Peltier Waterblock Design

[RadeonHDx]

So I was hoping that I could get some input from some of you guys on my waterblock design that I making to better utilize the power of my peltier and make it that bit more efficient. I'm tired so this thread will be quite quick and may have some typos.

So just some pictures of some Sketchup renders that I have done



The channels of the waterblock section are 1mm wide and 1mm high with a 1mm wall between each. Im not sure there is enough turbulence with just channels but I'll get to that later.



The two copper plates are 5mm thick and the plexi glass that I am using is 10mm which arrived today although I still need to find some copper with 5mm of thickness.



This is the top of the plexi block that will have 2 chambers, 1 for in and one for out. the water will go in one side, be pushed under the un milled out piece of plexi and then come out the other side after being pushed under the fins and go out the other tube. It is shown here in foxrena's block design http://www.overclock.net/g/i/117112...770k-on-cascade-tec-block/sort/display_order/.

Foxrena's final product

My render of the block top




Any opinions are welcome as long as they aren't to tell me that using a peltier is dumb . Im sure youll be able to help me but if you think this design will be adequate then I can get it cut in the next few weeks .

Thanks, Joe.

 

[S_I_N]

1st cooling with pelts is dumb ok had to say that cause ya asked us not to lol.
2nd i dont think that will work unless the pelt is just under the inlet side. Meaning your only cooling half the pelt with that design.

Lastly ok I have done just about every kind of cooling there is with the exception of dry ice and the likes. (done pelts chilled water phase change and super chilled water. Of then all the only ones I would attempt again would be super chilled water and pahse change. Cause the benefits are alot better and more noticeable. Pelts are just not worth the effort cost and power usage versus the end results. Bear in mind when I did pelts it was all the rage for super cooling and imho it gained me squat outside of the high costs and frustration that I couldnt cool the hot side enough to get any better than good air or water.

 

[RadeonHDx]

1st cooling with pelts is dumb ok had to say that cause ya asked us not to lol.
2nd i dont think that will work unless the pelt is just under the inlet side. Meaning your only cooling half the pelt with that design.

Lastly ok I have done just about every kind of cooling there is with the exception of dry ice and the likes. (done pelts chilled water phase change and super chilled water. Of then all the only ones I would attempt again would be super chilled water and pahse change. Cause the benefits are alot better and more noticeable. Pelts are just not worth the effort cost and power usage versus the end results. Bear in mind when I did pelts it was all the rage for super cooling and imho it gained me squat outside of the high costs and frustration that I couldnt cool the hot side enough to get any better than good air or water.

Would it be possible for you to go more in depth in regards to the super water chiller? I've thought about the possibility of phasechange but I would never be able to justify wi with the price and the huge amount of noise . If you mean a water chiller then I've looked at them too but I always thought that they were just as power hungry as a pelt.

 

[S_I_N]

basically it was a window shaker unit that was regassed for lower temps and modified to chill liquid ( I should of said liquid earlier as water would freeze in this bad boy) iirc the liquid temps was -42 to -45 and my cpu was at -35. Yes this method uses alot of power as well as being a tad noisy as well. But overall performance was superb. It requires a ton of condensation proofing as you not only have to proof the board for it but every inch of tubing used.

heres a couple pics of my old unit.

 

[RadeonHDx]

basically it was a window shaker unit that was regassed for lower temps and modified to chill liquid ( I should of said liquid earlier as water would freeze in this bad boy) iirc the liquid temps was -42 to -45 and my cpu was at -35. Yes this method uses alot of power as well as being a tad noisy as well. But overall performance was superb. It requires a ton of condensation proofing as you not only have to proof the board for it but every inch of tubing used.

heres a couple pics of my old unit.

I won't lie, this peltier concept is more about getting my rig to stand out and really make it one of the few builds that actually uses direct peltier cooling. The power really isn't an issue as the power isn't expensive and I think at the cost of water chillers that I would be saving money getting the pelt working, even with the high power requirements.

As you've cooled with pelts before, could you tell me what equipment you used and how you had it setup with some pictures possibly? I'm really interested in getting this to work and fabrication of parts really is no issue as my dad has a CNC and I can use it for pretty much what I want.

Also, do you think that this design for the waterblock is any good? As in, will the 65mm to 40mm adapter to concentrate the cold actually help or just be a waste of copper?

All this help is really appreciated and I'm glad I have someone as experienced as you to help .

 

[S_I_N]

the main issue with pelts is the point of diminishing return is alot lowerthan other methods. You can only cool the hot side so much to make the cold side effective. I didnt take any pics of the pelt era of my cooling years unfortunately. I do however recall alot of the info in my miniscule brain lol. I used meanwell psu's to power my pelts and commercially available (at the time) water blocks for the pelts themselves. for your pelt block to work you'll want the inlet in the middle of the block versus one side. Think of it as more of an inward to outward spiral versus your diagram of straight line channels. Your diagram suggests it coming in on one side then out the other side, this will make for uneven cooling on the pelt and in some cases might cause it to crack or fail prematurely due to uneven cooling. Not to mention at minimum you will want to konform coat your mobo and apply dielectric grease in the socket just in case you are able to reach sub ambient temps, or those instances where your idle alot and hit sub temps. You deffo wont have to worry about wrapping the lines as I did with the chiller because your water wont (most likely) hit sub ambient temps. I'm betting you will feel alot of heat coming off your rads to the point you'll be able to allow the rig to act as a space heater during the winter months. Summer months may be a challenge lol.

 

[RadeonHDx]

the main issue with pelts is the point of diminishing return is alot lowerthan other methods. You can only cool the hot side so much to make the cold side effective. I didnt take any pics of the pelt era of my cooling years unfortunately. I do however recall alot of the info in my miniscule brain lol. I used meanwell psu's to power my pelts and commercially available (at the time) water blocks for the pelts themselves. for your pelt block to work you'll want the inlet in the middle of the block versus one side. Think of it as more of an inward to outward spiral versus your diagram of straight line channels. Your diagram suggests it coming in on one side then out the other side, this will make for uneven cooling on the pelt and in some cases might cause it to crack or fail prematurely due to uneven cooling. Not to mention at minimum you will want to konform coat your mobo and apply dielectric grease in the socket just in case you are able to reach sub ambient temps, or those instances where your idle alot and hit sub temps. You deffo wont have to worry about wrapping the lines as I did with the chiller because your water wont (most likely) hit sub ambient temps. I'm betting you will feel alot of heat coming off your rads to the point you'll be able to allow the rig to act as a space heater during the winter months. Summer months may be a challenge lol.

Thanks for the info man. I see what you're saying about the block now and I'll knock out some new renders tomorrow (although I don't know how I'm going to make it look nice). I understand that you get very early diminishing returns but it's the only truly silent (that I know of that isn't chemical) way to get sub ambient temps nut that also doesn't have to have an insulated everything (I have actually already insulated my board).

Would you say that I should invest anymore time and effort into the build? This is a project that I'm pretty set on getting working properly and I really have tried to think of as many things as I can to get this to work. All I am missing at the moment to finialise it all is the power supply which I can get for sub £20.

I have spent probably close to £50 on the project at the moment and I don't want it to have been in vein which is the reason that I want this to work so badly.

Thanks for the help again, Joe.

 

[S_I_N]

if you want true silence the pelts aint for you as you'll have to either run a ton of rads for low fans but in my honest opinion your throwing good money after bad. why not just put in alot of rads with low fans and whalla low temps and quiet.

 

[RadeonHDx]

if you want true silence the pelts aint for you as you'll have to either run a ton of rads for low fans but in my honest opinion your throwing good money after bad. why not just put in alot of rads with low fans and whalla low temps and quiet.

After like a month of planning I think this has made me see what I should really be doing. I think I'm going to cut my losses and leave all the whole pelt idea. I now realize that I could have just bought another rad, or saved up for my GPU upgrade. Thanks for helping man, I'm so grateful for the help I get from you lot and the experience you have SIN . I really wish there was a quiet way to do sub zero that was cost effective (at least in the short term).

After all, I can always make an effective drink cooler .