Documente Academic
Documente Profesional
Documente Cultură
Dr Francesco Fornetti
Published by Explore RF Ltd, Bristol, United Kingdom
ALL RIGHTS RESERVED. This book contains material protected under International and Federal
Copyright Laws and Treaties. Any unauthorized reprint or use of this material is prohibited. No part
of this book may be reproduced or transmitted in any form or by any means, electronic or
mechanical, including photocopying, recording, or by any information storage and retrieval system
without express written permission from the author / publisher.
While the author and the publishers believe that the information and guidance given in this work are
correct, all parties must rely upon their own skill and judgement when making use of them. Neither
the authors nor the publishers assume any liability to anyone for any loss or damage caused by any
error or omission in the work, whether such error or omission is the result of negligence or any other
cause. Any and all such liability is disclaimed. For permission requests, email the publisher at
info@explorerf.com
Any unauthorised broadcasting, public performance, copying or re-recording of this material will
constitute an infringement of copyright.
ISBN: 978-0-9576635-2-7
Table of Contents
Contents
I
Table of Contents
II
Table of Contents
III
1.1 Making the transition from DC and AC to Radio Frequency
It has now been ten years since I was first introduced to the basic concepts of RF and
Microwave theory and I recall how excited I was as an undergraduate student to start learning and
understanding this mysterious and fascinating subject. Since then I have been to numerous
courses and have tackled a number of engineering tasks in the field. I have also had the opportunity
to teach the subject myself and invariably I can see that the students seem to struggle with exactly
the same concepts which I myself was a bit baffled by back in the days. This myth of RF Engineering
being a bit of a black-magic branch of electronics is not totally unfounded because it often takes
people years to master it and an inexperienced engineer may often be faced with seemingly
inexplicable phenomena.. I feel that the problem lies with the approach which is taken to teaching
RF & Microwave Engineering which, despite being rigorous and formally correct, focuses too soon
and too heavily on the maths when the physical concepts and phenomena should be given priority
and wide breath. Also, and more importantly, it fails to bridge the gap between what happens in AC
and DC circuits, which most students are familiar with, and what happens at Radio and Microwave
Frequencies. This lack of clarity, created by this lack of connection, is further exacerbated by the use
of similar terminology which has a very different meaning at DC/AC and RF.
With this book I will attempt to bridge this gap and make the sharp seam between low and
high frequency theory and techniques, which the classical teaching creates, a great deal smoother
and much easier to overcome.
Most textbooks and courses start with the concept of wavelength and how this quantity
allows you to determine whether you may use a lumped circuit model or whether you need to apply
RF techniques. I, on the other hand, will begin with revising the basic concepts of voltage, current
and impedance in DC and AC circuits and how they relate to Electric and Magnetic fields within such
circuits.
Subsequently I will clarify some mathematical concepts which are often used in the
treatment of the subject and connect them to the physical phenomena that they represent.
I will then introduce the wavelength not just as a simple formula but in terms of its physical
significance and explain why it is a good indication of what techniques need to be used to analyse a
circuit. From there we will move on to the concept of impedance at high frequency and how this
relates to voltages, currents and, more importantly at radio frequency, Electric and Magnetic fields.
A good understanding of these concepts forms the perfect platform onto which RF &
Microwave knowledge may be built and is key to understanding transmission lines, ubiquitous
building blocks of high frequency circuits.
Transmission lines, as most of you will know, are all about delivering power from a generator
to a load. The way we achieve maximum power transfer and maximum efficiency is however very
different for high and low frequency circuits. This is because, while at low frequencies (e.g. 50Hz),
power is conveyed predominantly as voltage and current and is quasi electrostatic in field form, at
high frequencies the power is predominantly conveyed as electromagnetic fields.
This text presents a detailed analysis of transmission lines terminated with different loads. The
concepts of incident and reflected voltages are illustrated both in the case of continuous wave and
pulsed signals. The applications of special lines are also illustrated together with the practical
implementation and simulation of the most common transmission line types.
1
CHAPTER 1 - Fundamentals of Electrical Circuits
After the conceptual and practical treatment of transmission lines, the maths which is used to
design, model and analyse such elements is illustrated. This leads to the mathematical definition of
reflection coefficient and impedance seen for a single-port network and then extends to the
definition of network parameters for two-port networks. Graphical tools for impedance
representation based on such parameters are then introduced (Smith Chart).
The basic concepts of impedance at high frequency are then revisited and enhanced to
include series and parallel passive networks comprising of both resistive and reactive elements.
The basics of impedance matching are then explained on a conceptual basis and some basic
examples are illustrated which describe L-section matching and series to parallel conversions.
The simplification of impedance matching through the use of Impedance and Admittance
Smith charts is then illustrated and a number of video tutorials based on simulations are made
available to the reader.
Finally the basic techniques for high frequency amplifier design are illustrated and a number
of video tutorials are provided to corroborate the students learning.
I firmly believe that Radio Frequency and Microwave Engineering is not a game of Sudoku
where you just apply rules to get the result that you want. To really be a successful designer you
must understand the concepts not just the procedures behind high frequency circuits.
In particular, an understanding of what happens in the time domain and what an impedance
represent is truly crucial when working with advanced amplifier topologies such as Doherty and
when trying to model devices and systems.
It is highly recommended that the reader who is not familiar with the AWR Microwave Office
simulation software watches the introductory video (video 1.1) before any of the others.
2
1.2 Voltage (V) and Electric Field (E)
Most Electronic Engineers are familiar with voltages and with the way they are used to
represent the behaviour of AC and DC electrical circuits. One should not forget however that Voltage
is also called Electric Potential and it represents the potential energy (per unit charge) associated
with electric forces. The Electric Potential Energy is analogous to the Gravitational Potential Energy.
This is illustrated by the example below
Just as the pumping of water to a higher level results in energy being stored, "pumping"
electrons to create an electric charge imbalance results in a certain amount of energy being stored in
that imbalance. And, just as providing a way for water to flow back down from the heights of the
reservoir results in a release of that stored energy, providing a way for electrons to flow back to their
original "levels" results in a release of stored energy.
When the electrons are poised in that static condition (up in the reservoir), the energy
stored there is called potential energy, because it has the possibility (potential) to be released and
do some work. This potential energy, stored in the form of an electric charge imbalance and capable
of provoking electrons to flow through a conductor, is often expressed in terms of voltage, which is a
measure of potential energy per unit charge of electrons.
We cannot assess the amount of stored energy in a water reservoir simply by measuring the
volume of water, we must also consider how far it will drop from its initial height. Likewise, the
potential energy available for moving electrons from one point to another is relative to those two
points. Therefore, voltage is always expressed as a quantity between two points. The analogy of a
mass potentially "dropping" from one height to another is why the voltage between two points is
sometimes called a voltage drop. Such a voltage drop V may be expressed mathematically as
shown by equation (1.2-1)
3
CHAPTER 1 - Fundamentals of Electrical Circuits
(1.2-1)
Equation (1.2-1) demonstrates how, under certain assumptions, electric circuit analysis may
be greatly simplified by using a scalar quantity, voltage, in place of the electric field, which is vector
entity and hence more complex to handle. In particular in DC and AC circuits the interest is in what
happens at the terminals of each component, for example the voltage at the terminals of a resistor
or capacitor or the voltage drop across it. Since a uniform and conservative field exists in resistors,
capacitors and inductors1, voltage may be used to characterise the effect of the electric field on such
components.
This simplification, which uses a scalar quantity (voltage) to represent the effects of an electric
field (vector) and is appropriate for AC and DC circuits, may not apply when higher frequencies are
used as we will see in due course. What should also be pointed out is that voltage is not a physical
quantity, it simply represents an effect of the real entity which causes it i.e. the electric field. The
presence of water in the reservoir is an effect of the pump pushing the water up to it, so the pump is
what physically changes the level of water in the reservoir, the reservoir filling up is just an effect
which may be quantified and used to represent the work done by the pump.
1
This is true only for ideal components however it is a good enough approximation in most practical cases
when only passive circuit elements are involved and the frequency of excitation remains low.
4
1.3 Current (I) and Magnetic Field (H)
As shown in Figure 1.3-1, the lines of the magnetic field are concentric circles for a long, straight,
current-carrying wire and their direction is given by the right hand rule. Most of you will be familiar
with the Biot-Savart law which allows the vector field B to be calculated. Its modulus is shown in
equation (1.3-1)
| | (1.3-1)
Where represents the distance from the wire and the magnetic permeability of space. In as
much as the B field is considered appropriate to represent magnetic fields in free space, when the
generated fields pass through magnetic materials which themselves contribute internal magnetic
fields, ambiguities can arise about what part of the field comes from the external currents and what
comes from the material itself. It has been common practice to define another magnetic field
quantity, usually called the "magnetic field strength" designated by H. In free space a very simple
relationship exists between the H and B fields
(1.3-2)
Equations (1.3-1) and (1.3-2) show that B and H have the same direction and orientation but
the magnitude of H is obtained by dividing the magnitude of the B field by the scalar quantity .
However when magnetic materials are present, this relationship no longer holds true and B and H
may be different not only in magnitude but also in direction and orientation. Because of this, the H
field is a more general representation of the magnetic field and hence, from here forth, by magnetic
field we will intend the H field, unless otherwise specified. This is also the field which is used at radio
frequency.
Equation (1.3-1) shows that a scalar quantity, current, may be used to describe the flow of
charge in an electric circuit. This, as in the case of voltage and electric field, greatly simplifies the
analysis of the circuit but it is appropriate only under specific assumptions which are generally
satisfied at low frequency. At higher frequencies however, a current alone may not be sufficient to
describe the effect of a magnetic field and hence the vector H may need to be used instead.
5
CHAPTER 1 - Fundamentals of Electrical Circuits
In this section we will explain how the values of resistance R, capacitance C, and inductance L, not
only provide a way to relate voltages and currents for resistors, capacitors and inductors but also a
link between the moduli of Electric and Magnetic fields. For the readers convenience the classical
V-I equations for resistors (1.4-1)(a), capacitors (1.4-1)(b) and inductors (1.4-1)(c) are shown below.
( ) ( )
( ) ( ) ( ) ( ) ( ) (1.4-1)
In the following three sections we will look at how we can approach the abovementioned circuit
elements from a field perspective.
1.4.1 Resistors
Let us now consider a simple resistor in the form a high resistance wire. This could be the filament of
a light bulb for instance (Figure 1.4-1) but also a heating element used in household appliances.
For continuity, the current inside the resistor will be the same as the current on the wires connected
to its terminals, let us call it .
Leads
I
I I
a b
Through equation (1.3-1) we can relate the current thought the resistor to the magnetic field inside
it.
(1.4-2)
Also, since the electric field inside the resistor may be considered uniform, it is easy to relate the
voltage across it to the respective electric field through equation (1.2-1)
(1.4-3)
6
1.4 V & I or E & H ?
If we now calculate the ratio of voltage and current from (1.4-2), (1.4-3), and (1.4-1)(a) we obtain:
(1.4-4)
(1.4-5)
Equation (1.4-5) shows that relates and at a specific point in space around the resistor
(identified by radial distance from its axis) in a linear fashion. This derivation is not entirely
accurate as one would need to take into account material properties to be fully rigorous.
Nonetheless it demonstrates how, in principle, the value of the resistance gives an indication of
the relative magnitudes of vector quantities and , not just of the scalar quantities and which
may be used to represent and in low frequency circuits.
1.4.2 Capacitors
(1.4-6)
(1.4-7)
where is the sheet charge density on each plate, is the absolute value of the charge on either
plate and is the surface area of each plate.
Now let us take a look at the magnetic field between the plates of the capacitor. It would
appear that, since no current, in its classical form2, can flow between the plates, we are unable to
use equation (1.3-1) to obtain the magnetic field from the current. This in fact is not the case
since a displacement current exists between the plates which is equal in magnitude to the current
flowing in and out of the terminals of the capacitor. This comes from Maxwells generalisation of
Amperes law and its proof is beyond the scope of this treatment3.
2
conduction current
3
For the reader who wishes to research this further, the forth Maxwell equation is used to come to this result
7
CHAPTER 1 - Fundamentals of Electrical Circuits
+ -
The magnetic field at a distance from the central axis of the capacitor may therefore be expressed
by equation (1.4-8)
(1.4-8)
(1.4-9)
Where is the voltage across the capacitor. Such a voltage may be easily calculated by means of
equation (1.2-1), since the Electric field inside the capacitor is uniform
(1.4-10)
( ) ( ) (1.4-11)
Where is the distance between the capacitor plates. By substituting (1.3-2) and (1.4-11)(b) into
(1.4-8) we obtain
( )
( ) ( ) (1.4-12)
Equation (1.4-12) shows that the capacitance relates the magnitude of the magnetic field and
the rate of change of the magnitude of the electric field inside a capacitor ( ), at a specific
point in space defined by the radial distance from its axis, in a linear fashion. It also shows how the
magnetic field is related to the first derivative of the electric field, just like the current though a
capacitor is related to the first derivative of the voltage across it (equation (1.4-1)(b)).
8
1.4 V & I or E & H ?
1.4.3 Inductors
Although inductors are usually wound around a cylinder or toroid made of specific materials,
at radio frequency, air-core inductors are quite common and in fact sometimes the best solution. For
simplicity we will therefore analyse the field of an air-filled solenoid since this gives us enough of an
insight into the behaviour of electric and magnetic fields to a first approximation.
Using Amperes law4, the field at any position inside the solenoid, may be easily calculated as
(1.4-13)
Where represents the number of turns for unit length and the current through the solenoid.
The electric field at a distance r from the axis of a solenoid may be derived from Faradays
law of Induction where the line integral is calculated over a circular path of radius and length
and the magnetic flux over the respective surface . This is shown in Figure 1.4-4.
2
A r
B E
B
Circular path
Figure 1.4-4 Electric and Magnetic fields inside a solenoid, E is perpendicular to the page and entering it.
(1.4-14)
(1.4-15)
4
Amperes law states that where the integral is calculated along an Amperian loop which
encloses the total current
9
CHAPTER 1 - Fundamentals of Electrical Circuits
( )
( ) ( )
( ) (1.4-16)
This shows how, at position , the magnitude of the electric field is related to the first
derivative of the magnitude of the magnetic field in a linear fashion, just like the voltage across an
inductor is related to the first derivative of the current, equation (1.4-1)(c).
Now, before we introduce the inductance , let us see how this quantity comes about. First
of all, we need to calculate the total induced voltage or for a solenoid. To this end we use a
very similar expression to (1.4-15) but, because we are looking at the total across the terminals
of the whole solenoid, we must include all loops and hence multiply the rate of change of flux
( ) by the total number of turns N.
(1.4-17)
(1.4-18)
(1.4-19)
is used to conglomerate several constants into one constant of proportionality which linearly
relates and .
Now if we substitute (1.4-19) into (1.4-16), we obtain
( ) ( ) (1.4-20)
( )
Equation (1.4-20) shows that inductance may be used to relate mathematically the
magnitude of the electric field and the rate of change of the magnitude of the magnetic field inside
an inductor at a specific point in space defined by the radial distance from its axis.
10
1.5 The significance of reactive components equations
In the authors experience, far too often mathematical expressions and operators are used
without their actual meaning in a physical or, for that matter, mathematical context being explained
adequately. In this section the author aims to give a more real and intuitive explanation of the
meaning and the application of derivatives and complex numbers. A good understanding of these
mathematical concepts forms the perfect platform onto which one can build knowledge of circuits
and systems at radio frequency.
1.5.2 Derivatives
The derivative of a function is too often seen as a mathematical entity which may be found
in a lookup table. One should not forget however that the derivate of a function at a specific point
represents the slope of the tangent to that function at that very point. Below are some derivatives
for simple and commonly encountered rational functions.
(1.5-1)
(1.5-2)
These may be found in mathematical tables however one should be able to work them out
quite easily just by finding the slopes of the tangents at various points for each curve and joining the
dots to find the expression of the derivative.
For , this process is illustrated in Figure 1.5-1 which shows that this graphical method yields
the same result as that shown by equation (1.5-1). This is also shown as an animation in video 1.2
For , this process is illustrated in (1.5-2), which yet again confirms the result shown by
equation (1.5-2). This is also shown as an animation in video 1.3
What is obvious from both the mathematical and graphical approach, is that the derivative
of a simple power function will always be an order of power lower than the function and hence
increase more slowly. This result may be generalised as
So it all seems quite straight forward for simple rational functions in that the derivate simply
increases at a lower rate than the function from which it was derived. However, often enough in
electronic circuits, we are interested in periodic functions and stimuli which vary in a sinusoidal
fashion, so let us derive graphically the derivative of the sine function. This is shown in Figure 1.5-3
and Figure 1.5-4 and also demonstrated with an animation in video 1.4
11
CHAPTER 1 - Fundamentals of Electrical Circuits
(a)
(b)
(c)
Figure 1.5-1 (a) tangent to at x=5 (b) tangent to at x=7 (c) interpolation of slope points shows the
function representing the derivative of
12
1.5 The significance of reactive components equations
(a)
(b)
(c)
Figure 1.5-2 (a) tangent to at x=5 (b) tangent to at x=7 (c) interpolation of slope points shows the
function representing the derivative of
13
CHAPTER 1 - Fundamentals of Electrical Circuits
(a)
(b)
(c)
(d)
Figure 1.5-3 Tangents to ( ) at (a) x=90 (b) x=180 (c) x=270 (d) x=360
14
1.5 The significance of reactive components equations
( )
( )
Figure 1.5-4 Interpolation of slope points shows the function representing the derivative of ( )
From Figure 1.5-4 it is apparent that, in the case of a sinusoidal function, its derivative is an
identical curve which is however offset along the x-axis. In the case of ( ), this offset is 90 and
hence its derivative is ( ).
In lookup tables however, this derivative is often stated as ( ) which is also correct by
virtue of the well know trigonometric equations. The author however prefers to think of the
derivative of a sinusoidal function in the latter from since it gives a better idea of what such a
derivative represents.
Let us now consider a capacitor and see how these mathematical concepts apply to the
physical world. Let us assume that the voltage across the capacitor may be described by a sinusoidal
function
( ) ( )
Where represent the angular frequency5 of the stimulus. According to equation (1.4-1)(b)
( )
( ) ( ) (1.5-3)
This means that, at any instant in time, the current ( ) will always be ahead of the
voltage. This may be seen in Figure 1.5-4 where the derivative of the ( ) function, reaches a
peak before the ( ) function does. This is what is meant by the current leading the voltage
in a capacitor.
A dual argument may be derived for inductors where the opposite applies i.e. the voltage
leads the current by . This latter demonstration is left to the reader.
This is discussed in more details in section 4.2.
5
, where is the frequency in Hertz
15
CHAPTER 1 - Fundamentals of Electrical Circuits
A complex number effectively allows you to represent two values ( and ) with one entity
and comprises of a real part and an imaginary part , as shown below
The letter represents the imaginary constant6 which is equal to . Because is multiplied by the
imaginary constant , there is no way to mix it with , which is the beauty of complex numbers.
The number may be plotted on a Cartesian graph where the abscissa represents real values and
the ordinate represents imaginary values as shown in Figure 1.5-5.
Equivalently, may be represented in polar form by means of its modulus and angle as shown in
Figure 1.5-6.
6
is sometimes called
16
1.5 The significance of reactive components equations
The polar representation makes life easier when manipulating equations and it also has the
advantage of being easily represented by simple exponential functions which have an imaginary
exponent. This is possible thanks to Euler formula which is shown below
( ) ( ) (1.5-4)
( ) ( ) (1.5-5)
So the complex exponential gives a very compact notation for our complex number!
One of the main advantages of complex numbers is that they allow phase shifts to be
easily represented. For instance, the angle of a complex number may be increased by by simply
multiplying it by the complex constant . As an example, let us consider a complex number with
equal real and imaginary parts (just for simplicity) and express it in both polar and Cartesian form
[ ]
17
CHAPTER 1 - Fundamentals of Electrical Circuits
Now if we plot this result on our graph, we can clearly see that the effect of multiplying a complex
number by is that the angle of such a number is shifted by .
1.5.3.2 Impedance
Now that we have reviewed complex numbers, let us look at the impedance for resistors
(1.5-6)(a), capacitors (1.5-6)(b) and inductors (1.5-6)(c).
(1.5-6)
( ) ( ) ( )
Where . Equations (1.5-6)(b-c) clearly show that the impedance of reactive components
depends on frequency as well as physical constants C and L.
Ohms law still applies to impedance and may be simply rewritten as
Hence
( ) ( ) ( ) (1.5-7)
Equation (1.5-7)(a) tells us that, in the case of a resistor, there is no phase shift between
voltage and current. In the case of a capacitor however, as shown by equation (1.5-7)(b), the current
is multiplied by to get the voltage. This means, as was shown in section 1.5.3.1, that the voltage
across a capacitor is behind the current. This ties in perfectly with what we saw in section 1.5.2
(page 15), where we found that, in a capacitor, the current leads the voltage by .
By a similar argument, equation (1.5-7)(c) shows that voltage across an inductor is ahead of the
current i.e. the voltage leads the current by . We will look at complex impedances and
admittances in more details in chapter 4.
18
1.5 The significance of reactive components equations
Lastly let us now look at some mathematical properties which are peculiar to exponential functions.
It may be useful in some instances to represent the imaginary constants and as exponentials in
order to make the 90 and -90 phase shifts which they represent more explicit and simplify
calculations.
This may be easily achieved by means of Eulers formulae as shown in (1.5-8).
(1.5-8)
( ) ( )
So the derivative of a basic exponential function is actually the exponential function itself! However,
if the exponent is multiplied by a constant , then this constant appears as a multiplicative factor in
the derivative
( ) (1.5-10)
So the derivative of a complex exponential is shifted in phase by with respect to the original
exponential.
19
CHAPTER 1 - Fundamentals of Electrical Circuits
Euler equation (1.5-4) may also be manipulated to give expressions for both sine and cosine
in terms of complex exponentials8 as shown in (1.5-11).
(1.5-11)
( ) ( )
Let us use these results to find the derivative of the cosine function. From (1.5-11)(a), we get
( )
( ) ( )
( ) ( )
[ ] ( )
This result ties in very well with the results obtained for the derivative of sinusoidal functions
illustrated in section 1.5.2, page 15.
What is also worth pointing out at this stage is that any arbitrary phase shift may be
introduced by using the respective complex exponential as a multiplicative factor, as shown
below
( ) (1.5-12)
A (or ) phase shift is just a special case of (1.5-12) which takes advantage of (1.5-8).
( )
8
For further insight into this formulae please refer to section 1.5.4.2.
20
1.5 The significance of reactive components equations
(1.5-11)(a)
Now, you will recall, from the physics of rotational motion, that if we are moving around a circle at
an angular speed , the angle which we travel in time is equal to . Equation (1.5-11)(a) may
therefore be rewritten as
(1.5-13)
The term may be represented in the complex plane, as a vector rotating anticlockwise as time
increases, with angular speed . Equally the term represents a vector in the complex plane,
rotating clockwise at radians per second. This is illustrated in Figure 1.5-9. So the numerator of
equation (1.5-13) is just the sum of these two vectors of equal magnitudes which are rotating at the
same speed but in opposite directions. From Figure 1.5-9 (a)-(f), it is apparent that the sum of these
two vectors always lies on the real axis. If we plot the modulus of this sum on the y-axis of an
ordinary Cartesian graph versus time, as shown on the right-hand pane of Figure 1.5-9 (a)-(f), then
we can see that we get a cosine-like function. This is also illustrated in an animated fashion in video
1.5
You may notice however that this function varies between -2 and 2, i.e. is twice the amplitude of a
cosine function. This is why equation (1.5-13) prescribes that we should halve it to get the cosine.
Understanding complex exponentials and what they represent is key to understanding the
mathematical representation of waves and their propagation though transmission lines and media.
As we will see in section 2.14, just as light shining through a glass is partly reflected and partly shone
through, electromagnetic waves and signals undergo the same process at radio frequency. The
reflected and transmitted signals, which travel in opposite directions, may be represented by
complex exponentials rotating around in opposite directions on the complex plane, in a similar
fashion to the ones shown in Figure 1.5-9.
For sine functions we may use a similar equation
(1.5-11)(b)
21
CHAPTER 1 - Fundamentals of Electrical Circuits
(a)
(b)
(c)
22
1.5 The significance of reactive components equations
(d)
(e)
(f)
23
CHAPTER 1 - Fundamentals of Electrical Circuits
As shown in Figure 1.5-10, the total impedance of this network is simply the sum of the individual
impedances. This impedance is more easily expressed in polar form as
| | | | ( ) (1.5-14)
( )
Equation (1.5-14) shows that, as more elements are added in series, the expression for the overall
impedance becomes more complex. Also, when elements are added in parallel, the maths
complicates matters further. This is why, at high frequency, we use a special graphical tool called
Smith chart which allows us to calculate the impedance of complex networks in a clever graphical
manner (section 3.6 and 4.4).
So, we have seen how impedance gives an indication of the relative values of voltage and
current and also of the phase difference between them. As you may recall, in sections 1.3 and 1.4,
we saw how voltage and current are a simplification of Electric and Magnetic fields and how the
values of R, C and L also give a relationship between the moduli of such fields. The impedance goes a
step further and, in addition to showing a relationship between the moduli of E and H, it gives us an
indication of the phase difference between them. We will clarify what we mean by phase difference
in this context in due course but, for now, you may think of such a difference as an indication of how
long it takes one field to respond to changes in the other field.
The Impedance also helps us describe how the frequency of excitation affects the
magnitudes and phases of electric and magnetic field.
24
1.5 The significance of reactive components equations
ACVS
M_PROBE
ID=V1
RES ID=VP1
Mag=1 V
Ang=0 Deg ID=RS
Offset=0 V R=50 Ohm
DCVal=0 V
CAP
ID=C1
C=20 pF
0.1 10
p1
0 0
-0.1 -10
p2
-0.2 -20
0 0.5 1 1.5 2
Time (ns)
9
The internal impedance of a radio frequency generator is often chosen to be 50
25
CHAPTER 1 - Fundamentals of Electrical Circuits
0.05 10
p1
0 0
-0.05 -10
p2
-0.1 -20
0 0.2 0.4 0.6 0.8 1
Time (ns)
Figure 1.5-12 and Figure 1.5-13 show voltage and current waveforms for the capacitor at 1GHz and 2
GHz respectively. The markers also show the maximum amplitudes of current and voltage and the
time stamps at which such amplitudes are reached. Let us start with the 1GHz case.
At 1 GHz the period of the signal is
is therefore a quarter of and hence a quarter of a full 360 period i.e. 90. The phase shift
between voltage and current is therefore 90. In other words the current leads the voltage by 90.
Let us now look at the relative amplitudes of voltage and current
This last equation demonstrates that the modulus of the impedance calculated algebraically is
indeed the ratio of and and that the represents the phase relationship between
voltage and current.
26
1.6 Animations and Video Tutorials
The reader may now wish to watch some illustrative videos and animations which give further
insight into the topics explained so far in this chapter. A brief summary of the content of this
material is presented below.
VIDEO
CONTENT
REFERENCE
This animation shows Eulers formula for cosine, eq. (1.5-13), from a graphical
1.5 view point. It shows how two vectors rotating around in the complex plane may
be used to represent a cosine function.
This animation shows Eulers formula for sine, eq.(1.5-11)(b), from a graphical
1.6 view point. It shows how two vectors rotating around in the complex plane may
be used to represent a sine function.
27
CHAPTER 1 - Fundamentals of Electrical Circuits
1.7 Conclusions
In this chapter we first revised the concepts of voltage and current and how they relate to
Electric and Magnetic fields. We also demonstrated how circuit elements properties such as
resistance, capacitance and inductance, not only relate V and I but also the magnitudes of E and H.
As we move further up in frequency, the assumptions which are made for AC and DC circuits and
which enable us to use the scalar quantities V and I in place of vector entities E and H, may however
no longer be valid. Therefore a new approach to the design and analysis of circuits will need to be
taken.
We also revised some relevant mathematical concepts, namely derivatives, complex numbers,
complex exponentials and Eulers formulae, from a different viewpoint and established a clear
connection between the mathematics and the physical phenomena that they are used to represent.
Moreover we explored the concept of complex impedances and explained why their imaginary parts
are indeed very real!
This gives us a solid launch platform to leap into high frequency circuit theory, transmission
lines and electromagnetic waves.
28
2.1 Introduction
2.1 Introduction
As mentioned in chapter one, whereas at low frequency Electric and Magnetic fields may be
considered quasi-static, at high frequency their rate of change is such that this assumption is no
longer valid. The simplified, low-frequency models, which assume that most of the power is
conveyed by currents, are therefore no longer suitable since, at Radio Frequency, most of the power
is conveyed by Electric and magnetic fields intertwined in such a way as to form an Electromagnetic
wave.
Let us look at an example to illustrate what we mean by electromagnetic waves.
The charged particle shown in Figure 2.1-1 is oscillating along a length of wire.
As the charge moves and assumes different positions, the electric field which it generates
also changes. If we were in free space and we looked at the electric field lines in the area
surrounding the charge, we would see their direction change at the same speed as the charge
oscillates. If we placed a test charge near the wire therefore, the force which this test charge would
experience would vary in time as the oscillating charge moves.
Also, the speed of the oscillating charge will be increasing as it leaves one end of the wire
and decreasing as it approaches the opposite end. This translates into a time-varying current which
generates a time-varying magnetic field as shown by equation (1.3-1) where the constant dc current
is replaced by ( ).
( )
( )
By Faradays law of induction, this magnetic field generates an electric field which, through
Amperes law, generates a magnetic field and so on. This mechanism, which Maxwells equations
beautifully describe, is shown in Figure 2.1-2.
29
CHAPTER 2 - Conveying Power at Radio Frequency
Figure 2.1-3 shows the Electric and Magnetic fields, which form the electromagnetic wave, from a
different angle.
Electromagnetic waves are ubiquitous in everyday life. Without them there would be no
mobile phones, no wireless internet, no satellite or cable TV and no Microwave ovens, to name but a
few!
EM Waves can either be guided through suitable structures, designed to contain their
electric and magnetic fields, or transmitted out into free space or other media. In this chapter we
will focus on the former topic and in particular on the fundamental building blocks of any radio
frequency and microwave circuits, transmission lines.
30
2.2 The true sense of Wavelength
When the frequency of excitation of our circuits is relatively low, the effects of the connecting
wires are limited to minor losses and can often be ignored. When working at higher frequencies
however, the effect of such wires can no longer be ignored. In fact they are so significant that we
do not use wires at Radio Frequency, but appropriately designed interconnecting structures called
transmission lines.
Now wheres the boundary between low and high frequency and when do we need to use
transmission lines in place of wires?
Well, this comes down to two factors: the maximum excitation frequency of our circuit and its
dimensions.
Let us consider a 100 km power line which carries a 50Hz signal, as shown in Figure 2.2-1.
Power Line
100 km
VS RL
Assuming that the signal will travel down the line at the speed of light10, it will get from the
transmitting to the receiving end in a time equal to
Now let us look at how quickly our signal changes. Well if the frequency is 50Hz then its period is
equal to
This means that every 20 ms our signal11 starts from zero, goes up to a positive peak, then a negative
peak and after 20 ms goes back to zero again and the same pattern repeats.
Now in the that it has taken the signal to travel to the end of the power line, the voltage at
the generator end has not changed by much. In fact, if we turn on our signal generator at time
, by time the signal reaches the end of the line at , the signal generator has
12
only gone through 6% of the signal period .
10
The speed of light in vacuum is approximately equal to
11
For simplicity assume a sinusoidal signal with an initial phase of zero degrees, ( )
12
This figure is calculated as
31
CHAPTER 2 - Conveying Power at Radio Frequency
This means that, at any instant in time, the signal that we observe at the generator end VS, and the
signal at the end of the power line VL , have a very little phase difference. This is shown in Figure
2.2-2.
Figure 2.2-2 Phase Difference between Generator and Load Signals for a 100km power line
This small phase difference over such a large distance is due to the fact that the speed at which our
signal changes is much lower than the speed at which is propagates.
However, if we made our transmission line much longer, then we would increase the phase
difference between VS and VL considerably. If, for instance, our line was of such a length that, by the
time the signal appeared at the load end, the generators signal had gone through a quarter of its
period, we would see a 90 phase shift between VS and VL at any instant in time.
This is shown in Figure 2.2-3.
Figure 2.2-3 Phase Difference between Generator and Load Signals for a 1500km power line
To find the length over which our signal suffers a 90 delay, we simply need to multiply the speed of
light by a quarter of the period as shown by equation (2.2-1).
32
2.2 The true sense of Wavelength
(2.2-1)
From this calculation it is apparent is that, at 50Hz, we need to travel a very long way in order for the
finite propagation speed of the signal to cause appreciable delays. However if the period of our
signal was much shorter, i.e. if our signal varied much faster with time, then the distance over which
such delays become appreciable would be considerably reduced, as we will see shortly.
Now one may ask, If we have a generator with a load at the end of it, what does it matter if
the signal gets to it with a bit of delay? After all, that is inevitable because of the finite speed of
propagation so whats the big deal?.
Well, that is a very fair point and indeed if you have a pure sinusoid travelling down a power line,
into some sort of load which is going to rectify your AC signal and turn it into DC then there is no
problem. However, suppose that you have additional elements connected along your line and that
the distance between such elements is of the order of 100s of kilometres as shown in Figure 2.2-4.
VS L C RL
In this case, at any instant in time, the phase offsets with respect to the signal generator are
45 across the inductor, 90 across the capacitor and 135 across the resistor due to the finite speed
of propagation of the signal alone. The inductor and the capacitor will also introduce phase changes
themselves which in turn will take time to propagate to other components. This complicates circuit
analysis considerably since the voltages across the various circuit elements are not cophasal. In a
small electrical circuit instead, you would have distances of the order of meters between the
elements and one may therefore assume that the signal propagates at infinite speed. This means
that voltages appear instantly across shunt elements and are cophasal which simplifies matters
considerably.
Let us stop for a moment and reflect on how we could avoid these problems. There are three
things we could do:
Option 1 is clearly impossible, but the other two are possible to some extent.
However, in a power line, you are unlikely to get this sort of scenario with components at such large
distances between one another so theres no reason to worry.
33
CHAPTER 2 - Conveying Power at Radio Frequency
One thing to notice however is that, just as deceasing the frequency of the signal would
improve the phase offsets and the issues that they may bring about, increasing the frequency would
have the opposite effect.
Let us now see what happens when the frequency of our signal is increased by considering a 100
MHz signal in an analogous circuit to that shown in Figure 2.2-1.
Earlier on we showed how to calculate the length that a line needs to be for there to exist a
90 phase difference between the generator and load signals. For a 100 MHz signal, this distance is
clearly shorter since the period of this signal is only 10 ns. This means that it takes a much shorter
time for the generator phase to reach 90. By using equation (2.2-1) and the period of our 100MHz
signal ( ), we can calculate this length
Figure 2.2-5 Phase difference between generator and load voltage at 100 MHz over a 0.75m line
In a similar fashion we may calculate the length of the line for other phase offsets
Figure 2.2-6 (a)-(c) shows the phase offset between generator and load voltages for these lengths of
line.
34
2.2 The true sense of Wavelength
(a)
(b)
(c)
Figure 2.2-6 Phase difference between generator and load voltage at 100 MHz over a 1.5m (a), 2.25m (b), 3m (c) line
35
CHAPTER 2 - Conveying Power at Radio Frequency
All these lengths give as an indication of how the finite speed of propagation influences the
relative phases of the generator and load voltages for a signal of a specific frequency. It is quite clear
however that they are all multiples (or fractions) of one another and hence we could just calculate
one of them and use its fractions or multiples to get the length associated with any phase difference.
It is customary to choose for this purpose and then use fractions of it to represent 90 and 180
shifts. What we called is commonly called wavelength and indicated by the greek letter . It
essentially represents the distance that a signal is able to travel over its period .
(2.2-2)
Now there are two things that must be pointed out. Firstly our signal may not always travel
at the speed of light. If, for instance, the signal propagates through some dielectric material then its
propagation speed will be lower. We should therefore use a generic speed 13 instead of c in the
above formula.
(2.2-3)
(2.2-4)
This is a nice form to express the wavelength because it intuitively suggests that gives an
indication of how fast our signal is travelling ( ) with respect to the rate at which it is changing ( ).
If this ratio is high, i.e. the signal travels much faster than it can change, then the range of
distances over which phase shifts are unimportant is much larger. On the other hand, if this ratio is
low, our signal may be changing very quickly hence small distances may introduce significant phase
shifts.
In general it is assumed that if the wavelength is much greater than the dimensions of the
circuit, the lumped elements approximation may be used for circuit analysis i.e. phase offsets due to
a finite speed of propagation may be ignored. As a rule of thumb, the wavelength should be at least
10 times greater than the dimensions of the circuit for this approximation to be valid.
If such a condition is not verified, then transmission line theory should be used when
designing and analysing the circuit. Transmission line theory and its practical applications will be the
subject of the following sections.
13
See Section 2.3 and 2.11 to see how the velocity of propagation for common transmission lines is calculated
36
2.3 Transmission Lines an Introduction
We will start our treatment of transmission lines with one the most common types, the
coaxial cable. A coaxial cable is a two-conductor cable made of a single conductor surrounded by a
braided wire jacket, with a plastic insulating material separating the two. The outer conductor
completely surrounds the inner conductor and the two conductors are insulated from each other for
the entire length of the cable.
If an ohmmeter was employed to check the cables resistance, it would show the two
conductors to be completely insulated from one other, with nearly infinite resistance between the
two. This is due to the fact that an ohmmeter would use a continuous direct current (DC) to perform
such a measurement. However, the cables response to short voltage pulses and high-frequency
signals would be quite different because of the effects of capacitance and inductance distributed
along the length of the cable. When the applied voltage changes rapidly, the cable presents a finite
impedance to the signal source (typically 50 or 75 ), and draws a current proportional to the
applied voltage. When such stimuli are used, this pair of wires becomes an important circuit element
with its own characteristic properties which we refer to as a transmission line.
Let us now consider a set of parallel wires of infinite length (Figure 2.3-2), with nothing
connected at the end. What would happen when we close the switch? Would there be no current at
all?
37
CHAPTER 2 - Conveying Power at Radio Frequency
Even if we assume that we are using wires with zero resistance, there would still exists some
capacitance along the cable due to the fact that any pair of conductors separated by an insulating
medium creates capacitance between those conductors (Figure 2.3-3).
When en electric filed is applied to our pair of wires, the capacitance which exists between them
means that a current proportional to the rate of change of voltage over time will be drawn. This is
described by the equation (1.4-1)(b), . This capacitance stores the energy provided by the
electric field created by the voltage source. According to the equation, an instant rise in applied
voltage (produced by a perfect switch closure) gives rise to an infinite charging current. However,
the current drawn by a pair of parallel wires will not be infinite, because there exists series
impedance along the wires due to inductance (Figure 2.3-4).
Infinite Length
Switch
DC
Source
Such an inductance comes from the fact that current through any conductor develops a magnetic
field of proportional magnitude and energy is stored in this magnetic field (Figure 2.3-5). This stored
energy comes from the electric field created by the source and this transformation of energy from
electric into magnetic manifests itself as a voltage drop governed by the inductance equation
(1.4-1)(c), . This voltage drop limits the rate of change of voltage across the distributed
capacitance, preventing the current from ever reaching an infinite magnitude.
38
2.3 Transmission Lines an Introduction
Figure 2.3-5 Electric and Magnetic fields as signal propagates down a transmission line
Assuming that there is air between our two conductors, voltages and currents will propagate down
our transmission line at nearly the speed of light, progressively charging the distributed capacitance
and drawing a constant current of limited magnitude from the battery. This is shown in Figure 2.3-6.
Since the wires are infinitely long, their distributed capacitance will never fully charge to the source
voltage and this pair of wires will continue to draw a constant current from the source so long as the
switch is closed, behaving as a constant load. When we consider our wires under these conditions
we cannot just see them as a pair of conductors, we must treat them as a transmission line.
So our infinitely long transmission line behaves as a constant load and its response to the applied
voltage is resistive rather than reactive, despite the fact that the line comprises solely of inductance
and capacitance (assuming that the wires have zero resistance). This is because, from the batterys
perspective, there is no difference between a resistor eternally dissipating energy and an infinite
transmission line eternally absorbing energy. The ratio of the battery voltage and the constant
current that an infinite line would draw is called the characteristic impedance ( ) of the line, and it
is determined by the values of our distributed capacitance and inductance, which in turn are fixed by
the geometry of the two conductors.
39
CHAPTER 2 - Conveying Power at Radio Frequency
40
2.3 Transmission Lines an Introduction
For a parallel-wire line with air insulation, shown in Figure 2.3-7, the characteristic impedance may
be calculated as shown:
(2.3-1)
Where,
Z0 = Characteristic impedance of line
d = Distance between conductor centres
r = Conductor radius
= Relative permittivity of insulation between conductors
If the transmission line is coaxial14 in construction (Figure 2.3-8), the characteristic impedance
follows a different equation:
(2.3-2)
Where,
Z0 = Characteristic impedance of line
d1 = Inside diameter of outer conductor
d2 = Outside diameter of inner conductor
= Relative permittivity of insulation between conductors
In both equations, identical units of measurement must be used in both terms of the fraction.
If the insulating material is not air (or vacuum) both the characteristic impedance and the
propagation velocity will be affected. The ratio of a transmission lines true propagation velocity and
the speed of light in a vacuum is called the velocity factor of the line.
The velocity factor is purely a factor of the relative permittivity (or dielectric constant) of the
insulating material which is defined as the ratio of a materials electric field permittivity to that of a
pure vacuum.
14
Coaxial lines are explained further in section 2.11.1
41
CHAPTER 2 - Conveying Power at Radio Frequency
The velocity factor of any cable type, coaxial or otherwise, may be calculated quite simply by means
of (2.3-3).
(2.3-3)
Where,
= Relative permittivity of insulation between conductors
= Velocity of wave propagation
= Velocity of light in a vacuum
Equations (2.3-1) and (2.3-2) show that a transmission lines characteristic impedance (Z0)
increases as the conductor spacing increases. If the conductors are moved away from each other,
the distributed capacitance will decrease (greater spacing between capacitor15 plates), and the
distributed inductance will increase (less cancellation of the two opposing magnetic fields). Less
parallel capacitance and more series inductance results in a smaller current drawn by the line for any
given amount of applied voltage therefore the characteristic impedance of the line also increases.
Conversely, bringing the two conductors closer together increases the parallel capacitance
and decreases the series inductance. Both changes result in a larger current drawn for a given
applied voltage, equating to a lesser impedance. A more detailed analysis of transmission lines
geometry and physical construction and their effects on characteristic impedance is presented in
section 2.11.
Ignoring any dissipative effects such as dielectric leakage and conductor resistance, the
characteristic impedance of a transmission line is equal to the square root of the ratio of the lines
inductance per unit length divided by the lines capacitance per unit length:
(2.3-4)
Where,
= Characteristic impedance of the line
L = Inductance per unit length of the line
C = Capacitance per unit length of the line
15
For a parallel plate capacitor , where is the distance between the plates.
42
2.4 Finite Length transmission line
We have determined that, if we had a transmission line of infinite length (Figure 2.4-1), we
would be able to measure a finite resistance between inner and outer conductor (usually 50 )
which is termed characteristic impedance of the line. But in reality, since the line cannot be infinite
in length, we would always see an infinite resistance between our two wires. Nonetheless, the
characteristic impedance rating of a transmission line is important even when dealing with limited
lengths.
If we had a finite length of line which was left open-circuited at the end, we would initially
observe the distributed inductance and capacitance progressively charging just as we had in the case
an an infinitely long line (Figure 2.3-6). From the point of a view of the battery, up until the time it
takes for the unterminated end of the line to be reached, the line would behave just like a constant
load, equal in value to the characteristic impedance of the line. However as the end of the line is
reached, the electrons have nowhere to go hence they pile up at the end and then travel back to the
battery at which point the current ceases and the line acts as a simple open circuit. This is shown in
Figure 2.4-2) and described in more details in section 2.5.1. All this happens very quickly! For a 1km-
long cable with a velocity factor of 0.66 (signal propagation velocity is 66% of light speed, or 200,000
km per second), it takes only 1/200,000 of a second (5s) for a signal to travel from one end to the
other. For the current signal to reach the lines end and reflect back to the source, the round-trip
time is twice this figure, or 10 s.
A signal propagating from the source-end to the load-end of a transmission line is called an
incident wave. The signal propagating from the load-end back to the source-end is called a reflected
wave.
High-speed measurement instruments are able to measure the time that it takes the signal to
reach the end of the line and come back to the source, and may be used for the purpose of
determining a cables length. This technique may also be used for determining the presence and
location of a break in one or both of the cables conductors, since a current will reflect off the wire
break just as it would off the end of an open-circuited cable. Instruments designed for such purposes
are called time-domain reflectometers (TDRs). The basic principle is identical to that of sonar range
finding: generating a sound pulse and measuring the time it takes for the echo to return.
A similar phenomenon takes place if the end of a transmission line is short-circuited as shown
in Figure 2.4-3 and explained in more detail in section 2.5.2.
Now, if we have a finite stretch of line with a characteristic impedance of 50, this will behave
as a resistor to a constant source of DC voltage for the brief time it takes for the signal to reach the
end of the line. If at this end we then connect a resistor equal to the characteristic impedance of the
line, the source will continue to see a 50 load. As we mentioned earlier, there is no difference from
the point of view of the battery between a resistor eternally dissipating energy and an infinite line
absorbing energy, it still sees the same constant load! Reflections are therefore eliminated.
In essence, a terminating resistor matching the characteristic impedance of the transmission line
makes the line appear infinitely long from the perspective of the source, because a resistor has the
ability to eternally dissipate energy in the same way as a transmission line of infinite length is able to
eternally absorb energy.
43
CHAPTER 2 - Conveying Power at Radio Frequency
Figure 2.4-2 This one km transmission Line behaves like a 50 resistor for 10s, then like an open (infinite resistance)
Figure 2.4-3 This one km Transmission Line behaves like a 50 resistor for 10s, then like a short (zero resistance)
Figure 2.4-4 A one km Line terminated with its characteristic impedance behaves like a 50 resistor
44
2.4 Finite Length transmission line
Reflected waves will also manifest if the terminating resistance is not precisely equal to the
characteristic impedance of the transmission line, not just if the line is left unconnected (open) or
shorted. In this case however, only part of the energy will be reflected. We look at this in more detail
in section 2.8.4 and 2.8.5 as well as section 3.3.
We may also simulate the time domain transients illustrated in this section by using the
transient simulation tools in Microwave Office as shown in video 2.1 and simulation project
TDR_video_2_1_and_3_4.emp.
45
CHAPTER 2 - Conveying Power at Radio Frequency
In this section we will be looking at what happens when we apply DC voltage steps and pulses
to open and short-circuited transmission lines.
Firstly we will be using a lumped model for the transmission line which uses lumped
equivalents to represent the distributed inductance and capacitance along the line (Figure 2.3-4),
but for simplicity we will consolidate the inductance in the top and bottom wires into one (Figure
2.5-1). The reader must bear in mind however that the phenomena are still happening along a
distributed capacitance and inductance, and that we only use this model because we do not know
how to draw the distributed case. Also losses are assumed to be negligible throughout this section.
Subsequently we will be looking at what happens at a more physical level, and charge
dynamics as well electromagnetic transients will be illustrated.
In this section we will be using the transmission line model shown in Figure 2.5-1. The
voltage stimulus is a DC step, created by closing the switch at time t=0s.
Let us assume that the battery in this circuit has an internal impedance Zs equal to the characteristic
impedance of the transmission line Z0, that the capacitors in the line are not charged before the
switch is closed and, since the line is open-ended, that the terminating impedance is infinitely large.
When the switch is closed, a voltage wave starts making its way down the transmission line
at the speed of propagation characteristic of the line (eq. (2.3-3)). This is shown in Figure 2.5-2.
46
2.5 Reflection of DC voltage in Transmission lines
E/2
Zs
E E/2
Zero
Figure 2.5-2 The switch is closed and the voltage wave starts propagating down the transmission line
Since Zs = Z0, one-half of the applied voltage (E/2) will appear across the internal battery impedance,
Zs, and one-half across the impedance of the line, Z0. This voltage wave applies a potential of E/2
across the first inductor-capacitor section and hence a current I flows through the inductor to
charge the respective capacitor up to a voltage equal to E/2 volts. This is again shown in Figure 2.5-2.
Once the first capacitor is charged, no further current will flow through it and, as the voltage wave
advances, the current I will flow through second inductor to charge the second capacitor up to a E/2
volts (Figure 2.5-3).
E/2
E E/2 E/2
Figure 2.5-3 As the voltage wave advances, it charges up the second capacitor
This process continues as the voltage wave continues to travel down the line charging each capacitor
through the preceding inductor (Figure 2.5-4).
E/2
Figure 2.5-4 The voltage wave continues to travel down the line
47
CHAPTER 2 - Conveying Power at Radio Frequency
E/2
Figure 2.5-5 Voltage approaches the open end of the line, which can be seen as a capacitor
As the voltage reaches the end of the line, the last capacitor, which represents our open-
circuit termination, also charges to E/2 through the last inductor (Figure 2.5-5). Now theres no
voltage difference across the terminals of the last inductor (Figure 2.5-5) and hence no further
current will flow from the battery into the inductor.
As we know, inductors oppose sudden changes in current while capacitors oppose sudden changes
in voltage. So, just like a capacitor would try to maintain the voltage across its terminals and
discharge gradually until the field and the voltage between the plates have dropped to zero, the
inductor will try to maintain the current flow, in the same direction as the initial current, by means
of the energy stored in its magnetic field (Figure 2.5-6).
Figure 2.5-6 The end of the line is reached, current from the battery stops, but last inductor keeps current flowing
However, just as in the case of the capacitor, where the electric field eventually dies away, the
magnetic field of the inductor collapses until no energy is left in it and current stops flowing into the
end capacitor. What happened to such energy? It has been used to charge the capacitor further
(Figure 2.5-6 and Figure 2.5-7)! But how much further? Well, since the energy stored in capacitors
and inductors is the same16, the capacitors electric field ends up storing twice this energy and hence
the voltage across its terminals doubles (Figure 2.5-7).
16
The transmission line has a characteristic impedance
48
2.5 Reflection of DC voltage in Transmission lines
E/2 E
Figure 2.5-7 The capacitor at the end of the line absorbs the energy of the preceding inductor and charges it to E volts.
The same process then takes place in the penultimate L-C section
Subsequently the same process takes place in the penultimate L-C section (Figure 2.5-7 and Figure
2.5-8) and carries on all the way back down the transmission line (Figure 2.5-9) until the voltage
wave reaches the generator end (Figure 2.5-10).
E/2 E
E E/2 E/2 E E E
E E/2 E E E E
49
CHAPTER 2 - Conveying Power at Radio Frequency
E E E E E
Zero current
We can think of the voltage wave travelling back down the line towards the generator as the
superposition of the incident voltage wave, of amplitude E/2, and a reflected voltage wave of
identical amplitude and polarity. This ties in well with the idea that a voltage wave is fully reflected
by an open termination. The current travelling back down the line on the other hand may be seen as
the superposition of the incident current and a reflected current of the same amplitude and
opposite polarity which add up to zero! This ties in well with the intuitive idea that, as carriers reach
a dead-end they travel back in the opposite direction towards the generator.
50
2.5 Reflection of DC voltage in Transmission lines
In this section we will be looking at what happens along an open circuited line when a pulse
stimulus is applied (Figure 2.5-11). In order for this behaviour to be observed in sufficient details, the
pulse width is must be lower than the time it takes a signal to travel to the end of the line at the
speed allowed by the transmission line.
Zs
As in the previous case, as the switch is closed, our voltage wave starts travelling down the
line, as shown in Figure 2.5-12 and Figure 2.5-13.
E/2
Zs
E E/2
Zero
E/2
E E/2 E/2
As the pulse moves down the line, current loops are formed which charge capacitors further down
the line and discharge the preceding ones thereby allowing the voltage to travel. This is shown in
Figure 2.5-14 and Figure 2.5-15.
51
CHAPTER 2 - Conveying Power at Radio Frequency
E/2
E E/2 E/2
Figure 2.5-14 Current loops charge capacitors further down the line and discharge preceding ones
E E/2 E/2
As the end of the line is reached, as in the previous case, there is no voltage across the
inductor to maintain current flow (Figure 2.5-15). The inductor however tries to maintain current
flow into the capacitor and charges it up to E volts (Figure 2.5-16).
E E/2 E
Figure 2.5-16 Inductor tries to keep current flowing as the end of the line is reached
This is only temporary however because once the current flow due to the energy stored in the
inductor stops, the capacitor will naturally discharge into the penultimate capacitor, which is at
lower voltage, until the voltages are even and back to E/2 across both capacitors. To do this the
current must flow in the opposite direction to the incident one (Figure 2.5-17). Current loops, with
direction of flow opposite to those created by the incident wave, move back down the transmission
line, creating a reflected voltage of the same amplitude and polarity and a reflected current of same
magnitude and opposite polarity travelling back towards the generator.
52
2.5 Reflection of DC voltage in Transmission lines
E E/2 E/2
-I
E/2
E/2
E E/2
-I
E/2
E/2
E E/2
-I
Eventually the pulse travels back into the source and, since Zs = Z0,is fully absorbed (Figure 2.5-20).
53
CHAPTER 2 - Conveying Power at Radio Frequency
E/2
E E/2
-I
The reflected current flows through each shunt capacitor in the same direction as the
current that charged it as the incident wave progressed. However, from the point of view of the line,
incident and reflected currents travel in opposite directions. This is illustrated in Figure 2.5-21.
INCIDENT REFLECTED
CURRENT CURRENT
Figure 2.5-21 Direction of flow of incident and reflected currents through shunt capacitors in an open-circuited line
Voltage is reflected from an open end without change in polarity, amplitude, or shape.
Current is reflected from an open end with opposite polarity and without change in amplitude or
shape.
54
2.5 Reflection of DC voltage in Transmission lines
Let us now look at the phenomena that take place along an open-circuited line from a more
fundamental and physical viewpoint. For simplicity, throughout this section we will stick with the
convention of current being the motion of positive charges. We will also assume that thermal
agitation is negligible. The green dots represent mobile charge carriers distributed evenly
throughout an invisible metallic lattice (Figure 2.5-22).
+
_
Figure 2.5-22 Open-circuited line with no stimulus, charge carriers evenly distributed throughout metallic lattice
As the switch is closed, charge is injected from the generator into the line. This makes the
charges bunch up and creates a sort of a wave of compression which travels, along the signal wire,
from the source towards the end of the line (Figure 2.5-23).
Vs
COMPRESSED
+_ + + + +
Vs
E Field
_ _ _ _
RELAXED
Figure 2.5-23 shows what happens shortly after closing the switch. On the top wire, the charges on
the left-hand side of the voltage wavefront are squeezed together, creating an excess of charge,
whereas on the right-hand side of it the charges are undisturbed. In the lower left portion of Figure
2.5-23, the bottom wire (also called the return wire) exhibits a deficit of charge carriers. This
happens because the battery, which stuffs positive carriers onto the top wire, creating an excess,
must draw those charges from the bottom wire, creating a deficit. In the region to the left of the
travelling wavefront, the excess of charges in the top wire, combined with the deficit of charges in
the bottom wire, creates an electric field (E-field) that points straight down, from the top wire to the
bottom one (Figure 2.5-23).
55
CHAPTER 2 - Conveying Power at Radio Frequency
Vs
COMPRESSED AT REST
Vs +_ + +
Electric Field
_ _
Figure 2.5-24 Charge profile in the voltage wavefront region
Vs +_ + +
Electric Field
_ _
RELAXED AT REST
Figure 2.5-25 Charge profile on the bottom wire corresponding to top wire region shown in Figure 2.5-24
Vs
+_
When the charges reach the end of the line, they hit the end of the wire (Figure 2.5-26)!
These charges carry with them a certain degree of momentum but there is nowhere left to go and
hence they will have to stop, but they cannot stop instantaneously. There is no way to notify the
source that a dead-end has been encountered. The source will therefore continue to pour charges
into the line, piling them up at the end, for some time (Figure 2.5-27).
56
2.5 Reflection of DC voltage in Transmission lines
2Vs
+_
Figure 2.5-27 Reflected voltage wave makes its way towards generator
These charges accumulate to a level twice as high as the incoming signal. That is the precise level of
compression required to arrest the momentum of further incoming particles. This is the so-called
"doubling effect" that happens at the end of an open-circuited transmission line.
There are two facts to remember about current. Firstly by convention it is defined as the
movement of positive charge carriers although it is electrons (i.e. negative charges) that are actually
free to move. Secondly charge carriers do not actually move at the speed of light, although the
knock-on effect of their movement does propagate at the speed of light.
57
CHAPTER 2 - Conveying Power at Radio Frequency
In this section we will be looking at what happens along a short-circuited line when a pulse
stimulus is applied (Figure 2.5-28). In order for this behaviour to be observed in sufficient details, the
pulse width must be lower than the time it takes a signal to travel to the end of the line at the speed
allowed by the transmission line.
Zs
As in the case of the open-circuit, let us assume that the battery in this circuit has an internal
impedance Zs equal to the characteristic impedance of the transmission line Z0 and that the
capacitors in the line are not charged before the switch is closed.
When the switch is closed, a voltage wave starts making its way down the transmission line
at the speed of propagation characteristic of the line (eq. (2.3-2)). This is shown in Figure 2.5-29 and
Figure 2.5-30.
E/2
Zs
+
E E/2
Zero
Since Zs = Z0, one-half of the applied voltage (E/2) will appear across the internal battery impedance,
Zs, and one-half across the impedance of the line, Z0. This voltage wave applies a potential of E/2
across the first inductor-capacitor section and hence a current I flows through the inductor to
charge the respective capacitor up to a voltage equal to E/2 volts. This is again shown in Figure
2.5-29.
Once the first capacitor is charged, no further current will flow through it and, as the voltage wave
advances, the current I will flow through the second inductor to charge the second capacitor up to a
E/2 volts (Figure 2.5-30).
58
2.5 Reflection of DC voltage in Transmission lines
E/2
E E/2 E/2
As the pulse moves down the line current loops are formed which charge capacitors further down
the line and discharge the preceding ones thereby allowing the voltage to travel. This is shown in
Figure 2.5-31.
E/2
+ +
E E/2 E/2
Figure 2.5-31 Current loops charge capacitors further down the line and discharge preceding ones
As the voltage reaches the end of the line, current I flows through the last inductor but it
cannot flow through the last capacitor since its terminals are shorted. This is shown in Figure 2.5-32
where the end capacitor is greyed out to signify that it cannot be charged.
E/2
The current however loops around to find some other place to flow! Since capacitors are
symmetrical elements, the penultimate capacitor, which is not charged at this instant in time, is
more than happy to let current flow through from its lower plate and be charged to a voltage equal
to E/2 (Figure 2.5-34).
59
CHAPTER 2 - Conveying Power at Radio Frequency
This current flows through the capacitor in the opposite direction to the current which initially
charged it to E/2 volts as the incident pulse travelled to the end of the line (Figure 2.5-33).
INCIDENT
CURRENT
REFLECTED
CURRENT
Figure 2.5-33 Direction of flow of incident and reflected currents through shunt capacitors in a short-circuited line
From the point of view of the line however, the reflected loop current travels in exactly the same
direction as the incident one hence, unlike the open-circuit case, the reflected current has the same
amplitude and same polarity as the incident one (Figure 2.5-33). The reflected voltage on the other
hand maintains the same amplitude but has opposite polarity to the incident one. This is the
opposite of what happens in the open circuited case and is shown in Figure 2.5-34.
-E/2
E -E/2
+
Figure 2.5-34 Pulse travelling back towards generator. Reflected current and voltage.
Subsequently current loops are yet again created which allow the pulse to travel back towards the
generator (Figure 2.5-35 and Figure 2.5-36) until the pulse is eventually absorbed by the generator
(Figure 2.5-37).
60
2.5 Reflection of DC voltage in Transmission lines
-E/2
E -E/2 -E/2
+ +
Figure 2.5-35 Current loops are created which allow the pulse travels back towards the generator
-E/2
E -E/2 -E/2
+ +
-E/2
E -E/2
+
61
CHAPTER 2 - Conveying Power at Radio Frequency
In this section we will be using the transmission line model shown in Figure 2.5-38. The
voltage stimulus is a DC step, created by closing the switch at time t=0s.
Zs
As in previous cases, as the switch is closed, our voltage wave starts travelling down the line at the
speed of propagation characteristic of the line as shown in Figure 2.5-39.
E/2
Zs
E E/2
Zero
Figure 2.5-39 The switch is closed and the voltage wave starts propagating down the transmission line
Since Zs = Z0, one-half of the applied voltage (E/2) will appear across the internal battery impedance,
Zs, and one-half across the impedance of the line, Z0. This voltage wave applies a potential of E/2
across the first inductor-capacitor section and hence a current I flows through the inductor to charge
the respective capacitor up to a voltage equal to E/2 volts. This is again shown in Figure 2.5-39.
Once the first capacitor is charged, no further current will flow through it and, as the voltage wave
advances, the current I will flow through the second inductor to charge the second capacitor up to a
E/2 volts (Figure 2.5-40).
E/2
E E/2 E/2
Figure 2.5-40 As the voltage wave advances, it charges up the second capacitor
62
2.5 Reflection of DC voltage in Transmission lines
This process continues as the voltage wave continues to travel down the line charging each capacitor
through the preceding inductor (Figure 2.5-41).
E/2
Figure 2.5-41 The voltage wave continues to travel down the line
E/2
Figure 2.5-42 The current loops around without charging the last capacitor and doubles in magnitude
This current from the last inductor will loop around and start flowing into the penultimate capacitor
but in opposite direction to the current that initially charged this capacitor (Figure 2.5-33). This will
discharge the capacitor and free the carriers trapped inside it. Such carriers will act to increment
the current flowing around the loop and, since the energy stored in the inductor and capacitor is the
same17, the loop current will eventually double as the capacitor is fully discharged. This is shown in
Figure 2.5-43.
17
The transmission line has a characteristic impedance
63
CHAPTER 2 - Conveying Power at Radio Frequency
E/2
E E/2 E/2 0
2I
Figure 2.5-43 The current loops around and flows through the penultimate capacitor so as to discharge it
This process of discharge continues down the line so that a current of 2I amps may be sustained.
This also results in the total voltage along the line progressively flattening down to zero (Figure
2.5-44 and Figure 2.5-45).
E/2
E E/2 0 0
2I
Figure 2.5-44 The process of discharge continues to sustain a current equal to 2I and take the voltage down to zero
E/2
E 0 0 0
2I
Figure 2.5-45 The process of discharge continues to sustain a current equal to 2I and take the voltage down to zero
Eventually the generator sees a voltage equal to zero volts, which is what one would expect
from a short circuit, and a current equal to twice the initial incident current (Figure 2.5-46). This
current may be thought of as the superposition of two currents, an incident and a reflected one,
equal in amplitude and polarity.
64
2.5 Reflection of DC voltage in Transmission lines
E 0 0 0
2I
Figure 2.5-46 The voltage seen by the generator eventually drops to zero
The reflected voltage has the opposite polarity but the same amplitude as the incident voltage.
The reflected current has the same polarity and the same amplitude as the incident current
65
CHAPTER 2 - Conveying Power at Radio Frequency
Let us now look at the phenomena that take place along a short-circuited line from a more
fundamental and physical viewpoint. For simplicity, throughout this section we will stick with the
convention of current being the motion of positive charges. We will also assume that thermal
agitation is negligible. The green dots represent mobile charge carriers distributed evenly
throughout an invisible metallic lattice (Figure 2.5-47).
+
_
Figure 2.5-47 Short-circuited line with no stimulus, charge carriers evenly distributed throughout metallic lattice
As the switch is closed, charge is injected from the generator into the line. This makes the
charges bunch up and creates a sort of a wave of compression which travels, along the signal wire,
from the source towards the end of the line (Figure 2.5-48).
Vs
COMPRESSED
+ + + + +
Vs +_
E Field
_ _ _ _ _
RELAXED
Figure 2.5-48 shows what happens shortly after closing the switch. The voltage has propagated
almost half way down the line. On the top wire, the charges on the left-hand side of the voltage
wavefront are squeezed together, creating an excess of charge, whereas on the right-hand side of it,
the charges are undisturbed. In the lower left portion of Figure (Figure 2.5-48), the bottom wire
exhibits a deficit of charge carriers. This happens because the battery, which stuffs positive carriers
onto the top wire, creating an excess, must draw those charges from the bottom wire, creating a
deficit. In the region to the left of the travelling wavefront, the excess of charges on the top wire,
combined with the deficit of charges in the bottom wire, creates an electric field (E-field) that points
straight down, from the top wire to the bottom one ((Figure 2.5-48).
66
2.5 Reflection of DC voltage in Transmission lines
Vs
COMPRESSED AT REST
Vs +_ + +
Electric Field
_ _
Figure 2.5-49 Charge profile in the voltage wavefront region
Vs +_ + +
Electric Field
_ _
RELAXED AT REST
Figure 2.5-50 Charge profile on the bottom wire corresponding to top wire region shown in Figure 2.5-49
On the bottom wire, the charge carriers are shoved in the opposite direction, peeling them away
from the uniformly-distributed mass of particles to the right, accelerating them toward the battery.
This is shown in Figure 2.5-50.
Figure 2.5-51 Voltage wavefront reaches the short circuited end of the line
At the time when the voltage wavefront reaches the end of the line (Figure 2.5-51), the top
wire has compressed charges all along its length. The bottom wire has a deficit of charges which
have been taken by the battery and shoved into the top wire. The charge carriers making up the
physical wire that creates the short-circuit (shown in green in Figure 2.5-51) experience two effects:
pushing from the top wire and pulling from the bottom.
67
CHAPTER 2 - Conveying Power at Radio Frequency
Bear in mind that the battery at this point is still drawing charges from the bottom wire, hence the
pull. Charges flush through the short circuit, rapidly discharging the excess of charge on the top wire
and repopulating the bottom. These charges experience double acceleration, doubling the current!
This is the "current-doubling" effect that occurs in a short-circuited transmission line. The voltage at
the end point remains zero since there is no difference in the carrier concentration of top and
bottom wires (Figure 2.5-52). This is the exact opposite of what happens in an open-circuited line,
where the voltage doubles, but the current drops to zero.
Vs
Zero Volts
+_
Figure 2.5-52 After the end of the line is reached, current doubles and voltage progressively flattens to zero
By the time the bottom terminal of the battery is reached, the charge carrier density once
again becomes uniform (zero voltage everywhere), but the carriers are still drifting like mad and
hence the doubled current carries on flowing. You may see a simulated animation which confirms
this behaviour in video 2.2.
In this discussion the charge carriers, once set in motion, drift freely along the conductor
forever. This is only possible if the conductors have no resistance. Any practical conductor has some
resistance associated with it but we for simplicity we assume that this creates a negligible loss.
In section 3.4 we will see what happens the line is terminated with impedances which are
either higher (section 3.4.1) or lower (section 3.4.2) than the characteristic impedance of the line.
68
2.5 Reflection of DC voltage in Transmission lines
A line is considered long if it is at least in length. For a 50 Hz AC power system, power lines
would have to exceed 1500km in length before the effects of propagation time became significant.
Cables connecting an audio amplifier to speakers would have to be over 7.5km in length before line
reflections would significantly impact a 10 kHz audio signal!
When dealing with radio frequency systems however, the lengths of transmission lines have
a much greater influence on the operation of the circuit. Consider a 100 MHz radio signal: its
wavelength is 3m if the signal is propagating at the speed of light. This means that even a line which
is 60-75cm in length would be considered long. If the propagation velocity of the line is lower, it
would be considered long for even shorter lengths. For instance if our line was characterised by a
velocity factor of 0.66, this critical length would shrink to 50cm.
The fundamental difference between short and long transmission lines may be summarised by
two points:
- When an electrical source is connected to a load via a short transmission line, the loads
impedance dominates the circuit.
- When a source is connected to a load via a long transmission line, the lines own
characteristic impedance dominates over the load impedance in determining the circuit
behaviour.
In other words, an electrically long line acts as the principal component in the circuit and its own
characteristics have greater influence than the load.
The most effective way to minimize the impact of the length of a transmission line on circuit
behaviour is to match the lines characteristic impedance to the load impedance. If the load
impedance is equal to the line impedance, then any signal source connected to the other end of the
line will see the exact same impedance, and will have the exact same amount of current drawn from
it, regardless of the length of the line. In this condition of perfect impedance matching, the length of
the line only affects the time it takes the signal to reach the load. However, perfect matching of line
and load impedances is not always practical or possible.
We will see the effect that different terminations have on the behaviour of high frequency
circuits in section 2.8 but we will start by looking at a simple example in the next section (2.7).
69
CHAPTER 2 - Conveying Power at Radio Frequency
Whenever there is a mismatch of impedance between transmission line and load, reflections
will occur. If the incident signal is a continuous AC waveform, these reflections will mix with more of
the oncoming incident waveform to produce stationary waveforms called standing waves.
Figure 2.7-1 shows how a triangle-shaped incident waveform turns into a mirror-image
reflection upon reaching the lines unterminated end. The transmission line in this illustrative
sequence is shown as a single, dotted line rather than a pair of wires, for simplicity. The incident
wave is shown travelling from left to right, while the reflected wave travels from right to left.
After the incident signal hits the end of the line, the overall voltage observed on the line will
be the sum of incident and reflected voltages. Figure 2.7-1 shows the step by step transients of the
total voltage along the line up until the reflected voltage reaches the source end. After this transient
time, what we observe on the line is the standing wave shown in Figure 2.7-2. This is representative
of the steady-state voltage along the line, i.e. the sum of incident and reflected voltage waves. This
standing wave oscillates in instantaneous magnitude, but does not propagate down the cables
length like the incident or reflected waveforms causing it. Note that the zero points of the standing
wave, where the incident and reflected waves cancel each other, never change position (Figure
2.7-2).
In the next section we will examine these phenomena in more detail for a range of terminations.
70
2.7 Standing waves and resonance
1
t T
t 0 2 t T
5
1 t T
t T 8
8 1
t T T
8
T
I
M
E
3
1 t T
t T 4
4 1
t T T
4
Incident Voltage
7
3 t T Reflected Voltage
t T 8
8
Total voltage along
the line
Figure 2.7-1 The sum of the incident and reflected waves is a stationary wave. T is the period of the signal
71
CHAPTER 2 - Conveying Power at Radio Frequency
Unterminated
Line
t T
RF Source
1
t T T
8
1
t T T
4
3
t T T
8
1
t T T
2
Figure 2.7-2 The standing wave does not propagate along the transmission line. T is the period of the signal
72
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
We will now look at what happens when we terminate an actual transmission line, with
characteristic impedance Z0 =50, in different impedances. The circuit which we will be analysing is
shown in Figure 2.8-1 and we will be using sinusoidal stimuli from now on.
Z0
Transmission Line
VS ZL
Figure 2.8-2 (a) and Figure 2.8-2 (b) show incident, reflected, and transmitted voltages for an open-
circuit and a short-circuit termination respectively.
Incident
Reflected
(a)
Incident
Reflected
(b)
Figure 2.8-2 Incident, Reflected and Transmitted voltage for a line terminated with (a) an open circuit (b) a short circuit
As is to be expected, if there is a short or open circuit at the end of the line, the entire signal is
reflected.
73
CHAPTER 2 - Conveying Power at Radio Frequency
Incident
Reflected
(a)
Incident
Reflected
(b)
Figure 2.8-3 Incident, Reflected and Transmitted voltage for a line terminated with (a) 25 (b) 100
However, when a finite load, whose impedance does not match the characteristic impedance of the
line, is placed at its end the reflection is not quite as severe. A fraction of the voltage, proportional to
the mismatch between ZL and Z0 gets reflected but the remainder does get delivered to the load.
This is shown in Figure 2.8-3.
In the following subsections the graphs will only show the incident and reflected voltages
along the line to avoid crowding the graphs. However the transmitted voltage is easily derived since
it will have an identical phase to the incident voltage and an amplitude equal to the amplitude
difference between incident and reflected voltage. This is demonstrated in Figure 2.8-3.
Now let us examine each termination in more details by taking a snapshot of incident and
reflected voltages along the line at different instants in time between zero and seconds.
represents the period of the signal travelling down the line.
A useful indicator of the proportion of the signal which gets reflected is the reflection
coefficient . This quantity is extremely useful at Radio and Microwave frequencies and we will
define it and describe it appropriately in section 2.14. is a complex number, however, for the time
being, we will just look at its modulus | | and interpret it as an indicator of the relative magnitudes
of reflected and incident signals. is defined in such a way that when the entire incident signal gets
reflected its modulus is 1 whereas when no reflection occur its modulus is zero. is clearly stated on
every graph in sections 2.8.1 to 2.8.5.
These graphs are snapshots taken from a transmission line animation, designed by the author, which
may be downloaded for free at http://docfrankie.com/index_files/matlab_anim.htm
74
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
In this section we will show snapshots of what is happening in our transmission line at
different instants in time when the line terminated with an open circuit.
In each figure, the top graph shows the incident wave (in red), which travels to the right-
hand side of the page, and the reflected wave (in green) which travels in the opposite direction.
The bottom graph shows total voltage (in blue), which is the sum of incident and reflected waves,
along with its envelope (dashed magenta line).
Note that if we travel an integer number wavelengths, the signals at the two ends of the line have
the same phase!
We have picked two points on the transmitted and reflected waves ( and ) and we will
now see how they move along a line terminated with an open circuit.
t=0s
At t=0s, the incident and reflected voltages are 180 out of phase hence their sum is zero!
It is apparent that when the signal finds an open circuit at the end of the line it gets reflected in its
entirety.
75
CHAPTER 2 - Conveying Power at Radio Frequency
t=0.1 T
Incident and reflected voltages are no longer 180 out of phase as the waves move towards one
another hence their sum is finite and shown in blue on the standing wave plot. represents the
period of the signal travelling down the line.
t=0.19T
When Incident and reflected voltages overlap the standing wave amplitude is at a maximum.
76
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
t=0.31T
As the waves move away from each other, their phase difference increases hence the amplitude of
the standing wave decreases..
t=0.45T
77
CHAPTER 2 - Conveying Power at Radio Frequency
t=0.5 T
..down to a point where it reaches zero (incident and reflected waves are 180 out of phase)..
t=0.61T
78
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
t=0.7 T
t= T
..and finally comes back to zero. Now we are back to where we started!
It is apparent that, although the location of the maxima and minima of the standing wave do
not change along the direction of propagation (hence the standing wave definition), the
instantaneous amplitude of the standing wave does change!
The nulls are spaced at /2 hence, if a line is terminated with an open circuit, one could just
measure the distance between two nulls (which is possible with a slotted line) and work out the
wavelength (and hence the frequency) of the wave travelling down the transmission line.
As we will demonstrate in the next section, a short circuit will also do the trick. It is important to
notice that, although the distance between two peaks could also be measured, it would not be as
reliable a measurement, since the standing wave pulsates and hence the peaks are not stable!
79
CHAPTER 2 - Conveying Power at Radio Frequency
In this section we will show snapshots of what is happening in our transmission line at
different instants in time when the line is terminated with a short circuit.
In each figure, the top graph shows the incident wave (in red), which travels to the right-
hand side of the page and the reflected wave (in green) which travels in the opposite direction.
The bottom graph shows total voltage (in blue), which is the sum of incident and reflected waves,
along with its envelope (dashed magenta line).
Note that if we travel an integer number wavelengths, the signals at the two ends of the line have
the same phase!
We have picked two points on the transmitted and reflected waves ( and )and we will
now see how they move along a line terminated with an short circuit.
t= 0s
At t=0s, incident and reflected voltages are in phase hence their sum is maximum!
It is apparent that when the signal finds a short circuit at the end of the line it gets reflected in its
entirety.
80
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
t=0.13T
Incident and reflected voltages are no longer in phase as the waves move away from one another.
Their sum i.e. the amplitude of the standing wave (shown in blue) therefore decreases. represents
the period of the signal travelling down the line
t=0.23T
As incident and reflected waves continue to move away from one another, the amplitude of the
standing wave keeps decreasing..
81
CHAPTER 2 - Conveying Power at Radio Frequency
t=0.34T
Until it reaches zero (incident and reflected voltages 180 out of phase) and then it flips
t=0.45T
82
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
t=0.5 T
t=0.63T
83
CHAPTER 2 - Conveying Power at Radio Frequency
t=0.74T
until..
t=0.79T
84
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
t= T
Then crosses over and reaches a peak again: we are back to where we started!
Notice that the nulls this time are in a different position to the open-circuited case, although the
amplitude and shape of the envelope of the standing wave is the same.
85
CHAPTER 2 - Conveying Power at Radio Frequency
In this section we will show snapshots of what is happening in our transmission line at
different instants in time when the line is terminated with a matched load which, in this particular
case, is 50 .
When the Line is matched, i.e. terminated with its characteristic impedance Z0 which is also equal to
the internal impedance of the source, no reflections occur and we achieve max power transfer.
There is no standing wave since the instantaneous voltage on the line is equal to the incident voltage.
Maxima and minima are equal to those of the incident wave and move along the transmission line. A
constant envelope (shown by the dashed magenta line on the standing wave plot) would therefore
be observed on a slotted line. (video 3.3)
86
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
When ZL< Z0 some of the power is reflected back to the source and hence a standing wave is
created along the line. Note that, in this case, the incident and reflected voltages DO NOT have the
same amplitude as in the short and open circuit cases. Instead of peaks and nulls we will therefore
observe minima and maxima. This is because part of the wave will indeed be transmitted to the load
and its amplitude will be equal to the amplitude of the incident voltage minus the amplitude of the
reflected voltage. The phase of the load voltage will be identical to that of the incident voltage.
In this section the line is terminated with a 25 load.
t= 0s
Incident and reflected voltages are in phase and hence they add constructively to maximise the
amplitude of the standing wave
t=0.25T
Incident and reflected voltages are 180 out of phase and hence the amplitude of the standing wave
is at a minimum
87
CHAPTER 2 - Conveying Power at Radio Frequency
t=0.5T
Incident and reflected voltages are back in phase after a time equal to half a period (and incident and
reflected waves having travelled half a wavelength each) so we are back to where we started!!
t=0.75T
88
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
t= T
89
CHAPTER 2 - Conveying Power at Radio Frequency
When ZL> Z0 some of the power is reflected back to the source and hence a standing wave is
created along the line. Note that, in this case, the incident and reflected voltages DO NOT have the
same amplitude as in the short and open circuit cases. Instead of peaks and nulls we will therefore
observe minima and maxima. This is because part of the wave will indeed be transmitted to the load
and its amplitude will be equal to the amplitude of the incident voltage minus the amplitude of the
reflected voltage. The phase of the load voltage will be identical to that of the incident voltage.
In this section the line is terminated with a 100 load.
t= 0s
Incident and reflected voltages are 180 out of phase and hence the standing wave amplitude is at a
minimum
t=0.25T
Incident and reflected voltages are in phase and hence they add constructively to maximise the
amplitude of the standing wave
90
2.8 Reflection and Transmission in unterminated, shorted, matched and mismatched line
t=0.5T
Incident and reflected voltages are again 180 out of phase after a time equal to half a period (and
incident and reflected waves having travelled half a wavelength each) so we are back to where we
started!!
t=0.75T
91
CHAPTER 2 - Conveying Power at Radio Frequency
t=T
As in the case of ZL< Z0 , when ZL> Z0, some of the power is reflected back to the source and
hence a standing wave is created along the line. It is apparent that the maxima and minima of the
standing wave are the same as the previous case. This is because the moduli of the reflection
coefficient are the same in both cases. Their signs however are opposite! This is reflected in the
phase difference between the standing waves observed for 25 and 100 terminations and hence
different locations of maxima and minima along the line. An insight into the physical origin of such a
phase difference may be found in section 3.4.
You may also be interested to see 2D and 3D animations of the voltage along a transmission line for
selected load impedances. These may be found at
http://docfrankie.com/index_files/matlab_anim.htm
92
2.9 Transmission Lines Applied to High Frequency Circuits
Z0
RS Transmission Line
ZIN
VS ZL
Figure 2.9-1 Input Impedance of a Transmission line of characteristic impedance Z0 terminated with load impedance ZL
( )
(2.9-1)
( )
Where is the length of the line and is the propagation constant. , which is equal to , is
used to account for the specific speed of propagation at which the signal travels in the transmission
line of interest. Note that, although the speed of propagation does not appear explicitely in the
expression for , at a specific frequency, and may interchangebly be used (eq (2.2-4)).
We will be looking at how equation (2.9-1) is derived in section 3.5 but for now let us just
use it to understand the behaviour of our transmission line.
First of all let us take a look at the ( ) function shown in Figure 2.9-2. Notice that the
periodicity of the trigonometric tangent is just and not 2 as in sine and cosine functions. This
periodicity is one of the features of this trigonometric function which makes it suitable to represent
the impedance of transmission lines, as we will in section 2.10.
93
CHAPTER 2 - Conveying Power at Radio Frequency
Let us now consider some significant values of (since is fixed for a specific transmission
line and frequency) and see how they affect the impedance of our line.
For , equation (2.9-1) becomes
( ) ( )
( )
( )
So we have found that, if a transmission line which is half a wavelength long is placed between
source and load, our transmission line will become transparent. This will be true for every value of
which makes equal to (or an integer multiple of ) and hence ( ) equal to zero. Let us find
the values of for which this is true.
94
2.9 Transmission Lines Applied to High Frequency Circuits
Secondly, since the wavelength changes with frequency, our circuit would only behave as expected
at a specific frequency and perhaps over a narrow band around it which is often undesirable. Thirdly,
when non-deal lines are employed, losses are incurred which are proportional to length. Lastly,
often enough, the load impedance will be some complex quantity and you may want to use equation
(2.9-1) to your advantage to turn the impedance seen by the generator into a more suitable value,
usually the same as the generators internal impedance. As you may recall, the maximum power
transfer theorem, which largely applies at Radio Frequency, states that the power delivered to the
load is maximized when RL=RS. Figure 2.9-3 and Figure 2.9-4 demonstrate this result at DC and AC
respectively.
DC Power Transfer
RES
ID=RS
R=50 Ohm 20
Power (mW) 15
RES 10
DCVS ID=R
ID=V1 R=RL Ohm
V=2 V
5
0
0 20 40 60 80 100
Load Resistance ()
RES
AC Power transfer (50Hz)
ID=RS1
R=50 Ohm 10
8
Power (mW)
6
ACVS
RES
ID=V1
Mag=2 V
ID=R1 4
R=RL Ohm
Ang=0 Deg
Offset=0 V
DCVal=0 V 2
0
0 20 40 60 80 100
Load Resistance ()
Additional details about how to setup the simulations shown in Figure 2.9-3 and Figure 2.9-4 are
given in the videos below
95
CHAPTER 2 - Conveying Power at Radio Frequency
The difference between the case of AC/DC circuits and RF circuits is of course that, in the
latter, a transmission line is used to connect RS and RL and its characteristic impedance must also
be equal to RS and RL in order for maximum power transfer to be achieved. Instead of looking at
three circuit elements however, we could just say that must be equal to RS (Figure 2.9-1) in order
for maximum power to be transferred to the actual load RL and, to this end, equation (2.9-1) helps
greatly! Let us get back to it and explore other interesting values of .
For , the modulus of becomes
( ) ( )
( ) ( )
Now that is a very interesting result, one that you will come across in one form or another, very
often!
( ) (2.9-2)
Equation (2.9-2) shows that, if our quarter-wave line is short circuited at the load end, i.e. ,
the impedance will be equal to infinity. This means that, at the generator end, we will see an
open circuit! Equation (2.9-2) also shows that if , i.e. the line is open-circuited, becomes
equal to zero. This means that at the generator end we will see a short circuit! This impedance
inversion is an extremely useful property of quarter-wave lines! Of course these statements are only
true at the frequency corresponding to the wavelength . Let us now take a look at one important
application of this and its practical realisation.
When we design an amplifier at Radio Frequency, much in the same way as we would at
audio frequency, we need to bias our active device i.e. supply the correct DC voltages for it to
operate in the desired class of amplification. At Radio Frequency it is extremely important to prevent
the amplifiers input and output signals from flowing back into the bias network since this may cause
the amplifier to become unstable, start oscillating and, often enough, burn out!
VDDDC
VOUT
D
VIN
G
S
VGGDC
Figure 2.9-5 A MOSFET Amplifier. The red arrows show RF signal power flowing back into the bias network.
96
2.9 Transmission Lines Applied to High Frequency Circuits
It would also take useful RF power away from the input and output of the amplifier thereby
decreasing its gain and efficiency.
The bias arrangement shown in Figure 2.9-5 is of course greatly simplified and it would normally be
implemented by a resistive network, fed by a single power supply (section 5.3.1), but this will do for
our example.
What we would like to avoid is RF power flowing back into the bias network. This may be achieved if
the RF signal sees a very high impedance at the point where the bias supply joins the signal path. The
RF current would then only be able to flow along the signal path.
The quarter wave line is ideally suited to this purpose. Let us see how.
Z0
Transmission Line
ZIN
As we have seen, a quarter wave line terminated with a short circuit, looks like an open circuit. In the
circuit shown in Figure 2.9-6, is therefore very high! So if we connected this transmission line to
the gate terminal (G) of our device for instance, there would be no chance of any RF current flowing
into it!
Now if we connect the power supply to the other end of the line as shown in Figure 2.9-7, then no
RF current will be able to flow back into the power supply itself. However the DC voltage will still be
able to reach the gate of the transistor since, at DC, a transmission line is just like a wire (section 2.3)
Gate
Z0
Terminal, G
Transmission Line
VGGDC
ZIN
Figure 2.9-7 Quarter-wave line terminated with an RF short circuit, feeds bias voltage to gate
Looking at Figure 2.9-7, you may be thinking, Is he mad? He will short the power supply to
ground!. Well, you would be right, were it not for the fact that the short circuit at the end of the
line (circled in red), is not a DC short-circuit but an RF one. This is another very important concept
which is encountered again and again in RF circuits. We cannot just use a wire connected to ground
to create an RF short because, as we have seen, the length of wire alone would change the
impedance seen by the line. Also, and more importantly, a wire would not be suited to contain the
electromagnetic fields which are created by high frequency signals.
97
CHAPTER 2 - Conveying Power at Radio Frequency
Nevertheless, there is a relatively easy solution to this problem. We need a circuit element which has
a very high impedance at DC and appears as a short18 at high frequency.
Well, a capacitor will do the trick beautifully since its impedance may be expressed as
| | (2.9-3)
Equation (2.9-3) clearly shows that, if , | | will be very high, whereas at high frequency,
provided a suitable value of capacitance is chosen, | | will be small.
Figure 2.9-8 shows the same circuit as Figure 2.9-7 with the practical realisation of the high
frequency short circuit clearly shown.
Gate
Z0
Terminal, G
Transmission Line
VGGDC
This technique allows us to feed a DC bias voltage to any of the terminals of an active device without
RF signals flowing back to the power supply. We will see the importance of avoiding this situation in
our amplifier section, where we will also discuss the limitations of this approach. Let us now look at
an actual, physical example of this arrangement.
/4
lines
GATE DRAIN
The capacitors shown in Figure 2.9-9 are very small and surface mounted so as to minimize any
impedance variation due to their physical dimensions or to parasitic elements. The value chosen for
them was 5.6 pF, which is suitable for the frequency of operation of the amplifier.
18
In practice a total short circuit is not needed for this purpose and a low impedance (<10 ) is usually
sufficient. We will explain this in more details in the amplifier section.
98
2.9 Transmission Lines Applied to High Frequency Circuits
So far we have only looked at voltage waveforms but of course there will also be currents
and they too will create standing waves which will have maxima and minima. The minima will repeat
every half wavelength and so will the maxima, however the phase of current and voltage standing
waves will be different. We will see the rigorous mathematical equations that describe this phase
relationship in section 2.14 but let us first of all look at it from a more intuitive viewpoint.
RF short
circuit
Figure 2.9-10 Voltage and Current Profiles along short-circuited quarter-wave line, VS is 1V peak and 1GHz in frequency
The quarter-wave line in Figure 2.9-10 is terminated with a short circuit hence the voltage at point
must be zero. The current on the other hand will be maximum since the low impedance seen at
allows maximum current flow. At point instead, a large impedance (equation (2.9-2)) will be seen
and hence no current should flow through. The current is therefore at a minimum while the voltage
is at a maximum.
The existence of a phase difference between voltage and current standing waves, which in the case
of open and short circuits happens to be 90, may also be understood intuitively from Figure 2.3-4
(section 2.3). The figure is shown below for the readers convenience.
This figure shows how lossless transmission lines may be modelled by large networks of series
inductors and shunt capacitors. As demonstrated by equations (1.4-1) and (1.5-7) these elements
introduce a phase shift between voltage and current, hence some degree of difference between
voltage and current along a transmission line is also to be expected. I would like to reiterate at this
stage that this phase difference is not always 90 as in the case of open and short circuits.
There are of course perfectly rigorous expressions which describe voltages and currents along a
transmission line, which we will illustrate in section 2.14.
99
CHAPTER 2 - Conveying Power at Radio Frequency
Let us now explore another extremely useful application of quarter-wave lines, the quarter-
wave impedance transformer. Consider the circuit show in Figure 2.9-11
Z0 Z1
ZIN RL
In order to get maximum power transfer to the load must be equal to . By applying this
condition the equation above we obtain
Hence if we are in a position to fabricate our quarter wave line in such way that its characteristic
impedance satisfies the equation below
100
2.9 Transmission Lines Applied to High Frequency Circuits
Now, referring back to Figure 2.9-1, let consider one last interesting length for the
transmission line, .
For , equation (2.9-1) becomes
( ) ( )
( ) ( )
Now, if our line is terminated with a short circuit, i.e. , equation (2.9-1) becomes
This means that, if we measure the input impedance of a line terminated with a short
circuit, its modulus will be equal to the characteristic impedance of our transmission line.
Also if there was an open circuit at the end of the line, i.e. , then
So, yet again, the modulus of gives us the characteristic impedance of the line.
This is a very useful trick when it comes to verifying the impedance of a transmission line. You may
for example design a line by using a formula to achieve a specific impedance value and then wish to
verify its actual impedance either by simulation or direct measurement. To this end you can choose
a frequency ( ) at which your line appears long
(2.9-4)
where is the speed of propagation along the line and is its actual physical length. By measuring
the input impedance of the line and taking its modulus you can then verify the actual impedance
of your line.
101
CHAPTER 2 - Conveying Power at Radio Frequency
As we have seen in previous sections, when a line is mismatched, the voltage at any point
along it becomes the sum of incident and reflected voltages and a standing wave is created. We have
also seen how the input impedance of the line (Figure 2.9-1) may be represented by equation (2.9-1)
for any termination. In this section we will try to give an intuitive explanation as to why in equation
(2.9-1), a trigonometric tangent is a suitable function to represent the line impedance.
Let us consider the case of a line with characteristic impedance terminated with a
load. Let us assume for simplicity that both signals propagate along our line at the speed of
light. Incident and reflected voltages will each be moving at speed in opposite directions towards
one another as shown in Figure 2.10-1.
t= 0s
Figure 2.10-1 Incident, reflected, total voltages and standing wave along a mismatched line at t=0 s
Figure 2.10-2 and Figure 2.10-3 show what happens along the line as time progresses. It is
apparent that by the time point goes past point (Figure 2.10-3), a whole period of the incident
voltage will have slid past a whole period of the reflected voltage. After such a time our total voltage
waveform and hence its envelope will start repeating again. This means that the period of the
standing wave is equal to the time it takes for and to reach the point where their paths cross. As
shown in Figure 2.10-3, the instant in time at which this occurs is equal to half the period of incident
and reflected voltages. A similar argument will hold true for the current. Since our impedance is
ratio of voltage and current and since the envelopes of these quantities vary in a periodic manner
with a period equal to half that of a sine or cosine, the trigonometric tangent, is ideally suited to
represent such an impedance.
102
2.10 Extra bits
t=0.25T
Figure 2.10-2 Incident, reflected, total voltages and standing wave along a mismatched line at t=0.25T s
t=0.5T
Figure 2.10-3 Incident, reflected, total voltages and standing wave along a mismatched line at t=0.5 T s
Another way to see this is through the theory of relative motion. If we were observing the
speed at which point moves away from us while standing at point , we would perceive the speed
of as twice the speed of light. That is because our frame of reference, which is anchored on the
incident voltage is moving at speed and point is also moving at the same speed but in opposite
direction. So relative to one another, incident and reflected voltages are moving at twice the speed
of light and hence we expect the change in their resultant sum to also change twice as fast as each
wave is propagating. This of course does not mean to say that there is anything physically changing
at twice the speed of light but that there are two quantities which are simultaneously changing at
such a speed.
103
CHAPTER 2 - Conveying Power at Radio Frequency
A coaxial line usually comes in the form of a coaxial cable with connectors at either side. A
typical cable structure is shown in Figure 2.11-1.
The metallic shield is used as a ground reference while the signal is carried by the centre core.
Figure 2.11-2 Cross-section of a coaxial cable showing inner radius and outer radius
Figure 2.11-2 shows a cross-section of the coaxial cable and clearly indicates the dimensions
which are used to calculate the characteristic impedance of the line, . The grey area between the
conductors is usually some type of dielectric but air would also work. The dielectric constant of
the material interposed between the two conductors does influence the impedance of the line
therefore this quantity is also included in the calculation. This is shown by equation (2.11-1).
( ) (2.11-1)
Note that in many textbooks, and indeed in Microwave Office, diameters are used in place of radii
for this calculation. However it is apparent that either may be used without the value of being
affected.
104
2.11 Transmission lines Design and Practical Realisation
We must also remember that, when a dielectric is present, the velocity of propagation is
also affected and it usually becomes a fraction of the speed of light, . Such a fraction may be
determined by means of the dielectric constant , as shown by equation (2.11-2).
(2.11-2)
A reduced speed of propagation of the signal also affects the wavelength as you may recall from
section 2.2 and equation (2.2-4). The effective wavelength for the signal travelling along our
coaxial line therefore becomes
(2.11-3)
We will not show the derivation of equation (2.11-1) however this equation is very useful to
understand how the characteristic impedance of the coaxial line is affected by the various
parameters. The first thing to notice is that it is not just or that affects the impedance value, but
their ratio!
Now let us try to reason a little. If we increase the radius of the inner conductor (thereby
decreasing ) we will facilitate the flow of electrons by providing a larger cross sectional area. The
resistance to current flow will therefore be lower which translates into a lower .
Also, increasing reduces the distance between the surface of the inner conductor and the inner
surface of the outer conductor. A shorter distance between the facing surfaces of inner and outer
conductors reduces the voltage between them and hence decreases the impedance. Remember
that such a voltage may be expressed as shown in equation (2.11-4) since the electric field19 is radial
as illustrated in Figure 2.11-3. is the charge density per unit length.
( ) ( ) ( ) (2.11-4)
Figure 2.11-3 Electric field inside a coaxial line. The line is effectively a cylindrical capacitor.
19
The electric field may be easily calculated in this case by using Gausss Law:
105
CHAPTER 2 - Conveying Power at Radio Frequency
When the radius of the outer conductor is increased, the opposite happens: the voltage
between inner and outer conductors increases and hence so does the impedance of the line.
Remember that ( ) is a positive and monotonically increasing function when . This condition
is always verified in this case since is always greater then .
There is also another way to look at this. We can think of our coaxial line as a cylindrical
capacitor. Much in same way as in the case of parallel plate capacitors20, as the distance between
the two conductors increases, the capacitance decreases. In the case of a capacitor, a lower
means higher impedance as shown by equation (1.5-6).
Now, as you may recall from Figure 2.3-4, a transmission line may be represented as a
network of capacitors and inductors so, if we reduce the capacitance value, thereby increasing the
impedance of the capacitors, we would expect the impedance of the line to increase also. This ties in
well with equation (2.11-1) shown in graphical form in Figure 2.11-4 where is plotted versus
at a fixed .
80
60
Z0, ()
40
20
0
1 2 3 4 5 6 7 8 9 10
b/a
Figure 2.11-4 Characteristic impedance of a coaxial line vs the b/a ratio,
Now let us re-examine Figure 2.11-3 and pick up again the analogy with a parallel plate
capacitor. Just as in that geometry, as the dielectric constant of the dielectric between the coaxial
line conductors increases, so does the capacitance . This in turn means that the impedance of the
capacitors which model the line (Figure 2.3-4) decreases and hence so does the characteristic
impedance of the line, . This is yet again confirmed by equation (2.11-1) plotted vs at a
constant in Figure 2.11-5.
From a field viewpoint, the dielectric reduces the electric field by the factor i.e.
This means that, as the dielectric constant increases, the Electric field decreases and consequently,
voltage (equation (2.11-4)) and impedance decrease also.
20
For a parallel plate capacitor , where is the distance between the plates
106
2.11 Transmission lines Design and Practical Realisation
70
60
Z0, ()
50
40
30
20
0 2 4 6 8 10 12
r
Core Dielectric
Dielectric OD
Type Impedance diameter diameter Application
Type mm
(mm) (mm)
PF cable television,
RG-6/U 75 1.0 mm (Polyethylene 1.64 4.7 6.86 satellite television and
Foam) cable modems
PE
Radio communication
RG-58/U 50 0.81 mm (Solid 2.3 2.9 5
and amateur radio
Polyethylene)
Radio communication
and amateur radio,
RG-213/U 50 70.0296 in Cu PE 2.3 7.2 10.3 EMC test antenna
cables. Typically lower
loss than RG58
Video 2.5 shows how to simulate a coaxial cable with specific dimensions and dielectric
material in Microwave Office. It also shows the effect that changing such dimensions has on the
characteristic impedance of the coaxial transmission line.
107
CHAPTER 2 - Conveying Power at Radio Frequency
Coaxial lines are extensively used at RF and Microwave frequencies since they are ideal
structures to contain the electric and magnetic fields generated at high frequency.
This is because they support the TEM21 mode of propagation which means that they keep Electric
and Magnetic fields perpendicular to one another and to the direction of propagation. The speed of
propagation is of course affected by the dielectric medium used inside the line but, since this
medium is homogeneous, such a speed may be easily calculated by means of equation (2.11-2).
Figure 2.11-6 Electric and Magnetic Field inside a coaxial line, the direction of propagation is perpendicular to the page
The problem with coaxial lines is that they are not very compact, cannot be easily integrated
with printed circuits and are hard to prototype. For example, you could not easily construct a
quarter-wave transformer (section 2.9) by means of a coaxial cable and you would need to use
connectors to join the lines.
An easier way to implement transmission lines with a specific characteristic impedance is to
use microstrip lines. Such lines are very simple to construct and their basic structure is shown in
Figure 2.11-7.
They consist of a large sheet of metal, usually copper, called ground plane, with a slab of dielectric
material on top. On top of the dielectric, a strip of metal is used to create the equivalent of the inner
conductor in a coaxial line which will carry the actual signal.
The obvious problem with this type of line is that both magnetic and electric fields will be partially
contained in the dielectric material and partially in the air surrounding the line. This is shown in
Figure 2.11-8.
21
Transverse ElectroMagnetic
108
2.11 Transmission lines Design and Practical Realisation
This means that the signal propagates at different speeds above ( ) and below
( ) the signal line and this influences the direction of the fields also. Nevertheless,
if the cross-section geometrical size of the microstrip line is much smaller than the wavelength of the
signal22, we may assume that the signal propagates approximately in a TEM mode. This type of
propagation is called quasi-TEM.
(2.11-5)
The characteristic impedance of the line may then be calculated by means of equation (2.11-6).
( )
(2.11-6)
{ [ ( )]
Also, the velocity of propagation along the line and effective wavelength become
( ) ( ) (2.11-7)
With microstrip you usually start with a specific characteristic impedance that you aim to
achieve, then pick a substrate with dielectric constant and height and calculate the width
that the line needs to be to achieve the desired impedance. A more useful formula than (2.11-6)
would therefore be one which allows us to calculate given , and . This formula does exist
but it is quite complex!
An easier way to design a microstrip line is to use the TXLine tool included in Microwave Office
shown in figure 2.11-9.
22
As a rule of thumb, in order to prevent higher-order transmission modes affecting the wave propagation,
you should limit the thickness of your microstrip substrate to 10% of a wavelength.
23
In some textbooks two different expressions are given for , when and however the
difference is usually very small and, due to the empirical nature of the equations, often negligible. We follow
the Pozar approach which only uses one formula for
109
CHAPTER 2 - Conveying Power at Radio Frequency
This tool gives us the width of the line , given the parameters of the substrate, the desired
characteristic impedance of the line and the frequency at which the line is to operate. Note that the
physical length of the line changes the phase of the signal but has nothing to do with its
characteristic impedance!
There are also useful graphs which offer a good first approximation to microstrip line design,
these are shown in Figure 2.11-10 and Figure 2.11-11.
110
2.11 Transmission lines Design and Practical Realisation
Figure 2.11-11 Effective dielectric constant as a function of W/h for different dielectric constants
Figure 2.11-13 and Figure 2.11-14, which are based on equation (2.11-6), show how the
characteristic impedance of the line varies with and respectively.
The narrower the width of the line , the smaller the cross section offered to the current to flow
through and hence the higher the impedance.
To understand the effect of the height of the substrate we need to recall how voltage is
actually worked out (section 1.2).
Figure 2.11-12 Integration path and electric field for voltage calculation
Consider Figure 2.11-12. Using equation (1.2-1) and assuming a uniform Electric field underneath the
signal line, we obtain
(2.11-8)
This shows that increasing the height of the substrate increases the voltage thereby increasing the
impedance. Of course this is a rather simplistic explanation and only accounts for the field away
from the edges of the signal line which is similar to that of a parallel plate capacitor. Nevertheless it
gives an intuitive understanding of how the characteristic impedance of the line is related to
substrate height h. In real life however this relationship is not as linear as equation (2.11-8) would
indicate. This is because, at the edge of the line, fringing effects occur and the field lines are partially
contained in air and partially in the dielectric (Figure 2.11-11). This is reflected in Figure 2.11-14
which shows that the characteristic impedance of the line and height of the substrate are related in
a non-linear fashion.
111
CHAPTER 2 - Conveying Power at Radio Frequency
Figure 2.11-15 shows how the effective dielectric constant affects the characteristic
impedance of the line. Figure 2.11-16 shows how there exists a quasi-linear relationship between
the nominal dielectric constant of the substrate and the effective dielectric constant .
Figure 2.11-15 Microstrip Line Impedance vs substrate height for =0.66 and =2
112
2.11 Transmission lines Design and Practical Realisation
As in the case of the coaxial line, the Electric field is inversely proportional to the dielectric
constant and hence as such a constant increases, the field and therefore the voltage and impedance
also decrease. This is shown in Figure 2.11-15.
We must remind the reader that all microstrip equations are approximate. Whats more,
they do not take the thickness of the metal into account, so we would not recommend relying on
them for critical designs on thick copper boards. Having a finite thickness of metal for the conductor
strips tends to increase the capacitance of the lines, which affects the and calculations.
Other parameters which one needs to take into account when selecting a substrate are loss
tangent and thermal conductivity. The loss tangent quantifies the substrate dissipation of
electromagnetic energy. The lower this value the lower the electromagnetic energy lost due to the
substrate material. The thermal conductivity gives an indication of how well heat may be extracted
from the substrate. The Electrical resistivity of the metal which is used for the line also plays a role.
The resistivity is a measure of how strongly a material opposes the flow of electric current.
Microstrip lines are very widely used in modern electronic circuits because they are compact
and can be easily and cheaply fabricated and integrated on printed circuit boards. Let us now see
how Microwave Office and its TXLine tool can help us designing these useful structures.
Video 2.6 illustrates how to design and verify a 50 microstrip line. It also shows how the different
parameters of the line affect its characteristic impedance.
113
CHAPTER 2 - Conveying Power at Radio Frequency
The reader may now wish to watch some illustrative videos which give further insight into the topics
explained so far in this chapter. A brief of the content of this material is shown below.
VIDEO
CONTENT
REFERENCE
Time domain reflectometry part 1. This video tutorial shows how to use the
respective simulation file (TDR_video_2_1_and_3_4.emp) to replicate a time
2.1
domain reflectometer. Pulse propagation and reflection in the case of matched,
open-circuited and short-circuited lines are also illustrated.
The video tutorial uses MWO to show the effect that the various parameters of a
2.5
coaxial transmission line have on its characteristic impedance.
The video tutorial uses MWO to show the effect that the various parameters of a
2.6
microstrip transmission line have on its characteristic impedance.
114
2.12 Video Tutorials and Animations
2.13 Why 50 ?
There are two explanations: one is more practical and one more scientific. In microwave
transmission systems of the 1930s, coaxial transmission lines were initially fabricated in England with
a standard British plumbing pipe. Using a commonly available centre conductor to this pipe led to a
50 characteristic impedance. According to some, that was the birthplace of the 50 standard.
Another explanation is that for minimum signal attenuation, the characteristic impedance of
a coaxial transmission line must be around 77. For maximum power handling capability the
optimum impedance is around 30 (Figure 2.13-1). The arithmetic mean between 30 ohms (best
power handling) and 77 ohms (lowest loss) is 53.5, the geometric mean is 48 ohms. Thus the choice
of 50 ohms is a compromise between power handling capability and signal loss per unit length, for
air dielectric.
However there are exceptions where using a different characteristic impedance makes more sense.
For example, in long-haul cable television systems, where the signal needs to be amplified
repeatedly to overcome cable losses, attenuation is more important than power handling capability.
Accordingly, the cable TV industry uses 75 as a standard, requiring special test equipment, cables,
connectors and other components for their operation [3].
Figure 2.13-1 Power handling and Attenuation vs characteristic impedance. Power is normalised to its maximum value,
attenuation to its minimum value.
115
CHAPTER 2 - Conveying Power at Radio Frequency
2.14.1 Introduction
The ability of electromagnetic waves to propagate comes from two fundamental laws of
physics which are part of Maxwells equations:
- Faradays law, which states that a changing magnetic field causes an electric field, and
- Amperes law, which states that a changing electric field causes a magnetic field
Once these fields are created, they sustain one another and propagate through space at the speed
of light (in free space). This concept was introduced in section 2.1.
What we will be looking at here, are the mathematical models which are used to represent
travelling waves both in free space and in transmission lines. This section by no means aims to
present a comprehensive treatment of this topic which could fill an entire book! But we would like
to give the reader an idea of the maths behind the concepts explained in previous sections and show
how it all ties in.
We will only be considering plane waves propagating in TEM mode (section 2.11.2) i.e.
waves in which electric and magnetic fields are in the same xy plane and perpendicular to one
another and also perpendicular to the direction of propagation z. This is shown in Figure 2.14-1.
x z
As show in Figure 2.14-1, the electric field is a vector, varying in magnitude with time and
whose direction is confined to the y-axis. The electric field, for a wave of frequency , may be
represented mathematically by equation (2.14-1).
( ) [ ] (2.14-1)
Now, as you may recall from section 1.5.4.2, and represent two vectors on the complex
plane, rotating around at the same speed but in opposite directions. Also the envelope of the
resultant sum of these two exponentials has a sinusoidal behaviour (section 1.5.4.2). In this case
however, and are also multiplied by different constants, and , which means that
the amplitude of these rotating vectors may not be the same. In addition, they are both multiplied
by the same exponential which is used to express the dependency of the field on the frequency
of the propagating signal. We may rewrite (2.14-1) as
( ) ( ) ( ) (2.14-2)
116
2.14 The dreaded maths
which shows more explicitly how the speed at which the vectors rotate is affected by both the
frequency of the wave and its speed of propagation . Although does not feature directly in
(2.14-2), if we rewrite (section 2.9) as
(2.14-3)
( ) ( ) (2.14-4)
( )
From (2.14-4) it is apparent that the speed at which the two vectors rotate is related to both the
frequency of the wave, i.e. the rate of change of the amplitude of the field, and the speed of
propagation, .
Equation (2.14-2) makes use of vectors in the complex plane to represent the wave. We may
however wish to represent the wave in the time domain, without resorting to such vectors. This may
be achieved by taking the real part of (2.14-2) by means of Eulers formulae (section 1.5.3.1) and is
shown by equation (2.14-5).
{ ( )} ( ) ( ) (2.14-5)
The magnetic component of our electromagnetic wave may be expressed in a form which is
similar to that used for the electric field in (2.14-1)
( ) )
[ ] (2.14-6)
Equation (2.14-6) is derived by using one of Maxwells curl equations24. In (2.14-6), the and
components are subtracted rather than added, therefore we may expect a phase difference
between magnetic and electric fields.
Note that in this case we cannot use voltage and current to calculate the impedance and
hence electric and magnetic field magnitudes must be used instead
| |
| |
You may be wondering why the exponential associated with the forward travelling field has
a minus sign whereas the one associated the field travelling backwards has a positive exponent.
There is a good explanation for this and if you are not prepared to take this leap of faith please do
refer to reference [8].
24
117
CHAPTER 2 - Conveying Power at Radio Frequency
Plane waves may propagate in free space or be guided along transmission lines. The
principles are very much the same and we will now make this connection and see how plane waves,
propagating in a TEM mode, form the basis of transmission line theory.
In section 2.3 we saw how a transmission line may be modelled by a network of series
inductors and parallel capacitors (Figure 2.3-4). This is a simplified model and does not account for
losses. We will look at how to make it more realistic in section 2.14.2.2 but for now lets keep it
simple and lossless. The model for a short segment of transmission line is shown in Figure 2.14-2,
where L and C represent inductance and capacitance per unit length respectively.
i(z,t)
v(z,t)
z
z
L z
v(z,t) v(z+ z, t)
C z
( )
( ) ( )
( ) ( ) ( )
(2.14-7)
Now taking the limit for in equation (2.14-7), we can obtain the differential form for the
voltage along the line as shown by eq.(2.14-8)(a). Following a similar procedure we can obtain an
analogous expression for the current as shown by eq.(2.14-8)(b).
118
2.14 The dreaded maths
( ) ( )
( )
(2.14-8)
( ) ( )
( )
In this case, where we assume that no losses are incurred along the line, the solution of (2.14-8) for
the voltage will be of the form25 shown in (2.14-9)
( ) [ ] (2.14-9)
The and components of this equation represent the incident and reflected voltages
respectively. This is the equation which is used to model the incident and reflected waveforms which
we saw in section 2.8.
For the current we get a similar solution as shown by eq. (2.14-10)
(2.14-10)
( ) [ ] [ ]
Again the and component represent the incident and reflected currents respectively.
Remember that , where represents the effective wavelength inside the line
(section 2.11, eq. (2.11-3))
As mentioned in section 2.3, eq. (2.3-4), in the lossless case may also be expressed as
is related to the amplitudes of incident and reflected voltages and currents by equation (2.14-11).
(2.14-11)
The minus for the backward wave is significant and is indicative of the fact that the current is moving
in the negative direction, away from the load.
Now, to simplify our analysis, we may assume that we are in sinusoidal steady-state26 at angular
frequency and hence omit the term. This makes (2.14-8) a bit simpler as shown by (2.14-12).
( )
( ) ( ) ( )
(2.14-12)
( )
( ) ( ) ( )
25
We are using the complex exponential representation for our solution as we did in (2.14-1) however the
time domain solution will be similar to (2.14-5) i.e. { ( )} ( ) ( )
26
Sinusoidal steady-state refers to the steady-state that gets established in a circuit when all the independent
sources are sinusoids of the same angular frequency
119
CHAPTER 2 - Conveying Power at Radio Frequency
The solutions of (2.14-12) for voltage (2.14-13) and current (2.14-14) are also simplified as shown
below
(2.14-13)
( )
( ) (2.14-14)
Figure 2.14-3 show a transmission line of length terminated with a load impedance . To simplify
the calculations, the origin of the z-axis, along which the signal propagates, is chosen to be at the
load end.
V ( z ), I ( z )
+
Z0, VL
-l
Figure 2.14-3 Transmission line terminated with a load impedance ZL
The impedance at the load end, i.e. at position , may therefore be expressed by means of
equations (2.14-13) and (2.14-14) as shown below
( )
(2.14-15)
( )
We can then define the load reflection coefficient as the ratio of reflected and incident voltages
as measured at the load terminals and also express it as a function of and , as shown by
(2.14-16)
(2.14-16)
As we will see in section 3.5 this value is very useful and along with the value of for the line, it
allows us to calculate the reflection coefficient at an arbitrary distance from the load.
( )
120
2.14 The dreaded maths
The reflection coefficient , which was just mentioned in passing in section 2.8, will feature
extensively in this treatment since it is a very useful measure for RF engineers.
The model analysed in the previous section is a rather simplistic model and in real life we
would need to use the one shown in Figure 2.14-4 (b). In this model a short length of transmission
line is represented by a lumped element circuit where and are
In particular, is used to account for the finite conductivity of the metal and gives an indication of
the loss in the dielectric material between the conductors.
i(z,t)
(a)
v(z,t)
z
z
R z L z (b)
C z
v(z,t) G z v(z+ z, t)
z
Figure 2.14-4 Voltage and current definitions (a) and lumped-element equivalent circuit (b) for an incremental length of
transmission line
( )
( ) ( ) ( )
( ) ( ) ( ) (2.14-17)
( )
121
CHAPTER 2 - Conveying Power at Radio Frequency
Now taking the limit for in equation (2.14-17), we can obtain the differential form for the
voltage along the line (2.14-18)(a). Following a similar procedure we can obtain an analogous
expression for the current (2.14-18)(b).
( ) ( )
( ) ( )
(2.14-18)
( ) ( )
( ) ( )
In this case, where losses are incurred along the line, the solution to(2.14-8) will be of the form
( ) [ ] (2.14-19)
( ) [ ] [ ] (2.14-20)
( )( )
introduces an exponential decay by adding a real part in the exponent of the exponential. In the
forward travelling wave for instance, in place of we would use where
122
3.1 Introduction
R z L z
C z
v(z,t) G z v(z+ z, t)
z
Figure 3.1-1 Lumped element equivalent circuit for an incremental length of transmission line
The expression for the characteristic impedance of our transmission line therefore becomes
(3.1-1)
Equation (3.1-1) clearly shows the characteristic impedance of lossy lines is frequency dependent.
This is different from the case of lossless lines for which the characteristic impedance is not
dependent on frequency (equation (3.1-2))
(3.1-2)
The other complication that the introduction of loss elements in our model introduces is that,
instead of using a purely imaginary exponent ( ) for our complex exponential (equations (2.14-13)
and (2.14-14)) we must use , which comprises of a real part to account for losses.
(3.1-3)
Now let us take a look at Figure 3.1-2. Notice how, in this figure, we have chosen the z-axis to point
in the opposite direction to the one which we used in section 2.14.2, whilst keeping the origin at the
load end. This is done so that we can use positive values for our transmission line lengths ( ) instead
of negative ones (- ). We will use this direction for the z-axiz from here on. On a mathematical level,
this has the effect of turning every into a in the equations, as you may see by comparing
(3.1-4)and (2.14-19).
123
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
Tx Line (Zo,g)
IL
ZL ZVLL
ZIN
z=l z=0
Figure 3.1-2 Transmission Line Terminated with ZL , incident and reflected voltages
Now let us examine the expression for voltages and currents along the line
( ) (3.1-4)
( ) (3.1-5)
(3.1-6)
( )
( ) (3.1-7)
(3.1-8)
( )
(3.1-9)
( )
Note how both incident and reflected waves must respect the fundamental relationship which exists
between voltage and current at ANY point along the line, i.e. for any value of .
( ) ( ) (3.1-10)
( ) ( )
124
3.2 Reflection Coefficient
Notice how the ratio of the total voltage and total current for the impedance is fixed by itself!
The reflection coefficient measured at the load terminals, , which is just the ratio of reflected and
incident voltages at z=0, may be expressed as
( )
( )
( ) (3.2-2)
You may say, This is all well and good but how can I perform a measurement right at the terminals
of ?. You would be right, most of the time you cant do that and there will be a length of line,
albeit short, between your measurement point and your load impedance. So the question is, If I
dont measure my and right at the terminals of the load how is the reflection coefficient
affected?
Well the answer is quite simple. Your reflection coefficient is a complex quantity which has
magnitude and phase. The magnitude will be fixed once you have picked but the phase will
change depending on where you measure. Let us remember that, at any point along the line, is the
ratio of and and hence ( ) may be expressed as
(3.2-3)
( )
You can see that the modulus of the complex number which represents depends solely on A and B
which are fixed for a specific termination. Its phase however depends on the point along your
transmission line where you are measuring your !
Now using (3.1-3) we can rewrite (3.2-3) as
(3.2-4)
( )
The part represents an exponential decay which accounts for resistive losses whereas the
accounts for phase shifts as mentioned in section 2.14.2.2 and in section 1.5.3.
125
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
Remember how in section 2.9 we expressed the physical length of the line in terms of
fractions of the wavelength ? Well, since the reflection coefficient is a complex quantity which
may be plotted on a polar plot, expressing the length of a transmission line in a form which helps
visualising its location on this type of plot may be very useful. We therefore define a measure, ,
termed electrical length, which is expressed in degrees or radians. Its definition is shown in equation
(3.2-5).
(3.2-5)
With this measure, we are effectively scaling the ratio between and by a or 360 angle. We
will see shortly how this can be very useful when using polar plots to represent the reflection
coefficient.
Lets see what correspondence there is between the physical length of a line (in fractions of
a wavelength) and its electrical length. Let us start with a line long (Figure 3.2-1). From (3.2-5)
we get .
= 180
Figure 3.2-1 Physical and electrical lengths for a line, =50 =100
Now remember that the reflection coefficient measures reflections so, if the transmission line
between measurement point and load has an electrical length of or 180, by the time the incident
signal reaches the load, gets reflected and comes back to the measurement point, it will have
travelled twice the length of the line i.e. or 360. This is why the electrical length is multiplied
by 2 in the exponent of the exponential in equation (3.2-4)!
This means that an electrical length of 180, gets the reflection coefficient to rotate by 360 and
hence takes us all the way around the polar plot thereby making travel back to the starting point. It
is just as if the line was invisible! This was also shown in section 2.9 and is further illustrated in Figure
3.2-2 and Figure 3.2-3.
126
3.2 Reflection Coefficient
Graph 1 S(1,1)
Mag Max Swp Max Schematic 1
90
75
105
1 1000 MHz
60
12
0
45
13
5
30
15
0
1000 MHz 15
165 Mag 0.3333
Ang 0 Deg
0
-180
-15
-165
-3
0
50
-1
-4
5
35
-1
-6
20
0
-1
-105
-75
0.2 Swp Min
-90
Figure 3.2-2 Reflection coefficient when transmission lines physical length , electrical length =0, =100
Graph 1 S(1,1)
Mag Max Swp Max
Schematic 1
90
75
105
1 1000 MHz
60
12
0
45
13
5
30
15
0
1000 MHz 15
165 Mag 0.3333
Ang 0 Deg
0
-180
-15
-165
-3
0
50
-1
-4
5
35
-1
-6
20
0
-1
-105
-75
Figure 3.2-3 Reflection coefficient when transmission lines physical length , electrical length =180, =100
In a similar fashion we can work out that the electrical length for a line is 90 (Figure
3.2-4) and that this line gets the reflection coefficient to rotate 180 around the polar plot (Figure
3.2-5).
Note that this rotation is in the clockwise direction and that this comes from the negative sign of the
exponent in equation (3.2-4).
127
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
= 90
Figure 3.2-4 Physical and electrical lengths for a line, =50 =100
Graph 1 S(1,1)
Mag Max Swp Max
Schematic 1
90
75
105
1 1000 MHz
60
12
0
45
13
5
30
15
0
1000 MHz 15
165 Mag 0.3333
Ang -180 Deg
0
-180
-15
-165
-3
0
50
-1
-4
5
35
-1
-6
20
0
-1
-105
-75
Figure 3.2-5 Reflection coefficient when transmission lines physical length , electrical length =90, =100
We can also work out that the electrical length for a line is 45 (Figure 3.2-6) and that,
over such a length, travels around the polar plot by 90 (Figure 3.2-7).
= 45
Figure 3.2-6 Physical and electrical lengths for a line, =50 =100
128
3.2 Reflection Coefficient
Graph 1 S(1,1)
Mag Max Swp Max
Schematic 1
90
75
105
1 1000 MHz
60
12
0
45
13
5
30
15
0
15
165
0
-180
1000 MHz
Mag 0.3333
Ang -90 Deg -15
-165
-3
0
50
-1
-4
5
35
-1
-6
20
0
-1
-105
-75
0.2 Swp Min
-90
Figure 3.2-7 Reflection coefficient when transmission lines physical length , electrical length =45, =100
Also note that must be different from to get a reflection! If it isnt, from (2.14-16)
shown below, we get a modulus of zero for the reflection coefficient.
| |
We are still rotating around the polar plot, but along of circle with a radius of zero! The rotation will
therefore be imperceptible and we will remain right in the centre of the plot.
There is another important point to make. In the cases illustrated above, our mismatch was
determined by a load impedance =100 which was greater than the characteristic impedance of
the line =50 This ensures that the ratio of B and A in equation (3.2-6), is positive.
(3.2-6)
Ignoring losses in eq. (3.2-4), the reflection coefficient for =100 can be expressed as
( ) (3.2-7)
However if we had chosen a termination , the ratio of A and B would have been
negative. For instance if we get
(3.2-8)
Now if you recall our treatment of complex exponentials (section 1.5.4), you will recall that we may
write
( ) ( ) (3.2-9)
129
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
(3.2-10)
If we ignore losses in eq. (3.2-4), we can rewrite the reflection coefficient for =25 as
( ) (3.2-11)
Comparing eq. (3.2-11) and (3.2-7), we can see that although the magnitude of is identical for both
terminations, when =25 the phase of the reflection coefficient is rotated clockwise by an extra
rad or 180. If we examine the simulation results for when =25 with those obtained when
=100, we can clearly see this. Compare Figure 3.2-2 and Figure 3.2-8, Figure 3.2-3 and Figure
3.2-9, Figure 3.2-5 and Figure 3.2-10, Figure 3.2-7 and Figure 3.2-11 and you will notice this 180
phase shift.
Graph 1 S(1,1)
Mag Max Swp Max Schematic 1
90
75
105
1 1000 MHz
60
12
0
45
13
5
30
15
0
1000 MHz 15
165 Mag 0.3333
Ang 180 Deg
0
-180
-15
-165
-3
0
50
-1
-4
5
35
-1
-6
20
0
-1
-105
-75
Figure 3.2-8 Reflection coefficient when transmission lines physical length , electrical length =0, =25
130
3.2 Reflection Coefficient
Graph 1 S(1,1)
Mag Max Swp Max Schematic 1
90
75
105
1 1000 MHz
60
12
0
45
13
5
30
15
0
1000 MHz 15
165 Mag 0.3333
Ang -180 Deg
0
-180
-15
-165
-3
0
50
-1
-4
5
35
-1
-6
20
0
-1
-105
Figure 3.2-9 Reflection coefficient when transmission lines physical length , electrical length =180, =25
Graph 1 S(1,1)
Mag Max Swp Max Schematic 1
90
75
105
1 1000 MHz
60
12
0
45
13
5
30
15
0
1000 MHz 15
165 Mag 0.3333
Ang 1.146e-010 Deg
0
-180
-15
-165
-3
0
50
-1
-4
5
35
-1
-6
20
0
-1
-105
-75
Figure 3.2-10 Reflection coefficient when transmission lines physical length , electrical length =90, =25
131
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
Graph 1 S(1,1)
Mag Max Swp Max Schematic 1
90
75
105
1 1000 MHz
60
12
0
45
13
1000 MHz
5
Mag 0.3333
30
15
0
Ang 90 Deg
15
165
0
-180
-15
-165
-3
0
50
-1
-4
5
35
-1
-6
20
0
-1
-105
Figure 3.2-11 Reflection coefficient when transmission lines physical length , electrical length =45, =25
Video 3.1 and video 3.2 illustrate this further with the help of Microwave Office.
132
3.3 Voltage Standing Wave Ratio (VSWR)
The figures in section 2.8 can help us understand the reflection coefficient a bit better.
Consider Figure 3.3-1 for instance. The quoted reflection coefficient, , has a modulus of 0.33. This
means that one-third of the incident voltage gets reflected and two-thirds are delivered to the load.
Figure 3.3-1 Incident, reflected and total voltages for a 50 line terminated by a 100 load
From Figure 3.3-1, it is also apparent that, instead of using the modulus of the reflection
coefficient as an indication of how much voltage gets through to the load, we could look at the
maximum and minimum amplitude of the envelope of the total voltage along the line (dashed
purple curve). This is commonly done and this measure, which represents the ratio of maximum and
minimum amplitude of the voltage standing wave, is called VSWR
Of course the maximum voltage along the line will be observed when incident and reflected voltages
are in phase and add up (A+B), Figure 3.3-2, whereas the minimum voltage will be observed when
such voltages are 180 degrees out of phase (A-B), Figure 3.3-1.
Lastly, let us point out that, as is to be expected, there exist a clear relationship between the
modulus of the reflection coefficient and the VSWR
| |
Time to play with this on Microwave Office and watch video 3.2 which illustrates this further.
133
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
Figure 3.3-2 Incident, reflected and total voltages for a 50 line terminated by a 100 load, and are in phase
134
3.4 Mismatched lines - A Physical viewpoint
Vs 1 V
Vs + __
3 S
Zero Volts
Vs +_
Comparative
Charge Spacing
Figure 3.4-1 Charge distribution and voltage profile for mismatched line with
135
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
We can see the total voltage along the line as the sum (or superposition) of incident and
reflected voltages of amplitudes A and B respectively. Since the overall voltage established on the
line is higher than the source voltage, we can infer that A and B have the same polarity and add to
increase the voltage along the line. To work out how much the voltage rises by, we just use equation
(3.2-6).
We can also look at things from a circuital point of view by using an equivalent electrical
model for our lossless line (Figure 3.4-2). As we get to the end of the line we encounter a load which
opposes greater resistance to current flow than the transmission line hence some current Ir is
reflected. Ir flows back into the last capacitor thereby increasing its charge and its voltage up
to E. This process continues in a similar fashion to the open-circuited case illustrated in section
2.5.1.1.1 and allows the higher voltage to propagate back to the source. Whist the current Ir acts to
increase the overall voltage along the line, it also acts decrease the overall current, since it has
opposite sign to the incident current. This is shown in Figure 3.4-2.
E 1 4
E E E
3 3
1
4
Ir= I
E 3
2E E E E ZL
3
I 1 2
I- I I
3 3
Figure 3.4-2 Reflected voltage travels back towards the generator and adds to the incident one
To determine the overall current, we need to work out the current reflection coefficient. The
expression for it is very similar to that for the voltage reflection coefficient (eq. (3.2-3)), but has
opposite sign. From eqns. (3.1-8) and (3.1-9)
( )
( )
136
3.4 Mismatched lines - A Physical viewpoint
Vs 1V
Vs - __
3 S
Zero Volts
Vs +_
Comparative
Charge Spacing
Figure 3.4-3 Charge distribution and voltage profile for mismatched line with
We can also look at things from a circuital point of view by using an equivalent electrical
model for our lossless line (Figure 3.4-4). As we get to the end of the line, we encounter a load which
opposes less resistance to current flow than the transmission line hence some additional current is
allowed to flow. The charges which make up this current must come from somewhere and that
somewhere is the last capacitor which discharges a little so as to allow an increase in current
through our load. This decreases the voltage across the capacitor to E. The current in turn
increases to I. This is shown in Figure 3.4-4.
137
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
E 1 2
E E E
3 3
2 1
E E E E I+ I ZL
2E 3 3
I 1 4
I I I
3 3
Figure 3.4-4 Reflected voltage travels back towards the generator and adds to the incident one
We may also simulate the time domain transients illustrated in this section by using the transient
simulation tools in Microwave Office as shown in video 3.4
The final remark of this section is on power. From the calculation below you may see that it is
inevitable to lose some power if the line is not matched!!
is the power delivered to the load when the line is matched. and are the powers
delivered to a 100 and 25 load respectively.
138
3.5 Reflection Coefficient VS Impedance
We have seen how, in the case of a lossless transmission line, the reflection coefficient may
be expressed as shown by eq. (3.2-4)
( )
( ) ( ) (3.5-1)
(3.5-2)
( ) ( )
Now using (3.5-1) and (3.5-2) we may work out the impedance at distance from the load
( ) ( ) (3.5-3)
( )
( ) ( )
This shows a direct correspondence between ( ) and ( )! In section 3.6 and 4.4 we will
see how this correspondence can be extremely useful and how it allows a conversion chart to be
superimposed on the polar plot of the reflection coefficient to convert its value to the corresponding
impedance. This conversion chart is called Smith chart and we will be using it extensively
throughout this treatment.
We should also point out two important and useful relationships expressed by eq (3.5-4) and (3.5-5).
(3.5-4)
( )
(3.5-5)
139
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
Using the equations analysed in this section, an expression for the impedance which does
not directly feature the reflection coefficient may also be derived, as shown in eq. (3.5-6). This is
essentially the same as equation (2.9-1) which we used extensively in section 2.9.
( ) (3.5-6)
( )
( )
( ) (3.5-7)
( )
( )
140
3.6 The Smith Chart
+j1
+j0.5 +j2
+j0.2 +j5
0.2
0.5
2
1
5
0.0
-j0.2 -j5
-j0.5 -j2
(a) (b)
-j1
+j0.5 +j2
+j0.2 +j5
0.2
0.5
2
1
0.0
-j0.2 -j5
-j0.5 -j2
27
This is always true for passive networks but not always true for active circuits!
141
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
The impedance chart shown in Figure 3.6-1 (b), is called Smith chart. We will look at how it is
constructed in more detail in section 4.4 but for now let us just look at how it works!
Remember how the complex number which represents any arbitrary impedance may be expressed
in Cartesian form (1.5.3.1) as
The values of and are the values which we can plot on (or read from) the Smith chart.
The circles which intersect the central axis of the chart (Figure 3.6-3) are used to represent the real
part of our normalised impedance, .
+j1
+j0.5 +j2
+j0.2 +j5
0.5
0.2
2
1
0.0
-j0.2 -j5
-j0.5 -j2
-j1
+j1
+j0.5 +j2
+j0.2 +j5
0.5
0.2
2
1
0.0
-j0.2 -j5
-j0.5 -j2
-j1
Figure 3.6-4 Smith Chart peripheral values represent reactance (the imaginary constant j is displayed for clarity)
142
3.6 The Smith Chart
The curves leading up to the values at the periphery of the chart are used to represent the reactance
of our normalised impedance, (Figure 3.6-4). Note how, in Figure 3.6-4, we have actually shown
the imaginary part of the impedance, which features the imaginary constant , not just . This was
done to increase clarity however usually on a Smith chart, the imaginary constant is not displayed
and the values of the reactance alone are shown.
Let us illustrate the usefulness of the Smith chart with an example. Suppose that we have
measured a reflection coefficient with a magnitude of 0.45 and an angle of 63.5 across an unknown
load. Also assume that our measurement system uses transmission lines with a characteristic
impedance of 50. Figure 3.6-5 shows this reflection coefficient on a polar plot. We now overlap our
impedance mask to transform our polar plot into a Smith chart (Figure 3.6-6).
63.5
( )
We will be talking about the Smith Chart in a lot more detail in section 4.4 and we will see just how
excellent a tool it has been for design engineers.
143
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
+j1
+j0.5 +j2
0.5
0.2
2
1
5
0.0
-j0.2 -j5
-j0.5 -j2
-j1
(b)
Figure 3.6-6 Impedance corresponding to reflection coefficient 0.4563.5
144
3.6 The Smith Chart
The reader may now wish to watch some illustrative videos which give further insight into the topics
explained so far in this chapter. A brief description of the content of these video tutorials is
provided below.
VIDEO
CONTENT
REFERENCE
This video demonstrates how to set up simulations to measure and display the
reflection coefficient for a simple circuit of the type shown below.
l ,
ZIN (l)
3.1
This tutorial illustrates the basic principles of operation of a slotted line and how
to set up a simulation that may emulate it. It also shows the voltage and
3.3
impedance profiles along the length of the line in both a matched and
mismatched case.
Time domain reflectometry part 2. This video tutorial shows how to use the
respective simulation file (TDR_video_2_1_and_3_4.emp) to replicate a time
3.4 domain reflectometer. Pulse propagation and reflection in the case of a line
mismatched with finite load impedances which are lower ( ) or greater
( ) than the characteristic impedance of the line, are shown.
It is recommended that the readers should try to set up similar simulations to those illustrated in the
video tutorials and explore this type of circuits. This will help them gain further insight and perhaps
find the answers to any doubts or questions that they may have.
145
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
In section 3.2 we saw how the reflection coefficient may be expressed as the ratio of incident
and reflected voltages or incident and reflected currents. These are not always easy to measure at
RF and microwave frequencies in particular when the line is terminated with open or short circuits!
However there is a quantity which is much more easily measured at high frequency and that is
power. Now our reflection coefficient may also be expressed in terms of incident and reflected
power as shown in (3.8-1)
(3.8-1)
We can easily see how this equation is equivalent to the one which we worked out for the voltage
(3.2-3).
(3.8-2)
In (3.8-2), is the normalised incident voltage wave and is the normalised reflected voltage wave.
These are defined as shown in (3.8-3) and allow us to make voltage waves independent of the
characteristic impedance of the line.
(3.8-3)
So far we have been looking at one port systems like that depicted in Figure 3.2-1 however
we often deal with two port networks such as filters and amplifiers. It would be good therefore to
define two sets of parameters, one for each port, which would allow us to understand what voltage
and current are doing at each port. We can therefore define and waves at both input ( , )
and output ports ( , ) as shown in Figure 3.8-1.
We can then work out a set of equations which shows how the reflected wave at port 1, , depends
on both incident wave at port 1, , and incident wave at port 2, . We can do the same for the
reflected wave at port 2, .
(3.8-4)
[ ] [ ][ ]
(3.8-5)
146
3.8 Two-Port Networks and S-parameters
b1 b2
a1 a2
ZS = Z0
ZTLin
OUTPUT
Two-port ZTLout
INPUT
= =
VS Network ZL
Z0 Z0
Figure 3.8-1 Two-port network with setup to measure and . Input and output transmission lines have negligible
electrical lengths.
But how do we work out these coefficients? Let us start with and . To measure these two we
keep the voltage source on port 1 as shown in Figure 3.8-1 and then set ZL to Z0.
Depending on whats in our network, we may get some power reflected back out of port one, and
the ratio between reflected and incident power at port 1 is our as shown in eq. (3.8-6). Note that
28
is the same as for a single port network!
(3.8-6)
The parameter measures the ratio between the power that comes out of port 2, in response to a
stimulus on port 1 (eq. (3.8-7)). This is not power reflected from an input at port 2 because we have
set by terminating the port in its characteristic impedance Z0. We can therefore see it as
power transmitted through the network, from port 1 to port 2 when port 2 is terminated with an
impedance ZL equal to Z0. If the two-port network was an amplifier, it would tell us the gain. If it was
a filter, it would tell us the insertion loss and allow us to profile the passband.
(3.8-7)
Now for and , we rejig things around a bit as shown in Figure 3.8-2.
28
Note that, in MWO, even for a single port network, an measurement is what must be set up to obtain !
147
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
b1 b2
a1 a2
ZS = Z0
ZTLin
OUTPUT
Two-port ZTLout
INPUT
= =
ZL Network VS
Z0 Z0
Figure 3.8-2 Two-port network with setup to measure and . Input and output transmission lines have negligible
electrical lengths
We have effectively switched our source to port 2 and terminated port 1 with a matched
termination i.e. ZL=Z0.
(3.8-8)
(3.8-9)
represents the reflection coefficient at port 2 (eq.(3.8-9)) and measures the power
transmitted to port 1 from port 2 (eq.(3.8-8)).
A note on the indexing of S-parameters. The first index represents the port which the power is
travelling to, the second index represents the port which the power is originating from.
S-parameters are effectively ratios of normalised voltage waves (eq. (3.8-2),(3.8-6) and
(3.8-7)) and if we square their magnitudes we obtain the normalised power. In the case of a lossless
two-port passive network, since power can only be transmitted or reflected we can say
| | | | (3.8-10)
So far we have assumed that the input and output transmission lines (ZTLin and ZTLout) which
we have used to connect to our network and measure incident and reflected powers (Figure 3.8-1
and Figure 3.8-2), are very small and that their electrical length is negligible. This however is not
often the case and we need to calibrate out such lengths if what we want is the S-parameters right
at the network ports. This process is called de-embedding or shifting the reference planes. Figure
3.8-3 illustrates this.
148
3.8 Two-Port Networks and S-parameters
ZS = Z0
OUTPUT
Two-port
INPUT
VS ZTLin = Z0 ZTLout = Z0 ZL
Network
1=l1 2=l2
Figure 3.8-3 Two-port network setup to measure and . Input and output transmission lines have electrical
lengths which cannot be neglected
As we saw in section 3.2, eq. (3.2-3), adding a length of line means that the phase of the
reflection coefficient changes according to the electrical length of such a line. This case is no
different, we just have two ports instead of one. The spatial coordinate (or position along the line) at
which the S-parameters are measured, is called a reference plane. Two reference planes are shown
for both input (port 1, port 1) and output (port 2, port 2) in Figure 3.8-3.
Suppose that we are characterising a transistor or a filter and that we are measuring the S-
parameters at x1=0 and x2=0. We would like to make our measurement independent of the lengths
of line that we have use to characterise our network so that, if we sell our product, in the datasheet
we will be able to specify the S-parameters right at the input and output ports!
This transformation may be achieved as shown below.
At the reference planes x1= and x2= (port 1 and port 2) we may write
( ) ( )
[ ] [ ][ ]
( ) ( )
( ) ( )
[ ] [ ][ ]
( ) ( )
Now we can relate waves at port 1 and port 1 by means of the electrical length that separates them.
( ) ( )
( ) ( )
149
CHAPTER 3 - Foundations of RF & Microwave Circuit Characterisation
We can also relate waves at port 2 and port 2 through the electrical length that separates them,
( ) ( )
( ) ( )
( )
[ ] [ ( )
]
which provides the relationship between S-parameters at two sets of reference planes.
Two-port S-parameters are an extremely important tool when it comes to transistor characterisation
and amplifier design, as we will see in chapter 5.
150
4.1 Introduction
4 Impedance Matching
4.1 Introduction
In this section we will take a step back first and revisit the maximum power transfer theorem
when passive components characterised by complex (capacitors, inductors) and real (resistors)
impedances (section 1.5.3) are present in the circuit.
As you may recall from Figure 2.9-4, in the case of resistive circuits, maximum power transfer
is achieved when load and source resistances are equal. What happens however when source and
load impedances are complex?
This case is similar to the resistive one, in that the real part of the source and load impedances must
be the same however the sign of their imaginary parts must be opposite. In mathematical terms the
load impedance must be the complex conjugate of the source impedance.
(4.1-1)
Lets see why this is. Our aim is to convey as much power to our load as we can. Any reactive
elements which the generator sees, will act to store some of the useful energy which could instead
be delivered to the load. If the load and source impedances are equal to the conjugate of one
another, then the reactive elements will be virtually invisible to the generator and all the useful
power will be delivered to the resistive load. This is shown in Figure 4.1-1, where eq. (4.1-1) is
satisfied when .
RS jXS RS
ZS -jXL
ZL
RL
RL
29
Figure 4.1-1 Source impedance driving its complex conjugate and resulting equivalent circuit
Often enough the signal generator has a purely resistive impedance which, in many RF and
Microwave systems is 50 The load to which it is connected however is often characterised by a
complex impedance. This could be the input port of a transistor for instance. This is shown in Figure
4.1-2. We may therefore need to add some additional elements between our source and load to
ensure that maximum power transfer is achieved. This is called a matching network and its purpose
is to make the load impedance look like the complex conjugate of the source impedance (Figure
4.1-3).
29
The impedance of reactive components is frequency dependent. Here we are assuming a sinusoidal steady
state.
151
CHAPTER 4 - Impedance Matching
CAP
ID=C1
RES
RS = 50 CL = -1j C=1e-6 uF
ID=R1
R=1 Ohm
ACVS
RES
ID=V1
ID=R2
Mag=1 V RL = 50 R=1 Ohm
Ang=0 Deg
Offset=0 V
VS ZL =50-j
DCVal=0 V
Figure 4.1-2 Load impedance (50 j ) does not match source impedance (50 )
IND
ID=L1
L=1 nH
RES RS = 50 LL = 1j CL = -1j
ID=R1
R=50 Ohm
ACVS
RES
ID=V1
ID=R2
Mag=1 V
Ang=0 Deg VS ZL = 50 RL = 50
R=50 O
Offset=0 V Matching
DCVal=0 V
Network
Figure 4.1-3 Matching network transforms load impedance, as seen by source terminals, into source impedance.
When it comes to impedance matching there are three main combinations of load and source
impedances which we may encounter:
Although it is quite easy to just regurgitate established procedures and the respective equations,
throughout this section we will try to explain things from a conceptual point of view, as we have
done throughout this treatment. To this end we will first start with a review of voltages and currents
across parallel and series combinations of resistive and reactive components (RLC networks).
Subsequently we will illustrate the techniques employed to design matching networks for purely
resistive impedances (section 4.3) and then for any arbitrary complex impedances (section 4.4).
152
4.2 Impedance and Admittance
4.2.1 Introduction
In section 1.5.4.3 we saw how the behaviour of a set of resistive and reactive elements may
be simply defined by its complex impedance. This complex number allows us to work out the ratio of
voltage and current amplitudes and also the relative phase of voltage and current for the series
connection. When looking at elements connected in parallel however, it is advantageous to use
another quantity, the admittance, to simplify circuit analysis. In this section we will be picking up the
concepts illustrated in section 1.5.4.3 and 1.5.5 and extending them to more complex passive
networks.
Before we proceed we must revisit a very important concept which will be used extensively
throughout this section, the relative phase of sinusoids in the time domain.
When in a sinusoidal steady state, we do not need to express frequency of our signals explicitly, all
we need to know is their amplitude and phase. So, much in the same way as we did for impedances,
we can use complex numbers to represent the phase and amplitude of our sinusoidal signals by
means of one single number. This is shown in Figure 4.2-1.
Im
A
A/2
Re
Figure 4.2-1 Phasor representation on the complex plane of sinusoids of different amplitude and phases
As we know, we may represent complex numbers as vectors in the complex plane. These
are usually referred to as phasors. This type of representation helps us work out the phase
difference between voltage and current and which of the two leads the other! Phasor
representations of voltage and current for resistors, capacitors and inductors are shown on the right-
hand-side of Figure 4.2-4, Figure 4.2-5 and Figure 4.2-6.
In the time domain, things are a little less clean-cut and I find that students often interpret
things the wrong way around. Let us take a look at the curves plotted in Figure 4.2-2. The sinusoids
represented in this graph may be described by equations of the type shown below
( )
where for the dashed, blue curve and for the solid red curve.
153
CHAPTER 4 - Impedance Matching
y = sin
( + )
1.5
--- 1= 60
1 ___ 2= 0
0.5
+ )
0
sin
(
-0.5
-1
-1.5
0 90 180 270 360
sin
( + )
Now the mistake that many students make is that they see that the red curve is further along the
abscissa than the blue curve. They assume therefore that the red curve leads.. This is WRONG!
There is an easy way to work out which one leads from this graph. The first one that reaches its peak
or, equivalently, crosses the x-axis, is the one that leads. In general, you just look at equivalent
points for each sinusoid across the time period, which is the same for both, and the one that reaches
them first is the leading one. These two sinusoids may be represented in a vector form as shown in
Figure 4.2-3.
y1
1- 2 = 60
A=1
y2
A
Figure 4.2-3 Phasor representation of sinusoid with identical amplitudes and a phase difference of 60
Let us now remind ourselves of current and voltage relationships for resistors, capacitors
and inductors. These are shown both as waveforms in the time domain and as vectors. As shown in
Figure 4.2-4, the current through a resistor and the voltage across its terminals are in phase. In the
case of capacitors, the current leads the voltage by 90 as shown in Figure 4.2-5. Finally in the case
of inductors the voltage leads the current by 90 (Figure 4.2-6).
154
4.2 Impedance and Admittance
V mA
Resistor IV
1 40 Vtime(M_PROBE.VP1,1)[*] (L, V)
Resistor IV
p2
p1
0 0
V
-0.5 -20 I
-1 -40
0 0.5 1 1.5 2
Time (ns)
V mA
Capacitor IV Vtime(M_PROBE.VP1,1)[*] (L, V)
1 40
Capacitor IV
RES M_PROBE
ID=Rs ID=VP1 Itime(ACVS.V1,1)[*] (R, mA)
R=50 Ohm 0.5 p1 20 Capacitor IV p1: Freq
p2: Freq
ACVS 0 0
ID=V1
Mag=2 V
Ang=0 Deg CAP
I
Offset=0 V ID=C1
C=5.6 pF -0.5 -20
DCVal=0 V
p2
-1 -40
0 0.5 1
Time (ns)
1.5 2 V
V mA
Inductor IV
1 40 Vtime(M_PROBE.VP1,1)[*] (L, V)
p2 Inductor IV
RES M_PROBE
ID=Rs ID=VP1 Itime(ACVS.V1,1)[*] (R, mA)
R=50 Ohm 0.5 20 Inductor IV p1: Fre
p2: Fre
ACVS
ID=V1
IND
0 0
V
Mag=2 V
Ang=0 Deg ID=L1
L=4.62 nH p1
Offset=0 V
DCVal=0 V -0.5 -20
-1 -40 I
0 0.5 1 1.5 2
Time (ns)
155
CHAPTER 4 - Impedance Matching
As explained in section 1.5.5, complex numbers, which are used to represent the impedance
of an electrical network, give us two main bits of information:
- the ratio of the amplitude of the voltage across the impedance and the amplitude of the
current through it
- the difference between the phase of the voltage waveform and current waveform.
This is shown in equation (4.2-1).
( ) (4.2-1)
Where the real part is termed resistance and the imaginary part is termed reactance.
Representing a passive electrical network with impedances is especially appropriate when only
series elements are present. This is because whether they are reactive or resistive, we can just add
such elements together and easily obtain a single complex number which greatly simplifies the
analysis of the network. For example the series R-C combination shown in Figure 4.2-7
R C
(4.2-2)
Let us now consider the series R-C circuit shown in Figure 4.2-8 and let us attempt to calculate the
impedance of this network by inspection of voltage and current waveforms alone!
M_PROBE
ID=VP1
RES CAP
ACVS ID=R1 ID=C1
ID=V1 R=50 Ohm C=5.6 pF
Mag=1 V
Ang=0 Deg
Offset=0 V
DCVal=0 V
156
4.2 Impedance and Admittance
If the electrical network shown in Figure 4.2-8 was purely resistive, we would have no phase
difference between voltage and current, if it was purely capacitive we would have the current
leading the voltage by 90 , but, since we have a combination of these two elements, we may still
expect a current lead but a phase difference somewhat less than 90 ! This is shown in Figure 4.2-9.
series_R_C
1 20
1.248 ns
0.9999 V
0.5 1.168 ns 10
17.39 mA
0.5 ns
0V Vtime(M_PROBE.VP1,1)[*] (L, V)
p2 series R_C
0 0
Itime(ACVS.V1,1)[*] (R, mA)
0.4177 ns series R_C
0 mA
-0.5 -10
Figure 4.2-9 Current and voltage waveform for a 1000 MHz stimulus
As we mentioned in section 1.5.5, we can work out the impedance of an electrical network purely
from the voltage and current waveforms. Consider the markers in rectangular frames in Figure 4.2-9,
which show us the voltage and current peaks. From these measurements we may write
Now consider the markers in circular frames in Figure 4.2-9, which show us the time interval by
which voltage lags current. Remembering that (section 1.5.4.2) and that we are in a
sinusoidal steady state at a frequency of 1000 MHz, we may write
| | | | | |
( )
Since the current leads the voltage in this case i.e. ,( ) must be negative and may
be written in polar form as
[ ( ) ( )]
157
CHAPTER 4 - Impedance Matching
From (4.2-2)
In the complex plane our impedance may be represented as shown in Figure 4.2-10.
Im
50
-30 Re
-28.4j
Z
Now let us the consider the case of the R-L network shown in Figure 4.2-11.
M_PROBE IND
ID=VP1 ID=L1
L=4.5 nH
RES
ACVS ID=R1
ID=V1 R=50 Ohm
Mag=1 V
Ang=0 Deg
Offset=0 V
DCVal=0 V
The voltage and current waveforms for this network are shown in Figure 4.2-12.
158
4.2 Impedance and Admittance
series_R_L
1 20
1.249 ns
0.9999 V
0.5 1.329 ns 10
0.5819 ns 17.4 mA
0 mA Vtime(M_PROBE.VP1,1)[*] (L, V)
p2 series R_L
0 0
0.5 ns Itime(ACVS.V1,1)[*] (R, mA)
0V p1 series R_L
-0.5 -10
Figure 4.2-12 Voltage and current waveforms for the R-L network
In this case we expect a voltage lead since an inductor is present, but also a phase difference of less
than 90 because a resistor is also present.
Following an identical procedure to that used for the R-C circuit, the reader may verify that the
impedance worked out from the current and voltage markers in Figure 4.2-12 is the same as that
derived algebraically, shown below
This impedance may be represented in the complex plane as shown in Figure 4.2-13
Im
28.4j Z
30
50 Re
159
CHAPTER 4 - Impedance Matching
Let us now consider a network that comprises of resistive, inductive and capacitive elements
connected in series as shown in Figure 4.2-14, Figure 4.2-15 and Figure 4.2-16.
As we have seen in previous cases, we will probably have a phase difference between voltage and
current which is somewhat less than 90 due to the presence of the resistor. However the voltage
may either lead or lag the current depending on the values of our capacitor and inductor.
In the circuit shown in Figure 4.2-14, the capacitor and inductor values are such that the
modulus of the reactance of the cap | | is greater than that of the inductor | |. The overall
impedance of the L-C series is , which means that the capacitor dominates, and the
current leads the voltage (Figure 4.2-14).
| | | |
V series R_L_C mA
1 20
0.5 p1 10
p2
0 0
-0.5 -10
In the circuit shown in Figure 4.2-15, the capacitor and inductor values are such that the
modulus of the reactance of the capacitor | | is smaller than that of the inductor
| | . The overall impedance of the L-C series is , which means that the inductor
dominates, and the voltage leads the current (Figure 4.2-15).
V series R_L_C
mA
1 20
0.5 10
p2
0 0
p1
-0.5 -10
160
4.2 Impedance and Admittance
In this case the overall impedance of the L-C series is zero and hence it acts like a short circuit! This is
true only at the one frequency that we are operating at (1000 MHz). As shown in Figure 4.2-16,
current and voltage are in phase, as they would be if no reactive elements were present.
-0.5 -10
This is a very useful property of R-L-C circuits which is called resonance. Resonance occurs when
| | | |
Now remember that these impedances are dependent on the value of the frequency but also on the
values of C and L. So once the values of C and L are fixed, if we sweep the frequency across a wide
range starting from DC, our series L-C will appear capacitive (| | > | |) up to the resonant
frequency
It will look inductive for frequencies higher than , since | | < | |, and it will be invisible (zero
impedance) at the resonant frequency.
If we consider the resistor in our R-L-C series as the source internal resistance (RS) and we
add a load (RL) at the other end of our R-L-C network (Figure 4.2-17), we can see our circuit as a
source with internal resistance RS and a load with resistance RL, with an L-C series resonant circuit
between them. This resonator turns out to have a band-pass effect across the frequency range as
shown in Figure 4.2-18.
161
CHAPTER 4 - Impedance Matching
P_METER3
ID=P1
W
1 I 2
IND V
RES CAP
ACVS ID=RS ID=Cs ID=Ls 3
ID=V1 R=50 Ohm C=5.6 pF L=4.5 nH
Mag=1 V RES
Ang=0 Deg ID=RL
Offset=0 V R=50 Ohm
DCVal=0 V
Figure 4.2-17 Measuring power transfer when L-C resonator is present between source and load
Series_Power
3
2.5
1.5
1 |Pcomp(P_METER3.P1,1)| (mW)
LC_series_bandpass
0.5
0
1 1001 2001 3000
Frequency (MHz)
As is to be expected, maximum power transfer between source and load is achieved at the resonant
frequency (Figure 4.2-18).
Video 4.1 illustrates the concepts explained in section 4.2.1, 4.2.2 and 4.2.3 by means of Microwave
Office simulations.
162
4.2 Impedance and Admittance
( ) (4.2-3)
If you compare eq. (4.2-1) with (4.2-3) you can see two major differences. Firstly the moduli of
admittance and impedance are the inverse of one another. Secondly, whilst the angle of the
impedance represents the difference between the phases of voltage and current ( ), the angle
of the admittance represents the difference between the phases of current and voltage ( ).
In a Cartesian form the admittance may be written as
Where the real part is termed conductance and the imaginary part is termed susceptance.
Now let us consider the parallel R-C circuit shown in Figure 4.2-19.
parallel_R_C
1 60
1.257 ns
0.999 V
p1
0.5 1.079 ns 30
0.5 ns
0V 40.5 mA
p2
0 0
0.3323 ns
-0.5 0 mA
-30
Figure 4.2-19 Parallel R-C circuit, voltage and current waveforms at 1000 MHz
Following a similar procedure to that used in the R-C series case, from the markers in the rectangular
frames in Figure 4.2-19 we may write
| |
The SI units of admittance are Siemens, expressed by the capital letter S. Nonetheless milli Siemens,
mS, are generally easier to handle and simplify calculations. This is because the ratio of current in
and voltage in , directly gives .
Now consider the markers in circular frames in Figure 4.2-19, which show us the time interval by
which current leads the voltage. Remembering that (section 1.5.4.2) and that we are in a
sinusoidal steady state at a frequency of 1000 MHz, we may write
163
CHAPTER 4 - Impedance Matching
| | | | | |
( )
Now, just as in the R-C series case (Figure 4.2-9), the current leads the voltage, and hence is
negative. However with the admittance we are looking at , which is positive! This is an
important fact that always applies: impedance and admittance have opposite phases! Therefore our
admittance in polar form works out to be
[ ( ) ( )]
Im
35.2j Y
60.4
Re
20
Let us now consider the parallel R-L circuit shown in Figure 4.2-21.
As before, by using the values which may be read out from the markers in Figure 4.2-21, we may
calculate the admittance of our inductive network.
| |
164
4.2 Impedance and Admittance
parallel_R_L
1 60
1.246 ns
0.9996 V
p2
0 0
0.5 ns
-0.5 0V p1 -30
Figure 4.2-22 shows a representation of the admittance of the R-L network in the complex plane.
Im
20
-60.5 Re
35.3j
Y
Figure 4.2-22 Representation of or in the complex plane
Remember that the angle of the admittance is and, since the voltage leads the current in an
inductive network, this angle is negative. This is just the opposite of what we had for the impedance
of an inductive network (Figure 4.2-13).
165
CHAPTER 4 - Impedance Matching
Let us now consider a network that comprises of resistive, inductive and capacitive elements
connected in parallel as shown in Figure 4.2-23, Figure 4.2-24 and Figure 4.2-25.
As we have seen in previous cases, we will probably have a phase difference between voltage and
current which is somewhat less than 90 due to the presence of the resistor. However the voltage
may either lead or lag the current depending on the values of our capacitor and inductor.
The analysis of this type of network is much easier if we use admittances instead of impedances!
In the circuit shown in Figure 4.2-23, the capacitor and inductor values are such that the modulus of
the susceptance of the capacitor | | is lower than that of the inductor | |.
The overall admittance of the L-C parallel is which means that the inductor
dominates and the voltage leads the current (Figure 4.2-23).
V parallel_R_L_C
mA
1 40
0.5 20
p2
0 0
-0.5 -20
p1
p1: Freq = 1000 MHz
Vtime(M_PROBE.VP1,1)[*] (L, V)
parallel_R_L_C -1 -40
Freq = 1000 MHz 0 0.5 1 1.5 2
p2: Freq = 1000 MHz Itime(ACVS.V1,1)[*] (R, mA)
parallel_R_L_C Time (ns)
V parallel_R_L_C mA
1 40
p1
0.5 20
p2
0 0
-0.5 -20
166
4.2 Impedance and Admittance
In the circuit shown in Figure 4.2-24, the capacitor and inductor values are such that the modulus of
the susceptance of the capacitor | | is greater than that of the inductor | |.
The overall admittance of the L-C parallel is which means that the capacitor
dominates and the current leads the voltage (Figure 4.2-24).
Now if we consider the circuit shown in Figure 4.2-25, we can see that in this case
| | | |
This means that the parallel combination of capacitor and inductor is resonant and its admittance is
zero! This means that the parallel L-C circuit acts as an open circuit at 1000 MHz and hence voltage
and current are in phase just as they would be if no reactive elements were present.
V parallel_R_L_C mA
1 40
0.5 20
p2
p1
0 0
-0.5 -20
Now let us consider the circuit shown in Figure 4.2-26. In this circuit we have a source and a
load impedance (RS and RL) connected by a parallel L-C circuit.
Once the values of C and L are fixed, if we sweep the frequency across a wide range starting from
DC, our parallel L-C will appear inductive (| | > | |) up to the resonant frequency30
It will look capacitive for frequencies higher than , since | | < | |, and it will behave as an
open-circuit (zero admittance) at the resonant frequency thereby preventing the signal from
reaching the load. We will therefore experience a band-stop effect, as shown in Figure 4.2-27.
This is confirmed by the voltage and current profiles at the resonant frequency, which are shown in
Figure 4.2-28. This figure shows that, at resonance, the voltage across and the current through the
parallel L-C are in phase but their amplitudes are very small!
30
Similarly to the previous case the resonant frequency satisfies the condition below
| | | |
167
CHAPTER 4 - Impedance Matching
IND
ID=Lp
L=4.5 nH
P_METER3 M_PROBE
RES ID=VP1
ID=RS ID=P1
R=50 Ohm W
CAP
1 I 2
ID=Cp V
ACVS C=5.6 pF
ID=V1 3 RES
Mag=1 V ID=RL
Ang=0 Deg R=50 Ohm
Offset=0 V
DCVal=0 V
Figure 4.2-26 Parallel L-C resonator between source and load impedances
Parallel_power
|Pcomp(P_METER3.P1,1)| (mW)
3
LC_parallel_bandstop
2.5
1.5
0.5
0
1 501 1001 1501 2000
Frequency (MHz)
Itime(ACVS.V1,1)[*]
Graph 1 (R, mA)
0.1 LC_parallel_bandstop 2
Vtime(M_PROBE.VP1,1)[*] (L, V)
LC_parallel_bandstop p1: Freq = 1000 MHz
0.05 1
0 p2 0
0.496 ns 1.498 ns
0.1819 mA 0.009094 V
-0.05 -1
Figure 4.2-28 voltage across resonator and current through it at resonant frequency (1000 MHz)
168
4.2 Impedance and Admittance
In this section we will define an important and useful quantity: the quality factor, Q.
Several quality factors may be defined for reactive components and the circuits which contain them.
They generally relate the energy loss versus the energy stored in reactive components during a
signal cycle. Non-ideal capacitors and inductors will introduce losses since they dissipate energy in
the parasitic elements which are introduced by their physical construction and packaging. The
unloaded Q, , of a reactive component is defined as shown below
This quality factor takes into account the energy dissipated in the component alone and not in the
rest of the circuit. For ideal components would of course be infinite.
For a real inductor we can represent the losses by means of a series resistor (Figure 4.2-29)
Figure 4.2-29 Series R-L, the resistance is used to model power dissipation in a non-ideal inductor
The lower the series resistance , the lower the voltage drop across it and the power dissipated in
it, hence the higher the Q!
For a real capacitor we can represent the losses by means of a shunt resistor (Figure 4.2-30)
Figure 4.2-30 Parallel R-C, the resistance is used to model power dissipation in a non-ideal capacitor
The greater the shunt resistance , the less current will flow through it and the less power will be
dissipated in it, hence the greater the Q.
The unloaded Q of individual components does have an effect on the overall quality factor of a
network and particularly on its frequency response!
The quality factor Q is also very useful when it comes to carrying out series to parallel conversions of
electrical network since it simplifies the maths considerably as we will see in the next section.
169
CHAPTER 4 - Impedance Matching
So far we have looked at the unloaded Q of reactive components based on the value of the
component reactance and of the respective parasitic resistance. However when our reactive
element is connected to an external resistive one, the Q is no longer determined by the parasitic
element internal to the reactive component, which is usually quite small, but by the external
resistive element connected to it. In this section we will therefore define Q for networks of resistive
and reactive elements. Let us now look at a few illustrative examples to clarify this.
For the R-C network shown in Figure 4.2-8, and repeated below in Figure 4.2-31 for the readers
convenience,
M_PROBE
ID=VP1
RES CAP
ACVS ID=R1 ID=C1
ID=V1 R=50 Ohm C=5.6 pF
Mag=1 V
Ang=0 Deg
Offset=0 V Figure 4.2-31 Series R-C in MWO
DCVal=0 V
the Q factor is defined as
For the series R-L circuit shown in Figure 4.2-11, and repeated below in Figure 4.2-32, the Q factor is
calculated in an identical fashion and has the same value,
M_PROBE IND
ID=VP1 ID=L1
L=4.5 nH
RES
ACVS ID=R1
ID=V1 R=50 Ohm
Mag=1 V
Ang=0 Deg Figure 4.2-32 Series R-L in MWO
Offset=0 V
DCVal=0 V
For parallel circuits, it is easier to use the real part (conductance, G) and imaginary part
(susceptance, B) of the admittance to calculate the Q. For the parallel R-C of Figure 4.2-19 and
repeated in Figure 4.2-33 parallel_R_C
1 60
1.257 ns
0.999 V
p1
0.5 1.079 ns 30
0.5 ns
0V 40.5 mA
p2
0 0
0.3323 ns
-0.5 0 mA
-3
Figure 4.2-33 Parallel R-C in MWO
p1: Freq = 1000 MHz Vtime(M_PROBE.VP1,1)[*] (L, V)
we may write Freq = 1000 MHz parallel_R_C -1 -6
0 0.5 1 1.5 2
p2: Freq = 1000 MHz Itime(ACVS.V1,1)[*] (R, mA) Time (ns)
parallel_R_C
170
4.2 Impedance and Admittance
For the series R-L network of Figure 4.2-21 the calculation is identical.
The Q which we have defined for series and parallel R-C and R-L networks is also a very
useful parameter to work out series to parallel and parallel to series conversions. These conversions
can be very useful when it comes to circuit analysis since they allow us to simplify it considerably. Let
us now go through an example.
RES
ID=R1
R=10 Ohm
CAP
ID=C1
ACVS C=7.95 pF
ID=V1
Mag=1 V
Ang=0 Deg
Offset=0 V
DCVal=0 V
Consider the R-C network shown in Figure 4.2-34. The impedance of the R-C network is
Now we can use specific formulae, the derivation of which is beyond the scope of this treatment, to
convert our R-C series network to an R-C parallel equivalent.
To this end we first calculate our equivalent parallel resistor value , which is derived from the
value of our series resistor and the Q of the series circuit . This is shown below
( )
Then we calculate the value of our equivalent parallel capacitor , by using and ,
as shown below
( )
ACVS
ID=V1 RES CAP
Mag=1 V ID=R1 ID=C1
Ang=0 Deg R=50 Ohm C=6.36 pF
Offset=0 V
DCVal=0 V
Figure 4.2-35 Parallel R-C network equivalent to the series R-C shown in Figure 4.2-34
171
CHAPTER 4 - Impedance Matching
If the parallel R-C circuit that we have worked out is equivalent to the series one, our signal
generator should see the same impedance for both networks. This is indeed the case as shown by
the waveforms in Figure 4.2-36.
Figure 4.2-36 Voltage and current waveforms for series R-C in Figure 4.2-34 (left), Voltage and current waveforms for
equivalents parallel R-C in Figure 4.2-35 (right)
To summarise, for each series R-C network there exists and equivalent parallel R-C network and vice-
versa. The same applies to R-L networks. Table 4.2-1 and Table 4.2-2 give all the formulae that we
need to carry out these conversions.
( )
Parallel to
Series
( )
Series to
Parallel
( ) ( )
Table 4.2-1 Formulae for parallel to series and series to parallel conversion of R-C and R-L circuits based on Q
Parallel to
Series
Series to
Parallel
Table 4.2-2 Fomulae for parallel to series and series to parallel conversion of R-C and R-L circuits based on reactance
172
4.2 Impedance and Admittance
Let us now talk about the Q factor of the series RLC (Figure 4.2-38) and parallel RLC (Figure 4.2-37)
networks which we examined in sections 4.2.3 and 4.2.5. Although in both of these cases we have
two reactive elements, their reactance or susceptance will be the same at the resonant frequency!
For the series resonant circuit shown in Figure 4.2-37
M_PROBE series R_L
ID=VP1
1
Where is the series resistance, is the reactance of the series inductor and is the reactance
of the series capacitor (Figure 4.2-37).
For the parallel circuit shown in Figure 4.2-38 parallel_R_
1
0.5
-0.5
Figure 4.2-38 Parallel RLC Resonator
p1: Freq = 1000 MHz
we may writeFreq = Vtime(M_PROBE.VP1,1)[*] (L, V)
1000 MHz parallel_R_L_C -1
0 0.5 1
p2: Freq = 1000 MHz Itime(ACVS.V1,1)[*] (R, mA)
Time (ns
parallel_R_L_C
where is the parallel resistance, is the reactance of the parallel inductor and is the
reactance of the parallel capacitor. Or, equivalently
Remember that, although the reactance of the inductor and the capacitor cancel each other out at
resonance, there is still current flowing between them! This means that you still get losses in the
resistor which allow you to define the resonant Q of the circuit! This is true for both parallel and
series connections.
173
CHAPTER 4 - Impedance Matching
In the case of two unequal resistive terminations we cannot carry out a simple resistive
matching as the one shown in Figure 4.3-1.
RS = 50
46
VS Matching RL = 4
Network
50 96
It is quite easy to see why. The current flowing through this circuit will be equal to VS/100. Now
since voltage is proportional to resistance we will have almost half of the source voltage dropped
across the matching resistor leaving very little voltage, and hence little power, to be delivered to the
load.
Also let us take a look at the impedances seen from source and load end. When we look into
the matching network from the source end, we will see an overall impedance of 50 However
when we look back from the load end into the matching network we will see an impedance of 96 !
This means that, if some of the power is reflected by the load into the matching network, it will see a
high impedance, and bounce back towards the load again. This may create oscillations which, in the
case of amplifier circuits for instance may be fatal!
Ideally we would like our matching network to dissipate as little power as possible. Reactive
elements such as capacitors and inductors are therefore ideal candidates! However bear in mind
that they do suffer from the effects of parasitics and that their real-life equivalents are lossy to some
extent.
In this section we will first look at L-section matching which makes use of two reactive
elements to create a match between two unequal resistive terminations. We will then look at the
frequency response of matching networks and ways to improve the Q.
The topics illustrated in this section are demonstrated with the aid of Microwave Office simulations
in video 4.2 and video 4.3
174
4.3 Matching two unequal resistive impedances
Suppose that we have a generator with an internal resistance RSOURCE=100 and a much
greater load resistance RLP=1000. To achieve maximum power transfer we want the generator see
a load of 100 .
Now if we put a capacitor in parallel with RLP (Figure 4.3-2), the source will see a greater load and will
supply additional current which will flow through the capacitor to charge it. Since the overall current
through the R-C parallel is higher than the current supplied by the source when only RLP was present,
whilst the voltage remains the same, the overall impedance seen by source is lower than RLP.
RSOURCE
VSOURCE CP RLP
Figure 4.3-2 The shunt capacitor CP reduces the resistive part of the load impedance
You may argue that, if in place of the capacitor CP, we put a resistor in parallel with RLP, we could
easily make the load impedance equal to that of the source. This is true, however the value of this
resistor would have to be 110 and we would end up having almost 90% of the current flowing
through it and just over 10% being delivered to the load! The advantage of using a reactive element,
like the capacitor, is that it stores but does not dissipate energy31.
So we managed to lower the overall impedance seen by the source by introducing a shunt capacitor,
but how much have we lowered it by? Well, to work this out, the first step is to determine the
equivalent series circuit (Figure 4.3-3), for our parallel R-C.
RSOURCE
CS
VSOURCE RLS
From Table 4.2-2 we may work out the resistive part of the equivalent series impedance, , as
shown below. represents the reactance of CP (Figure 4.3-2).
(4.3-1)
31
Actual capacitors are lossy as explained in section 4.2.6.
175
CHAPTER 4 - Impedance Matching
So, in order to make the real part of the impedance seen by the generator equal to 100, we must
use a shunt capacitor with a reactance .
Now we can determine the reactance of the series capacitor CS in our equivalent series R-C
network (Figure 4.3-3) by using the formulae in Table 4.2-2 yet again.
So we have determined the value of the reactance of the shunt capacitor ( ) which gives us the
right value for the resistive part of the load impedance ( ) but now we have also ended up with an
equivalent series reactance ( ) which we need to get rid of to ensure maximum power transfer to
the load. This can be achieved by an equal and opposite series reactance i.e. by a series inductor
(Figure 4.3-4) with reactance .
RSOURCE
LS
CS
VSOURCE RLS
Figure 4.3-4 We can resonate out the apparent series capacitance CS with a series inductor LS
At this stage we must make two important points. Firstly, although we have made use of the
equivalent series R-C for our parallel R-C network to make it easier to determine the value of LS, the
actual physical R-C network will still be a parallel R-C! This is shown in Figure 4.3-5.
RSOURCE
LS
VSOURCE CP RLP
176
4.3 Matching two unequal resistive impedances
Secondly, resonance and hence a perfect match will only occur at the one frequency at which the
capacitive and inductive reactances, and cancel each other out. At 100 MHz for instance, the
values of and yield CP = 4.78pF and LS =477nH.
To summarise:
- The shunt component of the L-section acts to reduce the impedance seen by the generator
to one whose real part is equal to the real part of the source impedance
- The series component of the L-section acts as to resonate out the reactive component
introduced by the shunt component (and possibly parasitics)
A general procedure may be given to work out the elements of the L-section matching network
shown in Figure 4.3-6.
QS=XS/RS
RS XS
VSOURCE XP RRPLP
QP=RP/XP
Figure 4.3-6 The general case for L-section matching XS and XP must be of different types
1) Calculate QP and QS, which have the same value, by means of the formula below
(4.3-2)
3) Decide which reactance is going to represent a capacitor and which an inductor. You may
chose either but they must not be the same!
4) Work out the values of capacitor and inductor at the frequency of interest
I know that you may feel that this whole algebraic procedure looks cumbersome and error prone but
do not despair! We will soon see how we can use the Smith Chart (section 4.4) to make the design of
L-section matching networks a great deal easier!
Note how the network which we devised earlier to achieve a match at 100MHz (Figure
4.3-7(a)) could be replaced with another L-network where the places of capacitor and inductor have
been swapped (Figure 4.3-7(b)).
177
CHAPTER 4 - Impedance Matching
100
LS=477 nH
VSOURCE 1000
CP =4.8 pF
(a)
100
CS=5.3 pF
VSOURCE 1000
LP =530 nH
(b)
Figure 4.3-7 L-C (a) and C-L (b) L-section matching
These two networks will behave in a very similar fashion at 100 MHz but the difference between
them lies in their frequency response across a wider range of frequencies.
dB
L S & CP
CS & L P
Figure 4.3-8 Ratio of Output and Input powers in dB for the circuits in Figure 4.3-7
Figure 4.3-8 shows that the circuit in Figure 4.3-7(a), has a low-pass response. This is quite
predictable since, as the frequency increases, the impedance of the parallel branch (capacitor)
diminishes thereby providing a low impedance path to ground and sinking current in there!
The circuit in Figure 4.3-7(b) instead exhibits a high-pass behaviour since, at low frequency, the
impedance of the parallel branch (inductor) is very low and sinks current into the ground but, at high
frequency, such an impedance is very high and hence current is allowed to flow into the load
thereby delivering power to it. A high-pass design is particularly useful when we need to block DC.
178
4.3 Matching two unequal resistive impedances
As we have seen in the previous section, a matching network behaves effectively like a filter
which is characterised by its own frequency response. This frequency selective behaviour is even
more pronounced in the case of parallel and series L-C resonators as shown in Figure 4.2-18 and
Figure 4.2-27. It is important to understand the frequency response of matching networks in the
context of the circuits in which they are used in i.e under loaded conditions. Such a response will be
influenced by three major factors:
1) The source resistance (RS)
2) The load resistance (RL)
3) The components Q
Depending on the complexity of the network it may be rather tricky to work out an analytical
expression for the loaded Q which takes into account all of the above factors. This is why we tend to
define the loaded Q of a circuit in terms of its frequency response as
Where represents the centre frequency and represents the 3dB bandwidth. This is shown
in Figure 4.3-9.
f1 f2
Gain (dB)
fC
Figure 4.3-9 3dB-bandwidth ( ) and centre frequency ( ) may be used to work out the loaded Q
The circuit associated with the frequency response shown in Figure 4.3-9 is shown in Figure 4.3-10.
179
CHAPTER 4 - Impedance Matching
M_PROBE M_PROBE
ID=VP1 RES ID=VP2
ID=R1
R=50 Ohm
ACVS
ID=V1 IND RES Xo . . . Xn
Mag=1 V CAP
ID=L1 ID=R2
Ang=0 Deg ID=C1
L=50 nH R=1e5 Ohm
Offset=0 V C=25 pF
DCVal=0 V
SWPVAR
Figure 4.3-10 Schematic corresponding to frequency response in Figure 4.3-9 ID=SWP1
VarName="RS"
Values={ 50,1000 }
To work out the frequency response we looked at the voltage across the load resistor (VP2)
UnitType=None
RS=50
and the voltage across the generator (VP1), we then calculated the ratio of these voltages and
converted it into dB.
( ) ( )
At this point we must specify what we mean by attenuation and gain. In this case, since we have only
passive elements present, the gain of the filter is mostly negative and may only reach zero dB as a
maximum. In actual facts, the signal is attenuated across the majority of the frequency band of
interest. However, attenuation is defined as a positive value which is the ratio of input over output
voltage hence we cannot use this term here! If we look at the ratio shown above, it is gain, albeit
negative gain, that we are looking at. It is matter of nomenclature but we must avoid confusion!
Note how we have used a very high value for our load resistor R2=100 k to simulate a high
impedance load. This was done to enable us to assess the impact of the source impedance alone by
minimising the effect of a load. If we now increase the source impedance R1 to 1000 , we can see
how the Q of the circuit changes dramatically as shown in Figure 4.3-11. Our circuit has become a
lot more frequency-selective!
Neither of the curves shown in Figure 4.3-11 however addresses the impact of a load
impedance on the circuit since our R2 at the moment is so large that it does not allow a significant
amount of current though it.
Q=1.1
Gain (dB)
Q=22.4
180
4.3 Matching two unequal resistive impedances
Let us now look at how the load impedance affects the overall Q of the circuit shown in
Figure 4.3-12. In this circuit the load impedance RL is no longer a very large value, as it was in the
schematic of Figure 4.3-10, and hence it draws significant current thereby affecting the circuit
response.
RS
VSOURCE L C RL
L C
The resonant circuit sees an equivalent resistance RP which is the parallel combination of RS and RL as
its actual load. By definition RP is smaller in value than RS and RL. Now let us consider the Q of this
resonant circuit, in a similar fashion as we did in section 4.2.8
(4.3-3)
here is either the inductive or capacitive reactance since these values are the same at resonance.
Usually however, source and load impedances are fixed hence the only control the designer has is
over the values of C and L. One must be careful not to end up with impractical values though!
So far we have assumed that our capacitors and inductors were lossless but in practice, the
finite component Q must be taken into account. Let us now look at the effect of component Q on
the performance of the circuit.
181
CHAPTER 4 - Impedance Matching
As explained in section 4.2.6, the losses of an inductor may be modelled by a series resistor (Figure
4.2-29) and the losses of a capacitor with a shunt resistor (Figure 4.2-30). The L-C resonator in Figure
4.3-12 when losses are included would therefore look as shown in Figure 4.3-14.
L C RCP
RLS
Figure 4.3-14 Resistors may be added to model the losses in non-ideal capacitors and inductors
We can use parallel to series conversions formulae (4.2.7) for the series R-L above so as to have all
elements in parallel.
RLP L C RCP
Figure 4.3-15 Equivalent parallel conversion for the circuit in Figure 4.3-14
In the schematic of Figure 4.3-15, RLP may be calculated by using the conversions formulae in Table
4.2-1 as
( )
Now recall how the Q is directly proportional (eq. (4.3-3)) to the equivalent parallel
resistance seen by the resonator (Figure 4.3-13). Now if RCP or RLP are small, they will considerably
decrease the overall equivalent parallel resistance RP since the resistance of a parallel must be lower
than lowest of the resistors that such a parallel comprises of. A poor component Q may therefore
decrease the circuit Q considerably.
In most cases the Q of the inductor alone must be included in calculations since the Q of capacitors,
and hence their equivalent shunt resistance is quite high and hence can usually be neglected.
182
4.3 Matching two unequal resistive impedances
VSOURCE X1 X3 RLOAD
RSOURCE X1 X3
VSOURCE X2 RLOAD
We will not be looking at the algebraic calculations to determine these networks since there is a
much easier way to design them which makes use of Smith charts and will be illustrated in section
4.4.
183
CHAPTER 4 - Impedance Matching
Let us first have a little recap on the concepts which form the basis of Smith Chart plots.
We have seen how, when a generator is connected to a load through a transmission line, depending
on the value of the load impedance (ZL) some of the signal may be reflected. We may quantify the
proportion of the incident signal that gets reflect through the reflection coefficient . This is shown
in Figure 4.4-1.
l ,
Figure 4.4-1 Reflection coefficient at the input of a transmission line in series with a load impedance Z L
Usually the internal resistance of the voltage source VS and the characteristic impedance of
the line Z0, are chosen to be the same value. This is done because in order to achieve maximum
power transfer we must have RS=Z0=ZL (section 2.9). The first part of this equality, RS=Z0, may be
easily achieved by utilising transmission lines throughout the RF system with identical characteristic
impedance to RS. In most microwave systems RS and Z0 are chosen to be 50 for the reasons
explained in section 2.12. The load impedance ZL however, may be any value and hence, to satisfy
the condition for maximum power transfer, we may need to carry out some impedance matching.
Now, although the reflection coefficient is a very useful quantity, we would ideally like to
work with impedances since they are more easily relatable to our physical circuit. To this end we
may use (3.5-3) which gives us a direct correspondence between ( ) and ( ). This in turn
simplifies the formulation of the condition for maximum power transfer which becomes RS=
(Figure 4.4-2).
( ) ( ) (4.4-1)
( ) ( )
( ) ( )
l ,
ZIN (l)
Figure 4.4-2 Impedance seen at the input of a transmission line in series with a load impedance Z L
184
4.4 Matching any two complex impedances - Smith Chart Matching
By transforming the reflection coefficient into an impedance we have reduced our level of
abstraction however we have made things a bit more difficult graphically. That is because, whereas
the magnitude of the reflection coefficient only varies between -1 and 1, thereby making it easy to
represent it on a polar graph (section 3.6), the magnitude of a complex impedance may be just any
value! Good old Phillip H. Smith of Bell Labs found a really clever way around this.
First of all, he recognised that in an RF or Microwave Systems the characteristic impedances
of the majority of the transmission lines are identical and equal to the internal impedance of the
generator. He therefore thought to normalise each impedance in his circuit to the characteristic
impedance of the RF system, Z0. Let us indicate the normalised impedances, and their respective
resistances and reactances with lower case letters and un-normalised ones with upper case letters.
For an impedance we define a normalised equivalent as
0 r
r=0 r=0.5 r=1 r=2
x=1
+jx
x=1 x=0.5
x=0.5
x=0
0 x=0
x=-0.5
x=-1
-jx x=-0.5
x=-1
185
CHAPTER 4 - Impedance Matching
0 1
| | | | | |
Figure 4.4-5 Constant normalised resistance circles and normalised reactance lines form the Impedance Smith chart grid
The mathematics behind the creation of this plot are beyond the scope of this treatment but
what you MUST remember when you look at a Smith chart, is that it is a polar plot of the reflection
coefficient upon which we have superimposed (Figure 3.6-2) a very cleverly designed mask to read
out impedance values directly without having to use conversion formulae!
Bearing in mind how the impedance chart is constructed (Figure 4.4-3 and Figure 4.4-4), it is
apparent that its top half represents impedances with inductive reactances and the bottom half
represents impedances with capacitive reactances (Figure 4.4-6).
Note that all points on the outer circle represent impedances which are purely reactive:
purely inductive in the upper half and purely capacitive in the bottom half.
0.5
0 0.25 1 2
-0.5
-1
Figure 4.4-6 Inductive and capacitive halves of the impedance Smith chart
186
4.4 Matching any two complex impedances - Smith Chart Matching
To de-normalise our admittance all we need to do is multiply by . Also note how an admittance
equal to the characteristic admittance of the system, has a normalised value of 1.
Next we consider lines of constant normalised conductance (Figure 4.4-7) and lines of
constant normalised susceptance (Figure 4.4-8). And we bend them in such a way as to fit in the
unity radius polar plot of the reflection coefficient as shown in Figure 4.4-9.
187
CHAPTER 4 - Impedance Matching
Figure 4.4-9 Normalised conductance circles and susceptance lines form the Admittance Smith chart grid
This looks remarkably similar to the impedance chart shown in Figure 4.4-5 but DO NOT BE
FOOLED! The zero point here represents a point with 0 conductance i.e. infinite resistance. The point
in the same location in Figure 4.4-5, which is also labelled 0, indicates a point of zero resistance!
Similarly, the point on the admittance chart labelled , represents infinite conductance and hence
zero resistance. The point at the same location on the impedance chart instead represents a point
with infinite resistance!
The other crucial difference is that on an admittance chart the top half is capacitive, since
capacitors are characterised by positive susceptances, and the bottom half is inductive, since
inductors are characterised by negative susceptances. This is shown in Figure 4.4-10.
0.5
0 0.25 1 2
-0.5
-1
Figure 4.4-10 Capacitive and inductive halves of the Admittance Smith chart
188
4.4 Matching any two complex impedances - Smith Chart Matching
Also all the points on the outer circle represent admittances which are purely reactive:
purely capacitive in the upper half and purely inductive in the bottom half. This is the exact opposite
of what we had for the impedance chart.
The first time I saw this, I thought to myself, How confusing! These two things looks the
same but represent completely opposite things!. It is however possible to combine the two in such
a way as to leave no ambiguities. We will see this in the next section.
ADMITTANCE 1 1
IMPEDANCE
0.5 0.5
0 0.25 1 2 0 0.25 1 2
-0.5 -0.5
-1 -1
-1 ADMITTANCE IMPEDANCE 1
-0.5 0.5
0 0.25 1 2
2 1 0.25
0
0.5 -0.5
1 -1
Figure 4.4-12 Admittance chart rotated by 180 (left), Impedance chart (right)
189
CHAPTER 4 - Impedance Matching
Now if we overlap the two, we get the chart shown in Figure 4.4-13.
-1 1
0.5
-0.5
0 0.25 1 2
4 1 0.5 0
-0.5 0.5
1 -1
By using the chart shown in Figure 4.4-13 we are able to design a matching network by using both
parallel and series elements. As we add reactive elements in series we will move up or down a
constant resistance circle, depending on whether it is an inductor or a capacitor that we are adding.
If we add reactive elements in shunt, we will be moving up or down a constant conductance circle
depending on whether we are adding an inductor or a capacitor. This is shown in Figure 4.4-14.
-1 1
0.5
-0.5
Series C Shunt C
0 1
1 0
Series L Shunt L
-0.5 0.5
1 -1
Figure 4.4-14 Adding series and shunt components on the Smith chart
190
4.5 Discrete vs Distributed elements
L3 L2 L1
ZIN C4 C3 C2 C1
RL
Figure 4.5-1 Short section of transmission line and equivalent electrical model
Figure 4.5-1 shows a small section of transmission line, of electrical length , terminated
with a resistance RL, and the electrical model for such a line. Let us assume that RL = 1 and that
each capacitor has an impedance and each inductor has an impedance of
.
Now, starting from the load end, let us consider the parallel of RL and C1. The modulus of the
impedance of C1 is much higher than that of RL. This means that C1 is quite small and hence draws a
very small current compared to the resistor. We may therefore ignore C1 and assume that the
impedance of the parallel is equal to RL. The impedance seen at point will therefore be
. A similar reasoning may lead us to assume that the capacitor C2 is also negligible and hence the
impedance at point will be . Now, when calculating the impedance at point , since
the impedance of C3 is still much larger than ( ) , we can again neglect the capacitor and
assume that . We could carry on repeating this process for a few more inductors but
we would eventually come to a point (i.e. to a transmission line length) at which we would need to
take into account the impedance of the capacitors. This simple example explains why we may state
that short transmission line segments terminated with low impedances behave like inductors.
32
By short we mean less than 90 in electrical length at the frequency of interest.
191
CHAPTER 4 - Impedance Matching
Now let us look at what happens when the terminating impedance is much larger, RL =
1000 while the parameters of the electrical model for the line remain unchanged.
Now the impedance of C1 and RL are comparable and the currents which they draw will be quite
similar. This means that we cannot neglect the capacitor when working out the impedance of this
parallel. Also since both the impedance of C1 and that of RL are quite large, the parallel will draw
quite a small current overall. This means that the inductor will have little current flowing through it
and hence its impedance may be neglected. Again we can carry on doing this for a few more
inductor-capacitor sections until we get to a point where the overall current drawn has increased to
a value which means that the inductors can no longer be neglected. We can infer therefore that a
short segment of transmission line terminated with high impedance behaves a capacitor.
Now let us take things a step further and make some quantitative considerations. The lowest
impedance that we can terminate our line with would be a short circuit as shown in Figure 4.5-2.
In section 2.9, we saw a mathematical expression for the impedance seen at the input of the line
(eq.(2.9-1)) which is shown below
(4.5-1)
Since, in the case of a line terminated with a short circuit , equation (4.5-1) becomes
Now, for values of which are between 0 and 90, is positive and hence will
represent an inductive impedance. We could therefore replace a discrete inductor with its
distributed equivalent, which is our short-circuited line of the appropriate length.
Notice however that, if we choose an electrical length for our line which is between 90 and
180, would be negative and hence we would have a capacitive impedance. This goes to
show the versatility of the distributed approach which allows you to realise both capacitors and
inductors with same topology by simply selecting the appropriate electrical length.
Now we have seen, how things work when we terminate our line with the lowest possible
impedance, but what happens when we terminate it with the highest possible impedance, i.e. we
have an open circuit at the end of the line?
192
4.5 Discrete vs Distributed elements
In this case, the impedance seen at the input terminal of the line, may be expressed as
Now, for values of which are between 0 and 90, is positive and hence will
represent a capacitive impedance. We could therefore replace a discrete capacitor with its
distributed equivalent, which is our open-circuited line of the appropriate length.
Notice however that, if we choose an electrical length for our line which is between 90 and
180, would be negative and hence we would have an inductive impedance. This goes to
show yet again the versatility of the distributed approach which allows you to realise both capacitors
and inductors with same topology by simply selecting the appropriate electrical length.
These short transmission line segments are usually employed as shunt elements which
branch out from the main path and are called transmission line stubs. Since we are dealing with
shunt elements, it is advantageous to use admittances instead of impedances.
The admittance of short-circuited stubs, which may be used as shunt inductors (or capacitors), may
be written as
Whereas the admittance of open-circuited stubs, which may be used in place of shunt capacitors (or
inductors), may be written as
Stubs are usually fabricated as open or short circuited microstrip lines since the planar
nature of this type of lines makes them ideal for the implementation of this technique.
Figure 4.5-4 shows the equivalence between shunt inductors and short circuited stubs. In this figure
, ( ) and the electrical length of the stub is lower than 90.
Figure 4.5-4 Equivalence between shunt inductor and short-circuited shunt stub with electrical length < 90
193
CHAPTER 4 - Impedance Matching
Figure 4.5-5 shows the equivalence between shunt capacitors and open circuited stubs. In this figure
( ), and the electrical length of the stub is again lower than 90.
ZC
ZOC
YC
YOC
Figure 4.5-5 Equivalence between shunt capacitor and open-circuited shunt stub with electrical length < 90
By equating the admittance of the lumped components to that of the respective stub, we
may obtain the electrical length which our stub needs to be to achieve the required value of
capacitance ( ) or inductance ( ).
For short-circuited stubs used as inductors we get
For example if we wanted to use a short-circuited transmission line stub, with a characteristic
impedance of 50, in place of a 4.6 nH inductor at 1 GHz we would need to use the electrical length
show below
Notice that, since the electrical length of a line is related to frequency, this is a narrow-band
approach. That is to say that the equivalence between and inductor of a specific value and short-
circuited stub of the respective electrical length is only valid at one frequency.
For open-circuited stubs used as capacitors we get
For example if we wanted to use an open-circuited transmission line stub, with a characteristic
impedance of 50, in place of a 3.2 pF capacitor at 1 GHz we would need to use the electrical length
show below
Distributed elements such as transmission stubs and series transmission lines segments can be
extremely useful when it comes to impedance matching at high frequency as demonstrated in video
4.13 and 4.14.
194
4.6 S-Parameters, Impedance and Smith Charts in MWO
Pincident
Electrical length length,
Electrical
Let us first consider Figure 4.6-1 which is essentially the same as Figure 4.4-1 where we have
added directional power meters and have specified the length of the line by its electrical length
instead of its physical length .
How can we implement this circuit in MWO?
195
CHAPTER 4 - Impedance Matching
The simulation equivalent of the circuit of Figure 4.6-1 is shown in Figure 4.6-2. Corresponding
elements in these two figures are shown by frames of the same type and colour.
First and foremost let us talk about the PORT element in Figure 4.6-2. This element
represents a signal generator, with internal impedance specified by the parameter Z, which is also
able to measure the power that flows out of its terminal and the power that is reflected back into it.
Its circuital equivalent is shown in the dashed frame in Figure 4.6-1. When we have more than one
port, some of the ports may simply behave as an impedance equal to Z ohm furnished with power
meters as shown in Figure 4.6-3. The latter are defined as passive ports whereas and the former as
active ports.
Pincident
Electrical length length,
Electrical
The second element if Figure 4.6-2 represents a transmission of line whose length is
specified in terms of its electrical length TLIN. For this element we need to specify the characteristic
impedance (Z0), which we have chosen to be the same as that of PORT 1, the electrical length (EL) in
degrees (20 in this case) and the frequency F0 which is used to calculate the electrical EL33.
Finally we have the load impedance ZL, which we have implemented with an element called
34
IMPED . This element allows us to specify explicitly the real and imaginary parts of a complex
impedance.
(EL)
196
4.6 S-Parameters, Impedance and Smith Charts in MWO
Now, if we want to measure the values of the impedance we must first measure ( ) as
seen by PORT 1 (Figure 4.6-4). To measure our reflection coefficient in MWO, we will use an S11
measurement which, as we have seen in section 3.6 is the same as the reflection coefficient for a
one-port network.
We would then normally use equation (4.1-1) to convert this value into an impedance
( ), however MWO allows us to carry out this conversion in several, simpler ways. The easiest
one would entail plotting the reflection coefficient on a Smith Chart and reading the value of the
impedance directly from the chart markings or by using a marker.
Be careful though! This impedance is not ZL! It is the combination of a line, with electrical length EL
and characteristic impedance Z0, and ZL!!
Remember that the impedance seen at PORT 1 may be derived from the phase and
magnitude of the reflection coefficient S11 ( eq.(3.5-3)). Now the magnitude of S11 is determined by
ZL alone and hence is not affected by the length of the line EL (section 3.1), but its phase is. In a
laboratory environment we would have apply a technique called de-embedding (section 3.8) in
order to work out the phase of S11 right at the load terminals and then calculate ZL from the
adjusted S11. This may of course be done in MWO also, however the simulator also allows us to
probe and measure impedances in many places in our circuit which in the real world would not be
directly accessible. This is very useful!
Now that we have illustrated some basic simulation concepts, we will delegate the teaching
of the remainder of the material on passive circuits to our video tutorials. These will explain in great
details how to analyse and design passive circuits at RF and Microwave frequencies. In particular the
reader will learn how to design various types of matching networks which are essential building
blocks of most high frequency circuits.
197
CHAPTER 4 - Impedance Matching
A brief summary of the content of the training videos which are relevant to this chapter is shown
below.
VIDEO
CONTENT
REFERENCE
Resistors, capacitors and inductors at RF. Time domain Voltage and current
waveforms at high frequency and their relationship to components impedance.
4.1 Comparison of mathematical impedance calculation and impedance values
obtained from voltage and current phase and amplitude relationships.
RC circuits and RL circuits.
198
4.7 Video Tutorials
Transmission line matching. Open and short circuited transmission line stubs.
Comparison between capacitors and open-circuited transmission line stubs.
Comparison between inductors and short-circuited transmission line stubs.
Demonstration of the fact that stubs of any type can behave as both inductors
and capacitors depending on electrical length. Clear demarcation of electrical
4.14
length boundaries.
Matching networks comprising exclusively of distributed transmission line
elements. Using I-Match for distributed matching. Comparison of various types of
matching networks (for identical terminations), in terms of insertion loss, return
loss and frequency profiles, using I-Match.
199
CHAPTER 5 - Amplifier Design
5 Amplifier Design
5.1 Introduction The transistor at Radio Frequency
In this section we will be looking at the most common methods employed to perform high-
frequency amplifier design. Modern amplifier design is almost invariably carried out with the aid of
simulation tools which employ specific models for the active devices at the heart of the amplifier.
The accuracy of the simulation results is chiefly governed by the accuracy of such models. These may
be obtained in a number of ways but they can all be grouped into two main categories: linear and
non-linear. Linear models may be employed under small signal conditions, when the amplitude of
the signals involved is such that the inherent non-linearities of the active device may be neglected.
These include linear electrical models such as the hybrid- (section 5.1.1) and S-parameter models
(section 3.7). Non-linear models may also be of circuital nature but they would include non-linear
elements such as diodes for instance. Other non-linear models are based on the actual physics of the
devices and are considerably more complex but often more accurate than electrical models. Other
characterisation methods exist but are beyond the scope of this chapter.
Although this chapter provides some concise reference material in written form, the
majority of the teaching is delegated to the video tutorials listed in section 5.5.
Let us first take a look at the small-signal35 model of a BJT in common emitter configuration.
This type of model is usually referred to as hybrid- Notice how this electrical model comprises
exclusively of linear elements. From this model it is apparent that amplifier design at high frequency
is a task that would be very difficult to accomplish without a thorough understanding of impedance
matching concepts and of the elements which are utilised to perform it.
r '
b c
r '
bb Cc
B C
I '
B C
B
r ' I
b e rce B
'
Ce E
E E
Figure 5.1-1 Small-signal, hybrid- model of a BJT
First and foremost this model reminds us of the fact that the bipolar junction transistor is a
current amplifier. We have a small current flowing from the base through to the emitter terminal
( ) and this gets amplified by a factor of thereby allowing a current equal to to flow
35
Small-signal modelling is a common analysis technique in electrical engineering which is used to
approximate the behaviour of nonlinear devices with linear equations. This linearization is performed about
the DC bias point of the device and can be accurate for small excursions about this point.
200
5.1 Introduction The transistor at Radio Frequency
between collector and emitter terminals (red ovals Figure 5.1-1). The resistance represents the
input resistance of the transistor and it is the resistance which occurs at the base-emitter junction of
the forward biased transistor. Typical values range around 1000 .
The output resistance between the collector and emitter terminals is usually rather high
in value. Since this resistance appears in parallel with the load and with the current generator
supplying the load, we want it to be as high as possible. This way the current through it would be
minimised and we would maximise the load current and hence the power delivered to our load. This
resistance is also responsible for the Early effect i.e. the IV curves sloping up in the active region of
the transistor as shown in Figure 5.1-2.
IC
IB5
IB4
IB3
IB2
IB1
VCE
Figure 5.1-2 I-V Characteristic for a BJT
The other important elements are and which allow feedback from the output to the
input to take place. The most significant of these two is the feedback capacitance , which is
formed at the reverse-biased collector-to-base junction of the transistor. As the frequency of
operation of the transistor increases, the impedance of decreases and thus has a much greater
effect on the transistor operation. A typical value for this component might be 3 pF. The feedback
resistance , which is the resistance appearing from the base to the collector of the transistor, is
a very large (> 5 M). Since the impedance of a parallel can at most be equal to the lowest of the
two, as the frequency increases the impedance of the parallel will be dominated by . Because
thruough this capacitance current can flow back to the input from our current generator instead of
going to the load, may act as to decease the gain.
The other elements in this model are:
- Base spreading resistance ( ) which is an inevitable resistance that occurs at the junction
between the base contact and the semiconductor material that composes the base. Its value
is usually of the order of tens of ohms.
- Emitter diffusion capacitance ( ) which is the sum of the emitter diffusion capacitance and
the emitter junction capacitance, both of which are associated with the physics of the semi-
conductor junction itself.
201
CHAPTER 5 - Amplifier Design
Now, having considered the parasitic elements which are internal to the transistor let us
look at the parasitics introduced by the connecting leads or bond wires. These usually introduce
some series inductance which may be modelled as shown in Figure 5.1-3.
r '
b c
LB r LC
'
bb Cc
B C
I '
B
r ' I
b e rce B
'
Ce
LE
E E
Figure 5.1-3 BJT electrical model, including lead inductance
Now let us use the models which we have illustrated in this section to take a closer look at the
impedances seen at the input and output terminals of the transistor.
LB r LC
'
bb
B C
I '
B
r ' I
b e rce B
'
CT
RL
LE
E E
Figure 5.1-4 BJT electrical model, including lead inductance and load resistance
Now let us simplify this model further by only including the elements that have an effect on the
input impedance. This is shown in Figure 5.1-5
36
is the load resistance
202
5.1 Introduction The transistor at Radio Frequency
LB r '
bb
r '
b e
Z IN CT
LE
E
Figure 5.1-5 Section of the BJT electrical model which affects the input impedance of the device
The primary contributors to the input impedance of a BJT are and , neither of which
the designer has any control over. The resistance on the other hand is a very small resistance
and hence has little influence. and can vary in size depending on circuit layout and the
designer may try and minimise them. However they may still have a significant effect when
operating in high frequency bands.
( )
r '
B bb Cc
C
I '
B
RS r '
b e
Ce
E E
Figure 5.1-6 Section of the BJT electrical model which affects the output impedance of the device
The output impedance of a transistor typically decreases with frequency. One would assume
that and are the determining factors in any output impedance calculation and that they alone
203
CHAPTER 5 - Amplifier Design
cause the output impedance to decrease with frequency. However, although and do have an
effect on the output impedance of the device, there is another, more subtle, mechanism that also
affects it. Let us assume that the transistor is in operation and that some of the collector signal is
being fed back to the base through . Some of the feedback voltage will appear across , causing
current to flow through this resistor. The BJT amplifies this current by a factor of thus increasing
the collector current. Since impedance is the ratio of voltage and current, this increase in collector
current appears as a decrease in collector impedance. and of course also act to reduce the
output impedance of the transistor through a decrease in their capacitive reactances, but the
mechanism that we have just described also plays an important role. In addition, the impedance
connected at the input terminals will also affect the output impedance of the transistor. This is
because if we increase , more of the feedback current will flow through thereby increasing
the collector current further and therefore decreasing .
204
5.1 Introduction The transistor at Radio Frequency
5.1.5 Gain
The gain that we are normally interested in for RF transistors is the power gain of the device,
rather than just the voltage or current gain. It is power gain that is important because of the
numerous impedance levels which are encountered in RF circuitry. When an impedance level
changes in a circuit, voltage and current gains alone no longer mean anything. Even a passive device
can produce a voltage or current gain but it cannot produce both simultaneously. That is what
transistors are for, to produce power gain.
The power gain of a transistor typically resembles a curve similar to that shown in Figure
5.1-7. This curve makes sense when we consider the equivalent transistor circuit of Figure 5.1-4.
Notice that what we have, in effect, is an RC low-pass filter with a gain which must fall off (neglecting
lead inductance) at the rate of 6 dB per octave. The maximum frequency at which the transistor
provides a power gain is called . The gain curve passes through at 0 dB (i.e. Gain=1), and at
the rate of 6 dB per octave.
Power Gain (dB)
6dB/octave
f max
Frequency
205
CHAPTER 5 - Amplifier Design
S21
C
S11 S22
B
S12
Figure 5.2-1 2-port S-parameters for a BJT
We have seen how the reflection coefficient for a one port network allows us to work out
the impedance of the network as seen from the measurements point (section 3.5). For a two port
network (section 3.8) we get a similar deal but in this case we have both an input and output
impedance. The input impedance may be derived from the input reflection coefficient, S11, whereas
the output impedance may be derived from the output reflection coefficient, S22.
The S21 parameter, which is always less than 1 for a passive network, helps us work the gain
of the transistor. Remember however that S-parameters are voltage-type quantities (section 3.7)
and that we are interested in power gain (section 5.1.5).
In order to get the basic power gain of the transistor therefore, we need to square the modulus of
S21.
| |
206
5.2 Two-port S-Parameter characterisation of a transistor
( ) (| | ) (| |)
In Microwave Office you have an option to display | | is dB, and the simulator carries out this
conversion according the formula above.
Lastly S12 allow us to measure the power flowing back into the input port from the output
port thereby allowing us to work much how much feedback we are getting.
207
CHAPTER 5 - Amplifier Design
o Biasing
o Stabilisation
o Impedance Matching
These are explained in detail in the following subsections and in their respective video tutorials.
V CC V CC
R B1 RC
E
RB2
In Figure 5.3-2 a different bias topology is shown which provides dc feedback. In this
network, should the collector current increase, the voltage at the collector terminal of the transistor
will decrease, due to a higher voltage drop across the resistor RC. This, in turn, will decrease the
voltage difference between base and collector terminal VBC thereby decreasing the base current.
208
5.3 Amplifier Design Stages
A lower base current will determine a lower collector current and thereby act to correct the initial
increase in collector current. A dual argument applies if the collector current decreases. (video 5.1)
V CC
RC
RB
The bias network shown in Figure 5.3-3 provides a further improvement since it may be
implemented with lower values of resistance and hence is more compatible with thin- or thick-film
resistor values which are often employed in radio frequency circuits.
V CC
RC
R B1
RB E
RB2
Figure 5.3-3 Voltage feedback bias network which allows lower resistance values to be used
Active bias topologies are also an option and usually provide better bias stability, however
they are more expensive to realise.
209
CHAPTER 5 - Amplifier Design
Figure 5.3-4 shows that if we were to design a transistor at 1.9 GHz for instance we might
run the risk of using terminations which would cause oscillations. However, yet again, by using
mathematical formulae based on S-parameters we may derive two very useful graphical aids which
may be plotted on a Smith chart: Input and Output stability circles.
37
If we have a non-linear model for the device, instead of an S-parameter set for the device, we can always use
a simulator to derive the S-parameters from the model.
210
5.3 Amplifier Design Stages
Input stability circles allow us to identify the region of the Smith chart which contains
terminations which, if connected at the input of the transistor, will make the magnitude of the
output reflection coefficient S22, greater than one.
The output stability circles help us identify the region of the Smith chart which contains
terminations which, if connected at the output of the transistor will make the magnitude of the
input reflection coefficient S11 greater than one.
Since oscillations are possible when either input or output reflection coefficients are greater
than one, these circles greatly help us avoid picking unfriendly terminations which may make the
transistor oscillate.
Microwave Office allows us to plot stability circles very easily as we will see in video 5.3 and
5.4 and clearly identify whether the inside or the outside of the circles contain the terminations
which may cause potential instability. For instance in Figure 5.3-5, the dashed line on the Input
Stability circle plot (SCIR1) clearly indicates that the inside of the circle contains the troublesome
terminations.
Similarly for the output stability circle (SCIR2), the set of terminations which may cause
potential instability is on the inside of the circle. This is again indicated by the dashed lines and
clearly marked in Figure 5.3-6.
211
CHAPTER 5 - Amplifier Design
There are several ways to make our transistor unconditionally stable, most of which involve
giving up some of the potential gain of the transistor in order to achieve a state where, no matter
what terminations we use for our input and output, oscillations will not occur.
Videos 5.3 and 5.4 illustrate these concepts further.
212
5.3 Amplifier Design Stages
| |
S21
Z0
S12
Figure 5.3-7 S-parameters and basic gain measurement for a biased transistor
Then we may look at the input impedance of the transistor, which may be obtained from the
S11 parameter, and design a matching network so as to transform the source impedance into the
complex conjugate of the transistors own input impedance.
213
CHAPTER 5 - Amplifier Design
| |
A conjugate match at the input allows us to improve the gain as much as possible by means of the
input matching network. In general however we may decide not to maximise this gain and apply
some selective mismatch in order to achieve a specific value of Gain. We can think of our input
matching network as a gain dial, which may increase the gain by a factor of whose maximum
value is . This is shown in Figure 5.3-8.
S21
Z0
Input Transistor
matching
S11 + S22
S
network Bias Z0
GS G0
S12
G0
GS
Figure 5.3-8 Gain tuning by means of input matching network with unilateral device
| |
| |
It is apparent that this expression is maximum when i.e. when a conjugate match is
achieved.
A similar argument applies for the output. If we design a matching network which
transforms the load impedance into the conjugate of the output impedance of the transistor we can
improve the gain by a factor of
| |
214
5.3 Amplifier Design Stages
Yet again, we may want to apply some selective mismatch to achieve a specific value of gain. Our
output matching network can therefore also be seen as a gain dial which provides a gain
| |
| |
whose maximum value is achieved when i.e. when a conjugate match is achieved at the
output.
The complete picture is shown in Figure 5.3-9.
S21
Z0
S12
G0
GS GL
Figure 5.3-9 Gain tuning by means of input and output matching networks with unilateral device
| | | |
| |
| | | |
(5.3-1)
Our unilateral matching video tutorial (video 5.5) will give the reader a greater insight into this
technique and its limitations.
Note that the gain is often expressed in dB. In this case eq. (5.3-1) becomes
215
CHAPTER 5 - Amplifier Design
216
5.3 Amplifier Design Stages
Swp Max
0.8
6000MHz
6
0.
0
2.
p1
4
0. p2
0
p3 3.
p4 4.
0
5.0
0.2
10.0
10.0
0.2
0.4
0.6
0.8
1.0
2.0
3.0
4.0
5.0
0
-10.0
p1: Freq = 6000 MHz
-0.
2
0
G = 11.381 dB
-5.
.0
-4 p2: Freq = 6000 MHz
.0 G = 10.381 dB
.4
-3
-0
p3: Freq = 6000 MHz
.0
G = 9.3815 dB
-2
.6
-0
Swp Min
-1.0
6000MHz G = 8.3815 dB
Picking a termination along an operating gain circle corresponding to a specific gain value is
only the first step of this procedure! Once a value for the output termination has been chosen, a
matching network must be designed to transform our load impedance into such a termination. Once
the output matching network is in place we must measure the S11 of the transistor yet again to
obtain its new input impedance. We must then design a matching network to transform the source
impedance into the complex conjugate of the transistors own input impedance. Only then the gain
that was specified by the circle on which we selected the output termination may be achieved!
Notice how this technique may be applied to both stable and unstable devices. In the case of
stable transistors, the circles will be fully contained within the Smith Chart whereas if the device is
potentially unstable, part of the circles may lay outside the chart.
This is illustrated in video tutorials 5.8 and 5.9.
217
CHAPTER 5 - Amplifier Design
6000MHz
6
0.
0
2.
4
0.
0
3.
0
4.
5.0
0.2
10.0
p1 p2
10.0
p3 p4
p1: Freq = 6000 MHz
0.2
0.4
0.6
0.8
1.0
2.0
3.0
4.0
5.0
0
G = 11.381 dB
-10.0
p2: Freq = 6000 MHz
2
-0. 0
-5. G = 10.381 dB
.0
-4
-0.8
Swp Min
-1.0
6000MHz G = 8.3815 dB
Picking a termination along an available gain circle corresponding to a specific gain value is
only the first step of this procedure! Once a value for the input termination has been chosen, a
matching network must be designed to transform our source impedance into such a termination.
Once the input matching network is in place we must measure the S22 of the transistor yet again to
obtain its new output impedance. We must then design a matching network to transform the load
impedance into the complex conjugate of the transistors own output impedance. Only then the gain
that was specified by the circle on which we selected the input termination may be achieved! Notice
how this technique may be applied to both stable and unstable devices.
In the case of stable transistors, the circles will be fully contained within the Smith Chart
whereas if the device is potentially unstable, part of the circles may lay outside the chart.
This is illustrated in video tutorial 5.10.
218
5.3 Amplifier Design Stages
<FREQ> |S11| <S11 |S21| <S21 |S12| <S12 |S22| <S22 (for Magnitude and Angle)
<FREQ> Re{S11} Im{S11} Re{S21} Im{S21} Re{S12} Im{S12} Re{S22} Im{S22} (for Real Imaginary)
<FREQ> 20log10|S11| <S11 20log10|S21| <S21 20log10|S12| <S12 20log10|S22| <S22 (for dB
Angle)
It may be typed in a text editor but will need to be saved with a .s2p extension. (video 5.8)
219
CHAPTER 5 - Amplifier Design
A comprehensive set of videos tutorials on amplifier design is available to the reader. These are
essential to the understanding of amplifier design and form the core of the amplifier section. The
textbook should just be used for a reference and a brief introduction.
VIDEO
CONTENT
REFERENCE
220
5.5 Video Tutorials
Operating Gain Circles Part 1. Writing 2-port S-parameter files (.s2p) and using
them for amplifier design. Amplifier design for a specific gain with a stable
5.8
device. Operating gain circles (GPC_MAX) and maximum available gain (GMAX).
Effects of this technique on input and output VSWR. Using sub-circuits in MWO.
Operating Gain Circles Part 2. Amplifier design for a specific gain with an
5.9
unstable device. Operating gain circles (GPCIR) and maximum stable gain (MSG).
Available Gain Circles. Amplifier design for a specific gain with an unstable device
5.10
using available gain circles (GACIR). MDIF transistor models.
221
References
[1] Gonzalez, G., Microwave transistor amplifiers : analysis and design. 2nd ed. ed. 1997, Upper
Saddle River, N.J. London: Prentice Hall ; Prentice Hall International.
[2] Bowick, C. and P. Elsevier Science, RF circuit design. 1997, Boston: Newnes.
[3] Besser, L. and R. Gilmore, Practical RF circuit design for modern wireless systems. 2003,
Boston, Mass., London: Artech House.
[4] Gilmore, R. and L. Besser, Practical RF circuit design for modern wireless systems. 2003,
Boston, Mass. ; London: Artech House.
[5] Halliday, D., R. Resnick, and K.S. Krane, Physics. 5th ed. ed. 2002, New York ; Chichester:
Wiley.
[6] Pozar, D.M., Microwave engineering. 2nd ed. ed. 1998, New York ; Chichester: Wiley.
[7] Maas, S.A., The RF and microwave circuit design cookbook. 1998, Boston ; London: Artech
House.
222