No, I'm saying that iSegway was venting at you, for nothing in particular that you posted. I didn't see anything by you that was particularly inflammatory (unless I missed out on what you edited). I don't see why you are being so apologetic. I didn't see you do anything wrong.
No, I'm saying that iSegway was venting at you, for nothing in particular that you posted. I didn't see anything by you that was particularly inflammatory (unless I missed out on what you edited). I don't see why you are being so apologetic. I didn't see you do anything wrong.
Oh. Man this is a mess! Anyway you're sortof right. The reason I removed some is that people who know more then me about this (like you) have already explained why the water-cooling needs air-cooling and such.
This directed to the very first post. Having water cooling is just a novelty in a computer. There are many dangers to water cooling. If there was a leak in a seal, and drips all over your motherboard, well ya, you know what might happen. and just like in the other posts, the heat from the water, has to go somewhere.
That's the odd part. I don't think he is willing to acknowlege that the heat has to be dissipated somewhere after it is in the water and how that will be accomplished (and consequentially how that affects his argument). He doesn't recognize the "liquid part" and the "air part" as integral components of a single thermal management system. Anyone who forces the recognition of that issue is summarily "ruining" his topic.
This thread has become so, I don't know, emotional I guess, that I wasn't going to debate it any longer. But I just thought of an example whereby NOBODY can negate the fact that a liquid-cooled computer is ultimately air-cooled. Here it is!
You put a new 2.0 Gig G5 Power Mac in a room and have it perform the most processor intensive task possible. (This is in some sort of lab.) Then you suck as much air out of the room as possible so there's almost none. What would happen? It would overheat, naturally. There would not be enough air flowing over the heatsink to remove the heat. I don't know what computers do when they overheat but its not positive.
Ok, now you do the same with thing with a liquid-cooled computer*!!!
Remember, no air. Hmmm. The water gets hotter and hotter and hotter. Where is all that heat to go? Finally, the poor computer, bless its heart, has to overheat just like the water-cooled G5.
*This assumes a realistic, self-contained source of coolant. Not the Pacific Ocean or even a bathtub.
So there. This doesn't argue for or against liquid cooling. Just that any design has to take into account how to dissapate heat into the air.
One thing that a computer just cant deal with is heat. Heat makes it unstable. Whether the heat is from the Memory, the CPU, the chipset of the motherboard or of the video card. Companies spend so much money developing new technology to improve their products. In return, the consumer buys their products to keep their computers running stable.
The CPU of the computer, whether it may be Intel or AMD, it generally the largest source of heat. There are two basic ways to keep the heat down. We refer to them as "air" or active cooling. This category consists of heatsinks and fans. The second category is "H2O" or water cooling. Water cooling is the more expensive way to do things, but also the more efficient. Overclockers primarily use water cooling because it does a much better job of keeping the temperature low than air. Throughout this article a detailed explanation of pros and cons will be explained in a way that beginning overclockers will understand.
Air Cooling
The majority of computer users go with active cooling. This is the most basic method. There is a lot of technology behind active cooling. For example, some companies choose to make their coolers completely out of copper. This is not always the best way to do do it. Thermaltake for example, their volcano 11, has a completely copper base with copper fins. Copper is a great conductor; it is much better than aluminum. Aluminum although, is also used commonly in heatsink designs. Aluminum is much lighter and it does the job well. A great example of an aluminum based heatsink would be Vantec's Aeroflow. The heatsink is made of aluminum, although they use a copper insert right above where the CPU core would be placed.
The largest factor in a great heatsink, I believe are the fins and the fan. Fins increase surface space allowing the heat to dissipate very quickly. The fan's job is to cool the fins as fast as i can. a fan with a very high CFM rating is usually the best way to go. An idea that Thermaltake used to increase the efficiency of the fan is the ducting mod. The center of the fan produces no airflow. With the ducting mod, the air is shot through a large plastic duct that hits every part of the heatsink evenly.
It is also very important to not have a scorching hot case temperature. It wont help much if the fan continually blows 90 degree air onto the heatsink. Most people believe that the most important way to cool a case is a large amount of intake fans. Well, that helps, although it is more important to remove the heat from the case. Scientifically, there is no such thing as the transfer of cold air. Heat is the basis for cold and warm. Heat is removed by exhausting it from the rear of the case. Hot air rises, so another great way to remove hot air from your case is a top exhaust fan. It works much like a chimney, without the smoke (I hope). My case utilizes the idea of a top fan. When i place my hand over it, all I feel is warm air. I guess that means that it is working.
To ensure maximum airflow within the case, there are many extra accessories. Round cables are commonly used, because they don't block a 2" spot. You can also buy wire sleeve and loom to keep all your wire in a nice tight bunch.
Pros
Air cooling doesn't have any risks like water cooling does. There is no risk of it leaking, and its very reliable because of this.
Downfalls
Everything sounds great so far, but air flow equals noise. There are a couple of ways to battle this. Your first option would be to use quiet fans, such as Vantec's Stealth series of fans. These have DBA's between 20 and 21 depending on the size you get. For most users, such as myself, like to use a rheobus or switched baybus. This allows you to use any fan you want, but alter the voltage it is given to reduce the speed which in effect reduces CFM and DBA. For about $15 USD these are cheaper than some of the high quality, low noise fans.
A large portion of overclockers use water cooling. Water cooling is a great way to keep your CPU temperature very low. There are a few companies that make these water cooling kits, such as Swiftech, Koolance, Thermaltake, and Innovatek. A basic kit consists of a pump, a water block (CPU Cooler) and a radiator. Video card GPU blocks and motherboard chipset blocks are also available.
There are a couple of ways to hook up the water circuit. The most common way, and they way I did it, was to hook the pump to the water block, then the output from the water block to the radiator, then back to the pump. The way a water cooling system keeps your CPU cool, is basically by cooling the water when it passes through the radiator. The radiator works much like a heatsink, in the way it use fins for surface space. The amount of water is also a factor in the surface space. A good pump is also very important. A pump with a high GPH is better than a pump with a low GPH because the faster it moves the water, the cooler it is. It is also very important to put a good fan on the radiator; a fan that pushes a lot of air. Radiators come in many sizes. Some are built to fit an 80mm fan, and some are fit to support up to two 120mm fans. The larger the radiator, the better for the most part.
CPU blocks are made in many different ways. They are mostly made of copper. There are many kinds of blocks, for example, Swiftech uses diamond pin technology which increases surface space, much like fins on a heatsink. Xoxide's X-Aqua CPU block uses a common design we like to call a swirl. The water goes around a small circular design. Another commonly used design is a maze. This causes turbulence within the water, which is a good thing. A lot of people have trouble understanding the difference between friction and turbulence. Friction causes heat, but turbulence does not. Friction is when the water rushes up against the tubing or another accessory in the water circuit. Turbulence is when the water gets thrown around upon itself, which continues refreshing water with itself. The dictionary definition that pertains to this topic is, "Departure in a fluid from a smooth flow." From my tests between a spiral block and a block with diamond pin technology, the block with diamond pin technology worked better. It is important to get a good water block because it has to lift the heat from the CPU then dissipate it.
To sum this all up, the way a water cooling system works is by continually circulating the water though a radiator.
Pros
Water cooling is very quiet. It doesn't require a 75 CFM fan that screams a 48 DBA at you. A decent water cooling kit will keep your CPU temperature below any heatsink on the market today, allowing you to overclock to the limit.
Downfalls
The biggest problem and the reason why not all overclockers use water cooling is because of the price. Some of the low end kits cost around $100 USD. The more advanced kits cost upward of $200. Another large problem with using water around your computer, is just that. Water is conductive and electricity and water don't always make the best mix.
After reading this article, you should have plenty of knowledge on how the entire system of cooling works. By now you should have also come to a conclusion on which method you prefer. If you plan to overclock and get a little more juice out of your CPU, you can still use a heatsink. Some heatsinks such as the Vantec Aeroflow are so well designed, they are able to keep you CPU at a decent temperature even while slightly overclocked. For a more experienced overclocker, who wants more than a couple hundred megahertz out of his or her CPU, water cooling would be the way to go. I was able to get my Athlon XP 1700+ up to 2.0ghz while still at 35degrees C. It is all preference. If this is a hobby, or you think it is worth buying a water cooling kit, by all means, go for it.
Having water cooling is just a novelty in a computer.
Have you been reading the articles I have been posting? Unless we come up with a room temperature superconductor, water cooling is inevitable in computers. Water cooling is not a novelty -- it is an inevitability(or some other method than "air cooling"(quotes are for the idiots)).
Quote:
There are many dangers to water cooling.
There is a laptop shown in this thread that uses water cooling... if it can be done with a laptop, that undergoes much more severe shaking, it can certainly be done with a desktop. We live with electricity constantly and the human body is made primarily of water. Isn't that dangerous?
Quote:
If there was a leak in a seal, and drips all over your motherboard, well ya, you know what might happen.
Do you know what might happen?
Do you know there are fluids that can cool your computer that don't conduct electricity and won't harm your computer if leaked?
Quote:
and just like in the other posts, the heat from the water, has to go somewhere.
Yes... and water and a radiator will remove that heat from the chip and transmit it to the air MORE EFFICIENTLY and QUIETLY than an "air cooled"(quotes are for the idiot/s) systems heatsink and fan.
My bicycle has tubeless ceramic rims and hydraulic disc brakes. Many of the other extreme downhill bikers I know also run hydraulic systems on their bicycles. It is typical for these systems to go years without ever springing a leak. This is while rolling down clifflike slopes and occasionally failing to land a 5 foot drop. The G forces on these things are incredible. Limbs and ribs will typically break in a crash while the hydraulic system remains intact. It is impossible to damage the hydraulic lines with your bare hands and the connectors are nearly as tough.
You see, an inexpensive leakless system is possible. It is even possible to design one such that leaks wouldn't spill onto any circuit boards. It is also possible to design them such that very little machining is required and where the fluid is contained entirely in reservoirs internal to a solid heatsink. Pumps aren?t even required in some designs.
Fluid-cooling has already been proven as cheap, reliable, and highly effective.
The catch is that fan-cooling is also cheap, reliable, highly effective, and runs 0.000000000000000% risk of leaking fluid inside your computer.
The earliest cars and motorbikes also had passive radiators (no pumps) where convection carries hot liquid up to cool and cold liquid moved about the hydraulic circuit without pumps.
I could actually see a sealed system in a laptop to augment the often heavily finned heatsyncs now employed inside most laptop cases. If the performance were only the equal of fans and no better, such a passive hydraulic convection would have the advantage of allowing a completely environmentally sealed laptop. NO vents for dust or spilled coca-cola to seep into... HIGHLY DESIRABLE!
Everything sounds great so far, but air flow equals noise.
Not exactly. Turbulence in air flow and high fan RPM = noise. But let's go with "air flow equals noise" for the sake of examining the consistency of this article's logic.
Quote:
Water Cooling
[...]
The amount of water is also a factor in the surface space. A good pump is also very important. A pump with a high GPH is better than a pump with a low GPH because the faster it moves the water, the cooler it is.
How loud is the pump? How much heat does it generate? Do these increase with the power of the pump?
Quote:
It is also very important to put a good fan on the radiator; a fan that pushes a lot of air.
And "air flow equals noise," so you're back to square one, right?
Quote:
To sum this all up, the way a water cooling system works is by continually circulating the water though a radiator.
Which can accomodate one 80mm or even two 120mm fans, because water cooling requires air cooling, and air cooling requires air flow, and "air flow equals noise."
The G5 just cuts out the middleman and puts the radiator right on top of the CPUs, with the same two 120mm fans that your radiator requires sitting right nearby doing the same thing they'd have to do in a water cooled system. The case is designed to control the airflow so that it's quiet, effective, and more or less guaranteed to move as much air as the radiators (heat sinks) need. You'd need a similar setup to assure that the radiator cooling fans were as quiet.
Quote:
Water cooling is very quiet. It doesn't require a 75 CFM fan that screams a 48 DBA at you.
Well, that depends on the radiator, doesn't it? The PowerMac G5 doesn't require a 48DBA fan either. The PowerMac G5 has 9 huge fans spinning at 10% speed to move air through broad zones specifically to achieve quiet operation (because it's not so much that air flow=noise as it is that turbulence and high RPM = noise). Compare that to the problem of cooling one hot spot (the radiator) with one or two fans. In addition, the G5's solution accomodates expandability by staying out of the way of all expansion ports (you want a PCI card? add one) and having lots of headroom to accomodate extra heat without stressing itself at all.
Quote:
The biggest problem and the reason why not all overclockers use water cooling is because of the price. Some of the low end kits cost around $100 USD. The more advanced kits cost upward of $200. Another large problem with using water around your computer, is just that. Water is conductive and electricity and water don't always make the best mix.
More precisely, the impurities in water are conductive.
But that's not the only problem, either. Some of the pictures shown have pipes running to specific chips on the GPU, for example. This works against expandability because it means that any time you replace the video card you have to mess with the cooling system. In the G5, you just replace the video card. Or add the PCI card. Or the hard drive. Or RAM. Or whatever you please. Liquid cooling is great for wicking heat away from specific places (especially hard-to-reach places, like the inside of an engine or an iBook or an overstuffed mini-ATX case) to specific places - which is why Apple and Hitachi both have used liquid cooling in notebooks - but it's a mess in a tower whose components and heat profile keep changing.
Reports from the field indicate that the PMG5 runs cool and quiet, at the expense of being large. But I'm not convinced that water cooling would help at all. It might become necessary if you tried stuffing a G5 into something like a Cube, but you'd still have to move X amount of air over the radiator; if you want to do it quietly, you need the channels it moves through to be fairly large and arrow-straight, and if there's too much heat for passive cooling to dissipate, then you need a fan, and then you're back to the problem of making fans run quietly - and the solution to both quiet airflow and quiet fan operation are embodied in the current G5.
Additionally, the pictures show other devices which are heatsunk and not connected to the liquid cooling system. This means that the case's existing air cooling system must remain to cool these devices.
Also, I don't see the water cooler doing much good for the hard drives, memory modules, north/south bridge etc. Just because a particular device is *really* hot doesn't mean that less hot devices don't need cooling as well. liquid cooling then becomes a supplement, not a replacement.
As I have pointed out many times, my pismo *does* use a liquid cooling system: it has a fluid-filled heat pipe to draw heat from the processor (located deep in the center of the laptop, with no room for a vaned heat sink) and direct it to a radiator near the vent and the fan. Just because there is no pump doesn't mean that it is not totally analogous to other liquid cooling solutions--the heat itself is the engine to circulate the fluid. Simple and cheap. Most importantly however: liquid cooling was used in my Pismo not because the processor was particularly hot, but rather because it was in an inaccessible location. *That* is where liquid cooling really shines: pulling heat out of hard-to-reach places and moving it to where it can be more easily dissipated.
Oh, and don't believe everything you read. You don't need to be an engineer or an expert in thermal management to write a article. Where I work, we make half-terabit internet/ATM switches (MUCH hotter than even the most pimped-out overclocked PC in a poorly designed case) and we've never needed to resort to complex, expensive exotic cooling methods (yet). Just because companies make PCs with liquid cooling (Hitachi, etc.), it does not follow that the decision was based purely on engineering necessity--it could just as likely have been a marketing mandate to distinguish the product with a degree of pose-value.
It seems that some of this dispute has sprung not from questionable technical details but in defining what we're talking about.
The reason that I might seem pro liquid-cooling is that I'm approaching this from a scientific rather than a business perspective.
Liquid cooling is feasible and effective although currently non-optimal for the vast majority of computing related uses. However, it shows tremendous promise. I can envision a day when computers no longer need fans. They can simply move heat to the exterior of the case via closed, passive liquid-cooling systems. These designs will likely use the case or portions of the casing as the primary heat sink. Drilling miniature passageways through this heat sink, filling it with liquid, and then sealing it, would be a cost effective way to increase the thermal conductivity of a passive cooling system.
I agree with most of the objections to the feasibility of liquid cooling; yet, only when these objections are made pragmatically with regard to currently shipping products.
To claim that liquid cooling will always be inherently inferior or impractical, is just plain foolish.
Interesting anecdote: I have a friend with a 1991 Grand Marquis. It can be counted on to overheat, without fail, precisely when, and only when, the gas tank falls to less than 1/5 full.
To claim that liquid cooling will always be inherently inferior or impractical, is just plain foolish.
I agree.
As an engineer, I must always seek the optimal solution to a given problem. "Optimal" does not always mean technically ideal; it often means a balance of technical adequacy, cost, manufacturability, etc.
The reason the vast majority of PC's don't use liquid cooling (yet) is that it is currently not the "optimal" solution in most cases. Now, in my Pismo, where the processor was not accessible, the design engineer decided that a liquid-filled heat pipe was the right solution. Some day we may see liquid cooling in a standard PC, but I don't think that will be within the next 5 years of so; good passive thermal design is still ahead of the curve. When you have to keep costs down and move product, a good engineer will not do something solely because it is cool! (no matter how much we might want to, sometimes! )
P.S. defiler: Since I see that you are from Pittsburgh, I work at Marconi up in Cranberry (the old FORE Systems site).
Liquid cooling is feasible and effective although currently non-optimal for the vast majority of computing related uses. However, it shows tremendous promise. I can envision a day when computers no longer need fans. They can simply move heat to the exterior of the case via closed, passive liquid-cooling systems.
Just out of curiosity, why are you restricting yourself to liquid cooling? Apple's already rolled out several sterling examples of passively-cooled systems, some of which use liquid in various places and many of which don't. If you can stick all the heat-producing components directly against a heat sink whose fins circumscribe a convection tunnel (the Cube) why bother with liquid?
However, since passively cooled systems are passive, you have to know which components are going to be how hot where - the cooling system and the machine layout are interdependent. Anything that disrupts airflow or introduces a hot spot in a place the cooling can't account for - or anything that introduces more heat in a known place than the system is designed for, like some Cube CPU upgrades - is going to gum up the works. Given this, what we'll see is what we already see from Apple, only better as technology evolves: AIOs (including laptops) will use passive cooling largely but not exclusively, because it's nice to have a fan around in extremis - better noise than a meltdown. Anything designed for internal upgradability will use passive cooling strategically (the heatsinks on the CPUs in the G5) but rely on active cooling - the fans in the PCI and drive bay zones can adjust to whatever circumstances they find themselves in, because they're active.
And, of course, you can obviate a lot of the issues, regardless of which system you eventually settle on, by doing the motherboard layout and case design right. If you design things so that expansion cards actually block airflow between the CPU and the system fan (a legendary design flaw in early Gateway desktops) then you've got more basic problems to deal with.
Amorph, I still think that a liquid cooling system would make a really neat laptop solution. FOr some of the reasons that you mention, I think it's good for laptops in a way where it probably isn't good for desktops. ie, no one is going to open up their laptop to install any significant heat generating components.
Why did car engines move to liquid cooling if not for the efficiency of the systems versus air cooling? (there is the emissions thing, but that aside)
packaging efficiency, liquid cooled engines can be mounted many ways that air cooled engines could not,a nd they could be made smaller because liquids could absorb and dissipate more heat within a closed ciruit than air could.
Now, for something as small as a computer, ther are no miracles, and you still need to dissipate heat from the liquid, usually by exposing it to air, but you could gain the key advantage of an environmentally sealed unit if some sort of liquid colled laptop were devised.
Liquid in the circuit could be exposed to air through vents or chambers which are completely sealed from the electronics inside the case -- think of hat tubes passing through a bulkhead. On the inside of the bulkhead lies the electrical componentry. On the outside lies a large heatsink exposed to vents and outside air. any water or dust that gets into the vents, still can't get into the electronics. You can't really do that with an air cooled system. You might need more circulation than a passive circuit could provide (hence a small pump), and you might need a special liquid or refridgerant, but if the goal is to use liquid cooling not to replace air cooling "just because" or for rabid overclocking, but to do something different, then I think it a worthwhile objective.
I'd love to have a reasonably priced, enviromentally sealed laptop.
Amorph, I still think that a liquid cooling system would make a really neat laptop solution. FOr some of the reasons that you mention, I think it's good for laptops in a way where it probably isn't good for desktops. ie, no one is going to open up their laptop to install any significant heat generating components.
Why did car engines move to liquid cooling if not for the efficiency of the systems versus air cooling? (there is the emissions thing, but that aside)
packaging efficiency, liquid cooled engines can be mounted many ways that air cooled engines could not,a nd they could be made smaller because liquids could absorb and dissipate more heat within a closed ciruit than air could.
Yes, but a laptop isn't an engine. (First, keep in mind that Apple already has used liquid cooling in their laptops. Fawkes pointed that out right near the top of the thread, and again and again. They didn't hype it, but they did it. Despite some of the rhetoric in this thread, this is not a religious issue, and it shouldn't be.) Laptops are flat, half-empty slabs, which means that if you lay them out right it's not very far from whatever component to the air that will eventually carry away the heat. Any elaborate cooling systems will be designed to nudge heat from one place to another not all that far away (as with the Pismo CPU). You can use air cooling to some extent (although it's not that easy, it's easier than in a solid block of metal like an engine), and direct passive cooling (heat sinks/radiators) to some extent, and closed-circuit liquid cooling to some extent - in other words, you should simply use whatever gets the job done. The water-cooled Hitachi is a gimmick to the extent that it uses water to accomplish what something else could have done better. How, for instance, is it any better than the almost completely passively-cooled (and far less expensive) iBook?
Also, engines get to have lots and lots of air blown over their radiators one way (movement) or another (a great big loud blower fan). O'Grady's "Extreme PowerBook" tales notwithstanding, laptops do not generally find themselves hurtling through the air at 65MPH, broad side to the wind, and so they can't assume any air movement that they aren't specifically engineered to create themselves. Water doesn't absorb heat easily or give it up quickly, so it doesn't work very well in a system at rest. Hence the extremely modest specs on the Hitachi, and the big loud fan that kicks in when your car idles.
Quote:
Now, for something as small as a computer, ther are no miracles, and you still need to dissipate heat from the liquid, usually by exposing it to air, but you could gain the key advantage of an environmentally sealed unit if some sort of liquid colled laptop were devised.
Again, why not just use really cool components and arrange them so that the heat they do generate goes directly to the (aluminum) case? The specs on most hardened laptops are pathetic by mainstream standards because as soon as you close the system to that extent, you have a worst-case scenario as far as heat dissipation goes. Liquid would not change that - nothing would, really. The problem always comes back to air cooling, and if you assume still air (which, in this case, you'd have to) then you don't have a lot to work with.
That said, the original clamshell iBook was almost there.
Oh, I just can't resist a bit of trivia... Introduced in 1985, the Cray-2 Supercomputer (1.9 GFlops, 2 GBytes memory) used a fluid called Florinert (used in heart transplants) for cooling. It was pumped throughout the entire computer and then cooled by a refrigeration unit (see picture below, just to the right of the woman--note the cool transparent liquid-filled bubbly thing with the Cray logo). The stuff is completely non-conducting, so the computer was essentially immersed in it. This ability was demonstrated to the press by plunging a running Mac 128 (just introduced, remember) into a pool of the stuff!
For you other geeks out there, the Cray-1 (133 MFlops), introduced in 1976, was made entirely out of *2* chips: 5-input NANDs and 1k memories.
Comments
Originally posted by Randycat99
No, I'm saying that iSegway was venting at you, for nothing in particular that you posted. I didn't see anything by you that was particularly inflammatory (unless I missed out on what you edited). I don't see why you are being so apologetic. I didn't see you do anything wrong.
Oh. Man this is a mess! Anyway you're sortof right. The reason I removed some is that people who know more then me about this (like you) have already explained why the water-cooling needs air-cooling and such.
You put a new 2.0 Gig G5 Power Mac in a room and have it perform the most processor intensive task possible. (This is in some sort of lab.) Then you suck as much air out of the room as possible so there's almost none. What would happen? It would overheat, naturally. There would not be enough air flowing over the heatsink to remove the heat. I don't know what computers do when they overheat but its not positive.
Ok, now you do the same with thing with a liquid-cooled computer*!!!
Remember, no air. Hmmm. The water gets hotter and hotter and hotter. Where is all that heat to go? Finally, the poor computer, bless its heart, has to overheat just like the water-cooled G5.
*This assumes a realistic, self-contained source of coolant. Not the Pacific Ocean or even a bathtub.
So there. This doesn't argue for or against liquid cooling. Just that any design has to take into account how to dissapate heat into the air.
Air VS. Water:
Author: Kevin (FoxCokk)
Editor: Dan (N3M3S1S)
One thing that a computer just cant deal with is heat. Heat makes it unstable. Whether the heat is from the Memory, the CPU, the chipset of the motherboard or of the video card. Companies spend so much money developing new technology to improve their products. In return, the consumer buys their products to keep their computers running stable.
The CPU of the computer, whether it may be Intel or AMD, it generally the largest source of heat. There are two basic ways to keep the heat down. We refer to them as "air" or active cooling. This category consists of heatsinks and fans. The second category is "H2O" or water cooling. Water cooling is the more expensive way to do things, but also the more efficient. Overclockers primarily use water cooling because it does a much better job of keeping the temperature low than air. Throughout this article a detailed explanation of pros and cons will be explained in a way that beginning overclockers will understand.
Air Cooling
The majority of computer users go with active cooling. This is the most basic method. There is a lot of technology behind active cooling. For example, some companies choose to make their coolers completely out of copper. This is not always the best way to do do it. Thermaltake for example, their volcano 11, has a completely copper base with copper fins. Copper is a great conductor; it is much better than aluminum. Aluminum although, is also used commonly in heatsink designs. Aluminum is much lighter and it does the job well. A great example of an aluminum based heatsink would be Vantec's Aeroflow. The heatsink is made of aluminum, although they use a copper insert right above where the CPU core would be placed.
The largest factor in a great heatsink, I believe are the fins and the fan. Fins increase surface space allowing the heat to dissipate very quickly. The fan's job is to cool the fins as fast as i can. a fan with a very high CFM rating is usually the best way to go. An idea that Thermaltake used to increase the efficiency of the fan is the ducting mod. The center of the fan produces no airflow. With the ducting mod, the air is shot through a large plastic duct that hits every part of the heatsink evenly.
It is also very important to not have a scorching hot case temperature. It wont help much if the fan continually blows 90 degree air onto the heatsink. Most people believe that the most important way to cool a case is a large amount of intake fans. Well, that helps, although it is more important to remove the heat from the case. Scientifically, there is no such thing as the transfer of cold air. Heat is the basis for cold and warm. Heat is removed by exhausting it from the rear of the case. Hot air rises, so another great way to remove hot air from your case is a top exhaust fan. It works much like a chimney, without the smoke (I hope). My case utilizes the idea of a top fan. When i place my hand over it, all I feel is warm air. I guess that means that it is working.
To ensure maximum airflow within the case, there are many extra accessories. Round cables are commonly used, because they don't block a 2" spot. You can also buy wire sleeve and loom to keep all your wire in a nice tight bunch.
Pros
Air cooling doesn't have any risks like water cooling does. There is no risk of it leaking, and its very reliable because of this.
Downfalls
Everything sounds great so far, but air flow equals noise. There are a couple of ways to battle this. Your first option would be to use quiet fans, such as Vantec's Stealth series of fans. These have DBA's between 20 and 21 depending on the size you get. For most users, such as myself, like to use a rheobus or switched baybus. This allows you to use any fan you want, but alter the voltage it is given to reduce the speed which in effect reduces CFM and DBA. For about $15 USD these are cheaper than some of the high quality, low noise fans.
--------------------------------------------------------------------------------
Water Cooling
A large portion of overclockers use water cooling. Water cooling is a great way to keep your CPU temperature very low. There are a few companies that make these water cooling kits, such as Swiftech, Koolance, Thermaltake, and Innovatek. A basic kit consists of a pump, a water block (CPU Cooler) and a radiator. Video card GPU blocks and motherboard chipset blocks are also available.
There are a couple of ways to hook up the water circuit. The most common way, and they way I did it, was to hook the pump to the water block, then the output from the water block to the radiator, then back to the pump. The way a water cooling system keeps your CPU cool, is basically by cooling the water when it passes through the radiator. The radiator works much like a heatsink, in the way it use fins for surface space. The amount of water is also a factor in the surface space. A good pump is also very important. A pump with a high GPH is better than a pump with a low GPH because the faster it moves the water, the cooler it is. It is also very important to put a good fan on the radiator; a fan that pushes a lot of air. Radiators come in many sizes. Some are built to fit an 80mm fan, and some are fit to support up to two 120mm fans. The larger the radiator, the better for the most part.
CPU blocks are made in many different ways. They are mostly made of copper. There are many kinds of blocks, for example, Swiftech uses diamond pin technology which increases surface space, much like fins on a heatsink. Xoxide's X-Aqua CPU block uses a common design we like to call a swirl. The water goes around a small circular design. Another commonly used design is a maze. This causes turbulence within the water, which is a good thing. A lot of people have trouble understanding the difference between friction and turbulence. Friction causes heat, but turbulence does not. Friction is when the water rushes up against the tubing or another accessory in the water circuit. Turbulence is when the water gets thrown around upon itself, which continues refreshing water with itself. The dictionary definition that pertains to this topic is, "Departure in a fluid from a smooth flow." From my tests between a spiral block and a block with diamond pin technology, the block with diamond pin technology worked better. It is important to get a good water block because it has to lift the heat from the CPU then dissipate it.
To sum this all up, the way a water cooling system works is by continually circulating the water though a radiator.
Pros
Water cooling is very quiet. It doesn't require a 75 CFM fan that screams a 48 DBA at you. A decent water cooling kit will keep your CPU temperature below any heatsink on the market today, allowing you to overclock to the limit.
Downfalls
The biggest problem and the reason why not all overclockers use water cooling is because of the price. Some of the low end kits cost around $100 USD. The more advanced kits cost upward of $200. Another large problem with using water around your computer, is just that. Water is conductive and electricity and water don't always make the best mix.
After reading this article, you should have plenty of knowledge on how the entire system of cooling works. By now you should have also come to a conclusion on which method you prefer. If you plan to overclock and get a little more juice out of your CPU, you can still use a heatsink. Some heatsinks such as the Vantec Aeroflow are so well designed, they are able to keep you CPU at a decent temperature even while slightly overclocked. For a more experienced overclocker, who wants more than a couple hundred megahertz out of his or her CPU, water cooling would be the way to go. I was able to get my Athlon XP 1700+ up to 2.0ghz while still at 35degrees C. It is all preference. If this is a hobby, or you think it is worth buying a water cooling kit, by all means, go for it.
Having water cooling is just a novelty in a computer.
Have you been reading the articles I have been posting? Unless we come up with a room temperature superconductor, water cooling is inevitable in computers. Water cooling is not a novelty -- it is an inevitability(or some other method than "air cooling"(quotes are for the idiots)).
There are many dangers to water cooling.
There is a laptop shown in this thread that uses water cooling... if it can be done with a laptop, that undergoes much more severe shaking, it can certainly be done with a desktop. We live with electricity constantly and the human body is made primarily of water. Isn't that dangerous?
If there was a leak in a seal, and drips all over your motherboard, well ya, you know what might happen.
Do you know what might happen?
Do you know there are fluids that can cool your computer that don't conduct electricity and won't harm your computer if leaked?
and just like in the other posts, the heat from the water, has to go somewhere.
Yes... and water and a radiator will remove that heat from the chip and transmit it to the air MORE EFFICIENTLY and QUIETLY than an "air cooled"(quotes are for the idiot/s) systems heatsink and fan.
My bicycle has tubeless ceramic rims and hydraulic disc brakes. Many of the other extreme downhill bikers I know also run hydraulic systems on their bicycles. It is typical for these systems to go years without ever springing a leak. This is while rolling down clifflike slopes and occasionally failing to land a 5 foot drop. The G forces on these things are incredible. Limbs and ribs will typically break in a crash while the hydraulic system remains intact. It is impossible to damage the hydraulic lines with your bare hands and the connectors are nearly as tough.
You see, an inexpensive leakless system is possible. It is even possible to design one such that leaks wouldn't spill onto any circuit boards. It is also possible to design them such that very little machining is required and where the fluid is contained entirely in reservoirs internal to a solid heatsink. Pumps aren?t even required in some designs.
Fluid-cooling has already been proven as cheap, reliable, and highly effective.
The catch is that fan-cooling is also cheap, reliable, highly effective, and runs 0.000000000000000% risk of leaking fluid inside your computer.
I could actually see a sealed system in a laptop to augment the often heavily finned heatsyncs now employed inside most laptop cases. If the performance were only the equal of fans and no better, such a passive hydraulic convection would have the advantage of allowing a completely environmentally sealed laptop. NO vents for dust or spilled coca-cola to seep into... HIGHLY DESIRABLE!
Originally posted by iSegway
Air Cooling
[...]
Everything sounds great so far, but air flow equals noise.
Not exactly. Turbulence in air flow and high fan RPM = noise. But let's go with "air flow equals noise" for the sake of examining the consistency of this article's logic.
Water Cooling
[...]
The amount of water is also a factor in the surface space. A good pump is also very important. A pump with a high GPH is better than a pump with a low GPH because the faster it moves the water, the cooler it is.
How loud is the pump? How much heat does it generate? Do these increase with the power of the pump?
It is also very important to put a good fan on the radiator; a fan that pushes a lot of air.
And "air flow equals noise," so you're back to square one, right?
To sum this all up, the way a water cooling system works is by continually circulating the water though a radiator.
Which can accomodate one 80mm or even two 120mm fans, because water cooling requires air cooling, and air cooling requires air flow, and "air flow equals noise."
The G5 just cuts out the middleman and puts the radiator right on top of the CPUs, with the same two 120mm fans that your radiator requires sitting right nearby doing the same thing they'd have to do in a water cooled system. The case is designed to control the airflow so that it's quiet, effective, and more or less guaranteed to move as much air as the radiators (heat sinks) need. You'd need a similar setup to assure that the radiator cooling fans were as quiet.
Water cooling is very quiet. It doesn't require a 75 CFM fan that screams a 48 DBA at you.
Well, that depends on the radiator, doesn't it?
The biggest problem and the reason why not all overclockers use water cooling is because of the price. Some of the low end kits cost around $100 USD. The more advanced kits cost upward of $200. Another large problem with using water around your computer, is just that. Water is conductive and electricity and water don't always make the best mix.
More precisely, the impurities in water are conductive.
But that's not the only problem, either. Some of the pictures shown have pipes running to specific chips on the GPU, for example. This works against expandability because it means that any time you replace the video card you have to mess with the cooling system. In the G5, you just replace the video card. Or add the PCI card. Or the hard drive. Or RAM. Or whatever you please. Liquid cooling is great for wicking heat away from specific places (especially hard-to-reach places, like the inside of an engine or an iBook or an overstuffed mini-ATX case) to specific places - which is why Apple and Hitachi both have used liquid cooling in notebooks - but it's a mess in a tower whose components and heat profile keep changing.
Reports from the field indicate that the PMG5 runs cool and quiet, at the expense of being large. But I'm not convinced that water cooling would help at all. It might become necessary if you tried stuffing a G5 into something like a Cube, but you'd still have to move X amount of air over the radiator; if you want to do it quietly, you need the channels it moves through to be fairly large and arrow-straight, and if there's too much heat for passive cooling to dissipate, then you need a fan, and then you're back to the problem of making fans run quietly - and the solution to both quiet airflow and quiet fan operation are embodied in the current G5.
Also, I don't see the water cooler doing much good for the hard drives, memory modules, north/south bridge etc. Just because a particular device is *really* hot doesn't mean that less hot devices don't need cooling as well. liquid cooling then becomes a supplement, not a replacement.
As I have pointed out many times, my pismo *does* use a liquid cooling system: it has a fluid-filled heat pipe to draw heat from the processor (located deep in the center of the laptop, with no room for a vaned heat sink) and direct it to a radiator near the vent and the fan. Just because there is no pump doesn't mean that it is not totally analogous to other liquid cooling solutions--the heat itself is the engine to circulate the fluid. Simple and cheap. Most importantly however: liquid cooling was used in my Pismo not because the processor was particularly hot, but rather because it was in an inaccessible location. *That* is where liquid cooling really shines: pulling heat out of hard-to-reach places and moving it to where it can be more easily dissipated.
Oh, and don't believe everything you read. You don't need to be an engineer or an expert in thermal management to write a article. Where I work, we make half-terabit internet/ATM switches (MUCH hotter than even the most pimped-out overclocked PC in a poorly designed case) and we've never needed to resort to complex, expensive exotic cooling methods (yet). Just because companies make PCs with liquid cooling (Hitachi, etc.), it does not follow that the decision was based purely on engineering necessity--it could just as likely have been a marketing mandate to distinguish the product with a degree of pose-value.
It seems that some of this dispute has sprung not from questionable technical details but in defining what we're talking about.
The reason that I might seem pro liquid-cooling is that I'm approaching this from a scientific rather than a business perspective.
Liquid cooling is feasible and effective although currently non-optimal for the vast majority of computing related uses. However, it shows tremendous promise. I can envision a day when computers no longer need fans. They can simply move heat to the exterior of the case via closed, passive liquid-cooling systems. These designs will likely use the case or portions of the casing as the primary heat sink. Drilling miniature passageways through this heat sink, filling it with liquid, and then sealing it, would be a cost effective way to increase the thermal conductivity of a passive cooling system.
I agree with most of the objections to the feasibility of liquid cooling; yet, only when these objections are made pragmatically with regard to currently shipping products.
To claim that liquid cooling will always be inherently inferior or impractical, is just plain foolish.
Interesting anecdote: I have a friend with a 1991 Grand Marquis. It can be counted on to overheat, without fail, precisely when, and only when, the gas tank falls to less than 1/5 full.
Originally posted by dfiler
To claim that liquid cooling will always be inherently inferior or impractical, is just plain foolish.
I agree.
As an engineer, I must always seek the optimal solution to a given problem. "Optimal" does not always mean technically ideal; it often means a balance of technical adequacy, cost, manufacturability, etc.
The reason the vast majority of PC's don't use liquid cooling (yet) is that it is currently not the "optimal" solution in most cases. Now, in my Pismo, where the processor was not accessible, the design engineer decided that a liquid-filled heat pipe was the right solution. Some day we may see liquid cooling in a standard PC, but I don't think that will be within the next 5 years of so; good passive thermal design is still ahead of the curve. When you have to keep costs down and move product, a good engineer will not do something solely because it is cool!
P.S. defiler: Since I see that you are from Pittsburgh, I work at Marconi up in Cranberry (the old FORE Systems site).
Originally posted by dfiler
Liquid cooling is feasible and effective although currently non-optimal for the vast majority of computing related uses. However, it shows tremendous promise. I can envision a day when computers no longer need fans. They can simply move heat to the exterior of the case via closed, passive liquid-cooling systems.
Just out of curiosity, why are you restricting yourself to liquid cooling? Apple's already rolled out several sterling examples of passively-cooled systems, some of which use liquid in various places and many of which don't. If you can stick all the heat-producing components directly against a heat sink whose fins circumscribe a convection tunnel (the Cube) why bother with liquid?
However, since passively cooled systems are passive, you have to know which components are going to be how hot where - the cooling system and the machine layout are interdependent. Anything that disrupts airflow or introduces a hot spot in a place the cooling can't account for - or anything that introduces more heat in a known place than the system is designed for, like some Cube CPU upgrades - is going to gum up the works. Given this, what we'll see is what we already see from Apple, only better as technology evolves: AIOs (including laptops) will use passive cooling largely but not exclusively, because it's nice to have a fan around in extremis - better noise than a meltdown. Anything designed for internal upgradability will use passive cooling strategically (the heatsinks on the CPUs in the G5) but rely on active cooling - the fans in the PCI and drive bay zones can adjust to whatever circumstances they find themselves in, because they're active.
And, of course, you can obviate a lot of the issues, regardless of which system you eventually settle on, by doing the motherboard layout and case design right. If you design things so that expansion cards actually block airflow between the CPU and the system fan (a legendary design flaw in early Gateway desktops) then you've got more basic problems to deal with.
Why did car engines move to liquid cooling if not for the efficiency of the systems versus air cooling? (there is the emissions thing, but that aside)
packaging efficiency, liquid cooled engines can be mounted many ways that air cooled engines could not,a nd they could be made smaller because liquids could absorb and dissipate more heat within a closed ciruit than air could.
Now, for something as small as a computer, ther are no miracles, and you still need to dissipate heat from the liquid, usually by exposing it to air, but you could gain the key advantage of an environmentally sealed unit if some sort of liquid colled laptop were devised.
Liquid in the circuit could be exposed to air through vents or chambers which are completely sealed from the electronics inside the case -- think of hat tubes passing through a bulkhead. On the inside of the bulkhead lies the electrical componentry. On the outside lies a large heatsink exposed to vents and outside air. any water or dust that gets into the vents, still can't get into the electronics. You can't really do that with an air cooled system. You might need more circulation than a passive circuit could provide (hence a small pump), and you might need a special liquid or refridgerant, but if the goal is to use liquid cooling not to replace air cooling "just because" or for rabid overclocking, but to do something different, then I think it a worthwhile objective.
I'd love to have a reasonably priced, enviromentally sealed laptop.
Originally posted by Matsu
Amorph, I still think that a liquid cooling system would make a really neat laptop solution. FOr some of the reasons that you mention, I think it's good for laptops in a way where it probably isn't good for desktops. ie, no one is going to open up their laptop to install any significant heat generating components.
Why did car engines move to liquid cooling if not for the efficiency of the systems versus air cooling? (there is the emissions thing, but that aside)
packaging efficiency, liquid cooled engines can be mounted many ways that air cooled engines could not,a nd they could be made smaller because liquids could absorb and dissipate more heat within a closed ciruit than air could.
Yes, but a laptop isn't an engine. (First, keep in mind that Apple already has used liquid cooling in their laptops. Fawkes pointed that out right near the top of the thread, and again and again. They didn't hype it, but they did it. Despite some of the rhetoric in this thread, this is not a religious issue, and it shouldn't be.) Laptops are flat, half-empty slabs, which means that if you lay them out right it's not very far from whatever component to the air that will eventually carry away the heat. Any elaborate cooling systems will be designed to nudge heat from one place to another not all that far away (as with the Pismo CPU). You can use air cooling to some extent (although it's not that easy, it's easier than in a solid block of metal like an engine), and direct passive cooling (heat sinks/radiators) to some extent, and closed-circuit liquid cooling to some extent - in other words, you should simply use whatever gets the job done. The water-cooled Hitachi is a gimmick to the extent that it uses water to accomplish what something else could have done better. How, for instance, is it any better than the almost completely passively-cooled (and far less expensive) iBook?
Also, engines get to have lots and lots of air blown over their radiators one way (movement) or another (a great big loud blower fan). O'Grady's "Extreme PowerBook" tales notwithstanding, laptops do not generally find themselves hurtling through the air at 65MPH, broad side to the wind, and so they can't assume any air movement that they aren't specifically engineered to create themselves. Water doesn't absorb heat easily or give it up quickly, so it doesn't work very well in a system at rest. Hence the extremely modest specs on the Hitachi, and the big loud fan that kicks in when your car idles.
Now, for something as small as a computer, ther are no miracles, and you still need to dissipate heat from the liquid, usually by exposing it to air, but you could gain the key advantage of an environmentally sealed unit if some sort of liquid colled laptop were devised.
Again, why not just use really cool components and arrange them so that the heat they do generate goes directly to the (aluminum) case? The specs on most hardened laptops are pathetic by mainstream standards because as soon as you close the system to that extent, you have a worst-case scenario as far as heat dissipation goes. Liquid would not change that - nothing would, really. The problem always comes back to air cooling, and if you assume still air (which, in this case, you'd have to) then you don't have a lot to work with.
That said, the original clamshell iBook was almost there.
For you other geeks out there, the Cray-1 (133 MFlops), introduced in 1976, was made entirely out of *2* chips: 5-input NANDs and 1k memories.