Do-it-yourself CPU Cooling 87
Ice_Hole writes "Well, looks like the serious overclockers are going to have a few more options (that is, other than Kryotech) in terms of extreme CPU cooling. Better yet - it is a build it at home project."
Slightly offtopic: Quiet CPU Fans (Score:1)
On the topic of do it yourself cpu cooling, are there any cpu fans that stick out in anyones mind? I have an overclocked Celeron 300a (@450) that is actually running nice and cool with the generic fan that I stuck on there. But for some reason over the last month the fan has become really, really noisy, and I want to replace it. Know of any fans that run quietly?
thanks,
/joeyo
Re:Slightly offtopic: Quiet CPU Fans (Score:1)
Pc Power and CoolingIs where I get all my fans and power supplies now. I've had too many porblems with poor quality fans. FreeBSD is ultra reliable now that my processors have the required cooling.
Sure they are a little nosiy, but better then a little hot. I got ear plugs anyway.
ack! attack of MSTML! (Score:1)
BBF - Bigger Better Faster (Score:1)
Any plans for a including them into 'off the shelf' systems? I supose some vendor will at some point but I doubt it will be one of the 'big boys' due to their tie-in's with Intel.
Re:Cool! (Score:1)
--
Re:Slightly offtopic: Fan underclocking (Score:1)
Re:Slightly offtopic: Quiet CPU Fans (Score:1)
I don't find them in the current catalog, but I've had great results with 105CFM fans I bought from Mouser... They have a 5 year warranty.
NEWS: Geek hits 800MHz, 5 die in explosion (Score:1)
Get A Faster Processor (Score:1)
Alternatively get a G4 PowerMac.
Or a SPARC 5.
Stop barking up the Intel tree, because there *are* alternatives.
--
Re:Cool (Score:1)
Fan Ducts and Heat Pipes (Score:1)
I'm making a dual celery box (300A @ 450 MHz) and I've been surprised by the almost Luddite state of ventilation technology in the PC industry. Apparently, conventional practice is to put a fan near each processor's heat sink, and maybe also near the video card and main board chip, and to let the power supply's fan blow the hot air out of the box (perhaps with auxillary fans blowing in or out). No attention seems to be paid to heat accumulation toward the top of the typical case, except that the power supply is usually there, and most of them have their fans blowing toward the exterior (opposite the ATX specification). If the processor gets too hot, the conventional approach seems to be to blow more hot air over it with a more powerful fan. The overall flow through the case seems pretty haphazard.
So the first obvious step is to blow outside air at the heatsinks, not the heated air within the case. I found that Intel has recently promulgated a new spec. to accomplish this in ATX and micro-ATX systems. Described here [intel.com], it entails a wide, flat duct over a horizontal fan that sits above the processor boards. Air is sucked from outside the case, through the duct, to the fan, and onto the hot corner of the ATX board where the processors, memory, and board chipset lie. They claim it will also cool the AGP video card nearby.
That should help, but what about the hot air in the case? I've decided to make a box with an extra panel inside, an inch and a half from the side panel and parallel to it, forming an air plenum separate from the main volume of the box. I plan to run ducts from this plenum to each hot spot, mostly enveloping the heatsinks, and perhaps also to the hard drives and power supply. This way I can blow air into the plenum and cool everything, with exhaust holes mostly around the top of the case. Yes, this is a parallel resistor network, and it will be necessary to tinker with the duct sizes to balance the flows according to the heat produced in the different locations.
I hope this will let me dispense with multiple fans in the case, and instead have one substantial squirrel-cage blower for the plenum. If you've read this far, it has probably already occurred to you that you could attach a hose to such a case and put the blower ELSEWHERE, to quiet things down. It might also be quieter due to the more laminar airflow (less turbulance).
Alternatively, one could SUCK from the plenum, with air entry holes strategically located in the outer case (e.g., in the front panel above and below the hard drives, etc.). This would be even better, since the heated air would be evacuated from the case directly, and in the summer it could be vented outside the room.
Has anyone gone to such lengths to achieve quiet, effective cooling?
On another note, what about heat pipes? Seen any lately? About fifteen years ago, I fixed an audio power amplifier for an EE who usually attempted his own repairs, but was left scratching his head at this tiny, futuristic, limited-production "concept amp" from Sony. (It's no wonder why: there was no signal voltage until the output stage- it was a current amp, not a voltage amp; quite unusual at the time). This amp had the output transistors on a small copper block, with a copper tube running to the rear, where it ran through a long stack of thin aluminum plates spaced about a quarter inch apart from each other. This was a heat pipe. The tube was concentric and full of refrigerant, which was boiled by the power transistors, forced through one tube as a gas, and condensed to liquid as it flowed through the heat exchanger plates at the rear. The heat input caused the circulation. This allowed the power transistors to be located on a small board with the other circuits, without heating them up very much. Other components could sit right on top of the thin power amp chassis without getting hot. The whole thing was almost weightless, but each channel put out 125 Watts. (It also had a rectified-line switching power supply to eliminate the power transformer).
Now, that was COOL. Why don't Intel and Alpha owners have heat pipe coolers like that? It makes the status quo look pretty retro, if you ask me.
Re:Condensation (Score:1)
One problem (Score:1)
They have the time, but not the cash.
Not all slashdot readers are $80k a year sysadmins or something.
Why bother with all this.... (Score:1)
If you don't know what Peltier devices are, they are used in portable refrigerators for cars.
One source of them that I know of is here [melcor.com]
Have fun,
GRH
I've been curious about something for a few years. (Score:1)
allow you to place the cpu into the other end
of the interface??
I.E. a fibre-optic channel or something...(is copper
wire 2 feet in length fast enough for cpu traffic??)
This would allow you to cool the cpu seperate from the internal
components of the rest of your computer...
You might even be able to stick the cup in
a liquid nitrogen bath (sealed) inside a cooled unit.
I can think of a number of ways.. just curious as
to wether or not anyone has ever done anything like this
or if anyone knows why this is a bad idea???
Re:I've been curious about something for a few yea (Score:1)
fiber optic??
I understand the slowdown with copper...
wouldn't the fiber-optic be fast enough? Even at 10 inches?
There's got to be a way
Re:BBF - Bigger Better Faster (Score:1)
Has anyone tried running a Sterling engine using a CPU as the heat source? I was thinking about building one, but dont have access to a machine shop anymore. I think that it would be super cool to have a 18th century reciprocating engine running your case fan.
-BW
Re:A really cheapass cooling (Score:1)
Now this is the sort of thing that should be making the top page of
:-)
Cool! (Score:1)
Sure, it's fun to get a little more speed from your core but, after accounting for your time and effort at what you should be making as a techie who can do things like this isn't it still more cost effective to just buy the faster chip to start with?
--
Re:peltiers=good (Score:1)
i used to have lots of fun with liquid N2!!
Re:Get A Faster Processor (Score:1)
This means that most software is developed for Intel processors - sure, you can run NT on a DEC Alpha, and run FX32! to run Intel compiled programs, but that's not always a suitable alternative.
You could get a G4, but they're stuck at 450MHz at the moment - and there's a lack of software (though not as bad as in my next example).
As for getting a Sparc, as much as that'd be nice, again, you run into the problem that there's simply not the market for these processors to warrent the development of a lot of software, especially games (and you'll find that a large number of overclockers are actually gamers).
It's all good and well to crow about the alternatives, but unfortunately, that's all they are at the moment in many cases - alternative ideas, not necessarily good ones.
Re:A really cheapass cooling (Score:1)
Modern OS's don't need that (Score:1)
Also note that using Waterfall and something like D.net or Seti@home are mutually exclusive. Depending on what priorities the programs are running at, either Waterfall with grab all the idle CPU cycles and turn them into HLTs, grinding the other program to a halt, or D.net or Seti will grab all the idle CPU cycles and keep Waterfall from getting to any.
Re:peltiers=good (Score:1)
Oxygen is your best friend and your worst nightmare at the same time. Pretty neat if you ask me.
Re:Liquid Nitrogen--WARNING (Score:1)
Also your chip would be so brittle from being held at this low temp that the slighest vibration could shatter the circuitry of the chip.
I would strongly advise against the use of liquid nitrogen, unless you really want to destroy your computer in a creative manner.
Travis
Re:Different ideas (Score:1)
I've seen several sites that step you through howe to build it yourself.
http://www.agaweb.com/coolcpu/advantages.htm
-Joe
Re:I've been curious about something for a few yea (Score:1)
Question: How long can you run your IDE ribbon cable?
Answer: 18 inches (hey, the guy that wrote the specs worked on the mars orbiter misson, he still uses imperial measurments)
Followup Question: Why only 18 inches?
Answer: If you make the cable longer, you will get really bad cross talk between the data lines.
You really don't want to be messing around with the length of the traces between the processor and the chipset. You won't like the results very much.
Re:Cooling in cold places. (Score:1)
Plato IV's predecessor, Plato III, ran on a CDC 1604 or 1800 or whatever. Core memory, cathedral doors, built like a bank vault. 18-bit machine, if I remember correctly.
Plato was housed in CERL, the Computer-based Education Research Lab, which in turn was housed in a very old building. The building air conditioner was the same one which had cooled the lab in that building that worked on radar during WW II. When it finally blew up, parts had been unavailable for decades. An entire new unit had to be ordered. This being a university (i.e. zero clout), delivery time was quoted in months.
Obviously no project could survive months of downtime. Now, it happens that the project director, Dr. Donald Bitzer, had his office across the hall from the machine room. So, a square tunnel was constructed of wooden lathes and plastic sheeting, about two and a half feet on a side. The CPU's cathedral doors were opened, exposing the core memory. The tunnel started flush against the face of the machine, ran across the machine room, out the door, across the hall (!), through Dr. Bitzer's office, and straight to his window.
It worked. The machine ran this way for over a month.
Waterfall? (Score:1)
I didn't run it, I don't know where he got it, and really I don't even know where to start looking for it... but I was curious if anyone else had ever seen it, used it, and/or found out why it never made it big.
Re:Plato System (offtopic) (Score:1)
This is to some great extent why my very brief stint at the U of I sucked so badly and why it was so very very brief.
Affect on manufacturers (Score:1)
All we need now, is a major manufacturer to release an overclocked computer
Re:Slightly offtopic: Quiet CPU Fans (Score:1)
Re:actually (Score:1)
Re:Fan Ducts and Heat Pipes (Score:1)
Re:Kryotech Death (Score:1)
Re:A really cheapass cooling (Score:1)
Cool (Score:1)
Does anyone out there have any pricing/ordering information yet?
Condensation (Score:1)
From the pictures at Overlocker Shootout [ocshoot.com] it appears this unit has something all the homebrew coolers have missed: condensation/liquid detection (the funky silver rectangular lined screen wrapped around the CPU unit in the picture). That should solve the condensation problem.
All important idea.... (Score:1)
$800 to $850 USD (Score:1)
((MY STORY ACTUALLY MADE IT ON SLASHDOT!))
Source for Peltier devices (Score:1)
I think they will export stuff.
Check them out
Re:Slightly offtopic: Quiet CPU Fans (Score:1)
Having used both of these products, my mid-tower case is both cool and unobtrusive with a PIII 450 running @ 560, with an additional fan keeping the overclocked video card happy.
Rafe
V^^^^V
Re:Cool! (Score:1)
Not if you do it on company time.
As I seem to do anything I can think of a way to get done on company time gets done on company time, over clocking is REALLY cost effective.
Beyond that, I think this stuff looks great at first look. And I think I might just be hopping in line to get one for myself.
Re:Different ideas (Score:1)
Dealing with condensation (Score:1)
Re:Liquid Nitrogen--WARNING (Score:1)
When liquid nitrogen flows through uninsulated tubing or piping, a layer of water ice and dry ice form. I used to work in a lab where we used liquid nitrogen to cool gas chromatograph tubes and never had an explosion. I can see where liquid nitrogen in an open container would absorb oxygen, but I can't imagine dunking an operating chip in that sort of container anyway.
And now for some verse:
Sir James Dewar
Is a better man than you are
None of you asses
Can liquify gases.
peltiers=good (Score:1)
Re:A really cheapass cooling (Score:1)
actually (Score:1)
now for something completely different (Score:1)
a chilling thought (Score:1)
people who are "brought back" to life from death or near-death fare much better, have much less brain-damage, if they've been chilled. for example, if you drown in frigid water (or perhaps suffer a heart attack having an affair with my mother-in-law (ouch! I'm in trouble now!)) they will attempt a resuscitation after a much longer period of time, I think because the tissue in the brain goes to sleep and doesn't run out of O2.
BTW, my pet ideas for CPU cooling are more broad-brush, once-and-for-all sorts of things:
Re:Different ideas (Score:1)
Try boiling some distilled water with a piece of solder and some copper wire in it. Then measure the resistance.
If you want to make the most efficient heat transfer conneciton, then put the cold side of the fridge coil in the water (immersed). Use a submerged propeller to agitate the water and force the heat transfer. If you are going to immerse in water, you may as well add automotive antifreeze, as well (I think this is non-conductive)
Re:Different ideas (Score:1)
The mineral oil may again swell plastic parts, causing problems. The liquid you are looking for is (was?) called Flourinert. It is a non-reactive CFC blend which is a colorless liquid at room temperature. It was expensive 10 years ago, I don't know what the status of it is now.
Cooling in cold places. (Score:1)
Poof... No more heat problems.
Re:Liquid Nitrogen--WARNING (Score:1)
(what can i say, my dad's a chemist)
Different ideas (Score:1)
Re:Different ideas (Score:1)
Re:Different ideas (Score:1)
Re:Different ideas (Score:1)
Kryotech Death (Score:2)
He was 33.
Actually, if I remember right, they were celebrating cooling an Athlon(then K7) up to a Gigahertz...
For some reason, this has always stuck in my mind as a weird reminder that even us crazy young techs are all sooner or later going to have no more toys to play with...
Yours Truly,
Dan Kaminsky
DoxPara Research
http://www.doxpara.com
Condensation Problems? (Score:2)
A temperature sensor embedded on the CPU would be necessary, for a failure in the cooling system with this kind of insulation would surely cause a fire.
Re:Poison gas problems (Score:2)
Re:actually (Score:2)
Do-it-yourself? (Score:2)
There's a review and pictures here [ocshoot.com]. Doesn't seem so fantastic, but I guess it's better than putting the CPU in the kitchen freezer.
Basically, it has condensation, you have to vaccuum it for dust, the CPU died after two hours of overclocking, and the 5.25 bay is gone.
Oh, it also makes 50 dbA of noise. Man, that must be annoying. :)
"There is no surer way to ruin a good discussion than to contaminate it with the facts."
I have another application for this device (Score:2)
Lord knows this world can use it.
There are now 6 billion of us lurking about
We'll place it on males to cool their libido.
Stop the population explosion!!
Re:Kryotech Death (Score:2)
He was 33.
Sounds like he died from overclocking.
Review of VapoChill (Score:2)
I know I certainly want one when I upgrade to Athlon from my current PIII 450 @ 560 (courtesy of Step Thermodynamics. [step-thermodynamics.com]) Their product has performed flawlessly for me, and is definitely recommended if you want to run @ 25%+ over factory spec.
Rafe
V^^^^V
Scarry thought (Score:2)
A DVD needed to distribute the OS.
*shudder*
Hey! My shortest
(and most meaningless, probably)
Nick.
Re:Cool (Score:3)
£445 for Case
$50 for the processor Kit
£20 for delivery in the UK
add VAT @ 17.5% for these prices...
not cheap...
Re:Do-it-yourself? (Score:3)
- They attributed the CPU "death" to the fact that they had already put the CPU through hell before using this cooling method, and they added peletier coolers which could have shocked the CPU.
- The Kryotech cooler made 50dbA of noise, not the vapochill (it stood at 35dbA).
They also don't mention condensation anywhere when it comes to the vapochill.
Overall, I thought the review was particularly glowing. Comments like "In my opinion VapoChill and a Celeron is the perfect combination.", "Athlon 800 for half the prize off a KryotechCool Athlon800!" lead me to believe that the reviewer actually likes this product better than many others.
All well and Good... (Score:3)
I wish I had a quote for the day -- me, myself and I
Half-assed "do it yourself ideas" (Score:3)
2. Spray some liquid nitrogen on the cpu. Sure you might shatter the cpu, but if it survives, you can overclock that baby like hell... not to mention having extra liquid N2 readily available in case you get a nasty wart on your typing fingers.
3. Set up office in one of those giant wind tunnels with the case open, and anchor everything to the floor. Now hows that for a good fan?
4. Failing that you could always rent a giant meat locker or other refridgerated area and run your computer from there.
5. Move to Antarctica. If you by chance can get a high speed internet connection there, gimme a shell
6. Chain a small sibling or a friend to your desk, and have them continuously blow on the cpu to cool it, keeping a whip handy should they whine about being bored, tired, or hungry.
Now that the cooling issue is settled, let's see how ridiculously fast we can overclock. Maybe we can push that Celeron 550 up to 1 gHz without it melting on us.
A really cheapass cooling (Score:3)
Liquid Nitrogen--WARNING (Score:5)
HOWEVER, one should note that liquid oxygen boils at 90.2 K. Standing liquid nitrogen exposed to air WILL condense liquid oxygen over time. And liquid oxygen is extremely explosive. (I have no facts to back me on this, but being a diradical, in high concentrations, liquid oxygen might really mess with the electrical signals in your computer. Anyone know if this is true?)
Liquid nitrogen is a safe liquid to play with as long as you don't freeze yourself AND as long as you don't leave it out to air. Indeed, in my organic chemistry laboratory, we are advised not to leave liquid nitrogen around for more than 24 hours (only possible with LARGE amounts or by constantly replacing it). Large dewers of liquid nitrogen are stored in tanks specially designed for this purpose. (I'm not sure how. Maybe the vents have oxygen-absorbing filters of some sort? Anyone?) But leave your liquid nitrogen around and open to air long enough, and a pale blue color will slowly form, and when that happens, the slightest shock and everything around goes kaboom. I don't know of any computers blowing up, but then, I don't konw of anyone who's tried keeping a computer cooled to those temperatures for prolonged periods of time.
I've read one too many posts about using liquid nitrogen to cool a computer. If you're going to do so, please seriously consider how you're going to deal with the liquid oxygen problem.
-----
I've never posted here. I just lurk. And I probably won't post again in the next 2 years, just as I haven't in the past 2 years. But I feel a need to post here. No references available offhand (I have to go to class soon), but chemistry/physics people, feel free to check the facts. And to make me somewhat less anonymous, my name's Jason Chen, and my e-mail is jschen@fas.harvard.edu.