Summary
This blogpost uses the Convolution Integral to examine the transient and steady-state values of forward and reflected waves in a simple "thought-experiment" model of an antenna system consisting a mismatched load, a transmission line, an impedance matching device, a second transmission line, and a source.
Transient and Steady-state responses are derived by convolving (using the Convolution Integral) a Stimulus with the antenna system's calculated Impulse Responses at various points of the system.
This convolution integral technique does not replace the traditional "frequency domain" method of calculating reflections. Calculations using the latter technique are much easier to perform, but they do not allow for transient analysis, and that is the virtue of the convolution integral -- it lets us examine the system in its transient state, and from that analysis gain a better understanding of the reflection and transmission mechanisms underlying the measured reflected and transmitted steady-state wave-forms.
If we define a "matched system" as being a system in which a matching device matches an unmatched load to a transmission line connected to the source, then the analysis of the model shows that, as the "matched system" transits through its transient state to steady-state, the Reflection Coefficient looking into the input of the matching device (looking from the direction of the source) goes from a non-zero value to 0.
That is, in the transient state there are reflections back to the source, but these reflections go to 0 in the steady-state.
However (again in the matched system), the model's steady-state Reflection Coefficient looking backwards from the load into the matching device's output port remains unchanged (from the value calculated for the impulse response), irrespective of source impedance, even in steady-state.
For a similar analysis of a quarter-wave impedance transformer, go here: Quarter-wave Transformer: Transient and Steady-state.
Antenna Tuners: Transient and Steady-state Reflections
I've long agreed with the "Theory of Multiple Reflections" as an explanation of the wave-phenomena that occurs on a transmission line between a mismatched load at one end and an impedance matching device at the other end (or within an impedance matching device, should that device be a distributed-element device such as a quarter-wave matching transformer, rather than a lumped-element device, i.e. Multiple-reflections and the Quarter-wave Transformer).
In the example shown below, there are steady-state Forward and Reflected waves on Transmission Line 2 (between the mismatched antenna and the matching network). But, because of the impedance transformation provided by the matching network, no reflections return to the transmitter.
The "Theory of Multiple Reflections" explains how these reflected and transmitted waves build from zero-state (when the transmitted signal goes from being absent to being present), and it reveals that, at startup, a portion of the initial forward wave from the transmitter is actually reflected back to the transmitter by the Matching Network even though, when steady-state is reached, this reflected signal becomes zero.
In this blog post I will look more closely at how the forward and reflected waves behave both during their transient states (at signal startup or shutdown) and during their steady-states.
Convolution integration can be daunting if a system's impulse response is complex. So, to keep my example's impulse response simple, I have created a model of ideal components consisting of:
- A 200 ohm resistive load.
- A half-wavelength length of "ideal" transmission line (no loss, infinite bandwidth, velocity factor equal to 1) between the load and the output port of a matching network. (Being a half-wavelength long, the steady-state impedance looking into the "input" of this transmission line equals the impedance attached to the far end of this transmission line. In this example, the steady-state impedance looking into this transmission line is the load resistance 200 ohms, irrespective of transmission line Zo (it just needs to be a half-wavelength long), but I will assume a Zo of 50 ohms.
- A matching network consisting of a transformer whose turns ratio (N) is 1:2. Therefore the voltage ratio is 1:2 (in to out) and the impedance ratio is 1:4. A 200 ohm load connected to the output of this transformer looks like 50 ohms at the transformer's input. (The transformer is also assumed to be ideal).
- A length of "ideal"50-ohm transmission line between the source and the transformer's input port.
- And an "ideal" Thevenin-equivalent source whose source resistance is 50 ohms (so that, if any reflections do arrive back at the source, they are absorbed and not re-reflected).
Here's the model:
With this integral, we can calculate both the transient and steady-state responses of a system (whose "impulse response" has first been characterized) to any input signal.
Electrical Engineers will be familiar with the concept of convolution. It is taught in basic circuit-theory courses, and it is based upon the concept that any signal can be represented by an infinite number of consecutive, independent, infinitely narrow impulses. (The amplitude of each impulse, at a particular time T, is equal to the original signal's amplitude at that same time.)
A system's time-domain response to a stimulus signal is then simply the integration of each individual impulse (representing the stimulus) with the system's impulse response.
Anyone who has designed a digital Finite Impulse Response (FIR) filter has, in fact, implemented a convolution integral whether they know it or not. The values of the FIR filter's coefficients are simply the values of the impulse response (in time) of a filter with the desired frequency characteristics. The FIR architecture performs the convolution integral of the input signal with this impulse response, and the output is therefore a time-domain representation of the input signal, now filtered.
For those interested in learning more about the Convolution Integral, there is much more on the web regarding this topic (for example, see: Analog Devices: Convolution).
Characterizing the System's Impulse Response:
I've chosen my model's components to simplify the calculation of its impulse response. Ideal components that are wide bandwidth and lossless makes this process straight-forward. For example, the half-wavelength of transmission line can be thought of as an ideal delay-line whose delay equals half the period of a sine-wave cycle. And the transformer, being ideal, will change the amplitude of an impulse but not its shape.
So, to derive the system's impulse response I simply "drive" a single impulse (of amplitude "1") onto the first transmission line and calculate what happens as it moves through my model.
At each impedance discontinuity the impulse will be divided into two parts -- one part continues on, while the other part is reflected backwards. The amplitudes of these two parts are determined by the Reflection and Transmission coefficients defined for that discontinuity (and they are such that energy is conserved).
The diagram below illustrates this concept of transmission and reflection:
To demonstrate transient and steady-state responses, I will define a stimulus that is limited in time: a 12-cycle sinusoidal signal. The stimulus is 0 before the start of the cycles and 0 following the completion of the 12 cycles.
The delay of the second transmission line (between transformer and load) is defined to be half the period of a single cycle of the sine-wave, above (i.e. it is a half-wavelength of line).
I have used MATLAB to calculate both the impulse-response coefficients and the convolution of this 12-cycle sinusoidal signal with the impulse responses as seen at different points in the model. These will be shown, below
Summary of the Steady-state Response:
The drawing below summarizes the steady-state of the model when stimulated with a 12-cycle sinusoidal drive signal.
(= Vi(0°)*TF)
In steady-state, the part of the driven signal that reflects from the input of the transformer back to the source (and that is occurring now) is 180 degrees out of phase with the reflections from the load (Vr) that are passing through the transformer from its output port to its input (i.e. this "now" reflection has a minus sign, the other reflections passing through the transformer have a positive sign). In the steady-state they cancel (0.6 - 0.6 = 0 volts), and no reflection returns back to the source.
Because of this voltage cancellation, overall there is no reflected wave from the transformer's input in steady-state. And thus the apparent Reflection Coefficient at the transformer's input is 0 in steady-state.
And this is my first point: if we define a "matched system" as being a system in which a matching device matches an unmatched load to a transmission line connected to the source, then this model's analysis shows that as it transitions through its transient state to steady-state, the model's Reflection Coefficient looking into the input of the matching device (looking from the direction of the source) goes from a non-zero value to 0.
That is, in the transient state there are reflections back to the source, but these reflections go to 0 in the steady-state.
However, even though it appears that no signal is reflected from the transformer's input in steady-state, the signal transmitted through the transformer (from input to output) is still the amplitude of the driven signal at the transformer's input (1.0 volt) times the transformer's Forward Transmission Coefficient TF (= 0.8). This value equals Vi(0°)*TF = 1V * 0.8 = 0.8V
Next, let's look at Vr2 and Vf2 on Transmission Line 2, between the transformer's output and the load:
First, we know that (by solving the infinite series), in steady state Vf2 converges to be 1.25 volts.
Note that Vf2, traveling from the transformer's output towards the load, consists of two voltages (both traveling in the same direction and in-phase) summed together. These voltages are the portion of the generator's "now" signal that has just passed through the transformer towards the load ( = Vi(0°)*TF = 1V * 0.8 = 0.8V, that is, the blue line pointing right in the figure above) and the portion of Vr2 that is re-reflected by the mismatch looking into the transformer's output port (the green line pointing right in the figure above).
The re-reflection of Vr2 occurs when Vr2, traveling from the load back to the transformer's output port, sees a Reflection Coefficient of 0.6 (Γ3) at the Transformer's output port. Part of Vr2 re-reflects back towards the load and the remaining part continues through the transformer towards the source (as discussed in the paragraphs above). Let's label the part that is re-reflected back towards the load as "Vr2(re-reflected)".
From the solution of our infinite series we know that Vr2's steady state value is 0.75 volts (which is the same as Vf2 multiplied by the load's Reflection Coefficient (0.75 = 1.25 * 0.6)).
So, Vr2(re-reflected) that is now traveling towards the load equals 0.75 volts times the transformer output's Reflection Coefficient (Γ3) of 0.6 (i.e. 0.45 volts ). This forward-traveling re-reflected voltage is in phase with the "now" forward-traveling 0.8V that has just passed through the transformer. These two voltages sum, and because they are in phase, the total Vf2 on the half-wavelength of Transmission Line 2 is therefore 1.25 volts (i.e. 0.8 + 0.45 volts).
(The portion of Vf2 that is not reflected by the load continues to the load (i.e. Vf2 times the load's Transmission Coefficient, or 1.25 * 1.6) where it is dissipated as power (i.e. radiated).)
And this is my second point: In the matched system represented by this model, the steady-state Reflection Coefficient looking backwards from the load into the matching device's output port remains unchanged from our original calculation of Γ3 that was based upon "local" impedances connected to the ports of the matching device (as seen be a single impulse), irrespective of source impedance, even in steady-state.
This latter conclusion should be intuitively obvious. If, in the steady-state, there are no reflections heading back to the source (because in steady-state it looks likes a matched system), then it does not matter what the source impedance is. With no signal coming towards it, the source is not reflecting anything back to the matching device's input, and so the source impedance has no effect on the Reflection Coefficient seen looking into the matching device's output.
A quick check...
These relationships between Vr2 and Vf2 can be shown mathematically using the equations derived for the infinite series, developed earlier. Unsurprisingly, the results equal the Reflection Coefficients at load and at the transformer's output port:
First, for Vr2/Vf2:
Transient and Steady-state when the Load Impedance is not Matched:
The model above matches a 200 ohm load to 50 ohms. What happens if the load value is changed from 200 ohms to something else?
Let's make the load 400 ohms. This should result in a steady-state SWR of 2:1 as seen by the source.
Here are the corresponding impulse responses and the results of convolving the impulse responses with the 12-cycle sinusoidal waveform:
The image below is a summary of the steady-state voltages for this mismatched load.
I calculated the steady-state voltages using the calculated values of the system's Reflection and Transmission Coefficients and the formula for solving a geometric series (this formula was presented earlier in this post).
You can see that some amount of power is reflected back from the transformer's input (which no longer is matched to the transmission line) to be dissipated in the Thevenin source's resistance of 50 ohms. This power (2.2 mW) is the expected power to be reflected if the forward power into the transformer is 20 mW and the steady-state SWR at the transformer's input is 2:1.
Therefore, if we are sourcing 20 mW of power and 2.2 mW is being reflected back (by the 2:1 SWR) and dissipated in the source, the remaining 17.8 mW must be dissipated in the load, as all other system elements are assumed to be lossless. This expected result is verified by the calculations.
Also, note that the total Vf2 signal (composed of both the blue line's forward voltage and the green line's forward voltage, which are both in-phase) sums to 1.5 volts (i.e. 0.7 + 0.8). Some portion of this forward voltage is reflected by the load's mismatch (Gamma of 0.778), resulting in a steady-state reflected wave Vr2 with an amplitude of 1.167 volts (that is, 1.167 volts = 1.5*0.778 )
Some portion of Vr2, when it arrives back at the transformer's output, is re-reflected back to the load. The amplitude of this voltage (in steady-state) is Vr2 * Γ3 = 1.167 * 0.6 = 0.7 volts.
And some portion of Vr2 continues through the transformer back towards the source. In steady-state, this voltage is Vf2 * Tr = 1.167 * 0.8 = 0.9336 volts. This voltage is summed with the "current" reflection's amplitude (-0.6 volts) that is out-of-phase with the 0.9336 volts passing through the transformer (thus the minus sign), giving a combined result of 0.333 volts.
Note that the transformer's Input and Output Gammas (Γ2 and Γ3) are still -0.6 and 0.6. They have not changed from the Gamma's calculated for the "matched load" simulation. Ditto for their Transmission coefficients. However, in steady-state, the apparent Reflection Coefficient for the transformer's input is 1/3, not -0.6.
The only Reflection and Transmission coefficients that have changed are the coefficients for the 400 ohm load (referenced to 50 ohms).
Effect of Mismatched Source:
What happens if the source impedance is not matched to the characteristic impedance of the transmission line it is driving, even though the load is matched via the transformer?
The impulse response can quickly become overwhelming to calculate, as I will show, below. But, despite the apparent complexity, we will be able to draw some conclusions that will simplify steady-state calculations.
When deriving the impulse response for the model presented earlier, the source resistance was matched to the transmission line, and thus its Reflection Coefficient is 0 and it will not re-reflect any reflections that arrive back at it. Any signal that arrives back at the source will be absorbed (and dissipated) by the Thevenin-equivalent source resistance, whatever that value might be.
If you have trouble visualizing this, first realize that the single-impulse generator, after it has generated its one impulse, is now zero volts, so it is effectively just a short circuit and thus the transmission line, looking back towards the source, simply sees the source's impedance (which is matched to the transmission line's Zo) as its load.
But what happens if the source is mismatched to the transmission line?
Each time a version of the impulse arrives back at the source, some part of it will be re-reflected back towards the transformer and some part will be absorbed by the source resistance.
Let's take a look at these reflections. The Lattice Diagram, below, shows (in red) the first reflection of the first reflected-impulse arriving back at the source...
Despite this increasing complexity, these reflections of reflections will all decay to 0, so long as the transformer matches the load to the first transmission line's characteristic impedance. That is, the steady state impedance seen looking into the transformer's input should equal the characteristic impedance of the transmission line.
Therefore, in steady-state, if the system is matched looking towards the load, all of the reflections back to the source should sum to zero. If so, there will be no reflected power returning to the source, irrespective of the value of the source impedance.
We can show that this decay-to-zero occurs for every impulse reflected off of the source. Let's consider the first reflection from the source, shown below:
You can see that this reflection, if we consider only it and not any other reflections traveling from the source to the transformer's input, creates a series of impulses traveling from the transformer back to the source that, over time, sum to zero.
All of the other impulses in this series will also re-reflect off of the source impedance, and their re-reflections will also sum to zero over time.. Using the Principle of Superposition, we can consider each impulse arriving back at the source individually and then sum their results together to get a final result.
And therefore, because each and every impulse arriving at the transformer results in a series of reflected impulses that sum to zero, the sum of all of the re-reflections from all of the impulses, when they arrive back at the transformer-input, will sum to zero.
And therefore (to repeat myself), in steady-state, all of the reflections back to the source will sum to zero and there will be no reflected power returning to the source, irrespective of the value of the source impedance.
This last point is important. In steady-state, the source's impedance does not affect the match of the load. The source impedance can be anything, but if the load is matched to the transmission line, the source will see no reflected power in the steady-state.
This also means that, in steady-state, the source's impedance does not affect Vf2, and therefore it does not affect Γ3 (the reflection coefficient looking into the output of the transformer).
Conclusions:
1. As with any other linear system, the Convolution Integral allows us to calculate the time-domain transient and steady-state responses of the system to a stimulus.
2. Given a sine-wave stimulus that is bounded in time, there are reflections back to the source during the system's transient states. These reflections decay to 0 in the steady-state.
3. In the steady-state, Vf2 is a combination of the forward signal arriving from the source (scaled by the discontinuity's Transmission factor) and the re-reflection by Γ3 of the backwards traveling Vr2 wave at the impedance discontinuity looking into the transformer's Output Port. The amplitude of this re-reflection equals |Vr2 * Γ3|.
4. In steady-state Γ3 remains unchanged from the value calculated for an impulse response, and, assuming that the transformer is matching the load to the transmission line's characteristic impedance at its input, Γ3 is independent of the source impedance. In other words, because there are no signals sent back to the source in steady-state, it does not matter what the source impedance is because there is no signal to reflect.
Other Transmission-Line Posts:
http://k6jca.blogspot.com/2021/02/antenna-tuners-transient-and-steady.html. This post analyzes the transient and steady-state response of a simple impedance matching system consisting of a wide-band transformer. I calculate the system's impulse response and find the time-domain response by convolving this impulse-response with a stimulus signal.
http://k6jca.blogspot.com/2021/02/the-quarter-wave-transformer-transient.html. This post analyzes the transient and steady-state response of a Quarter-Wave Transformer impedance matching device. I calculate the system's impulse response and find the time-domain response by convolving this impulse-response with a stimulus signal.
http://k6jca.blogspot.com/2021/05/antenna-tuners-lumped-element-tuner.html. This post analyzes the transient and steady-state reflections of a lumped-element tuner (i.e. the common antenna tuner). I describe a method for making these calculations, and I note that the tuner's match is independent of the source impedance.
http://k6jca.blogspot.com/2021/05/lc-network-reflection-and-transmission.html. This post describes how to calculate the "Transmission Coefficient" through a lumped-element network (and also its Reflection Coefficient) if it were inserted into a transmission line.
http://k6jca.blogspot.com/2021/09/does-source-impedance-affect-swr.html. This post shows mathematically that source impedance does not affect a transmission line's SWR. This conclusion is then demonstrated with Simulink simulations.
https://k6jca.blogspot.com/2021/10/revisiting-maxwells-tutorial-concerning.html This posts revisits Walt Maxwell's 2004 QEX rebuttal of Steven Best's 2001 3-part series on Transmission Line Wave Mechanics. In this post I show simulation results which support Best's conclusions.
Standard Caveat:
Also, I will note:
This design and any associated information is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
No comments:
Post a Comment