[Amps] FW: Water cooling
Fuqua, Bill L
wlfuqu00 at uky.edu
Mon Aug 22 21:17:00 PDT 2011
From: Fuqua, Bill L
Sent: Tuesday, August 23, 2011 12:16 AM
To: Manfred Mornhinweg
Subject: RE: [Amps] Water cooling
Here you go. If you want really effecient heat transfer use one of these heat sinks.
From: Manfred Mornhinweg [manfred at ludens.cl]
Sent: Monday, August 22, 2011 6:58 PM
To: Fuqua, Bill L
Subject: Re: [Amps] Water cooling
> If you want to look at what would be extreme compared to most
> electronic applications, just look at a car engine.
Yes, I have been thinking a lot about car engines. But it's hard to
translate that to my amplifier! In car engines the thermal power is far
larger, but also the allowable thermal gradient is far larger, the flow
rate is far larger, the contact area is far larger, and worst of all, I
don't really know how much contact area a car engine has! I really would
need to take apart one, cut open both the block and the cylinder head,
and measure them!
> Say you have a
> car engine with about 100HP. When full throttle it has to dissapate
> around 8 time that in heat, 800HP. That is around a 100kW.
I agree with the 100kW of thermal power, and not even at full throttle
but just at normal highway cruise! At full throttle it's way more! But I
do not agree with 800HP being 100kW, nor with 8 times as much heat as
mechanical power! 100kW is more like 135HP, and the thermal output of a
gasoline engine is about 3 times the mechanical power, or even slightly
less than that, and part of that (I don't know how much, but certainly
the smaller part) goes through the exhaust rather than into the water.
Even so I also reached the 100kW figure! My calculation was simple: At
highway speed my car burns around 15 liters per hour. One liter of
gasoline contains about 9kWh of gross energy. About 3/4 of that goes
into heat, so that is about 100kW!
> So a 3/8 inch
> copper refigeration tube( thin wall) mashed a bit to improve contact
I have tried before flattening copper tubes, and never had much success.
They tend to buckle in, or even rip open along the seam. I guess the
tubes I can get here are not the best.
> and silver solder which would be more conductive than copper
What solder is that? I mean, if it has enough silver in it to really
conduct better than copper, then it must be pure silver! That would be
way too expensive! Pure silver conducts heat just a tad better than copper.
I just looked up values for solders: Common 60/40 solder conducts heat
about one fifth as well as copper, and about one third as well as an
average aluminium alloy (pure aluminium is better than alloy). But the
"silver solder" made from 2.5% silver, 5% indium, and the rest lead, is
significantly worse than 60/40 solder! And silver solder containing 3%
silver, 0.5% copper, and the balance being tin, is slightly better than
60/40 tin /lead, but not much.
Then I found another source stating these solders to be about 7 times
worse than copper!
On the other hand, even that might be enough. After all, the thermal
path through the solder is short and wide.
In a catalog from an aluminium supplier I found some extrusion stock
that looks much like an Omega sign, but closed. Like a round tube
attached to a flat plate. Unfortunately there are no dimensions given!
So I will have to go to the store and look. But if the dimensions are
right, that might be a great alternative! The aluminium is probably
about as good as a combination of copper and solder. But I have to check
if the extrusion is sufficiently thick-walled.
If I actually find that stuff to have promising dimensions, I would buy
a piece, and test it with a row of cheap transistors mounted on it!
> The real question is, where does the heat go
> from here?
That's no issue for me. It would go into a closed container under the
desk, holding maybe 20 liters of water. In a typical operating session,
I hardly transmit more than half an hour total, spread out over two or
three hours, and that's in SSB. Average dissipation of a legal limit amp
in SSB is probably around 600 watts, so that would make just 0.3kWh for
half an hour of TX time! That would heat 20 liters of water by just 15
degrees Celsius, assuming the container does not loose heat in those 3
hours! It would cool down naturally, almost completely, before the next
And if I should embrace RTTY contesting or another such high duty cycle
activity, well, it would be time to run water hoses from the bathroom to
the shack! I live in the woods, and have plentiful free water from a
spring, so I could afford to actually just let the water run!
If there is any concern of running spring water through the amp, then
the spring water could run through a copper tube coil placed in the
container under the desk.
So I'm not worried about what to do with all that warm water. My concern
is how to calculate things, to be sure how hot the transistors will be
running! And also I have some concerns about corrosion, specially if I
go along and use that aluminium extrusion. To this add that I need the
water to have pretty low conductivity, because the two heat sinks (or
rather, "cold plates", to use the correct name!) will be at drain
potential, not at ground, and even better, those drains will NOT have
galvanic insulation from the grid!
By simply letting the water in the bucket stay at grid potential too, I
can easily avoid tripping the ground fault interrupter, but the water
will be exposed to the full drain-to-drain RF voltage, which will be
like 140V RMS at full power. I measured the resistivity of my spring
water, and found that the water in the short hose running from one
MOSFET group to the other would have a resistance like 200 kiloohms, so
in principle that's a non-issue too. But if over time the water
contaminates with metal ions, that could be bad news! Talk about a
slowly growing dummy load built into the amp! ;-)
Visit my hobby homepage!
More information about the Amps