Richard Clark wrote:

On Mon, 28 Feb 2005 22:53:17 GMT, gwhite wrote:

Obviously people don't have 100 W (or more!) network analyzers looking into the
output and pretending the device is similar to a linear small signal device.

Hi OM,

Well, it is more accurate to say that you don't, that is for sure.
Defining a solution by negative results can fill up a library without
any positive accomplishment. Obviously people don't have a nuclear
reactor, or lunar lander, or bank account to balance the national
debt. The joke of this, of course, is that no one needs a 100 W (or
more!) network analyzer, or nuclear reactor, or lunar lander, or bank
account to balance the national debt to explain a rather more trivial
problem. Which, by the way, has nothing to do with pretending at all.

You entirely missed the point. You don't know the output impedance because you
don't have a way of determining it by swinging the output full-scale. Even for
class A, large signals will/can have rail to rail swing. The device will not be
linear for large swings: sinusoidal input swing will not result in a sinusoidal
output swing. But "impedance" is a sinusoidal (s-domain) concept. So how can
you define an impedance--a sinusoidal concept--when the waveform is not
sinusoidal for an inputted sine wave? The point is that the output impedance is
time dependent ("causes" the non-sinusoid output for sinusoid drive), which
rather makes the concept questionable. As I wrote earlier, one might decide to
consider a time averaged impedance, but I'm not clear on what the utility would
be.

The suggestion that
requires load pull test equipment and that can be expensive
does not negate its existence which commonly proves what you choose to
dismiss as impossible. I have calibrated this gear (called an
artificial or active load), and the gear (called transmitters) it
tests and there are no differences in Physics based upon your
presumption of low-power/high-power demarcations.

There is no "presumption." Linear parameters and theorems totally ignore
practical limitations--this is a fact and you can look it up in just about any
text on circuit analysis. The simple linear model is perfectly okay for small
signal devices. It isn't okay for large signal devices. In any case, load pull
equipment does not make the pretense of defining output impedance of an active
large signal device. It does say what the load needs to be to acquire maximum
power out of the device.

To say
pretending the device is similar to a linear small signal device
is one of those assumptions forced into the argument.

No, it isn't. Thevenins and conjugate matching (for maximum power transfer) are
explicitly linear small signal device models. Their use in RF PA output design
is a misapplication.

There are any
number of ways to do something wrong.

We're talking about one of them. Misapplying small signal linear parameters to
the output of a large signal device.

Trumping none of these straw
men validates another wrong impression passing as theory.

This
returns us to the imposition of impossibilities to answer a rather
mundane concept, eg.
pretending the device is similar to a small nuclear device
pretending the device is similar to a mars rover
pretending the device is similar to the national debt of Lithuania

Who are you quoting and why?

So to return to a common question that seems to defy 2 out of 3
analysis (and many demurred along the way) - A simple test of a
practical situation with a practical Amateur grade transistor model
100W transmitter commonly available for more than 20-30 years now:
1. Presuming CW mode into a "matched load" (any definition will do);

Any definition won't do, and for this discussion the specific "won't do" is
using conjugate matching which is a small signal (linear) model.

2. Report the DC power consumed before hitting the key;
3. Report the DC power consumed while holding the key.

Hey, at least you're recognizing that DC power is important. Where in conjugate
matching ideas or Thevenins theorem do you see any concern of DC power? That's
right, you don't because they a simple small signal models where DC power and
voltage have no bearing because the signals are so small, relatively speaking.

Concurrently note:
A. Report Heat Sink Temperature for a previously idle/rcv condition;
B. Report Heat Sink Temperature after 10 minute key-down.

For a hypothetical "100W" model (again, a contemporary, common example
for Amateur use) available through standard commercial venues:
2. About 20W - 30W
3. About 200W - 250W
A. About 20 degrees C (or room temperature)
B. Well above 37 degrees C (or skin temperature)

Now, if we are to be any judge of efficiency (Thevenin does not have
to be invoked, condemned, or venerated); then it runs close to 50%
(±10%). Others can invoke their favorite deity to explain.

*You* brought up Thevenins and armchair philosophy regarding it, not me. I said
Thevenins was irrelevent, and now you appear to agree with me. Ken effectively
brought up conjugate matching, not me. The original comment I was challenging
was:

"...the antenna works as an impedance mathcing network that matches the output
stages impedance to the radiation resistance."

I simply wanted to make it clear that the "matching" done was not an issue of
"output impedance" per se. It is an issue of how the transistor is to be loaded
to extract maximum ouput power.

Now, if we are to be any judge of dissipation (no requirement for
advanced degree); then heat as a loss by virtue of less than 100%
efficiency is quite evident. Others can invoke photons to describe
why.

To forestall any armchair engineers, yes, this efficiency is System
efficiency. However, I would be surprised if a practical common
Amateur grade transistor model transmitter commonly available for more
than 20-30 years now has any configuration that does not apply supply
voltage directly to the final transistors; and instead adds a
significant current path outside of this load (citations to available
schematics would be compelling, but any argument without this would be
speculation). It takes very little effort to subtract out the power
drain of the receive mode (being very representative of the similar
power demand of supporting circuitry for transmit up to the driver
stage). Barring such amazing evidence of a significant power drain
not found in the finals, it follows that a simple computation of
efficiency has its merit and has been met.

Exactly. It is about DC to RF efficiency, as I've been pointing out since my
first post, and which you initially commented was "nonsense" but now seem to
agree with. "Impedance matching" meant in the normal sense of conjugate
matching for maximum transfer of power is a misapplied small signal
concept/model. I think that is all I've really been saying.