Cooling a water-reservoir

[asmodean]

I'm still in the designing stage of my water cooling and I have a question about the temperature of the water.

Some bluepaper specs so far for the reservoir:
About 20x14x40cm (WxHxD), made from sheet-metal or some such.
It'll hold about 10~11 liters of water. 2xTemp sensors (LCD) & waterlevel-window made out of plexiglass. Attached with rails under my case (Antec SX1030).

Would that amount of water keep the water temp low enough for me to just recycle it thru the pump to the waterblock? My CPU pumps about 70W of heat currently... If not, could I attach a low-power peltier (30W?) to the back of the reservoir (outside) and cool the with a heatsink? The heatsink would be 97x130mm. Sideprofile of the HS attached as an image.

Oh, I could use my PEP66 without a fan also.

[asmodean]

bump.

Thank you for replying. :)

[Ridenow]

Somewhere I saw a thing about a guy who used a 5 gallon bucket with not radiator or pelts. It ran fine for several hours until the water got hot, then it crashed. I think you need something to loose the heat, probably a radiator would work better than a pelt.

[asmodean]

Thanks for the info. A radiator is out of the question, as it includes fans... And the e-shop I'll be buying the stuff from has only one radiator which costs the same as the rest of the stuff. That would be about $200 combined.

[Bender]

If you are going for a silent water cooled system you might try coiling a lot of copper tubing and running the water through that. Why don't you want to use a radiator, is it noise? I am planning on doing a cheep watercooled rig myself. Try looking at www.aquastelth.com for ideas or very inexpensive parts.

[Kryten]

The copper pipe idea sounds ok, you may wish to consider a larger water resovior, a plastic 20 litre Jerry can would be verry cheap.

[Thelemac]

The only problem with using just a large reservoir for cooling is that it doesn't do any cooling. At all. It makes the water take longer to reach an equilibrium temperature. You would definately have to add some form of cooling ability to the water for it to lose the heat. The TEC on the side might work...though I really don't know how much heat that would remove, nor do I know how hot it would get. The copper tubing idea does have merit...and you could consider a couple of ideas to go with it:

Get a low rpm fan (assuming it's the noise you don't want) or two and have it blowing (sucking, maybe?) on it

Get a whole lot of the tubing and run it out and around your case and everywhere so that it can just passivley cool off.

I hope this helps you out, and good luck with your ideas. :)

[asmodean]

Thanks for the ideas, people!

I had actually tought about running tubing around my case originally, but now it's layered with bitume to reduce noise, so I cannot do that.

Here comes my newest idea:
I'll get two FOP32's (quite cheap without fans) and two peltiers, about 60W? I'll put the peltiers to the back of the reservoir and attach the FOPs. Then I'll put those Titan 80mm fans to good use, they're thermocontrolled.
And I'll run the tubing inside the reservoir. That could be a bad idea, though...

Tell me what you think.

[Thelemac]

Well, if you're cooling the reservoir then you don't really need to run the tubing in it at all, cause the water *should* already be cooled...since it came from the reservoir and all.

Also, I don't know how well the fop will cool the TECs...they might do a good job and they might not, I really don't know. You might look into that if you haven't.

Other wise sounds like a good plan! :)

[asmodean]

Hmm... Maybe those FOPs could cool 36W/16V TECs? Smallest I know I can get, here in Finland. I'd be running them at 12V, though.

[Thelemac]

Well, I stopped to think for a second...

I think that the FOP would be up to cooling the 60's...though I don't know how cold the water would be getting. I came to this conclusion by comparing 60W TEC to 60W Tbird...keeps it at ~+20-25 (something like that anyway) C...so then the cold side would be whateve the change is from that. Of course, I really don't know the safe operating parameters of TECs, nor do I even know if you can compare the cooling ideas like this.

I really wish somebody that really knew their stuff would look in on this and help out...cause I'd really like to know if I'm right or not. :)

[surlyjoe]

HI!! OK first you have to take into account the heat from the pump ,add that to the heat from the CPU and thats what youll need to deal with ,,The TEC's in the side of the tank will work as a radiator. but they will never reach maximum potential air cooled ,, youd better plan on at least 20-30% more watts than you think youll need, a good radiator will keep you about ambient , using a chiller and a rad is kind of iffy , cause the water always returns to ambient when it goes through the rad. as far as the coils go ,,they work fairly well but will restrict flow proportinatly to its length.. if you go that route , make sure you use tubing larger than the rest of the system to try and reduce back pressure,,,ALSO a large res with no rad. or fans (5gal plastic bucket) actually cools quite well just by radiating heat off the sides and surfae of the water, It wasnt great , but it kept up after it stabilized

[asmodean]

Thanks, surlyjoe!

Some extra info: the pump would be a Sicce Nova.
Sicce's homepage (http://www.sicce.com).

So, if the pump generates about 30W (wild guess), the whole thing would be generating approx. 100W of heat... Holy mandung (<- ha ha). But having two pelts generating about 60W each, I should be safe with those FOPs.
There's a pelt (ICE-71) that is rated 80W@16,1v, so @12v, it would be 59W, if it's linear. The deltaT is 71C, so if I can keep the hotside about 40C, the coldside would be about -30C, if I'm not mistaken.

And the tubing... I don't know what to do with that. Maybe I'll have my friend inline the reservoir with copper, that should make the heat transfer faster away from the water and help the pelts distribute their cold more effectively around the whole thing.

I'm starting to think that I'm getting too ambitious with this whole reservoir thing. Grr. But it's such a neat idea.

[surlyjoe]

if both pelts were running perfectly, you would be OK ,although that would be to keep it at ambient correct?? I would make sure I gave myself some head room!! Also, it takes a considerable amount of juice to run 2 pelts , its gonna run up your power bill if you leave it on 24/7 ,. something else to think about :P

[asmodean]

Now that was a good thing to know.
Those pelts are out the question then, as I'm folding 24/7. I would've had a 230W PSU just for them. Bugger. The f@h server is down at the moment, grr...

It would have been to keep water at ambient or close enough. Now it's the tubing thing I'll do. Maybe drill holes to the back of the thing and have the tubes bent to a U rotated 90 degrees, so they'll come out some 8-9cm. Then I'll use those extra Titan fans to blow some air across them. Hows that sound?

[surlyjoe]

I'd really think of going with a cheap rad. you can get a used heater core for less than 10$ and it will get you pretty close to ambient , I have 2 nidec blowers on one and its almost silent , the coils work , but not as well as a decent rad.

[asmodean]

One of the reasons I've been planning this reservoir is the aesthetics, as my GF doesn't want any "ugly" things sitting around... If she didn't object, I'd already have water cooling, using a bong. I don't think I can hide a radiator so that it runs well enough, especially a cheap one.

I'll write about that tubing in approx. 3 hours, as I gotta go to work. It's 7.50am here.

[asmodean]

Didn't have enough time on lunchbreak, so here we go:

Copper piping (better word), coiled like a spring, 4cm radius. Four of these: two atop each other and two side by side. Goes something like 20cm(?) inside the reservoir. Those thermocontrolled Titan 8cm fans blow cool air across the coils.

One possible addition is a lowpower peltier (must put one somewhere :)) between the coils. Cooled with the PEP66T using 2 stacked 60x60x10mm Titan thermocontrolled fans. Seems like I have those in abundance.
The intake for the waterblock would be touching (piece of CU pipe) the pelt, so it would cool it somewhat.

BTW, I ordered that watercooling kit on the lunchbreak, I'll propably get it on week 22. <need a new thread>

[asmodean]

Bumpity bump.

[bdf24]

What size peltier are you looking for. Other wise try experimenting running different sizes at maybe 7volts. That would cutt the wattage down quit a bit. wouldnt running a 85w tec @ 7volts drop the wattage down to around 39watts? I'm not sure how much your looking to cool the water?
You could also check out this link on the homepage, A peltier Controller!
http://www.overclockers.com/articles424

Just make sure (and i'm sure you already know this) that to keep the water at it's coldest before it hits the cpu, that you it touches the cpu block right after going thru the chiller. Or atleast after the copper coils. If you have the chiller before the copper coils it may bring the temp of the water closer to room temp again. But then again if thats all your looking for is to keep the water at ambient temp then it doesnt matter!

[asmodean]

I was thinking of a 56W TEC, rated @15,7v. I'd be running it @12v, so it should be 43W. That PEP66T should be good enough to cool it down, with 80mm thermocontrolled fan sucking or blowing.

I read that article, it's not for me... Besides, I'd have to order abroad, which is not a thing I'll do easily.

That picture I drew up is from the side, so there would be 4 coils. I think I'll put 2 coils for the water that comes from the cpu & the rest 2 for the water that goes to the cpu. That TEC would be just before the output to the cpu, so it should cool the water down atleast few degrees.

The sad thing is that I'm still waiting for the kit to arrive. I'll start making some serious drawings when I get the kit. It SHOULD come today or tomorrow... We'll see about that.

Thanks.

[Spewn]

Sticking the two 60W peltiers on the side of the resevoir would work perfectly. Well, I'm lying. They would cool the water just fine, BUT, they MAY cool the metal casing below ambient, and cause condensation. Especially when you first turn on the system, because the water wouldn't have had a chance to do any warming up yet(if there was any warming up to do). Anyway, just my two cents...I think the pelts would work fine.

[asmodean]

Condensation wouldn't be a problem, because the reservoir will be under the case, outside. Maybe this'll explain it a bit better: 4 coils, 2 atop each other, with the TEC in between. Oh heck, I'll draw a picture.. Done.
It's from above. The gray area would be the TEC. I haven't figured out how to put the output to the cpu yet.

Any suggestions?

Thanks.

[Wa11y]

Here's an idea that's just so stupid, it's probably kinda dumb. But it might work.

Ever see those "ice packs" that are some supercooled gel thing? Suppose you could take those copper coils that are outside of your resivoir, and enclose them in a box filled with that gel from those ice packs, then stick a pelt on the outside wall of the box?

Here's a drawing. Don't laugh. I'm better with pencil and paper.

[Spewn]

you would want the copper pipe to be coiled INSIDE the resevoir, and have the TEC(s) flush against the resevoir. Then have the end of the coil exit into the resevoir.

[asmodean]

Thanks Wa11y & Spewn!

I'll have to do some searching around the local shops to find that gel. Might be a nice idea.

I'll be having those coils outside, atleast. They'll be getting some cool air with 4 thermocontrolled 80mm fans.
I just might coil them inside the reservoir too.

Oh, that white area in my superiously artistic and well drawn picture is the PEP66T. It gets also one of those 80mm thermocontrolled fans.

That gel raises few questions, tho. How does it like the temperatures? Does it dry? What does it cost?

This'll be an expensive endeavor... I'll propably go broke, but so what.

Almost forgot, little changes... I'll be ditching the waterlevel meter, it's not very useful. I'll put in a flow-indicator instead.

[Spewn]

You wouldn't necessarily want to use ice pack gel. The only reason ice packs have gel in them is so they'll be flexible as they thaw. You should look into different materials that require less energy to change temp. I know alcohol is one, so using alcohol in that chamber might work well. The reason why you want something that requires less energy to change temp is because the peltier's are going to be transferring energy out at a certain rate. The faster this can cool the substance down, the higher the temperature difference between the copper coils and the cooling substance will be, and the more efficient the energy transfer will be.

[Wa11y]

The only problem with using alcohol is that because it cools at a lower temp, it also evaporates at a lower temp. I suggested that ice pack gel because it doesn't seem to evaporate, and keeps cold for a pretty long time. But that was just something I came up with at work, while I was bored. I don't actually know how it would work, and haven't done any testing or anything. It just seemed like a good idea.

And those ice packs shouldn't be too expensive, Asmodean. I think they're about $2 each, and I don't think it'll take more than two. Again, just guesses. But if you do decide to go with it, let me know how it works. It's always good to find out that some weird idea that pops into your head is a good idea. As rare as that is.

[asmodean]

I definitely will try the icepacks, if they're that cheap. Just remembered, they're available at sports stores.

[Wa11y]

Sports stores, WalMart, anywhere that sells medical supplies for home use.

[Spewn]

Wa11y (Jun 01, 2001 08:27 a.m.):
The only problem with using alcohol is that because it cools at a lower temp, it also evaporates at a lower temp. I suggested that ice pack gel because it doesn't seem to evaporate, and keeps cold for a pretty long time. But that was just something I came up with at work, while I was bored. I don't actually know how it would work, and haven't done any testing or anything. It just seemed like a good idea.

And those ice packs shouldn't be too expensive, Asmodean. I think they're about $2 each, and I don't think it'll take more than two. Again, just guesses. But if you do decide to go with it, let me know how it works. It's always good to find out that some weird idea that pops into your head is a good idea. As rare as that is.

You wouldn't want this stuff exposed to the air, so the evaporation of the alcohol wouldn't be an issue. You'd want it sealed in a box so that less heat would be absorbed from the air.

[Spewn]

Another note, the fact that icepacks stay cold for longer just means that either A) they're bad at absorbing heat, which isn't what you want here. Or B) they take more energy to warm up, which also isn't what you want here(since you want the peltiers to be able to easily cool the substance down).

[LimeyGreg]

The gel pac won't work as it will not transfer the "cold" from the peltier to the reservoir (no convection current, only conduction), they are designed to hold more latent heat than just water and that is why they stay colder longer. You need to get the cold side to pull the heat from the reservoir as quickly as possible - the copper tubes/pipes need to be inside of the reservoir in contact with your coolant so that convection and conduction can cool the water more efficiently.

Your 50 watt peltier will be putting out about 100 watts of heat (approx). 50w from the reservoir and 50w it generates itself because of the current flowing through the peltiers resistance. Just slapping a heatsink on it is not going to work, you are going to need a good sized fan to accomplish the task - if you don't get rid of the heat you'll just end up heating the coolant rather than cooling it and I doubt that is the effect you are looking for.

In essence you are trying to build an intercooler that is cooled with air rather than a liquid coolant, you'll need a much bigger heatsink to accomplish the task. A radiator is the most efficient at removing heat from your coolant as it has a large surface area, if you can fit a old automobile heater core in the comp with say a 120mm fan running on 7v and the peltier on the cpu I think you will have better results than the way you are trying to do it now.

If you want an aesthetically acceptable evaporative cooling system how about a nice indoor table fountain? If you have enough surface area then the water in the fountain could be used to cool your cpu. It would also mask the sound of any case fans you have running. Or, you could put a coil of copper tubing into the fountains' reservoir and then pass your computers coolant through it (indirect cooling).

[DE]

surlyjoe (May 19, 2001 07:08 a.m.):
more watts than you think youll need, a good radiator will keep you about ambient , using a chiller and a rad is kind of iffy , cause the water always returns to ambient when it goes through the rad. as far as the coils

Only if the rad is after the resevoir. Put the radiator after the block, get the water near ambient, then send it to a resevoir where the extra cooling would occur.

[asmodean]

There's been some changes to the plans... I took the measurements yesterday with my friend, who'll make a proto for next weeks end.
While taking those measures, I noticed that I can't use a TEC, there simply isn't enough space for it and the heatsink. Sad but true.

This seems to be the current "release candidate": 4 coils outside, 8 coils inside (if I can fit them). The outer coils will be cooled by 4 fans, of which 2 will blow and 2 suck.
Those 8 inner coils will be attached 2 each to 1 outer coil.
I'll split the outer coils for incoming & outcoming water.

I'll be sure to post some images of the proto.

Thanks for the ideas. I might be able to incorporate some of them, when I get the proto.

[jcover]

Spewn (Jun 01, 2001 07:16 a.m.):
You wouldn't necessarily want to use ice pack gel. The only reason ice packs have gel in them is so they'll be flexible as they thaw. You should look into different materials that require less energy to change temp. I know alcohol is one, so using alcohol in that chamber might work well. The reason why you want something that requires less energy to change temp is because the peltier's are going to be transferring energy out at a certain rate. The faster this can cool the substance down, the higher the temperature difference between the copper coils and the cooling substance will be, and the more efficient the energy transfer will be.

I know a gel which is usefull for you:it is the gel Agar Agar used by physicians. If you tried ??? That's not expensive.

[asmodean]

What is the gel used for? I've never had need for any of gels...