kamo

Distinguished
May 3, 2002
67
0
18,630
Hi, I have duron 550 @ 1.3V, fanless with a cheap heatsink (some titan 12$). I'm considering water cooling, is it possible to be 0 db with watercooling, I'm ignorant about water cooling, please help! Aren't there some pumps or something like that to make noise?

:wink: <font color=green> The second time is always better than the first :wink: </font color=green>
 

LCARS

Distinguished
Dec 21, 2002
192
0
18,680
The Eheim water pump only makes a VERY slight electromagnetic hum. about 1 or 4 db at most. You can't hear the impeller turning at all.

My advice is: Don't go cheap, and mount your radiator on the outside the case for best results.
 

svol

Champion
Well with watercooling the best is to have a slow and silent 120mm fan mounted for your radiator... but that will generate sound below 20dB when properly slowed down.

My PC eats so much money that I'm in 'desperate' need of it to buy PC3500 RAM, help Svol with his OC project!
--- PM me for information.
 

kamo

Distinguished
May 3, 2002
67
0
18,630
I hope a big radiator without fan will do the the job for the new xp1700, anyway the fan is the least problem :)
So, I'm looking for the biggest radiator and the slowest pump. Is there a pump with variable speed, btw?


:wink: <font color=green> The second time is always better than the first :wink: </font color=green>
 

scalar

Distinguished
Dec 26, 2002
46
0
18,530
From another thread, do you realize 0 db is no detectable sound at all? At 0 decibels, someone running a fingertip lightly across a sheet of paper 25 ft away from you would make more noise than the computer.

To operate at that level, the computer would have to be dead silent, and the only real way to tell if it is even running would be by looking at the power light.

0 db is probably an unattainable goal, since even if the fans managed to move air at 0 db, the computer would still be noisy any time the CD-ROM or floppy is accessed, etc.
 

kamo

Distinguished
May 3, 2002
67
0
18,630
Yes I read almost all the post from the latest month :)
I meant 0 db for the processor cooler, not for the whole system :) My cpu cooler is now 0 db(what elase could be, without a fan :)) and I was wandering if watercooling can offer 0 bd, obviously not, because of the water pump, then I asked myself "Has anybody tried watercooling without water pump?", tried to calculate how many watts can be dissapated without pump, but don't have any books with formulas, so I post a new thread, if someone can help me ...

:wink: <font color=green> The second time is always better than the first :wink: </font color=green>
 

svol

Champion
The user called kisrobert did this... PM him and ask about his system.

My PC eats so much money that I'm in 'desperate' need of it to buy PC3500 RAM, help Svol with his OC project!
--- PM me for information.
 

vagabond

Distinguished
May 5, 2002
735
0
18,980
Hey svol,

I read his post, wanted to dip my 2 cents in but I said no. Now reading this, is it feasable what kisrobert did? I mean I'm not well-versed on fluid dynamics and its thermal characteristics but correct me if I'm wrong, when the water pump in a car goes, doesn't the engine eventually overheat if it keeps running under a load?

I'm stumped. :eek:

<b><font color=blue>veni,vidi, and ended up in THGC<font color=blue></b>
 

kamo

Distinguished
May 3, 2002
67
0
18,630
Deeply and sincerly thank you Svol :)
I read the post! This forum is great! thank you all guys :)
Now I'm totally convinced about going watercooling, time for shopping :))))))))
Here is the link, Svol and Vagabond are talking about
<A HREF="http://forumz.tomshardware.com/hardware/modules.php?name=Forums&file=viewtopic&p=656829#656829" target="_new">http://forumz.tomshardware.com/hardware/modules.php?name=Forums&file=viewtopic&p=656829#656829</A>
60 degree without a fan and waterpump ... unbeliveble!


:wink: <font color=green> The second time is always better than the first :wink: </font color=green>
 

svol

Champion
In theory if the warm water can move fast enough to the radiator and the radaitor has a great surface with enough air touching it it should be possible. But the amount of heat you produce shouldn't be hiogher then the amount you can move and give to the air otherwise you get overheating just like a car.

But we aren't very far with thermal and fluid dynamics at school at the moment... in about 1-2 months I can tell you much more :wink: .

My PC eats so much money that I'm in 'desperate' need of it to buy PC3500 RAM, help Svol with his OC project!
--- PM me for information.
 

scalar

Distinguished
Dec 26, 2002
46
0
18,530
For the water to flow without a pump, you have to depend on "convection flows". Basically, warm water will rise, cool water will sink.

Convection systems usually use huge pipes or ducts, because the flow is going to be very slow, but also will involve a very large volume of water.

Convection cooling is not going to work very well with quarter-inch tubing. You'd probably have to move up to 1-inch or even larger to get enough mass-flow.

One of the big problems here is your heat-output radiator. Most CPU-cooler radiators use small-diameter pipe in the radiator, and that will slow down the convection flows. You'd likely have to use a radiator with 1" tube to match the water lines, or two or more of the usual small-tube radiators in parallel. The warm water should enter the top of the radiator and the cool water will exit the bottom.

Since this depends on very gentle water movement, any rough spots inside the CPU cooler plate will cause a lot of turbulence and slow the water down. Therefore, the CPU cooler plate should have large, smoothly polished channels inside.

Since convection flow depends on gravity to work, it is severely dependant on tube arrangement. If you have loops in your hose, or a high spot in the middle of a hose, that will very likely stop the whole thing from working. Also, you can't have air pockets anywhere in the system or it won't work.

Since cool water sinks, the cold-return should be lower than the hot-output from the CPU plate. This way as the hot water rises out from the CPU plate the cool water flows in underneath it.

Since there is no pump, you'd need to mount the water tank up high, possibly as high as or higher than the radiator in the top of the case. The tank is attached with a single line to the highest point in the loop. This way when the tank is full the convection loop is also completely full of water, without any trapped air in the radiator or lines.


The general flow works like this:

1. Turn on computer

2. CPU cooler plate slowly warms up.

3. Water in CPU cooler plate warms up.

4. The water in the cooler plate becomes less dense as it warms, and so it becomes buoyant and starts rising up the hot-output tube.

5. Because this is a closed loop with no air gaps, as this water rises it very gently pushes on the water throughout the loop. This is where large tube size and a smooth flow channel is important because when the computer is first turned on the water has to overcome the huge resistence to flow all along the convection path.

5b. If the tubing is too small or there's too much path roughness, the friction may be great enough that the water doesn't really start moving right away, until the CPU plate gets fairly hot and most of the hot-output column is warm. Maybe then there's enough buoyant water to get the flow moving... but if the path is still too small and rough, the CPU will just overheat because the convection flow is either not moving or is way too small to do any good.

6. If all is going well, the rising warm water pushes the cold water above it up, pushes the cool water through the radiator loops, and finally pushes some cool water down the radiator output and into the bottom of the CPU cooler plate, under the warmed water.

7. This cool water entering the bottom of the CPU plate then warms up and also becomes less dense, and the cycle continues.

8. As the water warm rises up to the top of the radiator, it warms the radiator fins, which start a second convection flow of air through the fins. The air takes away the heat, and the water cools, until it exits the bottom back to the CPU plate again.

8b. If the radiator is packed into a tight space, there may not be enough natural airflow through it to remove heat quickly enough. You need large volumes of air to naturally flow over the radiator. If the water cannot be cooled quickly enough, the convection cycle will stall out because the water is still hot and is still floating as it comes out the bottom of the radiator. This warm water balances the warm water over on the "hot" side, and the flow stops.



...in short, this is quite a job to do right.. :)
 

scalar

Distinguished
Dec 26, 2002
46
0
18,530
Yeah that was a long post. Sorry, but if you want the details, that's how it goes. :)


Anyway, you can probably make this concept work better by NOT using a usual tube-radiator to dump the heat.

Instead, you should basically have a water tank and a CPU cooler plate connected together, using big tubes like I mentioned.

The trick is to mount two or more peltier plates to the outside of this water tank, along with huge heatsinks on the "hot side" of the peltiers.

When turned on, the peltier will create about a 30-degree temperature differential. The wall of the water tank will get 15 degrees colder and the heatsink will get 15 degrees hotter.

The hotter the heatsink, the larger the airflow over it, so the peltier will help improve the cooling efficiency of the system, vs the usual finned radiator.

So, the cooled wall of the tank will cool the water in the tank, and the cooled water will sink to the bottom of the tank, "looking" for the lowest spot.

Since the lowest point in the tank leads down to the CPU cooler plate, the cool water will help "jumpstart" the convection flow.... the cooled water in the tank is trying to move down at the same time as the warmed water in the CPU cooler is trying to move up. Together they will get the cycle flowing more quickly.

So the warm water from the CPU plate rises up and enters the top of the tank. It is then is cooled by the peltier on the side wall, and the cooled water sinks to the bottom of the tank. Finally, it heads down out of the tank, into the bottom of the CPU plate.

This convection flow will likely work far better and start flowing much sooner than a usual finned-radiator system that does not use peltiers.

I don't think there's much concern for the water tank freezing or even condensation forming, because the free-air peltier heat sinks are not likely to lose heat quickly enough to drop the water tank temperature much below room temperature.

To improve efficiency, the space between the peltier's heatsinks and the water tank should be insulated with a thin layer of foam. This would prevent the heatsinks from heating the tank and causing an reduction in peltier cooling ability.

-Scalar
 

kamo

Distinguished
May 3, 2002
67
0
18,630
I have some questions here :)
1. Where can I get water element with hose output >1 inch!?
2. Dou know any heatsink with wide inner pipes?

I ran two tests on my comp each one with closed case and idle cpu without fan.
1. Normal case position - 81 degree
_________
|+++++|
|++cpu+|
|+++++|
|+++++|
|+++++|
2.86 degree
--------------------
|++CPU+++++++++++++|
|+++++++++++++++++|
|+++++++++++++++++|
--------------------
Even though the air is 'trapped' into the heatsink, it finds its way out
My point is that the hot water can sink(harder but possible), because of the small density of the hot water compared to cold water.
Anyway that's just a remark, I'm not trying to altercate :)
Good idea about petliers :), but it's too "home-made" i think :)
<A HREF="http://members.easyspace.com/coolk6/jacket.htm" target="_new">
Imagine Alpha 8045 with that!</A>

:wink: <font color=green> The second time is always better than the first :wink: </font color=green><P ID="edit"><FONT SIZE=-1><EM>Edited by kamo on 12/31/02 06:32 AM.</EM></FONT></P>
 

scalar

Distinguished
Dec 26, 2002
46
0
18,530
Well, you're trying to do something unusual and uncommon, so it's <b>has</b> to be a "homemade" job. Nobody makes convection-flow coolers, so you have to do the legwork on this yourself.

Heatsinks with at least 1/2 inch pipes are very common. They're called "heater cores" and are used inside every automobile in existence. Go wander through an auto salvage yard or an auto parts store catalog, and you'll find these cores.

Very large diameter plastic hose can be obtained from many hardware stores, though you'll have better luck at a store for agriculture businesses (dairy, grain, horses, pigs, chickens, etc) since those usually have a need for big hoses.

I'll be the first to admit that a CPU cooler with large diameter hose connections designed for convection cooling is going to be hard to find. This may very well be something you'll have to build yourself, or have a friend help you build. Do you have access to a machine shop, or know someone who does?

You may be able to get help from a local high school that has a metalshop with mills, lathes, etc. Some technical colleges offer a "community night" where people can come in and use the equipment for their own little hobby projects (like building miniture model steam engines, etc).

Is this starting to look like too much work? Well, that's the way it is with things you can't just go out and buy and slap into the case.

Woohoo! ASCII-art!

Convection flow CPU-cooler:
<pre> .''''.
: : <---- hot water rises .-----.
|'....'| out of top tube ----> | |
.--| |--. | |
| | | | __| |_
| '----' | | |
|------------| | |
| | | |
| | _____| |
| .''''. | | | |
| | | | | | |
| | | | cool water --> |_____| |
'--|.''''.|--' <- enters here |__________|
: :
: :
'....'
</pre><p>
 

LCARS

Distinguished
Dec 21, 2002
192
0
18,680
I was just at a nuclear reactor that was cooled by this very same principle. It worked very well, dead silent too.

<A HREF="http://www.geocities.com/lcars_e/reactor.html" target="_new">http://www.geocities.com/lcars_e/reactor.html</A>
 

kamo

Distinguished
May 3, 2002
67
0
18,630
Happy New Year pals :)
Scalar you're inborn artist ...
Thank you for your suggestions, they are valuable to me.
I have a friend , who could help. Although I don't have much spare time, may be I'll do something without petliers... It will be fun, if not else :)
What do you thing if the cooler is with just one big hose output on the top, i.e if I remove the hose output for the cold water ?It will be easier to make I think. Something else is bothering me, won't the weight of the water cause too much pressure with very wide hoses and break my socket/motherboard? I don't know if I told you than physics is not my speciality at all :)

:wink: <font color=green> The second time is always better than the first :wink: </font color=green>
 

knowan

Distinguished
Aug 20, 2001
991
0
18,980
Sounds like what you want is a heat pipe. Can I make a suggestion here: Don't use water. Instead use a phase change material which boils at roughly 40 degrees. As the liquid boils to gas it will move up to the radiator much quicker. Then it condenses back into a liquid and moves back down again. A phase change heat pipe will be much more efficient than a convection-propelled liquid heatpipe.

There was a thread in here a few months ago (I think it was started by Crashman) about how to build one of these, and even a link to a webpage of someone who actually built one.

Possible problems: You want the phase change to occur in the 35-40 degree range. This will let the gas turn back into liquid in your radiator at room temperature. BUT before you rush out and buy some methanol you also have to consider the fact that as the liquid turns to gas it will increase the air pressure inside your heatpipe, which will raise the boiling point of the liquid. It's a fine balance.

On the plus side you will only need one pipe, not 2 like you would in a liquid convection model. It also does not need to be a particularly wide pipe, but it shouldn't be too long or else the liquid might boil off before it reaches the bottom of the pipe (after condensing).

On the minus side you might need to incorporate some sort of a pressure release valve so that you can modify the internal pressure of the system until you get the temperature just right. And you will likely still have to build your own block and radiator, as the radiator must be straight (it order to get the condensed liquid to drip back down) and the block will need to have a large open area with one opening on "top" (actually the side) giving way to the pipe.

--------------
Knowan likes you. Knowan is your friend. Knowan thinks you're great.
 

knowan

Distinguished
Aug 20, 2001
991
0
18,980
<A HREF="http://forumz.tomshardware.com/hardware/modules.php?name=Forums&file=viewtopic&p=641231#641231" target="_new">http://forumz.tomshardware.com/hardware/modules.php?name=Forums&file=viewtopic&p=641231#641231</A>

--------------
Knowan likes you. Knowan is your friend. Knowan thinks you're great.
 

kamo

Distinguished
May 3, 2002
67
0
18,630
Yeee that's interesting, thank you :)
Now I have a so much to read :).
May be I will use water, because I'm novice and easy assemble is what I'm looking for.
Once again thank you

:wink: <font color=green> The second time is always better than the first :wink: </font color=green>
 

kisrobert

Distinguished
Dec 16, 2002
17
0
18,510
As I see this thread I saw my name several times...
I was not sure that the system will work without pump, but the system is working...
no need 1" tube and other wonders...
My system: XP2000+ not OC.
The pipes: 9mm internal diameter transparent plastic tube
The radiator: 11x12x20 cm (withxlengthxheight)
the home made block: aluminium 20 (or 25 not remember)mm thick, 60x80mm (withxlength) with 4 drilled holes, 2 connector (water in/out)
balance tank: a 1 liter plastic soft drink bottle. Will change to 0.2-0,05 liter.
About 0.7-1 liter water.
The bottom of the radiator is 30 cm over the CPU.
The case is open (need to drill two holes for the pipes, and I will close it)

Opertaing temperature
Idle:
Water 38-42 C (at the top of the radiator)
34-38 C (at the bottom of the radiator)
CPU 49-51C
Full load:
Water 45-47C (at the top of the radiator)
40-42C (at the bottom of the radiator
CPU 58-62C
Room: 22-24 C
 

Crashman

Polypheme
Former Staff
You could try making a passive heatpipe and mounting it outside the case!

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>