Re: Standing-Wave Current vs Traveling-Wave Current

I'm top posting this so readers won't have to scroll down to see it, but so I can include the original posting completely as a reference.


Keith, you've presented a very good and well thought out argument. But I'm not willing to embrace it without a lot of further critical thought. Some of the things I find disturbing are:

1. There are no mathematics to quantitatively describe the phenomenon.
2. I don't understand the mechanism which causes waves to bounce.
3. No test has been proposed which gives measurable results that will be different if this phenomenon exists than if it doesn't. (I acknowledge your proposed test but don't believe it fits in this category.)
4. I'm skeptical that this mechanism wouldn't cause visible distortion when dissimilar waves collide. But without any describing mathematics or physical basis for the phenomenon, there's no way to predict what should or shouldn't occur.
5. Although the argument about no energy crossing the zero-current node is compelling, I don't feel that an adequate argument has been given to justify the wave "bouncing" theory over all other possible explanations.

None of these make an argument with your logical development, although I think I might be able to do that too. But I'm very reluctant to accept a view of wave interaction that's apparently contrary to established and completely successful theory and one, if true, might have profound effects on our understanding of how things work. So frankly I'm looking hard for a flaw in your argument. And I may have found one.

A large part of the argument seems to revolve around a single point in a perfect transmission line, where the current is exactly zero. This is an infinitesimal point on a perfect line, so some anomalous things might be expected to happen there.

Let's consider a transmission line as a huge number of series inductors and shunt capacitors, each an ideal lumped device. In the ideal case, of course, there would be an infinite number of each, and each would have an infinitesimal value. However, the LC product and ratio must remain correct even in the limiting case. Each L and C is an ideal device, so the current into one terminal of an inductor has to equal the current out of the other. A consequence of this is that either we have a whole inductor with zero current, or the zero current point occurs between inductors, at a node to which a capacitor is connected. I think we'll get the same result using either scenario, but let's consider the second.

If we analyze this situation carefully, we'll find that the inductor on each side of the zero-current point does have a finite current, equal in amplitude and flowing in opposite directions. So for half of the cycle, both are putting positive charge in the capacitor, and for the other half of the cycle, both are removing charge. The capacitor voltage goes up and down as a result, as we can also see by looking at the voltage at this zero-current point. So current from both sides is contributing to the capacitor charge, and turning off either one would change the line conditions. Any change in the current from the inductor on one side would change the capacitor voltage, and hence the current on the other side. So there is an interchange of information from one side to the other. Each inductor is conveying energy to the capacitor, which is storing and returning it.

Ok, so let's break the capacitor into two, each being half the original value, and constrain each inductor to deliver charge only to "its" capacitor. The wire between the capacitors carries no current because the capacitors always have equal voltages, and can be cut with no effect.

When there was one capacitor, it shared energy from both sides. When we broke it into two, there was no mixing of energy from either side. Why might one be a better description of reality than the other? It looks to me like the argument devolves into speculation about how small the "point" is at which the current drops to zero.

It would be instructive to see what happens as, for example, the load resistance is increased toward infinity or decreased toward zero arbitrarily closely, but not at the point at which it's actually there. If the "bouncing" phenomenon is necessary only to explain the limiting case of infinite SWR on a perfect line but no others, then an argument can be made that it's not necessary at all. I suspect this is the case.

I agree with your argument about two sources energized in turn, and have used that argument a number of times myself to refute the notion of superposing powers. Once two voltage or current waves occupy the same space, the only reality is the sum. We're free to split them up into traveling waves or any other combination we might dream up, with the sole requirement being that the sum of all our creations equals the correct total. (And the behavior of waves you're describing seemingly go beyond this.) The advantage to the non-interacting traveling wave model is that it so neatly predicts transient phenomena such as TDR and run-up to steady state. I spent a number of years designing TDR circuitry, interfacing with customers, and on several occasions developing and teaching classes on TDR techniques, without ever encountering any phenomena requiring explanations beyond classical traveling wave theory. So you can understand my reluctance to embrace it based on a problem with energy transfer across a single infinitesimal point in an ideal line.

Roy Lewallen, W7EL

Keith Dysart wrote:
On Dec 30 2007, 6:18 pm, Roy Lewallen <w...@xxxxxxxxx> wrote:
Keith Dysart wrote:

I predict that the pulse arriving at the left end will
have the same voltage, current and energy profile as
the pulse launched at the right end and the pulse
arriving at the right end will be similar to the
one launched at the left.
They will appear exactly AS IF they had passed
through each other.
The difficulty with saying THE pulses passed
through each other arises with the energy. The
energy profile of the pulse arriving at the left
will look exactly like that of the one launched
from the right so it will seem that the energy
travelled all the way down the line for delivery
at the far end. And yet, from the experiment above,
when the pulses arriving from each end have the
same shape, no energy crosses the middle of the
line.
So it would seem that the energy that actually
crosses the middle during the collision is
exacly the amount of energy that is needed to
reconstruct the pulses on each side after the
collision.
If all the energy that is launched at one end
does not travel to the other end, then I am
not comfortable saying that THE pulse travelled
from one end to the other.
But I have no problem saying that the system
behaves AS IF the pulses travelled from one
end to the other.
You said that:

> What will happen? Recall one of the basics about
> charge: like charge repel. So it is no surprise
> that these two pulses of charge bounce off each
> and head back from where they came.

Yet it sounds like you are saying that despite this charge repulsion and
bouncing of waves off each other, each wave appears to be completely
unaltered by the other? It seems to me that surely we would detect some
trace of this profound effect.

. . .
Is there any test you can conceive of which would produce different
measurable results if the pulses were repelling and bouncing off each
other or just passing by without noticing the other?

There are equations describing system behavior on the assumption of no
wave interaction, and these equations accurately predict all measurable
aspects of line behavior without exception. Have you developed equations
based on this charge interaction which predict line behavior with
equal accuracy and universal applicability?

No equations. I expect that such equations would be more complex
than those describing the behaviour using superposition. Since
the existing equations and techniques for analysis are tractable
and produce accurate results, I am not motivated to develop an
alternate set with lower utility.

And yet the "no interaction" model, while accurately predicting
the behaviour has some weaknesses with explaining what is
happening. It is, I suggest, these weaknesses that help lead
some so far astray.

To illustrate some of these weaknesses, consider an example
where a step function from a Z0 matched generator is applied
to a transmission line open at the far end. Charge begins to
flow into the line. The ratio of the current to voltage on
the line is defined by the distributed inductance and
capacitance. The inductance is resisting the change in current
which causes a voltage to charge the capacitance. A voltage
step (call this V for later use) propagates down the line
at the speed of light. In front of this step, the voltage,
current and charge in the line is zero. After the step, the
capacitance is charged to the voltage and charge is flowing
in the inductance.

The step function eventually reaches the open end where
the current can no longer flow. The inductance insists
that the current continue until the capacitance at the
end of the line is charged to the voltage which will stop
the flow. This voltage is double the voltage of the step
function applied to the line (i.e 2*V). Once the
infinitesimal capacitance at the end of the line is
charged, the current now has to stop just a bit earlier
and this charges the inifinitesimal capacitance a bit
further from the end. So a step in the voltage propagates
back along the line towards the source. In front of this
step, current is still flowing. Behind the step, the
current is zero and voltage is 2*V. The charge that
is continuing to flow from the source is being used
to charge the distributed capacitance of the line.

The voltage that is propagating backwards along the
line has the value 2*V, but this can also be viewed as
a step of voltage V added to the already present voltage
V. The latter view is the one that aligns with the "no
interaction" model; the total voltage on the line is
the sum of the forward voltage V and the reverse
voltage V or 2*V.

In this model, the step function has propagated to the
end, been reflected and is now propagating backwards.
Implicit in this description is that the step continues
to flow to the end of the line and be reflected as
the leading edge travels back to the source.

And this is the major weakness in the model. It claims
the step function is still flowing in the portion of
the line that has a voltage of 2*V and *zero* current.

Now without a doubt, when the voltages and currents
of the forward and reverse step function are summed,
the resulting totals are correct. But it seems to
me that this is just applying the techniques of
superposition. And when we do superposition on a
basic circuit, we get the correct totals for the
voltages and currents of the elements but we do
not assign any particular meaning to the partial
results.

A trivial example is connecting to 10 volt batteries
in parallel through a .001 ohm resistor. The partial
results show 10000 amps flowing in each direction
in the resistor with a total of 0. But I do not
think that anyone assigns significance to the 10000
amp intermediate result. Everyone does agree that
the actual current in the resistor is zero.

The "no interaction" model, while just being
superposition, seems to lend itself to having
great significance applied to the intermediate
results.

Partially this may be due to poor definitions. If the
wave is defined as just being a voltage wave, then
all is well.

But, for example, when looking at a solitary pulse,
it is easy (and accurate) to view the wave as having
more than just voltage. One can compute the charge,
the current, the power, and the energy. But when
two waves are simultaneously present, it is only
legal to superpose the voltage and the current.
But it is obvious that a solitary wave has voltage,
current, power, etc. But when two waves are present
it is not legal to.... etc., etc.
The "no interaction" model does not seem to resolve
this conflict well, and some are lead astray.

And it was this conflict that lead me to look for
other ways of thinking about the system.

Earlier you asked for an experiment. How about this
one....

Take two step function generators, one at each end
of a transmission line. Start a step from each end
at the same time. When the steps collide in the
middle, the steps can be viewed as passing each
other without interaction, or reversing and
propagating back to their respective sources. We
can measure the current at the middle of the line
and observe that it is always 0. Therefore the
charge that is filling the capacitance and causing
the voltage step which is propagating back towards
each generator must be coming from the generator
to which the step is propagatig because no charge
is crossing the middle of the line.

Do you like it?

...Keith
.

Relevant Pages

  • Re: An update on my TIG inverter project
    ... >> One option that I have used in the past is to make a triangle - wave ... and fixed mid-rail DC voltage on the inverting input) to set the ... The capacitor will have to deal with whatever ripple ...
    (sci.electronics.design)
  • Re: Standing-Wave Current vs Traveling-Wave Current
    ... If both the capacitor and the inductance are ... at a node to which a capacitor is connected. ... half of the cycle, both are removing charge. ... The capacitor voltage goes ...
    (rec.radio.amateur.antenna)
  • Re: Capacitor charging configurations
    ... When connected together C1 will charge C2, and the steady-state voltage of each will be 0.5 volts. ... and in lamens terms we can describe it by saying the charge in C1, which produces 1 volt there, will distribute itself evenly among the total capacitance and result in half of the charge left in C1 and half in C2, thus producing 0.5 volts in each capacitor. ...
    (sci.electronics.design)
  • Re: Capacitor charging configurations
    ... When connected together C1 will charge C2, and the steady-state voltage of each will be 0.5 volts. ... and in lamens terms we can describe it by saying the charge in C1, which produces 1 volt there, will distribute itself evenly among the total capacitance and result in half of the charge left in C1 and half in C2, thus producing 0.5 volts in each capacitor. ...
    (sci.electronics.design)
  • Re: solving a deceptively simple-looking switched RC circuit (singular circuit ?)
    ... time t=T, two switches are thrown: one disconnecting the DC source, and the other, connecting the "charged" portion of the circuit to a big capacitor C3. ... You mention that C3 may have an arbitrary initial voltage at the moment it is switched into the loop made up of the other components. ... C3 capacitor "instantaneously" (need to model this current with a delta ... You need to solve for the amount of charge ...
    (sci.electronics.design)