# The Dark Side of the Conjugate Match

Discussion in 'General Technical Questions and Answers' started by KL7AJ, Mar 12, 2010.

Not open for further replies.
1. ### W5DXPHam MemberQRZ Page

Hi Dave, there is no doubt that, in your suggested experiment, the reflections cause a change in the generator load-line. The real question is what causes the change in the generator load-line when there is no transmission line? My question is: What besides reflections, among the known laws of physics, can cause a change in the load-line? Nobody has presented a mechanism for that change that doesn't involve information feedback from the changed impedance. It is a very simple question having to do with information transfer theory which cannot be transferred at faster than the speed of light.

Source------any length except zero------impedance change

Given: The source load-line doesn't change at the instant of the impedance change. There is a speed-of-light delay between the impedance change and the change in the load-line. Exactly what is the nature of the speed-of-light feedback from the impedance change to the source load-line? What mechanism of physics can explain such behavior except for reflections? I have asked that question a dozen times and the silence of the replies is deafening.

Seems to me, reflections win the argument by default, since there is no other phenomenon known to physics that can explain the speed-of-light delay involved in the process.

If the lumped-circuit model was correct, the load-line would change the instant that the impedance is changed, but we know it doesn't. Therefore, the lumped-circuit model is incorrect and must be discarded in attempting to answer the question of what mechanism of physics causes the speed-of-light feedback delay from the changed impedance. The lumped-circuit model does not support speed-of-light delays.

In the above simple example, the only energy that exists in the system is source energy, forward-traveling energy, and reverse-traveling energy. Of those three types of energy, reverse-traveling energy is the only one that could possibly transfer information from the impedance change back to the source.

If this were EM light energy instead of EM RF energy, the answer would be "obvious".

Light source------any length except zero------mirror (index of refraction) change

There is no conceptual difference in the feedback mechanism between this light example and the RF example above.

2. ### WB2UAQHam MemberQRZ Page

Cecil,

I have followed this discussion but maybe not down to the last detail.

I see the events taking place on a source / transmission line / load configuration as a "transient response" situation.

A few years ago I suggested this here on QRZ but I was shot down.

The signals upon reaching a discontinuity act like an L-C circuit. It is easier to visualize when thinking about a TDR setup. The pulse reaches a discontinuity and there is a "transient response". The distributed inductance in the transmission line results in a minus L (di/dt) situation (think of a voltage developed as a result of a sudden change in the current thru and inductor no different than a Kettering Ignition ).

I don't have the time (here at lunch as usual when I get a moment to look at QRZ) to expand on this so I hope I came across with a suggestion to consider a transient response approach as opposed to a steadstate approach. To me the transmission line sure looks like a sliding 50 Ohm generator sending a signal back into the source as is crashes into it

73, Pete

3. ### WR9HHam MemberQRZ Page

Information theory is not needed here. Imagine a hose emptying into a very large container. The source can put its entire output capability into the load without knowing or caring about what is happening at the load!! Essentially, the load in this example is an open circuit that is very similar to open space. I think we are making the whole concept of source-load interaction WAY more complicated than it really is because of feedline physics.

73
Herb/WR9H

4. ### W8JIHam MemberQRZ Page

I agree.

That aside, delay between time current rises on the input of an inductor and output of an inductor was measured accurately a long time ago. Since this argument has been going on for several years, and since all people except one seem to agree, the debate isn't worth wasting any time on.

73 Tom

5. ### W5DXPHam MemberQRZ Page

This is a technical challenge for you to step up the the plate and defend those delay measurement techniques of yours by applying them to a straight wire. If you cannot defend your techniques using a straight wire, I and others, will conclude that since those techniques are not valid for straight wires, neither are they valid for loading coils. If you can prove that your techniques are valid for a straight wire, I will not bother you again concerning this subject.

Tom, the Hamwaves inductance calculator was designed by people who are a lot smarter than either you or I. Witness their PhD degrees. What degree do you hold? How many extremely competent technical people were involved in the design of that inductance calculator? Have you even read that web page? It is you who seem to be outnumbered by technical experts. I'm just the messenger.

Here is an example of your faulty measurement techniques and misconceptions that doesn't even contain a loading coil. It is just a simple technical question about a plain 1/4WL wire vertical over mininec ground modeled by EZNEC.

Using exactly the same techniques that you used to calculate the delay through your loading coil, we make current measurements at points X and Y and they agree with EZNEC. There is a one degree phase shift between points X and Y. At 7.15 MHz, one degree is a delay of 0.389 ns and that is your calculated delay. However, that value requires a current flow speed that is 30 times the speed of light. Does EZNEC support faster than light speeds or is something wrong with your delay concepts?

It is a given that there are 30 degrees of monopole between points X and Y so the actual current delay has to be at least 11.7 ns, i.e. no faster than the speed of light.

Please explain the discrepancy between what you would measure using your previous coil measurement techniques and what we know to be the technical facts. What should I, and others, conclude if you don't respond with a defense of your delay measurement techniques?

Given the above technical information from EZNEC, which indicates almost zero delay in 30 degrees of monopole, why are you surprised when your loading coil exhibits the same near-zero delay as 30 degrees of straight wire? Why wouldn't a loading-coil with the same delay as 30 degrees of straight wire have an actual length in the ballpark of 30 degrees? You seem to assume that delay is related to phase in a standing-wave antenna and that is a misconception/myth.

6. ### W5DXPHam MemberQRZ Page

There's nothing wrong with your transient response thinking. It requires a reversal of energy flow at the load, i.e. reflections are part of the transient response. When (and if) those reflections reach the source load line, the load-line will change.

Herb, if you want to use Archimedes Principle on RF energy, rather than Maxwell's equations, be my guest. Just one question, how are you going to get those water molecules to travel at the speed of light?

7. ### W5DXPHam MemberQRZ Page

Tom, what you measured was not the delay. You measured the phase shift and then calculated the delay from the phase shift based on your false assumptions about phase shift being related to delay. What you have to technically prove is that the delay is related to the phase shift in a standing-wave current environment. You have never presented that proof and I challenge you to present that proof now. I have presented proof to the contrary a number of times.

Here is what I think is a diagram of your 3 ns coil delay setup with the coil replaced by 25 feet of wire. If this is not correct, please tell me what is correct and I will modify the diagram.

Again, I challenge you to perform that experiment with 25 feet of wire (instead of your 100 turn coil) and report the results back to us. You will find that your calculated delay in 25 feet of wire is in the ballpark of the same 3 ns delay reported through your 100 turn coil. Hint: A 3 ns delay through 25 feet of wire is impossible because it violates the speed-of-light limit.

8. ### W5DXPHam MemberQRZ Page

Please note the complete absence of any technical argument from W8JI. Here is why he is unwilling and unable to technically justify and defend his 3 ns "measured" delay through a 10" long 100 turn loading coil.

The traveling wave current changes by one degree in phase for each one degree in length. The phase change in traveling wave current can accurately be used to determine the delay through a loading coil or through a wire. This is what W8JI mistakenly thought he was doing.

The standing wave current phase doesn't change at all. Therefore, standing wave current phase cannot be used to determine delay through anything including a loading coil and a wire. The delay through a coil (or through a wire) in a standing-wave antenna is mostly unrelated to the phase shift in the current. EZNEC verifies that fact of physics (see the above graph). The two EZNEC files are available on my web page, one unloaded in order to generate standing-wave current and the other loaded in order to generate traveling-wave current. Click on "Currents" to see the current phase in each wire segment.

http://www.w5dxp.com/standing.EZ
http://www.w5dxp.com/travling.EZ

A mobile antenna is called a standing-wave antenna because ~90% of the current on the antenna is standing-wave current. Under those conditions, it is impossible to use the change in current phase to predict, calculate, or measure the delay through a loading coil or through a wire. When (and if) W8JI runs the last suggested experiment above, he will discover that fact of physics.

9. ### K5MCHam MemberQRZ Page

Cecil,

I recently made some measurements of the input and output currents of a "large" air-core inductor at the local university where I teach electrical engineering. My instrumentation included two Tektronix TCP303 current probes (good up to 15 MHz), two matching Tektronix TCPA300 current probe amplifiers, and a two-channel 100-MHz oscilloscope. The approximate dimensions of the coil I tested are:

mean coil diameter = 81 mm
axial length = 264 mm
wire diameter = 6.2 mm (coil is made of copper tubing)
23.5 turns
test frequency = 7.0 MHz

Plugging these numbers into the inductor calculator at http://hamwaves.com/antennas/inductance.html gives the following:

Zc = 1658 ohms
phase shift constant = 1.31639 radians/meter

Therefore, with a terminating resistor of about 1700 ohms I would expect to see a phase shift of about 20 degrees (a time delay of about 8 ns) at 7 MHz between the input and output currents. For various resistor values between 300 ohms and 19,000 ohms I measured the following "time delays" between the input and output currents (the third number in each row is the corresponding ratio of the amplitudes of the input to output coil currents).

300 Ω 5 ns 0.8
480 Ω 9 ns 0.9
670 Ω 12 ns 1.1
1.2 kΩ 14 ns 1.3
1.9 kΩ 14 ns 1.9
2.4 kΩ 14 ns 2.0
3.3 kΩ 12 ns 2.4
5.6 kΩ 6 ns 3.0
11 kΩ 3 to 4 ns 3.4
19 kΩ 2 to 3 ns 4.2

I was using the pi network (in the reverse direction) in my old Heathkit DX-60B transmitter as a matching network between the signal generator (a 40-meter Ten-Tec QRP rig) and the coil/resistor network. Although the trends are definitely in the correct direction, I am somewhat disappointed that the measured values are not in better agreement with what the online inductor calculator predicts.

Regardless of the accuracy of my measurements above, I completely agree with your comments regarding the phase information of standing wave current versus traveling wave current. No EE familiar with these concepts will disagree with your statement that "standing wave current phase cannot be used to determine delay through anything including a loading coil and a wire."

73, K5MC

10. ### W5DXPHam MemberQRZ Page

Thanks Mickey, for your time and effort. I assume your "delay" was calculated using the measured current phase shift through the coil. At 7 MHz, since one RF cycle of 360 degrees is ~143 ns, I am going to reverse engineer your "delay" figures back into current phase shift degrees. I hope they are correct.

300 Ω 5 ns 12.6 deg
480 Ω 9 ns 22.7 deg
670 Ω 12 ns 30.2 deg
1.2 kΩ 14 ns 35.2 deg
1.9 kΩ 14 ns 35.2 deg
2.4 kΩ 14 ns 35.2 deg
3.3 kΩ 12 ns 30.2 deg
5.6 kΩ 6 ns 15.1 deg
11 kΩ 3 to 4 ns 10.1 deg
19 kΩ 2 to 3 ns 7.6 deg

Here's the question: If the coil and frequency remains the same, why does the "delay" through the coil change with a changing load resistor?

Answer: The actual delay through the coil does not change. What changes is the percentage of standing wave current vs traveling wave current. As the load resistor value approaches the characteristic impedance of the coil, the standing wave current magnitude is decreased as a percentage of total current while the traveling wave current magnitude is increased as a percentage of total current.

The total current is always the phasor sum of the standing wave current and the traveling wave current and the magnitude of each can vary all the way down to zero.

Let me offer an explanation for the difference in the calculated value and the measured value. You didn't eliminate the reflections, you just reduced them. If you had eliminated the reflections entirely, your delay measurements would have approached the calculated value of 20 ns.

Your phase measurements, just like W8JI's, were accurate. The inaccuracy is the misconception that total current phase can be used to calculate delay when reflections are present. In addition to the above quoted assertion: If any percentage of standing wave current is present, a calculated delay based on total (net) current phase shift is inaccurate. That is what your "measured" delays above indicate.

To summarize: Your pi-net matching network did not eliminate reflections in the circuit. It formed a conjugate match which includes a reflected wave. That's probably why you couldn't achieve the 20 ns = 50.3 degree phase shift.

It appears that W8JI should expand his experiment to include a varying load resistor and report those eye-opening results on his web page.