TEC waterchiller

[Jorlen]

The reliable way would be to place a thermalprobe under the heatspreader of the cpu. I think heatsink reviewers out there do that.
I can try to do a idle calibration when I get home with realtemp. But the would only shift my temps so they are more correct at idle. I need to add probably 10 deg C on core one and 5 deg C on core two.
That takes care of idle temps. But that just puts my load temps more 30 degrees above water temps. Which would indicate my waterblock is the worst in history (it might be), but I seriously doubt that. I will measure the temperature of waterblocks copperplate. I bet its very close to watertemperature.
And if it turns out as I expect that leaves me with the conclusion that the integrated heatspreader(30C temp delta, eeeek) is crap or internal temperature measurements are useless.

[Uncle Jimbo]

I have considered fans. But I need to point out the keeping noise down is more important.
There is a rather large copper heatplate between the heatlanes on the heatsink and the hotside of the TEC. The plate is warmer then the heatsink by a noticable margin. Which to me suggests that thermalresistance between heatsink and TECs is far from optimal.
I knew this when I decided to include the hotplate. It comes with the heatsink. It made things a lot easier when coming up with a way to assemble everything. And the peltier is 40mm while the heatsink is about 60mm, which mean I would use only 2/3 of the small heatpipes in the heatlane without at heatplate in between.

Thermal resistance of heatsinks using passive convection goes down as they get warmer. I dont know if its obvious, but at constant airflow, energy flow between air and heatsink is in linear relation to temperature difference of the air and heatsink. But since the airflow increases with the higher temperature of the heatsink the C/W goes down. With this in mind I am still hoping to cool well without fans.
There are so many materials for the heat to transfer through in my setup:
CPU, thermalcompound, heatspreader, TC, Waterblock, water, waterblock, TC, peltier, TC, heatplate, TC, heatsink(which is complex as well, since it heatpipe-based), air.
Something about this is not optimal.

Jorlen -
Brilliant work - I especially like the idea of using the fan controllers on the MB to control the TEC. If you need any help on that, glad to provide it - I have designed a lot of big power chopper control circuits and can help with device selection etc. With high speed PCM it's a good idea to add a low pass filter on the output - any old burned up PSU has all the parts you need for any power PCM designs at these levels.

I know the Heatlane Zen is designed for passive, but for others reading the thread, heatsinks designed for forced air don't do well passive - the fins are usually too close to each other to promote air flow from the convection transfer. Forced air sinks often have spacing of 1/8 inch or less, while passive usually have 1/4 or 3/8 spacing with longer fins.

I'm not a fan of the Heatlane Zen, despite its cool looks. I played around with one a few years ago and was disappointed with performance. My testing was done with a controlled heat source at 40W, 50W and 60W in 25C ambient. Even with a 40W load, the base was at almost 70C. In comparison, a big typhoon with the same load held the base under 30C, and a Melcor HX5-301 holds it to 28C. The Heatlane Zen is big and hard to mount, and passive performance is mediocre - about 1 C/W. But I was able to get better cooling with a little airflow - I mounted a Vantec Stealth 120 to blow through it, with the fan at 9V turning under 1000 RPM, which was virtually silent. That airflow took the performance to .55 C/W.

I'm pretty sure you are using 12708 TECs. At 5V, you should see about 2.1A per TEC, with a 25C differential. That puts you at a CoP of 1.5, about as good as you can do with a 25C differential, which is all those sinks will do without active airflow. That means you are pumping about 15W per TEC, and a total heat load of about 25W.

Going to 6V probably takes you to the 2.45A level, and now you are burning almost 15W per TEC. You will move 20W per TEC, but efficiency has fallen to 1.36 CoP. Your hot side is probably around 60C - I'm surprised you got any improvement in cold side at all.

Just adding minimal airflow to those sinks should take them to under .6 C/W, which with 5V on the TEC takes your hot side from probably around 45C to more like 35C. At 5V you will move to CoP of 1.9, and you will move 20W per TEC or better, with no increase in power, still at 5V. If you can get the Vantec Stealth fans, give them a try - even at 12V they are only 28db, and at 9V I really can't hear any sound at all.

One other point - I had a hard time getting good contact with the Heatlane Zen. Are you sure that the heatsink makes contact all across the TEC? Also, correct and even pressure is important - there are several references to the right way to do it on this forum, including mine at http://www.xtremesystems.org/forums/...=184686&page=2

[Jorlen]

About the choice of heatsinks, they might not be the best heatsinks but they were about 1/6 of the price of the heatsinks that could condend. And when you need 10 of them price is an issue, I would not have started this project if the first thing I needed was to spend $500 on heatsinks. Same with the waterblocks and TECs. Even though the resulting pricetag did run away a bit, I think it would be hard to do it much cheaper. I will be on the lookout for cheap quiete fans. I think it would be smarter for me to cycle fans on and off and run TECs at low power. The idea of driving TECs harder when there is a load falls short when the thermal resistance is as high as it is without fans. So I will need fans to justify running TECs higher then 5V.

I am not completly satisfied with the way the Heatlane is clamped to the hotplate. The plate is to thin that clamps the heatlane to the hotplate. So the pressure on the heatlane is at the edges closest to the bolts. And not in the center. The pressure over the TECs is probably fine. Cant improve that without reworking the whole thing. Right now I have 2 M4 bolts and a very sturdy U-profile of 2mm stainless steel that secures the waterblock and TEC to the 4mm hotplate.

[Uncle Jimbo]

About the choice of heatsinks, they might not be the best heatsinks but they were about 1/6 of the price of the heatsinks that could condend. And when you need 10 of them price is an issue, I would not have started this project if the first thing I needed was to spend $500 on heatsinks. Same with the waterblocks and TECs. Even though the resulting pricetag did run away a bit, I think it would be hard to do it much cheaper. I will be on the lookout for cheap quiete fans. I think it would be smarter for me to cycle fans on and off and run TECs at low power. The idea of driving TECs harder when there is a load falls short when the thermal resistance is as high as it is without fans. So I will need fans to justify running TECs higher then 5V.

I am not completly satisfied with the way the Heatlane is clamped to the hotplate. The plate is to thin that clamps the heatlane to the hotplate. So the pressure on the heatlane is at the edges closest to the bolts. And not in the center. The pressure over the TECs is probably fine. Cant improve that without reworking the whole thing. Right now I have 2 M4 bolts and a very sturdy U-profile of 2mm stainless steel that secures the waterblock and TEC to the 4mm hotplate.

I understand about the sinks - any price times 10 is significant. Those have a lot of area, and with a little airflow should do fine as long as you are getting good contact everywhere. That baby looks about right. Leaving the voltage at about 5V per TEC and getting the cooling right is probably the best way to get efficiency up.

I saw your comment about modding the PSU. But as you suspect, the dimensions are pretty important - switching speeds are over 50KHz and even a little parasitic transfer can cause big problems which can wipe out the other components in your rig. There is a much easier way if you have the space. Get a PSU with much more capacity than you need, put it in a bigger box, and bolt on bigger sinks to the main cooling inside the PSU. That is usually easy to do and the sinks that need cooling are obvious.

An alternative is oversize the PSU, but instead of increasing the sinks, use a bigger fan running slower. I have seen some 'silent' PSUs that do this using a 140mm fan running 800 RPM - not totally silent but very quiet.

By the way, you had an earlier question about frequency of PWM power for TECs. You need to decouple the PWM with a low pass filter including a decent sized capacitor ahead of the TECs but it's very easy to do. Cycling TEC power using essentially a thermostat is OK too, as long as the hot side and cold side don't change temp much - as in a design with WC on both sides. But with air cooling, the hot side will have a bigger spread between powered and unpowered. The issue is that thermal cycling causes expansion and contraction of elements inside the TEC, which causes stress fractures, which cause the TEC to fail. You want to keep that cycling to a few degrees at most to avoid problems.

[Jorlen]

I have an alternative solution to the noise issue. Since the pump, HDD and PSU create about the same amount of noise it only makes sense to take care of all three. So my solution is to make a soundproof enclosure of the right compartment of my servercase. Where HDDs PSU anb pump will be. The picture below explains it pretty good. When running the sidepanel will seal the remaining side of the box and the hole where the hose and wires enter the left compartment will be tight. The exhaust of the powersupply will push air through the radiator into the chamber on the other side of the radiator. And since that chamber is separate the air will leave by passing through the radiator in the other direction above the PSU. This will limit the need of fans to just the one in the PSU. I do not know how much extra energy will leak through the sound isolation material and into the water via the radiator. I dont know if the water will stay below ambient in this setup. I feel confident that I can run my system without fans as long as I dont overclock to much. And if I add fans to the Heatlanes I should have plenty of headroom.

I have upgraded the CPU MB and memory, to a more overclocking friendly setup. But I have not had time to play yet.
DFI blood iron, E7200 and Cellshock PC9200.
Discovered a poorly placed capacitor stopped me from using the same waterblock on the NB chipset. And the memories are running a bit warm without any airflow. So aside from soundproofing, a ram watercooler and a new chipset WB is on the the todo list. After that I will look at adding fans.

[Vinas]

I have upgraded the CPU MB and memory, to a more overclocking friendly setup. But I have not had time to play yet.
DFI blood iron, E7200 and Cellshock PC9200.
Discovered a poorly placed capacitor stopped me from using the same waterblock on the NB chipset. And the memories are running a bit warm without any airflow. So aside from soundproofing, a ram watercooler and a new chipset WB is on the the todo list. After that I will look at adding fans.

The BI northbridge runs very cool. I'd recommend using a fanless HS like this: http://www.newegg.com/Product/Produc...82E16835109019. The only problem with that HS is that once installed it is very close to the video card -- almost touching it. In my rig I just placed a plastic cover behind the cooler and I have no fan on mine. Temps at 517 FSB never exceed 45C with that cooler. Although I do have a lot of airflow in my case.

[XSS]

Nice but much more Watt gg ?

[Jorlen]

Hello again. Been running my system since early may. Performance is still the same as when first assembled.

I have not taken the time to finish the complete plan that included soundproofing the case and cooling the inside of it with the cooled water passing a radiator inside. And minimizing airflow through the case to keep the noise and cool air inside.

What I have learned so far. Silicon sucks. I have lots of silicon tubing to get all that water around. Silicon is to pourus, it lets air in and water out. My reservoir makes this issue more of a pain. I need to keep it completly filled with water otherwise turbulence in the reservoir will create airbubbles that make noise as the rush through the pump and the rest of the system.
The volume that can vary is that inside the tube coming off the top of the reservoir. So I need to refill that tube every week or so. To keep everything running with minimum noise.

Other then that I like the system very much. Silent and effective.

[iNF3rN0]

silicone... i think maybe its the way ur applying it, maybe the surface isnt grease, oil, and dirt free when u made the res, silicone shouldnt leak atol when applied properly

if silicone is pourus and lets air in and water out, why would it be used in every glass aquarium?

[Jorlen]

Update again. It wasnt the silicone it was the waterblock that leaked. Its leaking very slowly and drips down on my graphics card. It broke. I fixed it now. Life is still to busy to give me time to continue on this project.

Edit: Its not that purous. And when you seal with it like in an aquarium. The water is exposed to a small surface of silicone, and the seal is quite thick.
And even for me with all the hoses it wasnt the issue.

[Jorlen]

My system is still up and running. It works exactly as I expected it to work when I first planned it. Very quiete when the pump is suspended properly.

Changes made since last post:
New CPU block XPSPC Delta v3
New Reservoir. DDC top+reservoir integrated.
There are 3 blocks now cooling CPU, MB and Gfx. Added the MB last.

Since I have an Ati HD3600 I dont need a fancy waterblock, so I use the same Koolance blocks as on the MB and TECs (they were almost free when I bought them).

Aside from the pump, moving parts still in use:
PSU fan
2 x HDs

This project cost way to much if you just look at the value it produces at thist stage.
What you can do with this, as I have said earlier, is to build the system inside a soundproofed box and transport the heat out with the water and have the large passive sinks cool the whole thing. Without TECs the passive cooling outside the box would be huge in order to keep the system within operational limits. So there is potential.
To bad I dont have time for this anymore. Maybe next year who knows.

[Naja002]

Very Cool. Glad it's worked out for you. Thanx for updating this....

[Machinus]

Awesome project. These are clever ideas but still nothing too complicated...why no one has built a sealed TEC chiller with feedback, I don't know...it would outperform everything else and be worth paying for.

[Jorlen]

What the TECs do is to allow smaller convection based cooling. A combination of the Zalman TNN-500A and TECs should be very effective. Using heatpipes instead of water to expand the cooling surface enough for passivly cooled TECs to come into play.