Using Radiators to Cool CPUs

dan writes "Overclockers Australia have a review up of the CPU Radiator Zen, a new approach to cooling your toasty CPU's. Rather than taking the traditional approach of a heatsink with lots of fins and a noisy 7,000rpm fan it uses radiator/heat pipe technology. The implementation of the unit is a bit flawed, but it is interesting to see where the technology is heading.. and if it can be done right I personally think this is where it will end up."

Has this...

[rmadmin]

already been done? I do believe that one of the Cray models had a 'liquid cooled CPU', or something to that effect. I never saw the specifics on how it was done, if it was just the CPU, or if it was the whole machine being cooled that way.

I also remember someone else (Penguincomputing?) having the '1st commercial liquid cooled PC', which was a 1.6 Ghz(2x800Mhz) Dual Athlon.

Either way, its really cool to see this same technology replicated for private use.

Re:Submerging circuit board in an inert liquid

[Sir_Real]

It was called Flourinert actually, and the reason it isn't used is that at high temps it turns into musturd gas... ;)

Basically it was the next thing coming.

[BuBu_]

Well guys if you sat down and read the article all the way through and saw what was going on you could basically understand that this isn't that much of an innovation, just tweaking an already proven practice. People have been using water cooling for years, and basically this just takes water coolings model and just makes it self-contained (at least how I understood it) the only flaw that they are going to run into is keeping the coolant cool at all times which will be hard since in water cooling setups there is a return pipe to the cooler/recycled water..

Over all I give em two thumbs up for at least tweaking a proven practice, but then again they need some more work to really get the idea going.

-bubu

Re:Submerging circuit board in an inert liquid

[Raptor CK]

You mean Fluorinert? [google.com]

I remember seeing an overclocking/cooling experiment with this somewhere. Cool stuff, until it turns to mustard gas.

Re:laptops already have this, no?

[x-ntric_one]

I know that there is a company called Thermacore does it for Dell laptops. I have been able to play around with their heat pipes and they are amazing. I believe Thermacore does product for any laptop product higher than a C600. So to answer the question.. YES! Are there anymore company's like this besides these 2?

They gave it a positive review?

[512k]

and it physically broke their CPU and their mobo..if it were me, I wouldn't have anything positive to say about it.

Also reviewed...

[Daniel Rutter]

...by, ahem, me, as part of my monster cooler comparison [dansdata.com].

The Zen review is on page four [dansdata.com].

Re:Silent?

[Chakat]

It looks like its still farily loud. It's got two fans on it, so it's going to make a good amount of noise. Unfortunately, it's not the fanless system I was hoping for when I checked it out.

Plus, it looks like its not good for OCers. The device is very poorly designed, and busted caused damage to both the CPU and Mobo.

Re:Submerging circuit board in an inert liquid

[xcjohn]

you may be thinking of flourinert. This is actually used. I remember seeing a setup down at the San Diego Supercomputer Center that cooled itself by litterally passing flourinert over certain parts. iirc they called it, simply, a waterfall... and was on one of the Cray's

Re:Submerging circuit board in an inert liquid

[Anonymous Coward]

It's also very expensive. Several hundered dollars a gallon. You'd probably be better off just buying a faster processor.

Health Issues

[squaretorus]

There is a documented link between low level noise and hearing and stress levels in those spending long times exposed to them.

The hum of a fan, whatever it is cooling, is often at a level that you might strain to hear clearly. It is these levels that can cause hearing strain. This is similar to eye strain when you need glasses and can give you monster headaches.

Many articles in New Scientist, among others, have covered this - normally relating to office environments.

Symptoms can be migranes, and a persistant ringing / humming sound when you are in a silent room / trying to sleep. Its worth checking out if you feel any of these because the long term stress levels can be harmful.

I don't know if its a problem for babies - but I know the effects are magnified many fold if you are exposed for long periods, i.e. all night. So I wouldn't leave the machine on 24/7 even if the baby doesn't seem bothered by it 'just in case'.

Re:The Turd Report 11/07/2001

[neal n bob]

amen to that. People are ignorant of the rules all the time here. Lucky for me there are a lot more women than men in my section of the building. There is a nice safe haven bathroom - it has an open window to the outside so it is very serene. But there is one guy who I swear forces himself to take 2 craps a day - grunting, stanky odor, the whole production. I usually try to do a fly by - if it is clear, I go take care of business. One time I went in while he was there - he was in the middle of a pause and I did not know it was him - then right as I sit down he starts grunting and groaning again. A tip for you kids - forcing one out is bad for your colon. You don't need to wait until it is poking its head out but you should not have to force it - you need to eat more fiber.

Re:Submerging circuit board in an inert liquid

[rschwa]

aha! here's the link:
Wacky flourinert fun! [octools.com]

Heat pipes != water and other basics

[enkidu]

First of all the radiator in you car and the type of heat pipes used in this CPU cooler are similar in principle but radically different in construction and efficiency. The radiator in your car is an open system relying on water+antifreeze. The heat pipe used in many industrial applications (including laptop cooling, my Dell CPx has one that I can see through a grill) is a SEALED pipe with a wick and liquid (usually alchohol) inside. The liquid vaporizes at the operating temperature at any place where there is heating going on and condenses where there is cooling going on. The cool thing about heat pipes is that the heat transfer happens REALLY REALLY fast. For home computing applications (<1 meter) it should be instantaneous. This allows you to move the heat away from where it can do damage to any place you want (within the limits of cost and space of course.). This is also why thin laptops don't fry their CPU's instantly: using heat pipes they can spread the heat sink around and away from the CPU.

Check out a NASA tech brief [nasatech.com], Thermacore [thermacore.com] a company that makes them and MIC [mic.com] another company that makes them for more information.

Re:another way to do this

[CausticPuppy]

Good idea in theory but that won't move heat quickly enough for powerful cpu's. Also, your case would get very warm and act like an oven around your hard drives and other components. You'd have to drastically increase airflow through the case in order to keep the rest of the system cool, which would defeat the noiseless aspect of the copper wire cooling method. With cpu heatsinks, the heated air is usually vented right out without circulating to the other components.
Without a real heatsink that has a large air-exposed surface area a relatively short distance from the chip, you'll wind up with an impressive heat gradient across the wire.
An Athlon chip will get up to roughly 600-700 degrees (F) within just a few seconds of power-on if no heatsink is attached. The copper cloth wire might bring it down a bit but you're still talking about having something exposed inside your case that's hot enough to melt wire insulation and probably catch dust on fire (after your system crashes of course).

What you don't know.... WAS : Hell, *I* know...

[firewort]

You never owned an air-cooled VW, did you?

Those radiators are oil coolers. Air cools the cylinders, which are finned like motorcycle cylinders. Oil does the rest of the job, besides lubricating, it soaks up much of the heat. Running the oil through a small radiator makes a large difference in some climates, but is usually unnecessary. Shoot, in Israel (commonly accepted as a pretty hot climate) they run without 'em just fine.

The reasons to add the radiator for oil-cooling are:
exposing the oil to a finned, air cooled radiator cools the oil off faster, leading to a cooler engine, and

having greater oil capacity means that the oil is more resistant to heating up, and adding the oil cooler adds more oil capacity.

And remember, the air-cooling on the VW is the same as it is on air-cooled porsches, a fan on the back of the generator (alternator) driven by a belt off the crankshaft. Pretty darn efficient.

Liquid cooled mainframes are obsolete

[Animats]

IBM mainframes used to be water-cooled. Cray mainframes used to be Freon-cooled. One Cray machine of the early 1990s actually ran immersed in Fluorinert. All those machines were plumbing nightmares. IBM considered it a big step forward when the water-cooled mainframes were replaced by air-cooled mainframes. Far fewer maintenance headaches. You don't want to go to liquid cooling on the desktop unless there's no alternative.

Solid-state Peltier-effect coolers are much more promising. They actually refrigerate, they have no moving parts, and they don't make noise.

A problem with one of the graphs.

[flatcat]

Anyone else notice that the graphic on page 2 that shows the processor under load is messed up? The coloration for the MC462+delta and zen radiator are swapped. This graph shows the Zen as the worst of the coolers.

Re:This got bad reviews, but look here!

[carm$y$]

How about utilizing the Peltier Effect - a.k.a. thermo electric cooling?

You still have to get rid of the heat somehow - and thus the peltiers come with bulky radiators plus noisy fans themselves.
As a bonus, you also have to handle somehow the condensation problem - the peltiers being able to refrigerate (or at least to go down to pretty low temperatures, close to zero Celsius).

So, it's a promising technology, but it's not ready yet.

Re:Heat pipes != water and other basics

[billanderson71]

Heat pipes have been used to cool laptops since the mid to late 90's. The heat pipe transfers the heat from the cpu to the shield around the bottom of the laptop (which is one reason that our laptop often does a good job of warming your lap). The advantage is that cooling fans can be eliminated, which prolongs battery life.

Most of the heat pipes used in electronics cooling have a copper envelope, copper wick, and use water as the working fluid (operating under a partial vacuum). The advantage is not that they are really fast (the velocity of the water vapor is on the order of meters/second) but that they are very nearly isothermal. (The temperature drop in the heat pipe is essentially negligible, with small temperature drops occurring due to the heat conduction through the copper walls and wicks). This allows you to transfer heat over relatively long distances before removing the heat.

A second advantage is that heat pipes can be used to reduce the heat flux. The heat flux (watt/cm^2) out of the chip is fairly high. On the other hand, air cooling is relativley inefficient, so a low heat flux is preferred. By using a larger condenser than evaporator, the heat flux can be adjusted to match the capacity of the cooling media. This will become more important in the future, as the heat flux from the chips continues to rise. Some experimental designs water designs have cooled several hundred W/cm^2, which is higher than chips should reach in the near future (high temperature heat pipes - have removed in excess of 50,000 W/cm^2)

The best introductory book is Heat Pipes, by Chiu, but it is out of print. Dunn and Reay have a reasonable book on heat pipes, but it is quite expensive (~ $100).

Thermal generator runs against thermodynamics

[SysKoll]

Atilla,

The average Intel CPU dissipate a waste heat much greater than the few watts absorbed by your average fan. So the idea seems reasonable.

Alas! The laws of thermodynamics often fly in the face of reasonable ideas. See, if you want to passively cool off the CPU, all you have to do is let it radiate its heat. But what you seem to wish for here is some kind of device that actively cools off that CPU, by taking some of that waste heat as its energy source. That's called a thermic engine. And here, thermodynamics get you: You can generate power from a heat source only if you have a cold "sink". All thermic engines work by getting heat from a heat source and moving it to a heat sink. E.g., for a car, the heat sink is the radiator.

Here, your contraption would use the CPU as a heat source and would require some sink, such as, oh, a radiator. Maybe with a fan. Which is exactly what we are trying to avoid.

So it's a nice catch-22. But think about it: if it worked, we would have big ships moving smoothly on all oceans, powered by the extracted heat of sea water and leaving a trail of ice cubes in their wake...

-- SysKoll