Post Rad Chiller Concept

[systemviper]

actually i was going to keep the hot side and cold side seperate.

so hotside

DDC-3.2 or D-tek DB-1 pump IF it tests good.

Pump -> hotside -> MCR320 -> T-line -> Pump

Cold side would be inline right after the radiator first, and then alone.


im not going to recycle the water, would make no sense to do that.

it will be fun to see what kind of temps that chiller will make at the CPU when it's not tied to the hot side,
you should have some great numbers/temps.

you lucky *^*&^& best of luck with that CRazy CHILLER.

Keep us posted,

[Blauhung]

actually i was going to keep the hot side and cold side seperate.

so hotside

DDC-3.2 or D-tek DB-1 pump IF it tests good.

Pump -> hotside -> MCR320 -> T-line -> Pump

Cold side would be inline right after the radiator first, and then alone.


im not going to recycle the water, would make no sense to do that.

hehe, yup, you'd basically just be adding the heat pulled out back to the loop, plus the heat created due to the low efficiency.

I just started following your thread as I'm thinking about doing something similar, only I'm wanting to hit sub ambient temps so the cold loop will be completely insulated from the air with no radiator. It almost seems as if I got the idea to do it right as a bunch of others on the forums here did too only I'm being all lazy about getting started.

I think you might find that with the ammount of wattage you have on their you might be better off adding the rad on the cool side to the hot side of the loop and keeping that cooler. I haven't read through the whole thread, but It looks like your going to be putting upwards of 400W worth of TEC's on there. That's more then enough to pull heat off most CPU/NB/VGA setups.

I didn't see if the math was fully fleshed out, but basically your onto a system that is modeled by a few sets of equations. The solution comes in finding the steady state temperature of the loop that balances the energy in and out of the two loops.

Cold loop
Qout=Qin
Qout=Qtec+Qrad
Qin=Qcpu+Qpump

Qtec=f(∆T,Qmax,∆Tmax,Vtec)
Qrad=f(Urad,∆T)
Qcpu=f(freqcpu,Vcore)
Qpump=f(Wpump,Epump)
Urad=overall heat transfer coefficient of the rad
Epump=efficiency of the pump


Hot loop
Qout=Qin
Qout=Qrad
Qin=Qtec+(1-Etec)Qtec+Qpump

Qrad=f(Urad,∆T)
Urad=overall heat transfer coefficient of the rad
Qtec=f(∆T,Qmax,∆Tmax,Vtec)
Etec=efficiency of the tec
Qpump=f(Wpump,Epump)
Epump=efficiency of the pump

If you were interested, I'm sure I could dig up all the individual equations, but most of this stuff at this point would be far better to calculate experimentally.

But, after laying all this out, you would come up with 2 equations for the energy balance on each of the loops which you could solve for the temp of the inner loop at steady state.

Edit: also keep in mind, that with the volumes and flow rates you will most likely be dealing with, you will only be seeing on the order of 1 or less degrees of temperature difference on the inlet and outlet of your tec unit. For any decent water cooling flow rates it's a very easy assumption to make that the change in temperature in the loop at steady state is almost 0.

[NaeKuh]

guys got the stuff back from martin.

He couldnt test it, he ran out of time and i didnt want to push it on him.

He gave me A LOT more then i could ask for by building such massive blocks. And btw, there are massive. Heavy as heck too.

But, the radiator is not a safty net as i wished for. Its a Football Line Backer blocking me from dropping coolant levels.

So i am going to try another route. This will take the radiator out completely on the cold side, and utilize a reservoir.

I was thinkn of an external box to fit this. The unit is kinda massive. I honestly didnt think it was that big. Anyhow this was how i was gonna do this.

The internals:

EK-400 res. Or the largest possible, i think its the 400.
Im gonna use the top as an inlet with a Y, i'll tell you in a sec on why.
The bottom has plugs as barbs. Botom and 2 side on the bottom. I was thinkn of temp probe at the way bottom.

Now i was thinkn of paralelling the setup. So each of the outlets at the bottom will goto a pump.

Loop1: T-line -> DB-1 or DDC->MCR320 ->Hotside >pump Very small loop, with Tline.

Loop 2: EK-Res -> pump-> Coldside -> EK.

Loop 2 will continuously circulate the water to make it colder. I will probably have to insulate the EK, and also use neoprene tubing.

The second bottom barb will go directly out to the system. This way the water will be chilled, and the system will be getting cold water.

The top will use a Y as a return for both loop 2 and the path that goes out to the blocks. Do you kinda get the concept

I figure loop2 will keep the water cold, also since your calculations show it takes a grip of energy for the water temp to go up, i figure the chiller has a smaller loop so therefore can do more work then the larger one on the same paralell setup.

I intend on having 2 psu's in there. A Pico PSU, for the pumps and fans, and the meanwell for the TEC's.

So whatcha think?

[evil-98]

sounds good

so the y splits in to two paths meaning the cold side will continue to circulate? so you kinda need like a reverse turbo timer, got to cycle the system for a while before starting up

hum still kinda confused pics would help

[scifikg]

sounds like a plan

I'd like to see thermocouple readings on the hot and cold plates to see how they compare to detaT max for the TECs.

Do you have a fluke or similar device?

[NaeKuh]

sounds good

so the y splits in to two paths meaning the cold side will continue to circulate? so you kinda need like a reverse turbo timer, got to cycle the system for a while before starting up

hum still kinda confused pics would help

correct.

The cold unit would be on its own setup.

I'll have 2 meanwell's inside or 1 meanwell, and 1 pico PSU. I dont like dragging cables from the pc -> Back..

The block loop, the pump will be internal on the computer side.

So yeah, the Y at the top is a double return. 1 for the cold loop, and also for the main block loop.

[Pete]

Erm im confused, do me a paint drawing please. Can be mega basic

Just do 2 loops, hot & cold, hot nice big rad and decent pump to get rid of the heat. Cold use a Tline only

[Cupcake]

Updates ?

[biglipzit]

guys got the stuff back from martin.

He couldnt test it, he ran out of time and i didnt want to push it on him.

He gave me A LOT more then i could ask for by building such massive blocks. And btw, there are massive. Heavy as heck too.

But, the radiator is not a safty net as i wished for. Its a Football Line Backer blocking me from dropping coolant levels.

So i am going to try another route. This will take the radiator out completely on the cold side, and utilize a reservoir.

I was thinkn of an external box to fit this. The unit is kinda massive. I honestly didnt think it was that big. Anyhow this was how i was gonna do this.

The internals:

EK-400 res. Or the largest possible, i think its the 400.
Im gonna use the top as an inlet with a Y, i'll tell you in a sec on why.
The bottom has plugs as barbs. Botom and 2 side on the bottom. I was thinkn of temp probe at the way bottom.

Now i was thinkn of paralelling the setup. So each of the outlets at the bottom will goto a pump.

Loop1: T-line -> DB-1 or DDC->MCR320 ->Hotside >pump Very small loop, with Tline.

Loop 2: EK-Res -> pump-> Coldside -> EK.

Loop 2 will continuously circulate the water to make it colder. I will probably have to insulate the EK, and also use neoprene tubing.

The second bottom barb will go directly out to the system. This way the water will be chilled, and the system will be getting cold water.

The top will use a Y as a return for both loop 2 and the path that goes out to the blocks. Do you kinda get the concept

I figure loop2 will keep the water cold, also since your calculations show it takes a grip of energy for the water temp to go up, i figure the chiller has a smaller loop so therefore can do more work then the larger one on the same paralell setup.

I intend on having 2 psu's in there. A Pico PSU, for the pumps and fans, and the meanwell for the TEC's.

So whatcha think?

Sounds overly complexed. Why don't you use a large rad for the hot side and then just loop the cold side like you said using ethanol or something and insulating and then just seeing how that goes...

[Scarlet Infidel]

I'm really getting frustrated that I can't get on with my chiller now (Im at university with no tools). My design is nothing like this but the principle is still the same.

I'm not one to colour water, but with this I'd be sorely tempted to use blue water on the cold side and red on the hot. Much like stripes make a car go faster, my calculations suggest this will increase the temperature difference :p

[n00b 0f l337]

Hey PM me again mate, I might have some time in next weeks to test for you if you have all the equipment.

[systemviper]

All that work, hmmmmmm
Time to let her rip....

I know how it is, it takes so much time to set this stuff up and tune it to get the most from it...Just love that big shoe design, should pull some major amps and put out some chilly willy

[NaeKuh]

Hey PM me again mate, I might have some time in next weeks to test for you if you have all the equipment.

Thanks nol, PM Sent.

I guess you can get to the bottom of things on which way will be the best setup for this unit.

[n00b 0f l337]

PM replied.

[Martinm210]

I just had to switch gears over the the kitchen project for a while, but glad this is still moving forward.

Not making any more, but maybe with some successful testing, we can convince one of the CNC mill guys like Iandh to start fabricating them..

[pak]

WTB update.

--pak

[rogueagent6]

WTB update.

--pak

I'll raise you.

[moditir]

actually i was going to keep the hot side and cold side seperate.

that smart of you, otherwise you system would actually be getting hotter because TECs create more heat then they do cold (even though cold is the absence of heat you know what i mean).

[ArtosDracon]

I like the tripple loop set-up; especially if you run the chiller loop for a while before kicking in the main loop both before the system fires up. Are you going to run all 5 87W pelts downvolted to 60W or three 87W as origonally planned? Keep one thing in mind when setting up your reservoir, the bigger the res the longer it'll take for pulldown but the lower your load will be on the pelts once the system is running and the smoother any temp curves will be.

[n00b 0f l337]

Any updates on when its hitting my hands?

[antho]

Nice project

Any update ???

[NaeKuh]

Any updates on when its hitting my hands?

ack sorry nol....

Lemme pm you...

[Clue69Less]

I'm still getting parts together for my rig. I need to find a good low restriction two-way ball valve. The plan is to have the quiet, stock clock mode to run the CPU loop through a 3x12 rad. Change the position of the ball valve and the water gets routed through a dual-block TEC heat exchanger sort of like the one shown in this thread. I've got four 168 watt TECs and two 320 watt Mean Wells and another 3x12 rad to cool the hot side of the TEc block. It will be interesting to see what kind of temps the TECs will provide at load.

[ArtosDracon]

I'm still getting parts together for my rig. I need to find a good low restriction two-way ball valve. The plan is to have the quiet, stock clock mode to run the CPU loop through a 3x12 rad. Change the position of the ball valve and the water gets routed through a dual-block TEC heat exchanger sort of like the one shown in this thread. I've got four 168 watt TECs and two 320 watt Mean Wells and another 3x12 rad to cool the hot side of the TEc block. It will be interesting to see what kind of temps the TECs will provide at load.

you actually need a wye diverting valve. Page 399 of the mcmaster catalog.

[Clue69Less]

you actually need a wye diverting valve. Page 399 of the mcmaster catalog.

I'll stick with my simple ball valve. It has 1/4" BPT and no restrictions when open. And it was cheap - 9 bucks. Sort of like McMaster-Carr 4093T23 but 1/4 BPT.