Hybrid TEC / Watercooled cooling

[BiTBiTE]

Hey guys!!

Just wanted to fill you in on a project I'm working on. Feel free to chime in with anything that comes to mind. 

Ive been experimenting the past day or so with the TEC modules I finally received; the first was a single 12706 then two 12715's. I'm really intrigued with them. I've sealed a board with an old Q6600 I had laying around and was going to mess around with one directly on the CPU, but then I wondered how it would do in conjunction with my watercooling loop in my main rig. 

I know about the heat dissipation issues of the TECS and am currently using a hyper 212 evo which it's doing the trick. I know (first hand) of the condensation issues,  but with the way it's going to be implemented it puts the TECS on the outside of the case with the evos outside the back. I have an all copper tube build which is modular so it wouldn't take much work to change out one pipe. 

It wouldn't be always powered, rather only when running benchmarks or games and such. 

 

I'm placing the peltiers after my res, before it enters the gpu then cpu. (My cpu/gpu are in a parellel config, which was more for aesthetics but I did try both parallel and series config and seen marginal difference at best.)

The point of the TECS is not to be the main cooling component, rather chilling the water before entering components. As the cold water enters the gpu it will be heated before entering the cpu, negating the need for insulation past the gpu entry. Whatever heat the liquid picks up by the components is then cooled by my Rad, then chilled again before entering the components.

 

I'm oc'd at just shy of 5.2Ghz on my 8600k, really hoping to push this chip to 5.3. It's already delidded with a Rockit copper IHS installed w/ liquid metal. Temps are good; breaking 74 after a few hours of ghost recon wildlands but considering the OC I'm happy. 

[Zardoz13]

I never really got into them but they are super fun to mess around with.

 

How do you plan to mount the peltier in the system?

[TheRandomness]

- Thread moved to Liquid and Exotic Cooling.

[BiTBiTE]

14 hours ago, Zardoz13 said:

I never really got into them but they are super fun to mess around with.

 

How do you plan to mount the peltier in the system?

If you look at my rig pic, take note of the flow meter on the bottom. The pipe to the left has a t line adapter; one side goes to a valve out the back which is the drain, the other goes to the components starting with the gpu. 

I'm going to make a new pipe (actually two) that goes from the inlet of the flow meter (right side), out the back and another in for the loop. The external pipes will let me keep the TEC chiller external to the chassis so that any condensation can be dealt with. To cool the peltiers I've opted on two hyper 212 evo which will stick out the back of the case, having 120mm fans to cool the heatsinks. 

I was going to watercool the TECS but I'm trying to keep it as aesthetically pleasing as possible. The evos will look amazing out the back and will add to the steampunk look that I have already. 

 

I tested one peltier on top of my flow meter (it's laying down currently) on top of the plexi face and I noted a ten degree difference after 3 hours of gaming. Note that I didn't even apply thermal, rather just set it on top of the meter. 

My idea is to buy another vortex flow meter, replace one side with 3mm copper plating to contact the TECS, then position the visible flow meter side so that it's visible. I think it would look totally Rad. 

[W-L]

16 hours ago, BiTBiTE said:

-SNIP-

Nice loop, I would probably say for the drain it may be best to keep the valve as close to the T as possible to prevent the loop from having a dead zone where fluid can stagnate as that may collect stuff overtime. 

 

As for peliter as long as you are not planning on going subambient and control the cycling of the units doing something like a hot side and cold side is probably the best. As usual TEC units are very inefficient and the gains from them is very marginal but if you do want to do something sub-ambient chiller units are a good alternative.

[BiTBiTE]

47 minutes ago, W-L said:

Nice loop, I would probably say for the drain it may be best to keep the valve as close to the T as possible to prevent the loop from having a dead zone where fluid can stagnate as that may collect stuff overtime. 

 

As for peliter as long as you are not planning on going subambient and control the cycling of the units doing something like a hot side and cold side is probably the best. As usual TEC units are very inefficient and the gains from them is very marginal but if you do want to do something sub-ambient chiller units are a good alternative.

Initially I was just messing around with old components and wanted to see how low I could get a cpu with a TEC. 

 

I've tried to find an equation to calculate exactly how much heat is dissipated with a certain TEC given the load put on it but I've come up empty handed. 

 

I know they're seriously inefficient, requiring double the wattage of the component trying to be cooled, but I'm basing this project on two key factors:

1. They are not intended as the sole cooling component, rather an assisted solution that won't be powered at all times (only for tests) 

2. With the setup I have in mind they will add tons to the aesthetics of the case. 

 

What I'm wondering is how much they will cool the liquid. If I was able to find the equation to calculate the cooling capacity of my loop I'd be golden. 

 

In the end, I'm doing this for the lulz. Ive had lots of projects but never posted anything over the years, from my ac pc, to my mineral oil computer and the like. I don't have any modding friends and the wife's growing tired of listening to me lol. 

[W-L]

7 minutes ago, BiTBiTE said:

-SNIP-

For TEC coolers usually a general yard stick of double the TDP of the total components is a good starting point. With that you should be able to get close to ambient temps if not slightly sub-ambient. If you want to have it as an additional assisted cooling solution a hot and cold loop would work well since you have to really keep the hot side down for it to work well or it tends to undo the work for the cold side. 

 

If you like mineral oil cooling you should check out stefan's build:

 

[For Science!]

18 hours ago, BiTBiTE said:

 

The point of the TECS is not to be the main cooling component, rather chilling the water before entering components. As the cold water enters the gpu it will be heated before entering the cpu, negating the need for insulation past the gpu entry. Whatever heat the liquid picks up by the components is then cooled by my Rad, then chilled again before entering the components.

Normally water moves fast through a loop in a way that you won't have "localized hot/cold spots" and so your TEC will cool the whole loop down as a whole as opposed to a specific component.

[BiTBiTE]

4 hours ago, For Science! said:

Normally water moves fast through a loop in a way that you won't have "localized hot/cold spots" and so your TEC will cool the whole loop down as a whole as opposed to a specific component.

Hm. Maybe altering the flow rate over the TECS would yield results? An external chilled res? 

 

I was just caught off guard because even with every article I've read saying that TECS didn't work in a watercooling loop, I still seen a notable difference in temps when utilizing a TEC placed on my flow meter. 

 

I've already sealed my test board so that one will receive a TEC direct on the Q6600 to see what happens. Max wattage of the Q6600 is 180w. I have a 400w tec in the mail, but it hasn't even left China yet!!

 

I'm going to do some actual testing on my day off. It's the only way to know. The worst that can happen is a learning experience that I'm more than up for. I'll have to order some stuff tho, which could take a while. I'll keep this updated as the testing goes/parts arrive. 

[caliusoptimus]

6 hours ago, For Science! said:

Normally water moves fast through a loop in a way that you won't have "localized hot/cold spots" and so your TEC will cool the whole loop down as a whole as opposed to a specific component.

I was thinking the same. Unless you crank the flow rate down super low the water is only going to drop 1degC or so.

 

My 2000btu chiller only drops 3degC-ish with modest flow rates.

 

(ps: that chiller keeps both my gpu and cpu supplied with 12C water under full load, which is about 1000w at the wall)

[BiTBiTE]

11 hours ago, caliusoptimus said:

I was thinking the same. Unless you crank the flow rate down super low the water is only going to drop 1degC or so.

 

My 2000btu chiller only drops 3degC-ish with modest flow rates.

 

(ps: that chiller keeps both my gpu and cpu supplied with 12C water under full load, which is about 1000w at the wall)

Essientially that's what I was thinking. That feeding the loop with chilled water (let's say your 12c) directly before the components would yield results in the short term. 

I ordered some materials last night, I have a day off coming up so hopefully I have some time to test the setup. 

 

Thanks for the insight guys!!! 

[SolarNova]

So heres some info u may find interesting.

 

The old Dell XPX 720 H2C used a tech bassed AIO to cool its CPU. I had one, which used a QX6850 with a slight OC. It worked. But it wasnt anything amazing.

 

Any system that uses a underpowered TEC and a RAD to cool, will only do well down to ambient temps, beyond that the rad is actualy harming the cooling system.

To get sub ambient u need ot use a sufficently powerfull TEC, and only the TEC, no added radiators in the loop. You can use a seperate waterloop to cool the hotside of the TEC, but u cant use a rad to 'help' the cold side, as it wont be helping.