not necessary, but sure as hell is fun to build and experiment! I have the controller almost completely done and I've started a rough build log. You can see it over at: My Build log of Ryan's style cooler!

The controller might be something to think about. Mine is a bit pricey because I over purchased and didn't know what I was doing. If you're interested shoot me a PM and I can talk you through getting the parts for quite a bit cheaper than I spent.... I'm betting parts should run you around or under $30-40.
 
I am OCAddict over there and I'll be watching your progress!

Very nice reservoir you built! Excellent work! :)

What are your plans for insulating it?

I hope your controller works well for you, if it does I may consider that myself? Ry
 


Thanks!

For insulation I have a friend who has some ceramic wrap that he was using to build a kiln. He said he may have some left over. If not... I'm not sure, I've seen this really amazing stuff, super thin that you can hold your hand on and put a blow torch on the other side and not feel the heat, might try to get ahold of some of that, if I can find it again.....

What did you use for insulation and have you been happy with it?

I'll be more than happy to help out if you're interested in building a controller similar, of course it's never enough... I'm now thinking about adding PWM control for the fans onto this controller since it has the channels available. Maybe put a micro rocker switch for each fan individually and one main switch for all of them combined, this way I can control the temps using fans also, not just TEC on/off which I suspect is going to be hard on the PSU!!! Or alternatively I could use POT's.... hmm. LOL, I'm never goig to have this thing fully completed, there is always going to be "Just one more" improvement that I dream up :pt1cable:

 


I used two different types of insulation since my reservoir was a 4" PVC pipe, I used fiberglass pipe insulation and then covered that with reflective aluminum duct insulation, and it works very well.

Insulating your flat surfaces will be a challenge but you can always use foam board that is used on home construction it comes in various thicknesses.

 
I've been posting over at overclockers forums with my progress but had a question for you Ryan.

Have you seen any troubles with your blocks freezing? I've been tormenting myself over this for a couple of days now. My dew point frequently runs down to -6c.... So I set my limit on my custom controller to 2c, but I'm worried that might be too cold for straight distilled water and I'll start to ice up in the blocks.

Have you added any type of solution to your water to lower the freezing point? I've been reading all kinds of stuff, and everything seems to lower efficiency of the water. I'm leaning towards an ethanol solution of 20% or so but then I REALLY have to ensure that the reservoir is fully sealed (some chips around the top of the fill port that slooooowwly leak water. maybe some pipe dope would fill them in enough...)

As for insulation, I found a friend who had some ceramic insulation for a kiln, put an oxy settling torch on one side and your hand on the other and you feel nothing. It's a half inch thick. I wrapped the res with it, leaving the top and bottom exposed, then I cut a piece just barely larger than the open space for the bottom and will tuck in in there and tape it all down. Planning on having another piece that it sits on and a dual layer on the top, one taped down and sealed except for the fill port and another piece that is solid that covers the entire top. I haven't gotten around to pictures but I will eventually 🙂 I also picked up a thermal camera for work so I'll be able to see if my cold is trying to escape and fill any leaks.

Making good progress on the controller also. I've built a really basic buck convertor to control the voltage level feeding the TEC's (singular for now, I'm prototyping it and once fully proved I'll build the other two circuits), by adjusting the PWM on my arduino I can change my voltages levels from 0 to 12v. I've got the temp probe in the res as a feedback so I will program it to adjust power as needed and be all automated. Of course the actual values will be a trial and error once we're up and running, but I should be able to get some rough estimates initially that will allow it to at least run.

I'm also going to do a sensor for the temperature of the heat sinks, using the feedback I will control the fan speed on them as needed. Again, all automated so once it's up and tweaked all I will have to do is turn it on 🙂 Of course if I ever need to add a TEC I'll have to upgrade my Arduino... or split some signals... I'll have all my PWM channels used up with this little project!

I'm not looking forward to sitting down and actually sketching a full schematic of what I have done + have planned.... man, it's easy to do on the fly, trying to make sense of it after the fact makes my head hurt....
 
Yes I did freeze a water block because the flow through the block wasn't enough, that's the very reason for the flow modifying of the water blocks and they do not freeze, your CPUs heat output is a balancing factor, to take into consideration.

I use steam distilled water with zero additives but a 6c actual water temperature can produce condensation sweating on the CPU cooling block so you're taking a condensation risk going that low, (that's according to the dew point where you live of course as I live in South Carolina USA), as a general rule I do not go any lower than 8c water temperature sometimes for stress testing, and normally run at 10c ~ 12c.

If your dew point allows you to go down to 2c actual water temperature then the block still won't freeze as long as it is flow modified enough and your pumps flow rate is enough, you do not want to run a low flow rate on your pump, I run my Swiftech MCP655s on level 5, my preference.
 


So is the water temp of 6c causing sweating because your ambient dew point is at that, or is it because of the temp difference on the CPU? I hadn't really factored in the dew point on the CPU as it heats up, I was just thinking ambient dew point...

My dew point will allow to me go as low as -5c° comfortably some days (like today.. the dew point outside is -15c°, in my house it's probably around -4c°). I can't even give an average in my house, I've seen it as low as -8c° and as high as 6c°. But this is all ambient, so I'm ok on the hoses, but now you've got me thinking about the CPU itself sweating.

Have you by chance tested the flow rate on your modded CPU block? Do you know what the overall flow rate is for it? (Even a rough guess - It took XX seconds to fill a 1 gallon jug..)
 


Actually I do not even know what my dew point is, as I run an AC in my office even in the winter time, when the sun is directly on my office, and it is definitely run in the summer and it tends to lower my dew point to some degree, but I still do not know exactly what it is.

I know from my testing how low water temperature in the reservoir I can go without condensation building on the outside of the CPU water block, but you'll have to run your own tests.



Where do you live to have these dew point numbers?

If those numbers are accurate you will be able to run a water coolant temperature much lower than I can, the CPU sweating is not the problem it is the copper base of the CPU water block over the size of the contact footprint of the CPUs heat spreader that is the main problem,

All my hoses or tubing is insulated so that's not the problem, like I said you'll have to run you own operational tests to see what happens.



No I have not run any flow tests past setting pump operation speed and checking for flow rate cooling improvements of the CPU but seeing as how the CPU water block itself is so efficient at transferring the cold to the CPU I just run on level 5.

The main reason being, it allows faster cold build up of the insulated reservoir, your insulated reservoir is your cold storage and where you build up the cold, from my experience the faster the flow rate the faster the cold build up in the reservoir.

Edit: Try not to over think everything you really do not need to know every exact detail of the chilled water cooling, (Unless of course you're OCD), all you really need to know is that it works!

After you conduct your own operational running tests under CPU (idle and load), you'll know what flow rates are best for your setup, and exactly how low of water temperature you can safely run and always be above and condensation problems.

You're in the learning phase now and no matter how much you think you've got it figured out, you will make changes as you learn what the cooling can do for you, and with your dew point range you'll get even further overclocking than I have.

I am of course assuming you're shooting for high CPU overclock goals in the 5+ghz range, if not?, there is absolutely no justifiable reason to even run this cooling. 😉

 
Sorry! I had this thread open in a window on my laptop at work and didn't get notified you'd responded!

- I live in western Colorado, high desert country.
- It's not that I *need* to know everything... it's just interesting stuff so I dig and dig until I get bored with the topic 🙂
- No clue how high I'm going to take my overclock! I'm sitting on 4.4 @ 1.28v and my CPU runs around 60c while gaming, but I was unable to comfortably hit 4.5 as the voltage increase made my temps bump up by almost 10c under load. I figure I'll get a couple of bench tests before I tear the loop apart, and a couple with the new cooling system to see the (if any) improvement, once I know the difference, then I'll start seeing how high of an overclock I can get.
- There is absolutely no justifiable reason to build this loop period, my computer at 4.4g will beat out 90% of the systems out there that are at 5g, this 5820k just kicks butt. But, I think it's really cool, I'm having a blast, I'm learning a ton, and why wouldn't I do something I think is cool that also lets me crank up an already kick ass computer :-D
- Thanks again for posting up your build, it was the spark that set me off on this road and I'm loving the hell out of the trip!

Oh, I'll never learn it all, I'm not that smart 🙂 I just like reading papers that don't make sense, then later on when you're working on something a light bulb goes off and you go "OOOOOHHHHHHH".

So, my TEC's came in last night, I managed to get one mount fabbed up today and being the impatient sort, I hooked up a TEC to the heat sink and water block. No load, but with NO fan the heat sink never really climbed above 32c°. Water on the top of the water block froze in a couple seconds (Unfortunately I'm losing quite a bit of efficiency me thinks, my peltier does not cover the entire block, and my heat sink does not cover the entire peltier ~.^ )

I did find a 40x35mm peltier that is 196w Qmax at 13.7v with 23.1 amps...... I bought one to compare to these dinky little guys (136w Qmax, 15.4v@15amps) I have now. Pretty sure that the Cooler Master Evo's aren't going to cut it with that bad boy, but we'll see before I go spending even more money on a bigger cooler. Might have to find a cooler capable of almost 400w!!!! Holy crap on a stick that's a TON of heat!
 


The peltier not covering the entire water block for cold side pickup is not a problem at all, but the heat sink to cool the hot side not fully covering the peltier is bad, really BAD!

You need a cold plate to make sure the hot side of the peltier is fully covered, or the uncovered part of the peltier will burn up!

A cold plate is the terminology of the peltier coverage plate you need but you will be using it as a hot plate!

These are cold plates!



You're cooling the hot side with a CMH 212 EVO, "Good luck with that!"

You do have a lot to learn, I just hope you don't destroy something in your learning, there are facts you cannot get around, first off you have to have a lot more cooling capability than the CPU is outputting overclocked.

You have got to keep the hot side of the peltier cool enough and fully covered, or else you'll destroy it.

Running any air cooled heat sink on a peltiers hot side without a cooling fan running is literally hardware suicide!

But after all it is your hardware!

In all my experimentation I have not destroyed any hardware, but I fully researched everything, before I did it!

I did go against some advice of those telling me I could not cool the hot side with an air cooling heat sink, but even I did not attempt it passively.

 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom