Well...

If you want to get into the nitty-gritty, there are but TWO forms of
heat transfer I am aware of to either heat or cool, radiation and
conduction.  Convection is not a means of heat transfer.  It is moving
a body with its associated energy from one place to another, e.g. the
air (a thermodynamic fluid) is heated by conduction from the heat
source and it then moves from one part of the equipment to another
either naturally [heated air rises as it has expanded and thus is more
boyant] provided there are inlet and outlet air vents in the equipment
and we are not in a vacuum or by force via a fan.  Technically fan
cooling is convection just as a "convection oven" is an oven with a
fan in it - they should called them "forced convection ovens" but they
don't.  I should have said "natural or free convection" as none of the
equipment I design uses fans but instead relies upon natural
convection to remove heat from components to the ambient environment
(which can be up to 85C).

Obviously, a fan does increase the exchange rate of the air so that
new cooler air replaces it and thus conduction again heats the new
air.  Of course if this fan system is in a sealed room you eventually
reach an equilibrium temperature which then depends on the radiation
out of the room and that depends on the temperature of the environment
outside the room.  Nome Alaska cools a room (nomatter how good the
insulation) far better than the Sahara.  I don't like fans; they are
noisy, they fail, their filters clog (or there are no filters), the
equipment becomes a vacuum cleaner and dirt is deposited everywhere
internally (they don't help the life of floppy disk drives).  The best
I ever saw in a fan cooling design was in Tektronix oscilloscopes (my
465 has this feature) - a temperature controlled variable speed QUIET
fan.  I hardly put up as an example of a good industrial design a PC
and it's fan!  These are throwaways - we design for the -40 to +85C
regime and 30 years life and as you do spacecraft, I'm sure your
designs are even more conservative (even with the Space Shuttle, a
service call is pretty expensive in space!).

Yes, of course, two surfaces that radiate (both sides of the PCB) are
better than one as is twice the surface area.  And the fact is that a
tin-lead plated copper plate is a lousy radiator - blacken it and it's
much better.  But then what is it radiating TO?  Blacken the walls of
the container (3M used to make a lampblack type paint that was ~90%
absorbing) and it can suck up a lot of radiated heat.  Of course then
it has to radiate that heat to the outside environment.

I find it hard to believe the general statement that holes in a PCB
will cool unless the air is flowing through the board and that depends
upon the position of the board doesn't it?  If you are going to all
that trouble, a clip-on heatsink sticking up in the breeze would be
far better than the PCB.

As for me, I'm not a big fan of "experimental" evidence.  I prefer
good theory backed by some experiments on the particular design
because there are too many variables in each design that influence
cooling.  I'd still like to see the calculations.  I've not got the
time to dust off my old physics or ME books but it's in there under
either radiation or heat transfer topics, I'm sure.

Regards,

Bob Landman
H&L


----- Original Message -----
From: Abd ul-Rahman Lomax <[log in to unmask]>
To: <[log in to unmask]>
Sent: November 16, 1999 1:27 PM
Subject: Re: [DC] Calculating thermal dissipation of external plane
heat sink area


: At 06:41 PM 11/15/99 -0500, Bob Landman wrote:
: >I have a comment also.  This is often done but I don't have any
: >calculations to back up the practice.  It seems to me that this
: >surface, whether 1 oz or 2 oz or whatever, is a black body radiator
: >and that is what is "cooling" the heatsink tab on the regulator.
It
: >isn't convective cooling so putting holes in the plane makes
matters
: >worse as you have reduced the surface area of the radiator.
:
: A practice which is verified by experiment does not need to be
"backed up"
: by calculations. I did not make up the hole idea; I took it from an
article
: which appeared, as I recall, in SMT magazine showing the increase in
heat
: disippation from the use of holes, from actual measurements. It was
: substantial, but I do not recall the exact figures. An increase in
power
: disippation for the same heat rise, as I recall, was something like
a
: factor of five or ten. Larger holes were better than smaller holes.
Maybe I
: can find the article....
:
: It does seem to me that holes would reduce radiative cooling;
certainly
: radiation normal to the surface would be reduced, but this effect
would
: only hold at angles to the surface such that the other side of the
holes
: could be seen (i.e., one could see, from that angle, all the way
through
: the board). The loss in radiation would rapidly decline as the angle
: deviated from normal to board surface. I don't think that it would
be a
: large effect, overall.
:
: In the contrary direction would be two effects: the absolute surface
area
: of copper would be increased (and thus there is an increased area in
: contact with air), and there is improved heat transfer between both
sides
: of the board (the idea I had in mind includes having copper plane on
both
: sides). Thus even radiative cooling might be increased, since there
is
: radiation from both sides of the board.
:
: I don't know why Mr. Landman would say "it isn't convective
cooling." I
: design boards for space flight where the environment is a vacuum
and,
: definitely, we have to be much more careful about cooling because of
the
: lack of atmosphere. Ultimately, all the cooling is radiative.
:
: For the rest of us, if convective cooling were not significant, why
in the
: world do I have a fan in my computer's power supply? The fan does
not
: increase radiation, and it does not increase conduction, in only
increases
: convection.
:
: If I have the terms right, there are three kinds of cooling:
radiative,
: conductive, and convective. Putting holes in the board may reduce
radiative
: cooling a little, but it will probably increase conductive and
convective
: cooling. (When I originally wrote "convective," I meant "through the
air"
: and thus was including both air conduction and convection.)
:
: [log in to unmask]
: Abdulrahman Lomax
: P.O. Box 690
: El Verano, CA 95433
: