On Sat, 14 Apr 2007 16:04:55 -0700, Roy Lewallen wrote:

Walt, before digging into your recent posting, I'd really like to get
one issue settled. I think it would be helpful in our discussion. The
issue is:

Can you find even one example of any transmission line problem which
cannot be solved, or a complete analysis done, without making the
assumption that waves reflect from a "virtual short" or "virtual open"?
That is, any example where such an assumption is necessary in order to
find the currents, voltages, and impedances, and the magnitude and phase
of forward and reverse voltage and current waves?

Roy Lewallen, W7EL

No Roy, of course not. I am not attempting to assert that reflection coefficients should be used in such an
analysis. I'm only asserting that it's another way of performing an analysis, one that I believe paints a more
visible picture of the how the pertinent waves behave in the circuit.

If I still haven't persuaded you that it's a viable way of analyzing the impedance matching function then I'll
back off and not pursue the issue any further.

Incidentally, you didn't answer my questions.

Walt

Walter Maxwell wrote in
:

....
Re your worked solution (above), I agree that the normalised
admittance looking into 30deg of line with load 16.667+j0 is about
1-j1.1547 (not the different sign).

Yes Owen, you're right. I added the y values at the last moment, and
didn't catch the errors. Both the line and stub signs are reversed.
Sorry 'bout that.

Ok.

I make the normalised admittance looking into the stub about 0+j1.15
(and the reflection coefficient about 0.5-98, how do you get 1+j1.15?

Normalized y looking into the stub directly is y = 0 + 1.1547, but
looking at the stub while on the line at the 30° point is y = 1 +
1.1547. To view the stub separately on the line the line is terminated
in 50 ohms, because the real component of the line impedance at the
match point is 50 ohms.

You have a junction where three current paths appear in parallel, we can
add the admittances of each of those paths.

We are agreed that admittance of the load+30deg line is 1-j1.15, and that
of the stub is 0+j1.15, so the only place the additional 1+j0 can come
from is the source+line branch.

If that is the case, then your explanation of the stub (which I assume to
be a steady state explanation because you are talking about frequency
domain admittances), depends on the source admittance (or impedance). If
the equivalent source impedance at the junction figures in the calcs, you
are saying that the VSWR on the line from source to junction depends on
the source impedance... I thought we got over that error.

My view is that the stub in shunt with the 30deg line+load results in an
equivalent impedance of approximately 50+j0 at the junction, irrespective
of what is on the source side of the junction.

Owen

Walter Maxwell wrote:
On Sat, 14 Apr 2007 16:04:55 -0700, Roy Lewallen wrote:

Walt, before digging into your recent posting, I'd really like to get
one issue settled. I think it would be helpful in our discussion. The
issue is:

Can you find even one example of any transmission line problem which
cannot be solved, or a complete analysis done, without making the
assumption that waves reflect from a "virtual short" or "virtual open"?
That is, any example where such an assumption is necessary in order to
find the currents, voltages, and impedances, and the magnitude and phase
of forward and reverse voltage and current waves?

Roy Lewallen, W7EL

No Roy, of course not. I am not attempting to assert that reflection coefficients should be used in such an
analysis. I'm only asserting that it's another way of performing an analysis, one that I believe paints a more
visible picture of the how the pertinent waves behave in the circuit.

We're certainly not communicating well! I have never questioned that the
use of "virtual shorts" is another way of performing an analysis, nor
that it helps visualize some of the things going on.

If I still haven't persuaded you that it's a viable way of analyzing the impedance matching function then I'll
back off and not pursue the issue any further.

Nor have I questioned that it's a viable way of analyzing the impedance
matching function.

If you'll read what I've written, you'll hopefully see that my only
point of contention is with your claim that waves reflect from a
"virtual short". They do not. And the lack of a single example of a
system whose analysis requires this to happen is evidence that they do not.

If you back off and not pursue the issue any further, you'll continue
with your belief that "virtual shorts" cause reflections. And I'm afraid
that will detract from the wealth of accurate and useful things you do
say. So please continue. But don't waste time arguing that the concept
of "virtual shorts" is a useful analytical tool. I've always agreed with
that, and haven't seen any postings indicating anyone else doesn't.

Incidentally, you didn't answer my questions.

I wanted to get an answer to mine, first. Now that I have, I'll answer
yours.

Roy Lewallen, W7EL

Walter Maxwell wrote:

Consider my two explanations, or definitions of what I consider a virtual short--perhaps it should have a
different name, because of course 'virtual' implies non-existence. The short circuit evident at the input of
the two line examples I presented---do you agree that short circuits appear at the input of the two lines? If
so, what would you call them?

I'd call them "virtual shorts". If they were short circuits, we should
be able to connect a wire across the transmission line at that point
with no change in transmission line operation. But we can't. While
things will look the same on the generator side, they won't be the same
beyond the real short. So they aren't short circuits.

Roy, I'd like for you to take another, but perhaps closer look at the summarizing of the reflection
coefficients below. I originally typed in the wrong value for the magnitude of the resultant coefficients.
With the corrected magnitudes in place, the two paragraphs following the summarization now make more sense,
because the short circuit established at the stub point leads correctly to the wave action that occurs there.

Summarizing reflection coefficient values at stub point with stub in place:
Line coefficients: voltage 0.5 at +120°, current -60° (y = 1 + j1.1547)
Stub coefficients: voltage 0.5 at -120°, current +60° (y = 1 - j1.1547)
Resultant coefficients: voltage 0.5 at 180°, current 0.5 at 0° WRONG
Resultant coefficients: voltage 1.0 at 180°, current 1.0 at 0° CORRECT

Repeating from my original post for emphasis:
These two resultant reflection coefficients resulting from the interference between the load-reflected wave at
the stub point and the reflected wave produced by the stub define a virtual short circuit established at the
stub point.

There's no need to repeat this. I'm well acquainted with transmission
line phenomena, and understand fully what's happening. I have no
disagreement with this analysis. I would draw attention to the fact that
the "virtual short" is, as you say, simply the superposition
(interference) of traveling waves. So there is nothing at that point
except the traveling waves which pass through that point.

The following paragraph shows how the phases of the reflected waves become in phase with the source waves so
that the reflected waves add directly to the source waves, establishing the forward power, which we know
exceed the source power when the reflected power is re-reflected. The same concept applies to antena tuners.

Sorry, I'm not going to divert onto the topic of propagating power,
either instantaneous or average. If that concept is required in order to
show that waves interact with each other, then it simply shows that the
concept is invalid. Let's stick to voltages and currents. If that's not
adequate, then I'll exit at this point, and turn the discussion over to
Cecil. That's his domain, not mine.

Again repeating for emphasis:
Let's now consider what occurs when a wave encounters a short circuit.

Ok.

We know that the voltage wave
encounters a phase change of 180°, while the current wave encounters zero change in phase. Note that the
resultant voltage is at 180°, so the voltage phase changes to 0° on reflection at the short circuit, and is
now in phase with the source voltage wave. In addition, the resultant current is already at 0°, and because
the current phase does not change on reflection at the short circuit, it remains at 0° and in phase with
source current wave. Consequently, the reflected waves add in phase with the source waves,

Ok so far. . .

thus increasing the
forward power in the line section between the stub and the load.

Again, let's leave power out of it, ok?

Keep in mind that the short at the stub point is a one-way short, diode like, as you say, because in the
forward direction the voltage reflection coefficient rho is 0.0 at 0°, while in the reverse direction, rho at
the stub point is 1.0 at 180°, which is why it's a one-way short.

The voltages, currents, waves, and impedances impedances on the line are
just the same as if there were a diode-short at that point. Which is why
it's a useful analytical tool. But all there really is at that point are
some interfering waves, traveling through that point unhindered.

You say that no total re-reflection occurs at the stub point. However, with a perfect match the power rearward
of the stub is zero, and all the source power goes to the load in the forward direction. Is that not total
reflection?

Not from the "virtual short" -- it only looks like it. The re-reflection
is actually occurring from the end of the stub and from the load, not
from the "virtual short". If, for example, you suddenly increased the
source voltage, there would be no reflection as that change propagated
through that "virtual short". (That is, after a delay equal to the
round-trip time to the "virtual short", you'd see no change.) The
apparent reflection from that point wouldn't appear until the change
propagated to the end of the stub and to the load (going right through
the "virtual short" unhindered), reflected from them, and arrived back
at the "virtual short" point. This is one of the ways you can tell that
a "virtual short" isn't a real short. Under steady state conditions, it
looks just like a real one. But it isn't. Waves which seem to be
reflecting from it are really reflecting from the end of the stub and
from the load -- they're passing right through the "virtual short", in
both directions.

Using the numbers of my bench experiment, assuming a source power of 1 watt, and with the
magnitude rho of 0.04, power going rearward of the stub is 0.0016 w, while the power absorbed by the load is
0.9984 w, the sum of which is 1 w. The SWR seen by the source is 1.083:1, and the return loss in this
experiment is 27.96 dB, while the power lost to the load is 0.0070 dB. From a ham's practical viewpoint the
reflected power is totally re-reflected.

Sorry, you're going to have to do this without propagating waves of
average power, or I'm outta here.

In my example using the 49° stub the capacitive reactance it established at its input is Xc = -57.52 ohms.
Thus its inductive susceptance B = 0.0174 mhos, which cancels the capacitive line susceptance B = -0.0174 mhos
appearing at the stub point.

My point is that the 49° stub can be replaced with a lumped capacitance Xc = -57.52 ohms directly on the line
with the same results as with the stub--with the same reflection coefficients.

That's fine, I agree.

In this case one cannot say
that the re-reflection results from the physical open circuit terminating the stub line.

I most certainly can! And do. I don't see how your example furnishes any
proof or even evidence of wave interaction. I can come to the same
conclusion without any assumption of wave interaction, and you have
agreed (in your response to my question about finding an example that
requires interaction for analysis) that this can always be done.

Various posters have termed my approach as a 'short cut'. I disagree. I prefer to consider it as the wave
analysis to the stub-matching procedure, in contrast to the traditional method of simply saying that the stub
reactance cancels the line reactance at the point on the line where the line resistance R = Zo. In my mind the
wave analysis presents a more detailed view of what's actually happening to the pertinent waves while the
impedance match is being established.

I'm sorry, I disagree. It's a less detailed view, and it conceals what's
really going on.

Roy Lewallen, W7EL

K7ITM wrote:
. . .
It's a useful visualization tool and design aid; it's a poor analysis
tool at best. At worst, it will lull you into building something that
just won't work, wasting time and resources.

In my opinion, the potential harm can be much worse. If it causes you to
buy into the notion that traveling waves interact in a linear medium,
that opens the door to a whole universe of invalid conclusions. We've
seen some of those promoted very vigorously in this newsgroup.

Roy Lewallen, W7EL

On Sat, 14 Apr 2007 16:36:11 -0500, (Richard
Harrison) wrote:

Richard Clark wrote:
"This conforms to my experience with many plumbing designs on the
microwave bench."

Good. Then, Richard Clark must also be familiar with the grooved
circular flange used in conjunction with a smooth flange to join
waveguide segments. This groove isn`t just used to hold a neoprene
gasket. It is also used as an electrical choke to keep the microwaves
within the pipe. It is approximately a 1/4-wave choke and its high
impedance across its open-circuit helps foil the wave escape. If virtual
open-circuits didn`t work, the "choke-flange wouldn`t work either.

Hi Richard,

You got the dimensions wrong for the wrong reason. It is a halfwave
shorted cavity (or transmission line sub-section). The cut channel is
quarterwave, but the distance from the middle of the longest side of
the waveguide is another quarterwave. The virtual short bridges the
joined, but non-contacting faces of the two waveguide sections. There
is no high Z action involved.

Further, there are two grooves, the outer one is for the neoprene
gasket.

For others, when two sections of waveguide are joined, there is always
the possibility that the two faces of the ends will not see a machined
match (like an engine head to the engine block - which even there has
problems of warp); hence the possibility of a break in continuity with
an open circuit. Rather than burn money as a solution, engineers
simply forced the problem by having the two faces separated and never
in contact!

What they did to offset this deliberate open is they machined a
circular groove around the flange of one section that was a
quarterwave deep, and a quarterwave away from the voltage node for the
waveguide's TE10 mode of transmission. The geometry of the groove and
its distance from the node was very simple to control in comparison to
guaranteeing fully flush mating faces (especially under the torsion of
ship's movement, or simple heat expansion). This is a hallmark of
engineering where the problem becomes the solution.

As always, a short is vastly more preferred than an open when casting
back into the transmission line and the design engineers went to the
additional length to see it incorporated into the solution.

Dare I call the choke joint a virtual gasket?

73's
Richard Clark, KB7QHC

Roy Lewallen wrote:
I don't see how your example furnishes any
proof or even evidence of wave interaction.

Are you saying that wave interaction doesn't exist?
--
73, Cecil http://www.w5dxp.com

Roy Lewallen wrote:
K7ITM wrote:
. . .
It's a useful visualization tool and design aid; it's a poor analysis
tool at best. At worst, it will lull you into building something that
just won't work, wasting time and resources.

In my opinion, the potential harm can be much worse. If it causes you to
buy into the notion that traveling waves interact in a linear medium,
that opens the door to a whole universe of invalid conclusions.

Here is how Hecht described interference in "Optics":
"... interference corresponds to the *interaction* of two or
more lightwaves yielding a resultant irradiance that deviates
from the sum of the component irradiances."

If traveling waves cannot interact in a linear medium, why
does Hecht say they do indeed interact?

To deny the body of laws of physics regarding EM waves from
the field of optics is an example of extreme ignorance.
--
73, Cecil http://www.w5dxp.com

Roy Lewallen wrote:
If it causes you to
buy into the notion that traveling waves interact in a linear medium,
that opens the door to a whole universe of invalid conclusions.

Roy, seems you are the one with the invalid conclusions. Here
is a java-script of "traveling wave interaction in a linear medium".

http://micro.magnet.fsu.edu/primer/j...ons/index.html

"... when two waves of equal amplitude and wavelength that are
180-degrees ... out of phase with each other meet, they are not
actually annihilated, ... All of the photon energy present in
these waves must somehow be recovered or redistributed in a
new direction, according to the law of energy conservation ...
Instead, upon meeting, the photons are redistributed to regions
that permit constructive interference, so the effect should be
considered as a redistribution of light waves and photon energy
rather than the spontaneous construction or destruction of light."

Does energy being redistributed in new directions really look
like a lack of interaction to you?
--
73, Cecil http://www.w5dxp.com

On Apr 14, 6:06 pm, Roy Lewallen wrote:
K7ITM wrote:
. . .
It's a useful visualization tool and design aid; it's a poor analysis
tool at best. At worst, it will lull you into building something that
just won't work, wasting time and resources.

In my opinion, the potential harm can be much worse. If it causes you to
buy into the notion that traveling waves interact in a linear medium,
that opens the door to a whole universe of invalid conclusions. We've
seen some of those promoted very vigorously in this newsgroup.

Roy Lewallen, W7EL

Yes, you're right, Roy. I guess I didn't consider that because I'm
not very likely to buy into it, but from the point of view of someone
just learning about linear systems, it's a danger.

The analogy may not be prefect, but I think it's a lot like the
usefulness of the idea of a "virtual ground" at the inverting input of
an op amp. But it's a virtual ground only under specific conditions:
strong negative feedback is active, and the non-inverting input is at
(AC, at least) ground potential. For it to be a useful concept
without too many pitfalls, the person using it has to be aware that
the conditions that make it a good approximation don't always hold.
Similarly for a "virtual short" on a line.

Again, though, it IS useful to me to think along these lines, when
looking to do something useful with stubs: I want to kill frequency
W, so I can put a stub across my line that's half a wave long at W,
shorted at the far end. At the same time I want to pass V, and the
stub I just put there to kill W has reactance X at frequency V. If I
put another stub with reactance -X at freq V across the line there, it
will let V through with minimum effect. Now go calculate how well it
will perform with particular lines.

So, to come up with a design to try, I do think about how stubs
behave, in a general sense, including things like "a half-wave line
shorted at the far end echos a short", but with the programs I have
readily available, it's silly to rely on approximations that drop the
line attenuation, when I want to know how my idea will actually work
when I build it.

Cheers,
Tom