Cecil Moore wrote in news:avVKh.3753$Qw.1263
@newssvr29.news.prodigy.net:

One fact to note is that the virtual impedance changes
all up and down a transmission line yet no additional
reflections occur while the Z0 is constant. Reflections
occur only at *actual* impedance discontinuities, e.g. at
a junction of two different Z0s.

Cecil, that is a simple statement of a scenario in which reflections
*may* occur, but *not always* occur.

Think about it and you will think of examples where a reflection does not
occur at the "junction of two Zos".

I am not quite sure what you mean by an "impedance discontinuity" beyond
the simple "junction of two different Zos" case.

The magnitude of a reflection (zero or otherwise) is *always* and *only*
related to whether the ratio of V to I for the "thing" (whether it is
another line, a lumped circuit or some combination) attached to the end
of the line is equal to Zo. The magnitude is calculated from V/I (Zl) and
Zo using a well known expression.

Should your "rule" be more correctly stated as Reflections
may occur only at *actual* impedance discontinuities, e.g. at a junction
of two different Z0s. Since it has "may" in there, it isn't a rule, is it
worth stating? It is just one of those "may"s that people like to parrot
until they become a Rule of Thumb (ROT).

Owen

Owen Duffy wrote:
The magnitude of a reflection (zero or otherwise) is *always* and *only*
related to whether the ratio of V to I for the "thing" (whether it is
another line, a lumped circuit or some combination) attached to the end
of the line is equal to Zo. The magnitude is calculated from V/I (Zl) and
Zo using a well known expression.

I assume you are talking about virtual reflection
coefficients based on virtual V/I impedances, something
that has gotten hams into conceptual trouble for any
number of years. Let's take a look at the S-Parameter
equations.

Port 1 | Port 2
---Z0---+---Z1---
a1-- | --a2
--b1 | b2--

b1 = s11*a1 + s12*a2
b2 = s21*a1 + s22*a2

As you probably know, s11 is a *physical* reflection
coefficient involving unequal impedances.
s11 = (Z1-Z0)/(Z1+Z0)
Z0 and Z1 are physical impedances, not virtual
impedances, i.e. *NOT* merely a V/I ratio.

a1 is the voltage wave incident upon Port 1.
s11*a1 is the reflection from Port 1.

If Z1 Z0, there exists an impedance discontinuity.
s11 0, and s11*a1 0, i.e. there exist reflections.
This is why I say: If there is a physical impedance
discontinuity, then reflections exist.

If reflections are to be canceled, then s12*a2
must be equal in magnitude and 180 degrees out of
phase with reflection s11*a1. Note that for
reflections to be canceled, they must first exist.

s12*a2 is the voltage not reflected from Port 2.

All this is covered in HP Application Note 95-1,
"S-Parameter Techniques" available from:

http://www.sss-mag.com/pdf/an-95-1.pdf
--
73, Cecil http://www.w5dxp.com

"Cecil Moore" wrote
...That's simply not true. When the load is connected directly to
the source, incident power is often still rejected, it just doesn't
have very far to "bounce". And since it is internal to the source,
the "bouncing" is difficult if not impossible to quantitize. etc
_______________

Does the lack of a technical response to Cecil's post (so far) mean
that his analysis and conclusions are understood and accepted?

Hopefully so.

RF

Richard Fry wrote:
"Cecil Moore" wrote
...That's simply not true. When the load is connected directly to
the source, incident power is often still rejected, it just doesn't
have very far to "bounce". And since it is internal to the source,
the "bouncing" is difficult if not impossible to quantitize. etc

Does the lack of a technical response to Cecil's post (so far) mean
that his analysis and conclusions are understood and accepted?

The "eliminate the transmission line" sword cuts both
ways. If the source cannot tell the difference between
driving a one wavelength transmission line and driving
a lumped circuit load directly, it follows that the load
cannot tell if it is being driven by a one-wavelength
transmission line or being driven directly by a source.
The incident signal looks the same in either case and the
load rejects (reflects) the same amount of forward power
either way. Except for the energy stored in the one-
wavelength transmission line, conditions are the same
in either case.
--
73, Cecil http://www.w5dxp.com

"Cecil Moore" wrote in message
...
Richard Fry wrote:
"Cecil Moore" wrote
...That's simply not true. When the load is connected directly to
the source, incident power is often still rejected, it just doesn't
have very far to "bounce". And since it is internal to the source,
the "bouncing" is difficult if not impossible to quantitize. etc

Does the lack of a technical response to Cecil's post (so far) mean
that his analysis and conclusions are understood and accepted?

The "eliminate the transmission line" sword cuts both
ways. If the source cannot tell the difference between
driving a one wavelength transmission line and driving
a lumped circuit load directly, it follows that the load
cannot tell if it is being driven by a one-wavelength
transmission line or being driven directly by a source.
The incident signal looks the same in either case and the
load rejects (reflects) the same amount of forward power
either way. Except for the energy stored in the one-
wavelength transmission line, conditions are the same
in either case.
--
73, Cecil http://www.w5dxp.com

in steady state... where your favorite s equations hold. this is true. it
is not true in the general case where you account for startup transients.

Dave wrote:
in steady state... where your favorite s equations hold. this is true. it
is not true in the general case where you account for startup transients.

Thanks Dave, since I was talking about steady-state,
I should have said so.
--
73, Cecil http://www.w5dxp.com

Richard Fry wrote:

Does the lack of a technical response to Cecil's post (so far) mean
that his analysis and conclusions are understood and accepted?

Hopefully so.

In my case, it's because I plonked him long ago.

Roy Lewallen, W7EL

Roy Lewallen wrote:
Richard Fry wrote:
Does the lack of a technical response to Cecil's post (so far) mean
that his analysis and conclusions are understood and accepted?

Hopefully so.

In my case, it's because I plonked him long ago.

For pointing out that an antenna is a distributed
network, not a lumped circuit.
--
73, Cecil http://www.w5dxp.com

Points well taken.

In my "Food for thought" essay
(http://eznec.com/misc/Food_for_thought.pdf) I use a voltage source in
series with a resistor for most examples. Among the calculations are
those showing the power dissipation in the resistor. I've used this
simple circuit a number of times to illustrate various points regarding
transmission line operation and the effects of traveling voltage and
current waves.

People who aren't willing to accept the points being made borrow from
the politicians' play book and immediately declare the source to be a
"Thevenin equivalent" and therefore any calculation of source power to
be invalid and meaningless. This handily diverts the discussion from the
fundamental topic to something more to the attacker's liking. It can
then proceed to endless arguments about the magnitude and linearity of a
transmitter's output impedance, and whether or not it constitutes a
"dissipationless resistance". The discussion has followed this path many
times, and I'm sure will do so many times more.

The essay shows that "reflected power" is NOT absorbed or dissipated by
the source resistor in my simple circuit -- which is NOT a Thevenin
equivalent of a transmitter or anything else (although, as I point out,
it is a reasonable model for some signal generators). What remains for
the people promoting the notion of waves of average energy propagating
like voltage and current waves to show is how their theories can explain
the resistor dissipation in the very simple circuit I used. (How about a
single equation showing the resistor dissipation as a function of
"reflected power"?) Only after that is done is it necessary to begin the
argument about what the output of a transmitter looks like.

Roy Lewallen, W7EL

J. Mc Laughlin wrote:
I teach my students that prior to analysis of an electrical, electronic, or
EM network/system one must ask and answer a critical question. The question
is: Is the network/system linear, close enough to linear for engineering
purposes, or not linear?

If linear, or essentially linear, one brings into play linear analysis.
Thevenin equivalents, which are only equivalent as far as what they do to
the outside world, are a part of linear analysis.

Most RF power amplifiers that deliver more than one or two watts are
non-linear circuits. Typically, the active device conducts for only a
fraction of each cycle. How else could one get DC power to RF power
efficiencies of over 50 %? Great care must be taken in modeling such
circuits.

A simple example: Consider a transformer fed bridge rectifier (very
non-linear) that is connected to an (old fashion) series L, shunt C filter.
In steady state, if L is large enough, one may model the rectifier as a
series of series connected voltage sources with harmonically related
frequencies (and a DC source). It is left as an exercise for the student to
decide on the sizes, frequencies, and phases of the sources. (Because of
the LPF properties of the LC network, one does not need many harmonics.)
Then one may apply superposition (the essence of a linear process) to
estimate the ripple on the load. However, the model just described is
invalid if L is too small or if L is non-linear. The model is insufficient
to predict the losses in the rectifier. This example is not likely to be
found in current electronic texts, but we all know for whom they are
written.

Techniques exist for dealing with many non-linear networks. They must be
used with great care. If one holds one's nose, one might find an
"equivalent" for a transmitter that suffices for describing what happens
outside of the transmitter, but not inside of the transmitter. Please do
not make conclusions about the "equivalent" itself. Please discriminate
between linear and non-linear networks.

Thus ends the lecture. 73 Mac N8TT

--
J. Mc Laughlin; Michigan U.S.A.
Home:

I teach my students that prior to analysis of an electrical, electronic, or
EM network/system one must ask and answer a critical question. The question
is: Is the network/system linear, close enough to linear for engineering
purposes, or not linear?

If linear, or essentially linear, one brings into play linear analysis.
Thevenin equivalents, which are only equivalent as far as what they do to
the outside world, are a part of linear analysis.

Most RF power amplifiers that deliver more than one or two watts are
non-linear circuits. Typically, the active device conducts for only a
fraction of each cycle. How else could one get DC power to RF power
efficiencies of over 50 %? Great care must be taken in modeling such
circuits.

A simple example: Consider a transformer fed bridge rectifier (very
non-linear) that is connected to an (old fashion) series L, shunt C filter.
In steady state, if L is large enough, one may model the rectifier as a
series of series connected voltage sources with harmonically related
frequencies (and a DC source). It is left as an exercise for the student to
decide on the sizes, frequencies, and phases of the sources. (Because of
the LPF properties of the LC network, one does not need many harmonics.)
Then one may apply superposition (the essence of a linear process) to
estimate the ripple on the load. However, the model just described is
invalid if L is too small or if L is non-linear. The model is insufficient
to predict the losses in the rectifier. This example is not likely to be
found in current electronic texts, but we all know for whom they are
written.

Techniques exist for dealing with many non-linear networks. They must be
used with great care. If one holds one's nose, one might find an
"equivalent" for a transmitter that suffices for describing what happens
outside of the transmitter, but not inside of the transmitter. Please do
not make conclusions about the "equivalent" itself. Please discriminate
between linear and non-linear networks.

Thus ends the lecture. 73 Mac N8TT

--
J. Mc Laughlin; Michigan U.S.A.
Home: