At 08:54 PM 11/16/99 -0500, Bob Landman wrote:
>Now it is Mr. Lomax who is not speaking "precisely".

And Mr. Landman went on to give what was, indeed, a precise definition of
convection. If it differed from what I wrote, it escapes me.

Now, to the matter at hand:

>I still maintain that punching a bunch of holes in a PC board with
>clad on both sides to serve as a heat sink is a worthless endeavor but
>do what you will.  Until and unless I see some math to back it up, I
>for one, will not use the technique.

Mr. Landman maintains this, not only without providing any theoretical
analysis, but also in the face of experimental evidence. Such a heat sink
has been tested and shown to result in a substantial improvement in heat
dissipation; further, such an improvement is not theoretically strange; it
would result from improved transfer of heat between the surface of the
board and the surrounding air (or other medium -- I've done PCBs to be
operated immersed in oil).

But it is not necessarily easy to calculate the effect without some
experimental basis. Such experiments, on the other hand, would not be
difficult to perform. Mr. Landman seems to consider the matter complex:

  I've done more than my share of
>optical and temperature measurements (critical fluctuations of CO2 at
>its triple point [where it is a gas/liquid and solid] is one such
>experiment and it took THREE boxes within boxes with three temperature
>control loops to do it {to hold the temp to a millidegree} but before
>we did the experiments, we had an idea where we were headed and used
>math to get there.
>
>That is THE way to be "precise", in my opinion.

Some kinds of precision are totally unnecessary. This is engineering, after
all; we are not trying to determine the mass of an electron to ten decimal
places; if we could know the thermal resistance of the assembly within a
few percent, I'd expect that to be adequate.

As I mentioned, the method of drilling holes in a PCB I first saw in an
article in SMT magazine some years back. Perhaps two years, I'm not sure.
The authors, if I recall correctly they were from Motorola, had mounted a
particular SMT device on a bare board and also with various hole patterns
in the board. Then they cranked up the power dissipation of the device,
again, as I remember, until it reached a certain steady temperature; this
was repeated for each pattern; they recorded the power dissipation for
each. Assuming that the other experimental variables were the same, this
experiment would conclusively demonstrate the effect -- or lack of same --
of holes in a PCB, and would also provide, together with other information
such as the ambient temperature, a measure of the thermal resistance of
each assembly to the air.

This kind of data is exactly what the original writer wants to know. *How
much power can he dissipate through a PC board heat sink.*

It's a *lot* easier to measure such a thing than it is to accurately
calculate it, unless one has equations representing the experimental and
theoretical work of others who came before *and* one understands all the
issues.

I wish I had the formulas for the original question. I merely wrote what I
wrote because the object is usually to get the device cool rather than to
know precisely *how* cool it will be, as long as it is cool enough. And,
unless the board is going to be in a vacuum or will be potted, those holes
will help.

Even if the board is in a tight box, the holes will help with heat transfer
to the air within the box and thus to the outside of the box and, as I
noted, *theoretically* the holes will not substantially lower the radiative
cooling. (If the holes were 25% of the area of the board in the vicinity of
the device, the radiation would not decrease by 25% because only the
radiation normal to the board surface would decrease by that; at lower
angles the radiation would approach that of the undrilled surface and at an
angle low enough such that the bottom of the hole was no longer visible,
there would be no decrease at all. And the total radiation is the integral
of the radiation over all angles.... Without actually doing the
calculation, I'd guess the decrease would be perhaps 10%; obviously it
would depend on the hole size vs. board thickness.)

If I'm right, then, perhaps we would lose 10% of radiative cooling but we
will gain in conductive cooling, since the surface area increases, and I'd
expect improved convection as well. From the experiments, we obviously gain
much more than we lose. How much?

The article, even if we find it, was rather poorly written (or poorly
edited). It would be nice to have some better data, for sure. And with
better data, we could even come up with some math to satisfy Mr. Landman....

[log in to unmask]
Abdulrahman Lomax
P.O. Box 690
El Verano, CA 95433