These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

Out of Pod Experience

 
  • Topic is locked indefinitely.
 

The laws of thermodynics... Whos got the guts to figure this out...

Author
Xui Meili
Perkone
Caldari State
#1 - 2011-11-14 18:05:34 UTC
I am building myself a water cooled system. Currently my focus is only the CPU as the heat blocks haven't changed much in regard of fitting. (perhaps I am wrong, but look at video card heat sinks, they got a new one out every time a new video card is put on market... waste of money)

Anyways, I am trying to make a passive setup, without using a Rad as a way to cool the water. More of a holding tank that will contain X amount of water, with an aeration system going through the water to help draw heat out of the water.

The question I have for just a CPU block, how much water you guys think I would need to successfully setup a passive water system, that would be able to keep the water at a 90 degree (30 Celsius) temps under load?

My idea is to make an acrylic box that will hold roughly 2 liters of water, (Im even of thinking of recycling a 2 liter coke bottle).

The question comes down to would that be sufficient amount of water to allow cooling to occur at a fast enough rate (including air bubbles running through the water) to keep a system cool enough alone? Or am I stuck just forking over more money for this overly priced computer stuff. Im not cheap, but for a stupid water pump for this application they want 60 bucks, when I can go to a local hardware store and get it for 20-30 bucks. Over inflated pricing ftw =T
Sidus Isaacs
Center for Advanced Studies
Gallente Federation
#2 - 2011-11-14 23:00:43 UTC
why build a passive one? would not making passive massive copper tube be better?
Taedrin
Federal Navy Academy
Gallente Federation
#3 - 2011-11-15 00:32:45 UTC
I'm not an engineer, but I think what you are asking for is impossible. Let's do some Internet Science! (Someone who actually knows what they are talking about, please correct me if I'm wrong)

Use Newton's Law of Cooling:

q = h * A * dt

q = rate of heat transfer in Watts (We need 100 watts for a processor which consumes 100 Watts of power)
h = Convection coefficient (for free standing water, this is 20 to 100 W per m^2*K)
A = surface area being cooled in m^2 (your CPU block in this case)
dt = temperature difference between surface being cooled and the bulk fluid (in Kelvin).

100 = 100 * A * dt
OR
dt = 1/A

Let's say that your CPU block can give you .01 m^2 of surface area for cooling (NO idea if this is accurate, but this is equal to 100 cm^2)

If this is the case, your fluid (water) will need to be 100 degrees colder than the conductive surface of the CPU block.

Let's say the conductive surface is smaller.... say 10cm^2 or .001 m^2. This means that your bulk fluid will need to be 1000 degrees colder than the conductive surface of the CPU block - a physical impossibility in this case since if your CPU is at a dangerous temperature (like 90C), the water will have to have a negative temperature (which is meaningless in classical physics.

Let's say the conductive surface is larger... say 1000cm^2 or .1 m^2. This would mean that you would only need a temperature difference of 10 C - which is much more doable. So the question then becomes, how big of a cooling surface do you have on your CPU block?

Chances are, however, that you are going to need to drastically increase your convection coefficient by using forced convection from a water pump.
Zindela
Aegeonix Systems
#4 - 2011-11-15 02:01:10 UTC  |  Edited by: Zindela
From what I understand, a passive-liquid cooling system would be very hard to build efficiently.

What you'll need is some way to ensure constant flow over the CPU, could be convection driven, but at 30C, you're not going to be getting much flow. You also need a way to dissipate all the heat that the water pulls off the CPU. Again, if you don't want to buy a radiator, you'll need a holding reservoir with a fairly large surface area in order to dissipate as much heat as possible, as quickly as possible. In that case, your lower limit of temperature is theoretically the room temperature. However, at temperature differences of about 10C(assuming room temperature is about 20C), your convection flows are going to be very slow. With slow convection--> low volumetric flow, ---> not much heat transfer---> higher temperature on your CPU.

Which, actually will lead back to higher flow rates, more heat transfer, and therefore lower CPU temperatures... haha. The system would be self-regulating, up to a certain point, however I am VERY skeptical that it could keep a CPU at 30C under load without a pump.


I could design a passive system without much trouble, but without some fairly detailed, time consuming mathematical stuff, I couldn't give you any idea of how practical it would be. And I think your reservoir would need to be significantly bigger than an empty 2 liter. Off the cuff guess- several gallons.


If you're planning on gaming for long periods of time, you'd need to calculate exactly the flow rates through your piping, the heat transfer from the CPU to the coolant, the rate of heat dissipation from the coolant reservoir. There are probably a couple variables I'm forgetting. You'd also need to be in a stable climate controlled room, because seasonal variations in room temperature would affect how much heat can be dissipated from the reservoir.

My recommendation - don't go passive.


Honestly, you don't need to buy the flashy built-for-purpose pumps and radiators, if you can build them yourself, you can probably save a bunch of money, while getting something that doesn't look as good.

Edit: If you buy/build a radiator, it's much easier, and may actually be practical.

Disclaimer: I'm studying electrical/computer engineering, not mechanical. I've taken a few classes that have dealt with thermodynamics, but I'm not an expert. if a Mech person can add something or correct what I've written that would probably be good.
Taedrin
Federal Navy Academy
Gallente Federation
#5 - 2011-11-15 03:42:45 UTC
Actually, I found this website of a guy who apparently did exactly this: he created a fanless and pumpless passive water cooling system, built using parts bought at Home Depot. Problem is it's old and such a system might not be effective at cooling today's CPUs which generate roughly twice as much heat in a smaller package.
Lithalnas
Dirt 'n' Glitter
Local Is Primary
#6 - 2011-11-15 04:40:54 UTC
I am building myself a water cooled system. Currently my focus is only the CPU as the heat blocks haven't changed much in regard of fitting. (perhaps I am wrong, but look at video card heat sinks, they got a new one out every time a new video card is put on market... waste of money)

Ok i guess, video cards do change over time and I tend to think that the AM3 and LGA 1155 bolt patterns will be around for a while. Cooling the CPU via water cooling is a good idea in an age of 250W chips.

Anyways, I am trying to make a passive setup, without using a Rad as a way to cool the water. More of a holding tank that will contain X amount of water, with an aeration system going through the water to help draw heat out of the water.


If you are trying to do something with no pump, its not going to work. If you wanted to do something with a drop tower to do something more passive, its not going to work well. Fans, even slow ones, are exponentially better. I am thinking that using a PVC pipe stuffed with fishtank bio balls and using a 200mm fan to pull air up while water trickles down.

The question I have for just a CPU block, how much water you guys think I would need to successfully setup a passive water system, that would be able to keep the water at a 90 degree (30 Celsius) temps under load?


Most blocks are fairly good but a few are exceptional. I would go over to the [H]ardOCP forums and look at what other people are doing. I would suggest an EK copper waterblock(somepeople think their nickle block is flawed) or even the swifttech one that's on newegg isn't half that bad.
http://www.frozencpu.com/cat/l3/g30/c85/s139/list/p1/Liquid_Cooling-Water_Blocks_CPU-CPU_Water_Blocks_-Page1.html

My idea is to make an acrylic box that will hold roughly 2 liters of water, (Im even of thinking of recycling a 2 liter coke bottle).
I love ghetto mods but i would throw a silver shaving in there as a biocide. Residual stuff may encourage growth. Note that all your options where water is exposed to open air, you need to be really ontop of your maintenance. Not to mention you are going to rust out the fins of you waterblock due to electrolysis.

The question comes down to would that be sufficient amount of water to allow cooling to occur at a fast enough rate (including air bubbles running through the water) to keep a system cool enough alone? Or am I stuck just forking over more money for this overly priced computer stuff. Im not cheap, but for a stupid water pump for this application they want 60 bucks, when I can go to a local hardware store and get it for 20-30 bucks. Over inflated pricing ftw =T'

The pumps you see, like the swiftech 655 or the 35x are both good pumps that are made for long life and low noise.

but to answer your question on how much of a flow you need, well we would have to talk specifics, tell me what processor you want to use and what kind of overclock you want to have. After that tell me what pump you are going to use. If you use one of the before mentioned ones your GPM is roughly 4 or 5. Then its down to trying to figure out the size of the drop tower or the max wattage of what you can extract from the waterblock.

https://www.facebook.com/RipSeanVileRatSmith shoot at blue for Vile Rat http://community.eveonline.com/devblog.asp?a=blog&nbid=73406

Zindela
Aegeonix Systems
#7 - 2011-11-15 06:50:58 UTC
Taedrin wrote:
Actually, I found this website of a guy who apparently did exactly this: he created a fanless and pumpless passive water cooling system, built using parts bought at Home Depot. Problem is it's old and such a system might not be effective at cooling today's CPUs which generate roughly twice as much heat in a smaller package.



Yeah, I found that website as well. Only real difference is that the guy on the site used a radiator, the OP doesn't want to. However, a radiator in a passive setup will drastically reduce the space needed vs an open air cooling reservoir method like mine.

Still, in the interest of saving space, saving coolant, and actually getting cool temperatures, I say get a pump and radiator.
Sturmwolke
#8 - 2011-11-15 11:20:48 UTC  |  Edited by: Sturmwolke
I don't think there's an easy way to sustain cooling with what the OP has in mind.
Water needs to evaporate (aka change phase) in an open system, you'll still need to fill up periodically (which will become a real hassle long term).
(it's the basis for the so called Zeer pots - Google around for videos
here's another good informative link on Zeer pots - http://practicalaction.org/zeerpots)

For closed systems, it's already been answered by the few posters above.

Personally, if you're doing DIY cooling, might as well go all the way. Big smile
Do it with a (cheap) car radiator and pump.

Quick Google:

http://www.xtremesystems.org/forums/showthread.php?189928-Yes-Another-Car-Radiator-Thread....Major-56K-Warning!
(holy crap, he really went big)

http://www.instructables.com/id/Computer-liquid-cooling-with-Car-parts/?ALLSTEPS

http://homegardensupply.kitchenpopular.com/cool-computer-cooling-with-car-radiator-extreme/
(using an aquarium pump)

http://photonics.engr.uga.edu/watercooling/index.html
(water cooling 101 - very informative)
Xui Meili
Perkone
Caldari State
#9 - 2011-11-17 19:31:31 UTC  |  Edited by: Xui Meili
I apologize for the misconception of the word passive.

Here are the items that follow.

16 liter (4 gallons roughly) container
1 pump at 150 gph
1 air pump to push bubbles through water to help remove heat from water.

Possibly, cut 2 holes from one fan blowing in 120MM and another with that blowing out to to help create a circulation of air on the surface of the water as the bubbles do their thing.

So far, thank you for all those who replied, its helpful as this is more of a custom idea, rather than going out and just throwing money around. Its a lot cheaper too, just before I go dump about 100$ and find out it doesn't work, just trying to get thoughts and info of those who may actually have a better grasp of thermodynamics. In my head, I feel it will work, and I know their is math out their to do it, but honestly, its beyond me =T

Edit: As I have spent more and more time just thinking about it. I think I now truly understand the fundamentals of a radiator... Its a way for heat to be transferred without the effects of large amounts of evaporation since it is contained in the tubing of said radiator.

All the stuff I have read so far is helpful, but they don't really ever say WHY is works that way. Kinda had to do some more thinking. Its fun though =D