Iceprobe chiller in stock Nano Cube (no drilling)

[artarmon42]

I've got a copper heatsink and fan (which sucks from sideways and blows up).

 

The "current" problem is that I'm not pushing enough air through it(it'd doesn't stop the peltier from cooling the probe, it just becomes dangerous as the copper fins get super hot).

 

I am not going to try to solve the "current" problem, because I believe there is a bigger problem in lack of venting. Even if I got the necessary flow through the heatsink, there's currently no where for the hot air to go (besides mixing with the air already being heated up by the MH).

 

I think the correct solution is to dissipate the heat outside of the Cube. It's looking more and more likely that a liquid-cooling mechanism needs to be employed.

 

The good news is that I already have the Iceprobe all setup. Replacing the heatsink/fan with a waterblock will be simple.

[SeanCallan]

Well what I mean is, why not above the heatsink have a 60mm fan exhausting air from the rear of the cube, through the top vent?

 

Are you going to mount a waterblock on top of the iceprobe?

[artarmon42]

In the DX, the splashguard goes all the way to the back. There are no vents that open directly to the water surface. That's why I was saying, for the MH & Iceprobe mods an OG cube mighe be better.

 

Need to think of ways to get around the problems I found today. Gut feel is that a waterblock on the iceprobe would work but would be an expensive solution (using a true liquid-cooling system with coolants etc). I'd like to find a cheaper workable solution...

[SeanCallan]

I wish I wasn't taking so many senior level courses, no time for a job, no real play money.

 

I'd get myself an iceprobe and try it, but I can't justify that right now Wish I could you out more though, I'll put some thought into how you could over come this though.

 

I have fun into a snag though, the hoods are different, the side vents are behind the reflector, so actually there's no way for the scrollfan to blow straight on. On the reflector there is only 2 small openings that I'd have to run cords through aswell as exhaust.

[artarmon42]

Can you post a photo?

[SeanCallan]

Yeah give me a couple minutes

[SeanCallan]

 

 

[artarmon42]

Ah yes, that will be a problem.

Probably too dangerous with that reflector

[SeanCallan]

I wonder if I can get a new reflector or if its worth it to try and buy a DX hood...

[Chronicles]

Are the rear fans not 60mm? I'll wait till monday till I can see the back of this thing. They do have rack mount sink/fan combos that are 60mm.

 

 

-Fan dimension: 60 x 60 x 10 (mm)

-Rated Voltage: 12VDC

-Rated Speed: 4800 RPM

-Air Flow: 15.7 CFM

-Noise Level: 26 dBA

-Heatsink dimension: 64 x 63 x 15(H) (mm)

 

 

I'd say next step is to figure out some way to mount it somewhere, and give us the dimensions of the area. Only other solution is to get some solid copper pipe (can be just a crushed plumbing pipe) and thermal epoxy it to a sink somewhere else (maybe run it from teh probe, up inside the splashgaurd?). Anyway the more simple solution is cut the voltage down on the TEC, pretty sure that's what chris is doing.

 

Anyway go look @ his thread where it's mounted. 60mm fan with a low profile rackmount sink. Which will pull far more then the ~5cfm the iceberq does. Wish my cube was here, I have so many sinks/fans sitting around, I could make one to the size that's needed.

[ccjung]

I spoke with artamon on the phone yesterday and in the spirit of research, (and I want to see the comparison), im going to be sending him a heatsink to use with the probe.

 

I placed the heatsink in the stock fan location. I recommended to him, that he cut a hole in the splashguard and mount the probe as if the splashguard was the glass, and have the heatsink mounted in the stock DX fan location.

 

This will be an interesting comparison, but here is what i noted:

 

The probe has 2 modes of heat transfer, first, there is conduction from the TEC to the probe, and then convection from the probe to the water. My biggest concern is that there isnt enough flow making contact with the probe for it to effectively convect heat away.. This is why they most commonly bundle the probe with a HOB filter/pump in their "nano kit".

 

In the swiftech, the water is physically in contact with the TEC so there is a single mode of heat transfer from the water, convection...

 

You are also correct, i have cut the voltage significantly from rated voltage. Although the mode of heat transfer has been quite effective for our purposes, I believe the contact time between TEC and water is too short. A lower rated pump would probably allow more heat to be transferred from the system. The heatsink is running at a very manageable temp (80-85F) when the fans are on. This hopefully will help when we start thermally loading the system with lights.

 

Come on over to my thread chronicles.. I need someone to bounce ideas off of.

 

Chris

[artarmon42]

Yup, 60mm would fit. As you can see from my picture of the sideways view once it is installed, that is the "perfect" fit. Any lower and you'll have to jam the heatsink past the back lip and risk the waterline going above the nut onto the silicon nut/neoprene (which is around where it would make contact with the screws holding it all together).

 

Where would I get that type of server 60mm fan from?

 

FWIW, I'm not convinced that even more cfm is going to solve the lack-of-venting problem, but will try for the "good of the community" :angel:

[ParaDiddle]

Do you guys think that the stock IceProbe is enough to counteract the MH if it was in an external sump/fuge with more water flow? In a 12g Cube with about 7-8 gallon Fuge.

[artarmon42]

So I was thinking about it overnight, and have these thoughts... Feel free to contribute/contradict as necessary, to help get to the bottom of things.

 

1) The probe idea is definitely very inefficient. Chris's idea of passing tank water directly onto a peltier is definitely more efficient. I'm not sure about water contamination over time using Chris's technique (assuming the peltier's polymer surface doesn't break down with saltwater, there is risk of creep to the electrical contacts). Much more efficient, but need safeguards against the inherent risks of direct water contact.

 

2) The dissipation of heat on the heatsink/fan side is necessary only for safety. Unlike conduction, as long as there is a current, the Peltier will keep heating up the hot end. Dissipation, or lack of it, does not change the Peltier's efficiency to cool the cold side.

 

3) Lack of heat dissipation (which heats up the air between the MH and the water) will further contribute to the overall system's temperature. I believe the only way to make the system safe (as noted in point (2), heat dissipation from the Peltier does not have a direct relation to the cooling effects) is to dissipate it out of the system. I looked a heatpipes (interesting technology, recommend people google it) and that seems to be the best technique. However, commonly available heatpipes are build for PC overclockers (they don't channel heat below a certain temperature) and are probably not calibrated for our use. The simply alternative as I previously mentioned is to drill the back for direct venting, which then allows a greater heat dissipation technique (e.g. Chronicle's higher CFM server fan) to work.

 

4) Here's the punch line... Given points (1) and (2), I believe the 51W peltier is simply insufficient to cool against the heat generation from a 70W MH. I am not a scientist nor have the necessary tools, but I believe if you do the calculation of the heat generation from the MH (hindered by fans in the splashguard, air transfer and then water with a high rate of circulation) vs the cold generation (hindered by the distance it travels down the probe, the probe's insulation and then water with a similar high rate of circulation) would clearly point to the MH as the winner.

 

So in summary...

I don't believe a full-powered 51W peltier is sufficient to work against a 70W MH. The hinderance of the Iceprobe's thermal transfer inefficiencies make this technique even more inefficient.

Further, the lack of natural venting and limited options for a high CFM heat dissipation device, further heats the tank and poses a serious safety issue.

 

 

If people have thoughts or things they'd like me to try, I'm happy to help.

 

ps: No, I'm not going to buy a $200 titanium heat exchanger just because you'd like to see how it works

[Chronicles]

It doesn't work like that. The point is you're not cooling the hot side of the TEC, thus the cold side can't get cold. I'll explain more later, but we're leaving for easter. Just look at the cooler/fan that came on it, and wonder why it's so big.

 

If you believe it's the TECs fault and can't cool, hook your chiller back up totaly and stick it in there and see what happens.

[artarmon42]

About how a pelt works...

Ummm, I think that's what I said (point (2))

:-*

[SeanCallan]

I'm so sure I'm ready pass tank water onto a peltier that's not designed for saltwater. I think eventually we'll see a break down in materials and salt working it's way in 'dangerous' areas.

 

I think your iceprobe idea would work perfect in an OG hood. you can single out an entire rear vent to simple pull hot air out. Once I figure out how I can do a MH, I'll work on an Iceprobe solution.

 

In larger tanks it's common to mount fans to blow across the water to try and eliminate MH heat reaching the water. This that in mind, I wonder if it's possible to have a scrollfan push air from the back across the water and out the front? I don't have a DX so I don't know how much clearance you get between the backwall and the splashguard.

 

I wonder if it would be beneficial to change one of your scroll fans around. Have one exhausting and the other one directed at the MH and the other exhaust fan. This way you would get a significant flow of air across the bulb and hopefully you can exhaust the heat out the other side. If at all possible upgrading the exhaust scroll would be ideal. I think you really want to up the flow across the bulb and try to eliminate heat at the source. Note most pendants use 1 or more PC fans to try and cool the bulbs to help their longivity and to decrease downward expelled heat.

[Chronicles]

It's not working because your cooler is too small PERIOD. The unit you bought retail, has a big ol'sink on it for this exact reason. Hook everything back up, open the top and stick it in the back and watch your temps drop. The purpose of cooling the hot side is not for safty, it's so it can continue to work as a heat pump.

[artarmon42]

Ah yes, I get it now.

 

I was thinking the current drives the heat exchange singularly, but upon further reading the lack of dissipation on the hot side lowers the conversion efficiency.

 

The data I collected from the test showed that the "cold" water was getting warmer over time. I assumed (incorrectly) it was because the Peltier could not keep up with the MH heat, rather than the insufficient heat dissipation.

 

Thanks for the education Chronicles!

 

 

 

 

 

But it would seem that the root problem still remains... dissipation of heat in a tight area without adequate venting (I'm sure removing the hood would work, but that wasn't the goal). Do you think that server fan pushing 50 CFM (assuming I dremel a rear vent) will do it? And calculating backwards (based on the heatsink/fan's power) is it possible to use a more powerful Peltier?

 

BTW, where can I get one of those server fans to try?

[SeanCallan]

Try www.newegg.com

[artarmon42]

I just placed an order at nexfan.com

 

When it arrives, the new heatsink will be this. I had been struggling to look for a 60mm heatsink (so that the binding screws will fit) that it just dawned on me that the only "real" restriction is the height (being 25-30mm). That 60mm limitation is not applicable because I can use a thermal adhesive to hold it together (I hope).

 

 

 

Anyone seen these before ? I wonder if they might cool better than what I bought. What puts out more heat? The latest GPU or a Socket 940 CPU?

[SeanCallan]

Would any of these work?

 

http://www.thinkgeek.com/pcmods/cooling/652e/

 

http://www.thinkgeek.com/pcmods/cooling/5e55/

 

 

This one is neato:

 

http://www.thinkgeek.com/pcmods/cooling/6937/

[artarmon42]

I was considering the Zalman one, but because it's a circle and the contact (to go onto the Peltier) is in the middle, it is too big. The unit I got is 70mm, but I will mount it on one edge (rather than the center of the square), so that it is offset and a part will extend over the Nano Cube's fake wall.

[Chronicles]

I've used those VGA silencers before, pretty big though. The higher GPUs produce 70-90w of heat, I know my 6800Ultra is crazy hot. That cooler from nexfan is a good pick. 6 times the airflow of your iceberq . I'd kinda like to try out that big bay cooler I posted in your other thread, cut a 5" hole in the back and just slide it in so it's flush like a normal 5.25" drive.

 

If you need to secure it to the TEC without screws, just get yourself some arctic silver epoxy. Has the properties of arctic silver and will hold like no other. Also if you don't know, arctic silver is a thermal compound. Basicly drops temps between 5-7C vs normal compound, which is about 8-13F.

 

 

Oh and look @ this one with your scroll fans:

http://www.orbitmicro.com/products/accesso...ns/DC1U-B04.htm

 

Air Flow 8.1 CFM

 

Not so good. Are you sure your scroll fans are rated higher?

[artarmon42]

I actually bought that big bay cooler thing as well. Not sure how I'll mount it... will figure that out when I get it

 

And I also picked up some Arctic Alumina and Arctic Alumina Adhesive. From what I read, Alumina is the same properties as Silver but cheaper because it uses ceramics instead (plus I fugre one less metal compound in tank, one less thing to rust... although if water gets there, I'm SOL anyway!).

 

The scroll fans I got were rated around 10CFM. Scroll fans push LESS air (that conventional fans) but without dead spots (not that big a deal in this scenario) and push it FURTHER. I'm using it to exhaust the hot air out the side, rather than just circulating air and hope it fits it's way out. Exhausting it further will hopefulloy help get the hot air dissipated in the room before any gets sucked back in. Plus both the top/back vents of the hood suck air IN, so hot air needs to go sideways and not up.

 

As I said, I just put an order in at nexfan for different (higher CFM) fans and plan to rework my airflow. I think I'm going to suck air in from the side vents, and exhaust air out from the top/back vents. That way more cool air hits the MH first, and thermal flow (hot air goes up) should help cool.

 

(Why did I arrange to suck air in from the top originally? Back in the Nano Cube OG days, that vent was open to the water and sucking air out caused alot more evaporation).