Slashdot Mirror
Liquid Coolent System For PCs
Deadric writes
"Looking for a cheap way to cool your computer system? Really
want to overclock that Celeron? Just want to make sure your
processer keeps cool? Go check out this page and see if
this home-made liquid cooled do it your self process is
for you. "
A week from now we'll hear the story of how some kid set his house on fire trying this, hehe.
This is the coolest link that I've seen in a long time, just wish I had heard of this before I cooked my petium 200
hah, wow i really like the idea of liquids flowing around stuff that was never meant to even come close to water. Sounds like a good idea to me, not. I'm sure if it wasn't home made and it didnt' require crossing your fingers i might be compelled to get it. Or maybe not.
- ben
Just a thought, but what if you were able to distill water REALLY well, and then submerse your whole motherboard in it. Would the water pick up minerals from the components? Or would the water harm the equipment some how (I can't imagine floppy drives, etc. spinning too well under water)?
I imagine we'll be seeing open sided cases with a regular box fan mounted on one side of the case. Seriously though, why risk overclocking your processor for the tiny bit of performance increase you might see (if your lucky).
I think you spelled coolant wrong. I think it's spelled with an 'a'...
Has underwriter's labs tested this setup yet? I just don't know if it is really safe.
Especially considering that an overclocked PPC 750 is considered to be running hot when it is running at 31C. In fact, it seems that a 350MHz PPC750 overclocked to 400MHz seems to have a running temperature of 19C (assuming that the case it is in has decent airflow.)
What amazes me is that people stick with shitty intel hardware when there are better solutions around
Back in my more hardware oriented days, i remember coming across a certain model of Compaq pentium based laptop that featured a liquid filled cooling system.
;-)
heh.. its odd seeing a radiator on a laptop
Instead of messing with the copper sheet and solder, you could use something called BalsaFoam. It's a synthetic resin foam used for industrial design modeling, and other things including underwater insulation. It carves very easily and resists heat up to 225 degrees fahrenheit.
You could just take a block, carve a hole out to fit the heatsink, drill some holes, insert some in- and out- tubes, and glue it to the heatsink. Apply liberal amounts of aquarium sealant to plug any leaks. Voila, you're done. Fewer seams means fewer potential leaks.
It would have an added benefit in that the foam would act as an insulator. It wouldn't get very cold, helping prevent condensation.
Check it out at http://www.bfoam.com/Balsa-Foam.htm
I'd guess it could also be used to make a mold, which could then be used to create a solid block of clear acrylic plastic which could be used as the enclosure.
Keeps all the condensation off
This has nothing to do with studying electronics. It is basic chemistry. Ionic molecules that dissolve in water conduct electricity. Pure H2O does not.
hello???? last I heard, pure H2O does not conduct electricity........
this may seem trivial, but the second and third pages have a discrepancy ... the tubes run into the top on one, and the sides, with valves on the other.
how cool (no pun) would it be if we could miniturize a ac, so like forcing freon through the grooves on the heatsink.
wouldn't work with the pII's with that build-on-fan crap either.
humph.
I have heard that you can get these but of the "you give them your first child" kind for pc that will be used in extream conditions like the space shuttle or alien space ships.......
I wonder if you can also get forced air and and a larger intercooler for max hourse power (i mean cpu's)
Where did you las see a motherboard running on a 110 V power supply? Most motherboard run on 12 V and 5 V.
I saw this relly neat component a while ago when doing a tour thing at NIST in Boulder, CO. It's about the size/shape of a matchbook, and has two wire leads. It uses some kind of crystal viabration to move heat from one side to the other. If you hold it and send current through it, one side immediately gets cold and the other hot. Reverse the current and heat is moved the other way. I think it was called a thermoelectric module or something like that. Can anyone help on this?
Some company advertised in Popular Science a cooler that just used a fan and one of these (which converts to a heater at the flip of a switch).
Here's what you do: make a bigger one of these fitted to the top of the processor. Sandwich it between the processor and the heatsink, and maybe make the heatsink bigger. Keep the fan on top of the heatsink and use an AC adapter to power the TEM.
Charlie O'Keefe
thegrappler@usa.net
Someone (perhaps you) mentioned this on Slashdot a few days agao (ie. an overclocked PPC running at 18 or 19 degrees Centigrade). I just don't see how this could be, assuming a room temperature (~ 74F or ~23C) environment. 19C is ~66F, which just seems like too low a figure. Anyone have any confirming data?
Most of these heat sinks, I believe, are made of aluminum, so corrosion from tap water may be a problem. Further more, mineral deposits may build up inside. There is a reason why automotive radiator coolant has added anti-corrosives added.
Another problem is galvanic corrosion. There is probably going to be serious corrosion due to contact between the aluminum and the copper.
The solder may pose problems as well. Generally copper needs an acid flux to get good adhesion to the solder. This is why plumbing solder has an acid flux core. It's also a lead-free silver alloy so that it's safe for drinking water. Electronics solder is a tin/silver/lead alloy with no flux. I don't see a reference to the type of solder used. I hope for his sake he used plumbing solder.
The silicone sealer used to seal the tank section to the base is probably safe, but there are actually two types of silicone sealer. The older type is acidic, and the newer type is basic. I suspect the latter will not bond well to copper. "Fish tank" sealer probably is the acidic type.
Others have made reference to possible condensation problems, which is pretty serious if the ambient humidity is high. Keeping the coolant water at room temperature would probably help.
Using a pond pump and that tubing size seems like overkill, even for this. Also, there are such things as hose barbs which will help prevent leakage.
He's got cajones, tho', give him that much. Either way, it'll be spectacular...
That thing is not worth anyones time. It only
cools one side of the cpu, just like a heatsink.
The only benefit is if water colder than ambient temperature is used, but that might cause
condensation.
What a hardware freek really needs is a non-
conductive liquid to submerge the entire mobo in.
vane0026@tc.umn.edu
I'd do that in a second to my trusty Alpha Multia/UDB. That way I could run RC5 on it without the case melting.
Except there in one teeny tiny problem. The heat sink BOLTS on to the processor. None of that rubber band stuff for me.
I guess it is a intel world.
I got my system running DUAL 300 Celerons at 450 (that's right kiddies.. 900 Mhz)...
Anyways, to keep mine k00l, I plan on getting a thermal electric cooler from a typical 12VDC camping cooler. A couple fans on both side of the heat sinks and my case should be at about -3 C
Anyone else try this?
I thought about doing one of these, but then I couldnt take my machine to a lan party w/o dragging the water pump etc.
..
http://www.agaweb.com/coolcpu/
http://www.cpu-central.com/projectk.htm
http://larkin.nuclearwinter.com/cool/
http://members.easyspace.com/coolk6/
The last one is the one that I was going to do until I thought about the mobility aspect of it
NLucent w/o his cookies @ work.
Well, my first reaction was, why not *buy* copper tubing?
My second reaction was, why not just pull the fan, get *real* thin copper tubing, or, better yet, squish some, and run it through the fins of the heat sink.
My third thought was, you *don't* want the water to be too cold, or you'll get condensation on the outside, and dripping water on your mb isn't a Great Idea (so, *don't* use cold tapwater in January in Chicago...).
mark
http://www.d128.com/features/cooling/index.html
Forget water, stick a loudspeaker on the end of a tube filled with a Helium/Argon mixture and plug it in.
"The prototype, which differs considerably in appearance and size from what a commercial device would look like, is essentially a 3-foot-long aluminum tube, with inside dimensions ranging from 6 inches at the heart of the system to 3 inches elsewhere. At one end of the tube is an acoustic driver; at the other end is a cap. The driver, which has a vibrating diaphragm similar to a loudspeaker's (but sturdier and more powerful), sends pressure pulsations throughout the system, making gas particles oscillate back and forth at a single frequency.
''Fluctuating pressures inside the cavity are accompanied by fluctuations in temperature,'' Mongeau explains. ''When you compress a gas, it becomes warmer, and when you decompress it, it becomes cooler. The gas particles within the device become alternately hot and cold, dynamically, at a typical frequency of 200 oscillations per second.''
The gas particles in the ''stack,'' a parallel-plate structure near the driver, transfer heat to and from the solid material as they oscillate. The end result is that heat is pumped toward the driver, cooling the side farthest from the driver."
Thermoacoustic Project
ho ho! that is too funny. if you consider that
the pentium ii needs almost double the transistor
count to do the same thing - hence higher power
requirements, bigger die sizes, and overall less
effiecient to achieve the same result (if you can
grant that the PPC can give at least the same
performance, if not better for the same Mhz).
instead of fixing the problem with liquid cooling
as a work-around solution, it would be better to
fix the problem at the source: inneficient processor design yielding too much heat, because
of too many transistors to get the same functionality - this becomes ridiculous. only
the Intel world still runs on old legacy design CISC processors. intel needed to get the ARM risc
stuff because they'll have to transition to a new
more efficient RISC architecture (like the successful 68k to PPC transition back in '94) eventually to keep up in the processor arena.
its the same old quality/quantity thing again:
german engineers measure engine power in efficiency (torque); americans engineers measure
power in volume/capacity. intel engineers measure
power by how many transistors they can cram on a chip; RISC vendors measure chip power by how FEW transistors they can use to make it do a given task (i.e. efficiency). ditto: M$ measures "good software" by #lines of code; linux programmers measure good SW by elegance of code. good design is elegant and efficient down to the very core - right down to the way you design your microprocessor and file system architecture.
I have an NDA with INTEL and they tell me they are working on a sealed flourocarbon cooling system for their next chip release
Why not just run plastic tubing through the heatsink a la radiant floor heating?
Posted by mr_log:
I saw this a long time on a newsgroup (the one that is now comp.arch.hobbiest).. it's still an
interesting idea.
What about that liquid they emmerse crays in? I want to get a tub of that to put my motherboard in. =)
Posted by Mr. Assembly:
Peltier cooling would be easier and drier to implement. Also, your PC wouldn't become a boat if you sprung a leak. But I admire this marvel of engineering. While this seems a straight forward way to implement liquid cooling, it is not necessarily the best.
The finned heatsink is not neccesary, as it is meant to be used with 'air' and not liquid. What matters is heat transfer and surface area. Small diameter tubing could be tightly coiled and soldered/welded to a thin copper or brass plate (easier than the finned heatsink). The finned heatsink is thus replaced by circulating fluid. It would be important to use grease between the plate cpu as this removes tiny air pockets that prevent heat transfer. Traditional clips could be used to attach this assembly to the cpu. The tubing could then be run outside the case, and then if you sprung a leak, it would be 'outside' your pc. This method also lends itself to other cooling methods (i.g. using a refrigerator as a cooling source).
Posted by antivert:
Yours was about the 10th comment on the subject, and it's completely off topic - mentioning how efficient your PowerPC is has nothing to do with thermal cooling of processors. So.
I'm not always on topic, and I think it's good that we branch out into several topics - it's the order of nature. =) But really it just seemed as if you (and the others who posted similar comments) were taking a very thinly veiled (or not veiled at all) poke at the Intel users. We're not intel users because we think Intel is the first and last word in processors. Or at least most of us aren't. The fact is that we really don't care about how efficient your PowerPC is, and we're having great fun running water into our systems. Plus, I don't even use an intel processor. Nah nyah. =)
Distilled, PURE H2O is an insulator, and a good one.
Problem is, it dissolves minerals readily, and even a slight ion content makes it much more conductive. One tiny little screwup, and *POOF*
retrorocket.o not found, launch anyway?
Show me a decent 3D card that works with a PPC, and I'll switch.
Esp. LinuxPPC.
Linux has one available 3D card, the Voodoo/V2. Which is only supported on Intel platforms.
retrorocket.o not found, launch anyway?
The anodizing process basically leaves a thick layer of oxide on the aluminum, that's it.
The material created by anodizing isn't that great, but apparent immersing it in boiling water causes it to change form to a wonderful protective coating. (I think that's it...)
Look around on metalworking sites, you might be able to find a HOWTO for anodization.
retrorocket.o not found, launch anyway?
How many times do I have to hear that the PPC is more thermally efficient than a Pentium in a single thread? I'm kind of getting sick of it. Some of us have reasons for using x86 architectures. It's not our goddamn fault that some of the applications we like just AREN'T AVAILABLE on your oh-so-perfect processor.
I happen to LIKE Quake/QuakeII, etc., especially with a good 3D accelerator. Considering how far behind PPC architectures lag behind in these arenas, I'll stick with my thermally inefficient x86, as much as I'd like to be able to switch to PPC, it's just not practical right now.
Don't give me that "Oh, we've got the Rage 128 now." excuse, how long did it take for that to happen? Still, the R128 has its dirty little secrets ATI doesn't tell you about. Did you know it takes a 30% or so performance cut if you enable trilinear filtering? And no, the Riva TNT doesn't, in fact, I don't think ANY modern cards do.
And "We're getting Quake II". How long has Quake II been out with no PPC port?
And don't even get me started on LinuxPPC and 3D - Zero 3D acceleration whatsoever. And one thing - Good luck EVER seeing support for the 3D portions of ATI chipsets. Can you say ANAL? Trying to get specs from ATI is like trying to squeeze apple juice from a moon rock.
Apple's love affair with ATI has doomed them to be eternally in the bottom of the barrel for 3D graphics performance until now. Now, they're just average.
Personally, thermal trouble is a small price to pay for being able to run the applications I want to.
retrorocket.o not found, launch anyway?
Can't speak for others, but my motherboard doesn't support chips as slow as ppro-166s, and last summer I couldn't afford anything faster with a good cache size. Good fans and things were okay. (until the cheep sleeve bearings failed. I now can afford and bough pcpower and cooling fans, and they are much nicer)
Also gonna be required on PIII
I had a similar idea.
I was gonna throw my computer in my pool to keep it nice and cool, but then I thought, "Hey. I'm gonna have to take off my clothes and put on my swim trunks (or just leave them off if I'm feeling frisky) just to change a cd or floppy!"
Then I got the bestest idea in the whole wide world! I have an air compressor in my bedroom. I have it hooked up to a similar contraption as his. The air going into the compressor is cooled by an air conditioner. The only bad part is I have to replace my PII-400 every couple of days (you wouldn't beleive how quickly air tears through silicon). Hopefully a filter will help.
The Sega Dreamcast has a liquid cooling system for its Hitachi processor.
It's a bit of a weird idea -- copper tube filled with distilled water sinks the heat.. and if *that* overheats, the thing bursts,
soaking the motherboard with water, whereupon you have to send the unit back for repair...
You want trivia? just ask...
Wow !!
...
There must be some kind of magic here
Can you tell us how you do to keep your CPU temp below the room temp without using active cooling ??
Did you turn the computer ON before ?
Btw, my K6-2/300 is about 25C overclocked to 600Mhz and powered OFF !!
Mac fanatics are soooooo funy sometimes =)
...cause if that thing springs a leak your box will be FUBAR
While I am rather impressed by the intrepid young mans ingenuity (sp?), I don't understand why he doesn't just buy a Peltier Cooling Chip (i.e. a chip when you run current through it, one side gets cooler, and the other side gets hotter).
Seems to me that he could glue one o' these chips right onto his Celery chip, and it should be able to keep his processor chip cooler, without all the water.
Of course then he would have to cool the heat sinks on the Peltier chip itself, but still....
Hmmm.. maybe I am violating one of The Laws o' Thermo...
Comments anyone??
Aha,
Now that I have discovered the "Flat" mode at Slashdot
I see from above postings that someone else has suggested this too.
However, the questions still remains: would a Peltier Chip Work?
if I could just buy the board, no operating system, then I would be interested
For the hundred billionth time, you can. IBM sells them, and I believe Motorola does too. And of course you can always get them used on eBay or many other places.
this may seem trivial, but the second and third pages have a discrepancy ... the tubes run into the top on one, and the sides, with valves on the other.
Yes. The second page showed the "normal" setup. THe third page showed a modified version.
PPC's are NICE but I don't want to put money into Apple's coffers anymore than I want to put the stuff in Microsoft's. Since the primary source of obtainable PPC machines are Apple Macs, I won't buy one. Not to mention that I can pick up a motherboard just about anywhere for the Intel architecture- I can't do that with the PPC and it's hard to do it for the Alpha.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
CAn you say flaming motherboard?? Mother boards run on +12v, -12v, and 5v's dc if you want to get picky. The POWER SUPPLY can take either 110v or 220v ac, but either way they still output the same volts in dc. The processes is just more efficent(read: less heat from the power supply) if you use 220v.
---------------------------------
About 6 months ago I got this idea and started to do some serious research on the internet and I have amassed a fem book marks of sites with some more informative content. I think the most informative site is Water cooled CPU's
. Basicaly he addresses most of the points brought up in this discussion. In brief; Alcohol sucks, use Vasaline to stop condensation on the pins, and water works really well.
---------------------------------
Um, I actually had to do that this summer when I got my PII 400. Hopefully the new case I just bought will do better next year. :)
includes cascaded peltier design http://larkin.nuclearwinter.com/cool/
"There are two major products that come out of Berkeley: LSD and UNIX. We don't believe this to be a coincidence."
Nothing like a little liquid nitrogen to keep things cool. Maybe I'll try to over clock my 386/25 to a few terahertz now. Now I just need to find a motherboard.
Andrew
"You never know when some crazed rodent with cold feet might be running loose in your pants."
-Calvin
Why not just use a block of aluminum, drill some holes all the way through the side. Put threaded nipples in the holes, and presto you have a dependable water cooled heatsink. Its similar to the water cooled heatsinks on R/C boat engines.
Kriston J. Rehberg
http://kriston.net/
Kriston
I remember seeing a PC - might've been in Popular Mechanics, might've been on the web - which came in a refrigerated case. Like your fridge, you know, just sealed so as to prevent condensation. The promo material hinted at some verrrrry impressive overclocking speeds.
Anyone know what I'm talkng about?
theresa host of PCI Voodoo cards that can be used with LinuxPPC - theres also the PermediaII and the
forthcoming PermediaIII - and the Riva
alan
Why not just submerge the whole motherboard in mineral oil, like they do for high voltage transformers? This could keep your Voodoo IIs from overheating trying to run Unreal if you put the video card in too. The only hitch would be keeping connectors (mouse, keybd, SIMMs, ect) from getting contaminated with oil.
0 1 - just my two bits
:)
I was just wondering why they made their own coper tubing instead of buying it at a hardware store. I would think that would eliminate a large % of the probibility of leaks.
Harlequin
There's a company here in Japan that sells water-cooling kits for Socket7 and Celeron CPUs, and they even have a version that uses a peltier chip between the CPU and the water-cooling unit... I think the kits were called "Poseidon" or something similar.
A magazine here did a review of the kits a while ago, and they found that even with the peltier chip, the improvement is generally only one bus-speed higher. Try to cool the CPU too hard and you get condensation, too...
assuming a room temperature (~ 74F or ~23C)
;-) Yes, overclocked PPCs can run at ~ambient temperature.
Turn your thermostat down. Works for me in Minnesota in the winter
BD - whose thermostat is set at 64F
I hope he doesn't get the hot and cold water taps mixed up.
Kryotech made a modified PC cooling system case, and Anandtech reviewed it a while ago.
m ?document=343
http://www.anandtech.com/html/review_display.cf
So, this cooling has been done, better. Well, if it wasn't for the fact the cooling system was so damn expensive and it was limited to Socket 7 CPUs.
actually pure distilled water (H20) is extremely nonconductive. it conducts so little electricity it requires a voltmeter that can read voltages in the range of 1x10^-9 v to even detect it.
2H20 -> H30+ + HO-
that's the only reason distilled water will conduct any electricity at all. maybe you're thinking tap water.
Dogma: Dead (mostly because your Karma ran it over)
Ok Ok
Water contery to popular belief is NOT I repeat NOT a good conductor, (note I am an Electrician)
now for a little experiment fill a plastic bucket with plain tap water, get 1 hair drier, turn on hair drier, now drop the hair drier in to the bucket.
Now what do you expect will happen to the hair drier, blow up, blow a fuse, kill every one in sight. nah it just sits there churning water, and you can put your hand in it and lo and behold you dont feel a thing, you can even measure 240volts, on your arm with a volt stick..
BTW I have done this.....
But if your Interested about such things, about electrical safety. Mail me
carnage_visors@bigpond.com
Electricity should never be played with, the above experiment was done to show a bunch of aprentices, some of the common misconcepions about water and electricity.
"sometimes I wish I was blind I thought I saw a whole lot more than this"
OK lets just say that u actually get this thing not to leak. Ever hear on condensation? Rust is a bitch when it comes to the pins on the processor :)
Natas
Natas of
-=Pedophagia=-
http://www.mp3.com/pedophagia
Also Admin of
http://loki.linuxgames.com
Just after the release of the Pentium II 300, Intel was in Europe doing some intersting demonstrations.
By freezing a Pentium Pro, and then sticking it in, the system was running at a full 700-800mhz! The cooling seriously affects how the chip processese information, and it seem that freezing it seems to give a performance boast.
Anyone ever do this at home?
-Michael J. Lu
"The little secret that haunts Corporate America...a techonology that won't go away."
But...
Higher clock speed means less time for each transistor to accumulate enough electrons to switch. Clock it up too high, and not enough electrons flow into the transistors to make them switch... and, no boot!
Does freezig help this problem? Does super-cooling somehow increase conductivity in ordinary silicon?
To get a processor to run at, say 682 MHz -- that's 5.5 times 124, the max speed on most high-end Super7 boards -- by how much would you need to raise the voltage for an AMD K6-2? How far could you go before you fried the chip entirely? How much cooling might this require?
wouldn't it just be easier to drill some power cord holes in a fridge or something? =]
The things one has to do for those inefficient Intel chip designs... ;-)