Water Cooling Power Resistors

[MrAl]

Hello there,


In another thread we got to talking about testing power supplies with large power resistors and
hacklesup suggested immersing a power resistor in distilled water to keep it cool while testing.
The results were great and very surprising and that's what this thread is about.

Yesterday myself and a friend started testing another power supply which was from an older
computer so it was an AT type. It turned on without any jumpers or load so we tested it
a bit and found that a load on the +5v supply is a good idea in order to use this supply as
a general purpose bench supply because that keeps the +12v supply up high enough when it
is loaded also. Since i needed to test my ATX power supply too, we started thinking about
the distilled water idea again and decided to get out to the store to purchase some.
Since it was a grocery store i ended up buying a bunch of other stuff too, but at least i
got the distilled water so when we got back we were ready to do the tests.

The vessel used to hold the water was a clear plastic container 3.5 inches in diameter by about
6 inches high but only filled to about 2.5 inches with water because that's all it takes to
immerse a power resistor (or two) in the water.

Ok, but first the conductivity tests to make sure everything would work as planned. Each test
was performed by immersing two electrodes into the water and measuring the current with a
12v high current Lead Acid battery.
1. Tap water: This tested quite high, drawing 10ma at 12v with an electrode spacing distance of 2 inches.
2. Distillled water: This tested low but not zero: 30ua at 12v, same spacing.
3. Bottled 'spring' water: This tested almost like tap water: 8ma at 12v, same spacing.
4. Carbon filtered tap water: 6ma at 12v, same spacing.

So now we know that the current with a 12v supply is only 30ua, which is low enough so that
it will not cause excessive plating of the resistor leads or any electro etching of the leads either.
Interesting, the tap water may have worked too if the test didnt last too long.

Now on to the more interesting tests...
All temperatures quoted here are with the temperature probe clamped to the surface of the resistor
near the lateral center of the resistor body, and using distilled water.

1. A 5 ohm, 10 watt power resistor powered from 12v, which means it has to dissipate 28.8 watts
which is almost three times the power rating.
The result was that the temperature went up to 170 degrees F and stabilized. The stabilization was
due partly to the huge surface area of that container and partly due to evaporation which was evident
from the slight condensation forming on the inside walls of the vessel. It took about 10 minutes
to stabilize.

2. Two 1 ohm 10 watt resistors powered at 12v also. That's 72 watts into resistors that are together
rated to 20 watts, which means they are getting powered 3.6 times their power rating.
The temperature went up to 200 degrees F and may have still been rising very slowly but i ended the
experiment partly because time started to become an issue (getting late) and partly because i didnt
want to take a chance on ruining the resistors (which i doubt would really happen though).
The fact is though, that these underrated resistors lasted quite a long time in the water (maybe 20 minutes).

3. Here is the real kicker to this experiment:
This last test was done using ONE 1 ohm resistor, rated for 10 watts, and driven from the same 12v
power source. That's 144 watts into a 10 watt resistor! That's over 14 times the power rating!
The water temperature at the start of the test was about 150 degrees F, so this test
would have worked even better with cold water to start with, but even so it actually worked to some degree.
What happened is about 5 seconds after power was applied (12v) the resistor started to show bubbles near
the center of the body. This was because the resistor got so hot so fast that it caused localized super
heating and the water of course boiled at that center point. Still, it allowed time to measure the power
supply voltage, which actually measured 11.6v at that point. Total run time was about 30 seconds, but
that was long enough to measure the power supply voltage which is probably all that is needed. I want to
eventually test ripple too though, so all i have to do is have the scope probe connected before i power
up, then power up, note the scope peak to peak reading, then turn off again, all within about 10 seconds.
That will be long enough to perform the test, so what this means is we dont really need huge hugh
power resistors for quick tests like this.
Of course an overnight burn in run would be out of the question though, because the water would
evaporate at the very best, and that would mean the resistor would eventually be working in free air
again which would probably burn it up.

The short story is that the distilled water conducted much less than tap water and so worked very
well for cooling the resistor, at least temporarily, and double distilled water or super pure water
would be even better although probably a waste of money.
Oh yeah, that 30ua extrapolates to 300ua rms at 120vac so i wouldnt be doing this test with higher
voltages because there would be a shock hazard even if it wont kill you.


It would be interesting to hear other people's tests cooling resistors for testing power supplies
too if anyone wants to try some of these tests also.

Next test:
Immerse a Pentium IV 2GHz CPU in the distilled water, run a chess program, measure temperature
rise...
(just kidding)

[SETEC_Astronomy]

Next test:
Immerse a Pentium IV 2GHz CPU in the distilled water, run a chess program, measure temperature
rise...
(just kidding)

I'm guessing water wouldn't be a good idea but there are hobbyist who submerge their PCs in mineral oil for cooling. There are kits to make "fishtank" PCs. Not sure if you knew that or not, either way it's interesting.

[MrAl]

Hi SET,

No, i didnt know people were experimenting with this kind of technology.

You know what is interesting is that part of the problem that makes new design CPU's
hard is how to get rid of the extra heat due to the extra power needed to reach
higher and higher frequencies. With some kind of immersed solution like this (perhaps
just the CPU keeping the amount of oil to a minimum) faster CPU's would definitely
be possible. Perhaps the manufacturer could use an oil filled CPU, where the CPU
chip is on the bottom and the oil fill chamber on top and some big fins and a fan.
Using the oil it would be possible to use convection currents to distribute the heat
over a much much greater area making it easier to cool with fans. Of course coolant
might be a better idea, but whatever it is it has to be nearly fool proof so that there
is no increased safety risk.

I checked out the fish tank solution and it is very cool. We were thinking of doing just
the power supply at one time (yesterday). I would hate to have to clean up the mess
though if something leaked out.
I like the aquarium stone too on the bottom, that was a nice touch.
What else is nice is the boards would probably not be subject to growing 'whiskers'
as readily.
I did a few calculations and found that a 10 gallon aquarium (just for reference) would
support up to about 250 watts without a fan. A larger aquarium (that would be needed
for the whole computer) would probably work up to 350 watts or more without a fan,
with the temperature rise about the same as it is in free air, once equilibrium is reached.

Before i did something like that though i think i would want to know more about the properties
of mineral oil, as to how easy it catches on fire, burns, and also how it reacts to various
plastics used for electrical connectors and wire insulations...stuff like that.
I wonder what would happen if the power supply arced over inside while it was immersed
in mineral oil. Fire, explosion, or just quiet damping of the arc? Apparently it would damp
the arc as i have read that it is used in high voltage transformers.

[Bob Scott]

This reminds me of the Heathkit "Cantenna", the 1,000 watt load in a 1 gal paint can. They filled the can with oil. I don't know what type of oil, but I suspect it was mineral oil. Mineral oil is also sold as "baby oil" and is not poisonous. I think it is sold in some form as oral medication. If you look up the heat properties of mineral oil like I did a couple of years ago, you'll find that has a specific heat around 0.7 that of water. I have no idea what its R factor is (as in insulation from heat conductivity).

I have never found water to be perfectly non-conductive. It always measures at least a couple of megohms, although I can never be sure whether the meter is measuring actual resistance of the water, or is reacting with the metal of the probes electrolytically to generate a few microvolts. If it does, the voltage is interfering with the resistance reading. No matter how far apart I place the probes in a glass of water, the reading is constant.

I'd switch to oil. If water corrodes some metal even slightly, it gets contaminated with ions and the corrosion accelerates.

As for power resistors dissepating huge amounts of power when liquid cooled, I suppose experimentation will give varying results. I think that you will find that those neat looking "Dale type" aluminum housed (heat sunk) power resistors will not survive your experiments. I found that the innards of those resistors instantly boil, blow out the plastic end caps with dramatic smoke and boom of a cannon, even when overloaded slightly and mounted to a heat sink. Yes, the engineer in charge of our lab gained notoriety for his dramatic pyrotechnical design experiments. The smell, we called "essence of Ohmite", drifted through the whole building.

[Edit] This is a coincidence. I researched the mineral oil suitability while my brother and I were designing computer cooling systems for Coolit Systems of Calgary, 8 years ago. I was for mineral oil. He was for a non-toxic form of glycol. He won.

[Edit2] According to one HAM buy/sell advertisment, the Cantenna uses mineral oil.

Bob

[Janitor Tzap]

Heres an article from Toms Hardware website on Oil cooled PC.
http://www.tomshardware.com/reviews/str ... ,1203.html

I remember some one asked me about using distilled water for cooling.

Basically the water itself isn't conductive, but it is all the minerals, and contaminates that are.
And getting all those things out of the water, is far to time consuming and labor intensive.

I know of people who have used freon for cooling circuits.
But that has its own problems.
In that it the circuit and freon have to be completely sealed up in a Thermos type bottle.
Any place where wires are coming out of the bottle can be problematic for leaks.

There is some stuff that 3M makes that the CRAY II used to cool it's central processing tower.
Like freon, it required a closed system, but not a Thermos type bottle.
The CRAY II used a circulating pump to move the fluid.
There was and article in Nuts&Volts about it a few years back.
I remember it stated that the liquid had to be replace at regular intervals.
{I wonder how expensive that was.}


Signed: Janitor Tzap

[MrAl]

Hi Janitor,


Well, i didnt bother to get the minerals out of the water, i just went down to the store and
purchased some distilled water. It worked good enough for 12v anyway.

Nice site there too, very interesting. I guess there are quite a few people out there that
like to try this kind of thing. I'll probably stick to resistors myself though. Maybe higher
power transistors at some point too.

[CeaSaR]

MrAl,

I posted some resistor info over on your Computer PSU repair thread.

Hope it's helpful,

CeaSaR