Peltier Porn (cheap fun!) - Now a custom-built case project, loaded with TECs!

Discussion in 'Extreme and Water Cooling' started by IMtech, Sep 16, 2011.

  1. IMtech

    IMtech (Banned or Deleted)

    Joined:
    Jul 1, 2011
    Messages:
    1,430
    Location:
    Cheeseland, France
    I thought I may get people attention's with a title like that! :lol: ;)

    Here is continuation of my journey from my other thread "Building a series of standalone Cooling & TEC Controllers (Relay & PWM Based)".

    In this thread, I will post the testings result of my various peltier designs for my project TECKOOL, & also some fun projects that I will come up with along the way... :Paranoid: :lol:

    This will only include the cooling side of things, as the electronics to drive the TECs is discussed in the other thread.

    I will start with the puppies:


    Click to view full size!


    This is 10x 150w TEC modules... quite tasty! :weirdo:


    Click to view full size!


    This is one of the 3 PSU I have for testing TECs... they all are rated at 180w@12v, so plenty enough to drive 2 peltiers at 60% power or so! :thumbup:


    Click to view full size!


    The card I will use today for my testing. A simple MSI 9500GT 512MB DDR3. Was quite a nifty card when I got it back then when new, but it's now definitely not able to play the latest games... so why not try sticking a peltier on it? :leet:


    Click to view full size!


    A heatsink from an old 4850... funny that the holes line up perfectly with the ones on the 9500GT's PCB! :thumbup:
    All I had to do was to make the holes on the PCB a bit bigger, which I did with a 2.5mm drill bit and a bit of jiggly around while drilling. This definitely did the trick, and I was able to get M3 screws through.
    The other problem was on the heaqtsing itself. Once again, I pulled out the drill and 2.5mm brill bit, made the 4 holes bigger, and then tapped them with a M3 tap. This now made that heatsink work with the 9500GT! :thumbup: :)


    Click to view full size!


    The TEC fits perfectly on the heatsink! :thumbup:


    Click to view full size!


    Ghetto thermal paste application! :lol:


    Click to view full size!


    TEC is now affixed to the heatsink...


    Click to view full size!


    Front shot of the final assembly :)


    Click to view full size!


    Back shot. I was able to use the backplate from the ATI card, and it fits perfectly! :lol: :leet:


    Click to view full size!


    Bit of a blurry shot of the TEC and heatsink sandwich


    Click to view full size!


    The final assembly, ready for testing! :thumbup:


    Click to view full size!


    Overview of my temporary "test-bench"! :lol: :p


    Click to view full size!


    Close-up of the 9500GT and TEC PSU


    Click to view full size!


    34671 3DMark 2001SE - 715 Core 1558 Shaders 490 Memory - 3.8Ghz AMD X6 CPU


    Overall, temperature still bump the 100c mark, due to the TEC not being PWMed at the moment, and only air-cooled, but I will try this again next week when I hopefully get my first controller prototype finished! :thumbup: :)

    All in all, was a fun experiment to try and have my first play with peltiers. :)
     
    Last edited: Nov 2, 2011
  2. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,349
    Location:
    St Clair 2759
  3. OP
    OP
    IMtech

    IMtech (Banned or Deleted)

    Joined:
    Jul 1, 2011
    Messages:
    1,430
    Location:
    Cheeseland, France
    Wow! This is very similar to my controller, except it is definitely not as powerful or polyvalent as mine, and does not have as many buttons etc... but good find! Definitely shows there is interest in that kind of controllers! :thumbup: :)

    The price is pretty close to what I'm looking to price mine. I'm looking to have a $80 pricepoint or so for the controller, which can itself control 900w worth of TECs and 150w worth of Fans on the one board... ;) :D

    I am still anxiously waiting for that UPS delivery... come on UPS I want my Waterblocks already! :lol: :leet:
     
  4. Hive

    Hive Member

    Joined:
    Jul 8, 2010
    Messages:
    5,252
    Location:
    ( ͡° ͜ʖ ͡°)
    Nice to know TEC's are coming back, they do have some potential in them :lol:.

    for your next run, grab a 1-2MM thick sheet of copper plate about the size of the peltier and place it with thermal grease between the GPU and cold plate of the peltier, this will give you much greater heat transfer potential to the core.
     
    Last edited: Sep 19, 2011
  5. SiriusDragon

    SiriusDragon Member

    Joined:
    Feb 10, 2009
    Messages:
    2,353
    Location:
    Sydney
    I've seen some of the boutique PC retailers in the UK use TECs in their systems. eg. the Scan Jellyfish (http://www.scan.co.uk/shops/dreampc/jellyfish).

    The main problem is that they still didn't perform that well from what I read and sucked tremendous amounts of power in the process.

    But if the OP can overcome those weaknesses then that's fantastic :thumbup:
     
  6. OP
    OP
    IMtech

    IMtech (Banned or Deleted)

    Joined:
    Jul 1, 2011
    Messages:
    1,430
    Location:
    Cheeseland, France
    I had the same idea, but don't have any copper here atm. This is definitely on the cards if I have a play again with direct TEC GPU cooling. All my designs from now on with involve the heatexchanger, when the parts ever get here... :lol: ;)

    Quite a nice case in that link. Shame they only achieved 9c difference or so... :(

    But then we'll see how I fare with a 6x TECs block... according to my calculations I could take the CPU at nearly -2c at full TEC power. 40% power would yield a temp of 20c for a 150w load! :)

    The controller development board finally arrived today. I am posting details about it now in the other build thread. :)

    What I am seeing with TECs tho, is that they definitely perform a lot better when PWMed! :)

    Edit: Forgot the obligatory Peltier Porn pic... :p


    Click to view full size!
     
  7. SiriusDragon

    SiriusDragon Member

    Joined:
    Feb 10, 2009
    Messages:
    2,353
    Location:
    Sydney
    Argh! This forum has children browsing it, they don't want to see such smut! :upset:

    :p

    I've heard that once TECs reach a certain heat load, they fail horribly. ie. you need to match the number and power of the TECs exactly to the heat load that they'll be under.

    I have a TEC or 2 sitting around at home and I'm not sure what todo with them.. I've considered putting them on a Q6600 or similar but I'm not sure if they're powerful enough for that!
     
  8. OP
    OP
    IMtech

    IMtech (Banned or Deleted)

    Joined:
    Jul 1, 2011
    Messages:
    1,430
    Location:
    Cheeseland, France
    Man... I didn't think of the childrens... :upset: If anything, you can clearly see they are on a leash and are toilet trained... :lol: :p

    TECs can fail, if you allow the delta between the hot and cold plate to go over 65-70c... other factors can be excessive temperature on the hot side, which cracks and makes the TECs useless... :(

    My controller once again solves all those issues, which is gonna be put to the test this week end when I'm hopeful I will have some code down by! ;)


    Oh, and right now I feel like a kid in a candy shop... the water blocks have finally made it! :) :) :)

    There is a couple for my use, and the rest will go to the shop. I will use 3 of the heat-exchanger plates for my own CPU cooler, so I may have a 900w cooler available very soon if anyone wants to test/review it... :thumbup:

    Obligatory WB Porn: :leet:


    Click to view full size!

    Heat Exchanger blocks. Each one allows to fit 3x TECs lengthwise, and I will use 2 or 3 per assembly, depending if it needs 450w or 900w worth of TECs! ;)


    Click to view full size!

    CPU and GPU blocks. Enough to test with I think! ;) :leet:


    Click to view full size!

    GPU Block side view


    Click to view full size!

    CPU Block side view
     
  9. SiriusDragon

    SiriusDragon Member

    Joined:
    Feb 10, 2009
    Messages:
    2,353
    Location:
    Sydney
    I suppose by fail I meant not cool as effectively. eg. If you have a 150w TEC and a 200w cpu, the TEC may make things worse or not make any sort of a difference in temps compared to just a normal water loop.

    eg. The Coolermaster V10 has a TEC but it's quite low power and supposedly makes zero difference to how well the V10 works.

    I dunno, it's been a while since I read all this so maybe stuff has changed. :confused: I read on XtremeSystems that this was why they weren't as popular any more - the heat load of the components they were cooling outstripped the TEC's ability to effectively remove the heat so in the end you were no better off.
     
  10. flu!d

    flu!d Ubuntu Mate 16.04 LTS

    Joined:
    Jun 27, 2001
    Messages:
    13,252
    I remember TEC's from the original Athlon/P3 days....

    Melted PCB's and CPU/GPU's due to the fact that someone forgot the fact that failure to remove the immense heat from the hot side of the peltier results in the cold side being....

    ....Well, not so cold anymore....

    Get it right though and the results were quite impressive.

    Gonna watch this thread, could be exciting!
     
  11. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,349
    Location:
    St Clair 2759
    IMtech - check this out.... ;)

    This isnt to detract from your efforts at all, its for ideas and inspiration. I particularly like this part:

    http://www.arqtik.com/products/24tcon.html

    :thumbup:
     
    Last edited: Sep 19, 2011
  12. OP
    OP
    IMtech

    IMtech (Banned or Deleted)

    Joined:
    Jul 1, 2011
    Messages:
    1,430
    Location:
    Cheeseland, France
    That. Plus if the hot side is not cooled properly, it will crack and totally fail.

    The V10 is basically a joke, as it is not really a great performer at all compared to the power it uses... :thumbdn:

    TECs are good, but only if used properly... :thumbup: :leet:


    Thanks man! :)

    You're right tho. The better the cooling of the hot side, the colder the cold side will be... :)

    Yeah I saw that one from the link you posted in the other thread. :thumbup:

    There is so many differences in his and my design, I would not know where to start. But if my design does end up being successful, it will be superior, due to the ability of PWM-controlling up to a 900w TEC load, along with 150w worth of fan on a seperate fan channel. It will also feature a LCD screen and 4 physical buttons for navigating menus etc...

    So it is definitely a similar design, as in it controls the TECs according to temp, but this is pretty much where the similarities end... ;) :)

    Look, at the end of the day, if having 2 separate persons/businesses putting effort into TEC revival, creating a proper controller and marketing it out there, this can only be good for the end user. More choice, at hopefully a great price, due to competition! :thumbup: :leet:

    I have also built the 3xTEC heat exchanger tonight, as a quick test. I do not have a second pump for the cold loop, which should be here soon, but the design definitely worked, and took the hot side to over 60c before I shut it down, while the cold side was definitely getting colder and colder as well. :thumbup:

    I will have more pictures tomorrow, as I'm really buggered now and need to get some sleep, as I need to be up at 5am... :tired: :Paranoid:
     
    Last edited: Sep 24, 2011
  13. OP
    OP
    IMtech

    IMtech (Banned or Deleted)

    Joined:
    Jul 1, 2011
    Messages:
    1,430
    Location:
    Cheeseland, France
    Alright... here is what a few people have been waiting for... the Peltier-based Heat Exchanger assembly! :leet:

    First, I would like to point out that I have done testings today with the TECs, and blew 2 out of my 3 12V Power supplies, so the tests here are with the 6 TECs running off the one 150w PSU.
    This is the equivalent of running the TECs at 1/6 of their 12V potentials, due to the limited amperage that went to them in today's test.
    I am looking for a big PC PSU now, most likely 1kw single-railed, to do further testings.

    So all the results I will post tonight are only about 15% representative of what the system can potentially do at 100% power, which as you will see will be quite frightening! :shock: :leet:

    Here is the first block getting a TEC layer:


    Click to view full size!



    Click to view full size!



    Click to view full size!


    As you can see, the thermal paste alone can hold it... ;)


    Click to view full size!


    First sandwich is done...


    Click to view full size!



    Click to view full size!


    Here is the whole double TEC sandwich :)


    Click to view full size!


    And here it is with the prototype mounting:


    Click to view full size!


    With the hot-sides & return tubings fitted:


    Click to view full size!


    Shoot of the back side:


    Click to view full size!


    Overview of my temporary test setup. I am still missing a pump, so I cannot do real-world CPU and GPU tests yet. But I should have a new pump by the end of the week! :thumbup: :)


    Click to view full size!


    Here is a single 120mm rad, which comes from a Ford Laser heater box, with 2 fans in push-pull, handling cooling for the two hot sides of the assembly.


    Click to view full size!


    We have condensation on the middle cold side! :leet:


    Click to view full size!


    I attached a sensor to the side of it for better temp monitoring:


    Click to view full size!


    After about 20mins or so, to allow for the whole system to stabilise, this is the temps that showed.
    Left is the hot side temperature for both hot plates in series, straight from the rad to the pump section.
    The top right is the sensor I attached to the side, and the lower right is of a loose sensor inside the cold-side's barb, which was unfortunately not making contact with the cold-side itself, so it was pretty useless...

    2.6c at only 12V@150w is pretty good in my books! :)


    Click to view full size!


    Unfortunately shortly after this pic the third PSU started to buzz, and I turned it off in fear of blowing my last proper 12V PSU... :( :Pirate:

    So this definitely confirms my initial calculations, in that my system definitely works! :thumbup: :)

    Stay tuned for more updates by the week-end, including real-world CPU and hopefully GPU tests results! :)
     
    Last edited: Sep 20, 2011
  14. Hive

    Hive Member

    Joined:
    Jul 8, 2010
    Messages:
    5,252
    Location:
    ( ͡° ͜ʖ ͡°)
    fucking peltiers... HOW DO THEY WORK :lol:
     
  15. OP
    OP
    IMtech

    IMtech (Banned or Deleted)

    Joined:
    Jul 1, 2011
    Messages:
    1,430
    Location:
    Cheeseland, France
    ROFL! :lol:
     
  16. SiriusDragon

    SiriusDragon Member

    Joined:
    Feb 10, 2009
    Messages:
    2,353
    Location:
    Sydney
    AX1200 :thumbup: Single rail at 100 amps!

    Those results are really really nice. Making me wonder if I made the right choice with an aquarium chiller as opposed to a heat exchanger setup with some TECs and quiet fans on radiators :Paranoid: Cause with full power... sub zero should be possible maybe?

    Anyway, good stuff :thumbup:
     
  17. tuchan

    tuchan Member

    Joined:
    Aug 23, 2006
    Messages:
    615
    Location:
    Adelaide
    Having had plenty of success with the larger 437 watt peltiers over the years, this thread is of much interest.

    Keep up the great work:D:thumbup:
     
  18. OP
    OP
    IMtech

    IMtech (Banned or Deleted)

    Joined:
    Jul 1, 2011
    Messages:
    1,430
    Location:
    Cheeseland, France
    Thanks guys! :)

    Sub-zero looks like it may be attainable with a 150-200w heat load per heat exchanger... but I'll have to confirm that with more testing very very soon! :)
     
    Last edited: Sep 21, 2011
  19. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,349
    Location:
    St Clair 2759
    What brand were the now dead power supplies? Have you looked for something in the meanwell range? Also when you scale a pelt to say 60% load you reduce the current supplied to it not the voltage right? If you dont use the pc psu and use an auxillary psu then you can use higher voltages, such as the 24v one I linked, or a 36v one I was looking at , I found a 36v 27a power supply too but its a bit dear to buy for an experiment ...
     
  20. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,349
    Location:
    St Clair 2759
    Jack from Arktiq said his direct to die unit will cool an 1155 cpu to sub ambient up to 1.4v , so if you have a good cpu you might get 5ghz within its limits, so now a dual pelt block is the next step.

    Will your array controller throttle down with light load so its not guzzling power when your not gaming or benching?

    :)
     

Share This Page

Advertisement: