water cooling v. fan\heatsink

  • Thread starter Thread starter taragem
  • Start date Start date
In alt.comp.hardware.pc-homebuilt Mac Cool said:
Frank McCoy:


Are you using one like you describe on your computer?
Click to expand...

Hell no!
WAY too expensive.
My fan-cooled heatsink keeps my CPU at a comfortable 47C and my chassis
is at an even cooler 40C. Why would I put a huge device like a heatpipe
in a system working that well?

If I was building a super-dooper Gaming machine with quad processors
pulling 75 watts, I might think of such as being preferable to a water
cooled and pumped device. It would be more efficient, quieter, and
simpler; though probably not cheaper.

SOME motherboards actually come with heatpipe devices built in/on.
But usually that's for the fancy chipset, not the CPU.
 
A watercooling setup will dissipate heat from simple convection current of
the water through the pipes. I demonstrated this on mine by turning off the
pump and fans, and running the system at an underclocked speed.

Cpus temps hovered fairly high, upper 50s, but usable, and the large
external car oil cooler I used was definitely getting warm, so convection
was definitely occurring. I ran the system like this for a couple of years,
turning the pumps/fans back on when overclocking for gaming.

An improved setup designed to encourage convection might work quite well.
Some company does produce a very tall radiator device but I believe it still
incorporates a pump.

rms
 
Heat-pipes are relatively simple devices.
A source of heat, a heat-sink, and a pipe or other connection between
them, a volatile liquid/gas in the device, with a capillary cloth or
material running from the sink to the source.

Water, while usable, is a fairly poor "volatile liquid" to use for a
heat-pipe. You'd prefer a liquid that at "normal" pressure would easily
vaporize at the heat-source temperature; yet condense at the heat-sink
temperature (usually slightly above room-temperature).

There are many liquids that fill the bill:
Gasoline
Propane
Alcohol
Ammonia in water
Various refrigerant liquids.

People tend to shy away from the first three, even when used in a
completely closed device, because of their flammability, and the fourth
because of it's toxicity. That's why Freons were invented (and used
today).

The basic design of a heat-pipe is very simple:
Heat boils a volatile liquid at one end, and it's condensed at the
other. A relatively small cloth, rope, or other such item returns the
now-condensed liquid back to the heated end by capillary action. No
moving parts except the liquid itself.

Yes, water *can* be used; as water *will* boil at reduced pressures.
The common and easiest way to do so is to partially fill the device with
the desired liquid (again, yes, water will do) and bring the *whole
device* to the boiling-point; expelling liquid, air, steam, and other
gas until only the desired liquid and it's gas (steam) is left in the
device; then seal it while still hot and before it starts to suck air
back in.

This can be used to produce "hand boilers" that boil water in the palm
of your hand from the heat therein. Like I said, water *can* be used.

A person can make a demo of this with an old clear lightbulb and a
bunsen-burner. Carefully remove the screw-base without breaking the
bulb. Carefully either nip off, break off, or melt open the seal. Heat
the bulb with a flame for a few seconds to drive out some of the air
inside; then stick the tip into a cup of distilled water; holding it
with a pot-holder so you hands don't get burned. Take the bulb, now
about 10% filled with water, and place over the bunsen-burner in a rack.
Heat until the water is boiling furiously and you figure only steam is
coming out the nib. Reduce heat, and seal. To seal, the *preferred*
way is to melt the original nib closed with the bunsen-burner. However,
wax or epoxy *can* be used if the demo isn't going to be permanent.

Let cool. You'll now have a globe with a small amount of water in the
bottom that will actually *boil* in the palm of your hand from the heat
of your body, until the temperature equalizes.

A heat-pipe works the same way; except it has a heat-sink on the other
end taking away heat as fast as produced on the boiling end; with some
method of returning the liquid to the boiling end ... usually a small
bit of cloth or water-loving fabric that easily wets in the working
fluid. However, in *some* systems where the heatsink is *always* above
the heat-source, gravity also works quite well.

http://www.cheresources.com/htpipes.shtmlhttp://technology.grc.nasa.gov/tops/TOP300155.pdf http://electronics-cooling.com/articles/1996/sep/sep96_02.php

The transfer is nowhere near as efficient as a heat-pipe.
In a really good heatpipe design, the temperature difference between
heatsink end and heat-source rivals or even sometimes beats pumped
liquid designs; but without the problem of pumps, seals, and extra heat
input by the pump itself. Also not energy source other than the heat
differential itself is needed to power the device.




Butane, pentane, propane, all make good working liquids for that
temperature-range. Water *can* be used (see above); but it's "normal"
heat-of-vaporization isn't ideal for the job. Some people though get
scared when working with flammables like those.

OTOH, the machinery to work constantly with freons these days and not
let waste escape to the atmosphere, get rather expensive. (Ask any
air-conditioning auto-mechanic.)

As for a "pressurized system": ALL such systems are pressurized, or at
least SEALED so that there's nothing inside but the desired liquid and
its gas. At whatever temperature you work it at, the liquid/gas *will*
reach equilibrium, where additional heat will cause some of it to boil,
while reducing or removing heat will cause some to condense. That's how
the system works.

And work it will, no matter what the working-fluid, over a *wide* range
of temperatures. Even water. The only worry being:
A. Freezing of the working liquid.
B. Not enough liquid vaporizing fast enough.
C. The working pressure getting too high for the design.

A. In this design is something you won't have to worry about.
(If it freezes, nothing is damaged.)
B. This is where your choice of liquid/gas is chosen.
C. You need to make sure you have a big enough heatsink to keep the
temperature low enough that this doesn't occur.

Finally, you want to make sure your wick and distance run *can* return
enough liquid to the high-end fast enough; or you'll have the source
boiling away all the liquid faster than it can return; with all the
liquid at the heatsink end and only gas at the source (back to your
original convective design) ... a definite no-no for a CPU cooler.

Gravity as a backup helps a lot.

However, when properly designed, heat transfers from source to sink of a
heatpipe with almost unbelievable efficiency; and pretty much at the
speed of sound. It's *almost* a heat-superconductor. Almost.

Like stated above though, that does you no good if your heatsink isn't
big enough to cool the device. All a heatpipe does is *transfer* heat
(like a watercooled device) not get rid of it. You still need a nice
large external heatsink for that. The bigger the better.

You just don't need to have it sitting right on top of the CPU any more.
Click to expand...

It seems to me that I asked you what time it was and you told me how
to build a watch! Maybe that is not a good analogy, because I asked
you how a hobbiest could build a heat pipe and you gave me tons of
background info, but nothing useful in building a practical heat pipe
system.

My discussion of working fluids was in the context of the convection
liquid cooled system. If it could be done by forming vapors, as is
done in a heat pipe, then it will work pretty much the same. One pipe
from the CPU block will collect and carry away the vapors and the
other will return the liquid to the CPU block.

The problem with trying to make a heat pipe is that the construction
is difficult. I have never heard of anyone making one themselves. If
you know a practical way of doing this, I would be very interested.
 
A watercooling setup will dissipate heat from simple convection current of
the water through the pipes. I demonstrated this on mine by turning off the
pump and fans, and running the system at an underclocked speed.

Cpus temps hovered fairly high, upper 50s, but usable, and the large
external car oil cooler I used was definitely getting warm, so convection
was definitely occurring. I ran the system like this for a couple of years,
turning the pumps/fans back on when overclocking for gaming.

An improved setup designed to encourage convection might work quite well.
Some company does produce a very tall radiator device but I believe it still
incorporates a pump.
Click to expand...

That is what I have thought. By definition a convection system will
have a higher temperature drop since it is the temperature drop that
drives the heat flow. In a pumped system the heat in the liquid is
pushed along by the pump. The other parts of the system still
passively conduct heat however. The heat flow from the CPU through
the heatsink to the liquid is passive. The flow from the liquid
through the radiator to the air is still passive. The flow from the
air around the radiator to the bulk of the air in the room is either
passive or with a fan is now driven. Driven heat flows can approach
zero degree temperature drop, limited only by the speed at which the
medium can be pumped. But the temperature drops across the passive
portions of the heat path will not change.

In a totally passive system the resistance to heat flow should be
reduced by using large, open pipes or advanced technologies such as
heat pipes (if practical). One that I would like to try out sometime
is a chimney on the PSU instead of a fan. A six inch tube perhaps 5
foot high should provide an adequate air flow equivalent to an 80 mm
fan at low RPM. That should be easy enough to test. Maybe I'll try
that later today... I need to go to the hardware store anyway.
 
In alt.comp.hardware.pc-homebuilt rickman said:
My discussion of working fluids was in the context of the convection
liquid cooled system. If it could be done by forming vapors, as is
done in a heat pipe, then it will work pretty much the same. One pipe
from the CPU block will collect and carry away the vapors and the
other will return the liquid to the CPU block.
Click to expand...
Use just *one* pipe; and a wick in the same pipe to bring the liquid
back. A lot more reliable and simpler.
The problem with trying to make a heat pipe is that the construction
is difficult. I have never heard of anyone making one themselves. If
you know a practical way of doing this, I would be very interested.
Click to expand...

Easy:
Get a hunk of soft copper pipe and three standard "all copper"
heatsinks.

Cut all the fins and such off one of the three heatsinks, leaving only
the copper base and mounting-block where it mounts to the CPU. Holding
the heatsink in a vice and using a hacksaw works OK. Using a metal
band-saw is easier, if you have one.

Feed a soft cotton "wick" from one end of the copper tube to the other.
A wick made from about two or three layers of old cotton T-Shirt sewn
together and then cut to about 3/8" width will do fine. When in place,
it should fill about 1/4 to 1/3 of the available space in the tube
without blocking airflow. (You should be able to blow easily through
the tube, almost as if the wick wasn't there.)

Bend the copper tubing until it runs from where the CPU (and copper
block are to where you want the radiator (made from the other two
heatsinks) will be located. At the beginning, leave an extra inch or
two of overlap *past* where it fits over both sets of heatsinks. You'll
be mounting the tube full-length on each one.

Flatten the end of the tubing where the CPU block is to be attached with
a hammer. Not enough to close it off or even pinch tightly; but enough
to make a good and comparatively wide connection to that block. If
necessary, buy a slightly larger diameter pipe, solder that to the end,
and flatten the larger pipe. Pinch off the end with a cutting-pliars.

Flatten the other end of the pipe in a similar manner; but do *not*
pinch off the end. When you flatten the pipe-ends, you should keep in
mind the orientation of the CPU-block one end will be mounted on, and
the other end that will have the two other heat-sinks mounted on it.

Using a plumber's torch, flux, and plumbing solder, solder the flattened
end to the CPU block; and seal the pinched end at the same time with the
same solder. (Remember to *clean* the copper with sandpaper,
emery-cloth, or steel-wool before trying to solder it.)

Do the same thing at the other end; only this time solder the pads of
the two large finned-devices one on each side of the flattened tube.

At this time take a little extra time to plan and mount attachments to
the device for where it will fit on your computer system. Solder them
on too.

Fill the tube (from the sink-end, which you were supposed to leave open)
with isopropyl alcohol. Get it from your drugstore. Or better-yet, buy
it as "ISO-Heat" or similar gasoline-antifreeze.

Now find a *big* pot that will almost contain your new
heatpipe/heatsink. Fill the pot up with water, so that *with the pipe
inside the pot* it almost but not quite covers the open end of the
heatsink.

Take the heatpipe out of the pot.
dump out the (possibly contaminated) liquid inside into a measuring
container. Take careful note of how much liquid you removed. (Some
will remain inside, wetting the wick.)

Bring the water in the pot to a full boil.
Turn it down to "simmer".
Pour fresh isopropyl alcohol into the heatsink ... about 1/3 the amount
you dumped out and measured.

Using tongs, *carefully* put your heatsink loaded with alcohol into the
(just barely) boiling pot. Don't let water get in to mix with the
alcohol. Watch *closely*; and you'll soon see it begin to express gas
and perhaps even steam. After about ten seconds of this, pull the
heatpipe/heatsink combo out of the pot with tongs and *quickly* solder
the open end shut before it stops steaming.

Done.

Perhaps a *tiny* bit more complicated than designing and building a
water-cooled/pumped device; but not (to my notion anyway) seriously so.

Yes, fancy commercial heat-pipes use expensive metal foam or sintered
material as wicks ... But plain old cotton works fine too. You aren't
planning on the thing even getting up to boiling temperatures. That,
after all, is how early experimental heat-pipes were built.

Yes, it's a *lot* more expensive than just buying a commercial heatsink
with fan. But it gets the heat *ouside* the case without relying on
pumps, fans, or other power-consuming gadgets. The outside heatsink can
then be mounted to use normal air-convection to cool it.

Of course, it will probably be just as cheap, or possibly even cheaper,
not to mention better built, to buy a *commercial* heat-pipe/heatsink
combo made to fit your motherboard. They, doing it by hundreds, will do
it much cheaper than you ever could; and still sell it to you at about
the same price or less than you could build your own.

There's all these:
http://www.google.com/products?q=heat-pipe+CPU+cooler+PC&hl=en&um=1&sa=X&oi=froogle&ct=title
http://www.eastluna.com/hardware/details.php?id=2694&id2=b
http://www.frozencpu.com/products/5...ket_478_754_775_939_940_AM2_-_SCKTN-2000.html
http://www.pcper.com/article.php?aid=298

But they all seem to be designed to just be a bigger and more efficient
heatsink; not one going external.

Here's a passive set that comes with the motherboard ... but only for
the chipset; not the CPU itself.
http://www.pcstats.com/articleview.cfm?articleid=2092&page=2
http://www.pcper.com/article.php?aid=349&type=expert&pid=3

Here's somebody that supplies a whole passive "silent" CPU system using
heatpipes.
http://www.deltatronic.de/int/config_silent_pc.html

Hmmm ...
It seems that most people using heatpipes for CPU cooling use them to
enhance the efficiency of heatsinks with fans, *not* to take the heat
outside the case and disipate it there. I presume the main reason being
the varied differences in cases and motherboards. Thus water-cooled
designs with flexible tubing working better for such uses; allowing far
more flexibility in where and how they are mounted.

The few that *do* use heatpipes for completely passive systems, do so
for their own specially designed cases in which the heatsink *is* the
case.
 
Back
Top