2

3
Copyright © 2016 by

The American Radio Relay League, Inc.

Copyright secured under the Pan-American ​Convention

International Copyright secured

All rights reserved. No part of this work may be reproduced in any form except by

written permission of the publisher. All rights of translation are reserved.

Printed in USA

Quedan reservados todos los derechos

ISBN: 978-1-62595-049-9

First Edition

4
 CONTENTS
Foreword
Preface
About the Author
Introduction
About the ARRL
Part 1 Antenna Physics
1 Development of Antenna Physics
Early History
Modern History
Maxwell, the Turning Point
1901 to Present
Enter the Antenna
2 Fundamentals
What is an Antenna?
The Sphere as a Measurement Tool
Analogy Between Light and Radio Waves and
the Concept of Aperture
Defining the Critical Fundamental Terms
The 1⁄2 Wavelength Dipole Antenna
Relationship Between Aperture and Gain
Empirical Calculation of the Aperture of an
Isotropic Antenna
Free Space Path Loss
Refinements of the Aperture Concept

5
3 Radiation of Radio Waves
Modern Expression of Maxwell’s Equations
Maxwell’s First Equation, Gauss’s Law for
Electricity
Maxwell’s Second Equation, Gauss’s Law for
Magnetism
Maxwell’s Third Equation, Faraday’s Law of
Induction
Ampere’s Law
Maxwell’s Fourth Equation
The Wave Equation
Wave Equation for Electromagnetic Waves
Relationships Between the E and B fields
Impedance of Free Space
Fields, Power, and Energy
Radiation Resistance
Relationship Between Radiation Resistance
and Feed Point Impedance in Linear Antenna
Structures
Antenna Pattern Formation
PADL
Current and Voltage Distribution
Antenna Polarization
4 Transmission Lines
The Guided Wave
Radiation Loss Rr
Characteristic Impedance

6
 Characteristic Impedance of Two-Wire
Transmission Line
Characteristic Impedance of Coaxial Cable
Velocity Factor
Types of Guided Waves
Transverse Modes
Losses in Transmission Lines
Reflected Waves
VSWR and Standing Waves
5 Antennas Using Multiple Sources
Driven Arrays
Mutual Impedance
Quadrature Array
Parasitic Arrays
Part 2: Applied Antenna Physics
6 Dielectric Effects Upon Radio Waves
Reflection and Absorption
Conductors and Dielectrics
Wave Propagation in Dielectrics
Snell’s Law
The Earth as a Dielectric
Effect on Wave Polarization
Horizontal Polarized Wave Reflection
Vertical Polarized Wave Reflection
Assumption of an Infinite Plane
7 Vertical Antennas

7
Ground Mounted Vertical Antennas
Radiation Resistance of a Ground-Mounted
Base-Fed Vertical Antenna
Effective Height
Example of the Effects of he on Rr
Directivity of a Ground-Mounted Vertical
Antenna
Ground Influences on Ground-Mounted
Verticals
Experimental Method to Derive Rl
Feed Point R vs Maximum Rr
Verticals Higher than 1⁄4 l
Raised Vertical Antennas
Raised Vertical Antennas Over Earth
Short Vertical Antennas
Data for Various Ground-Mounted Unloaded
Vertical Physical Heights
Ground Proximity to Current Maximums
The Near-Ground Inverted-V Antenna
Near-Ground Phased Inverted-Vs
Broadside Bobtail Curtain
Single-Element Vertical (Monopole Antenna)
Multielement Vertical Arrays
Gain Comparison of Vertical Arrays
Conclusion
EIRP Calculation for Ground-Mounted Low
Frequency Vertical Antennas

8
 Ground Losses
Vertical Element Losses
Matching Losses
Summing the Losses and Determining
Required Transmitter Power Output
8 Yagi-Uda and Cubical Quad Antennas
Gain and Aperture of a Yagi-Uda Antenna
Stacking Yagi-Uda Antennas
Design and Optimization of a Yagi-Uda Array
The W8JK Array
Multiband Yagi-Uda Arrays
The Cubical Quad
Quad Parasitic Elements in Vertical Arrays
9 Specialized Antenna Configurations
Small Loop Antennas
Larger Small Loops
Helix Antennas
Small Helix Antenna
Aperture Antennas
Small Antennas
Traveling Wave Antennas
Antennas for Navigation and Location
Determination
Patch Antennas
Phased Array
Aperture Distribution
Polarization Diversity

9
 MIMO Antennas
Other Configurations
10 Noise, Temperature, and Signals
Boltzmann’s and Plank’s Constants
Boltzmann Noise in Circuits
Random Processes
Radio Receiver Sensitivity
Calculating a Minimum Antenna Gain
Specification
Noise Temperature
Antenna Temperature
A Nobel Prize
Atmospheric and Galactic Noise
Photons in Radio Waves
Atmospheric Noise Considerations at MF and
Low HF
Receive Antenna Parameters for MF and Low
HF
Shannon’s Limit
Extreme System Design
Physics Defines the Ultimate Limits to Radio
Communication

Appendix A — You Already Know Calculus

Appendix B — Summary of Rr Terms Described
by Kraus
Appendix C — References

10
Appendix D — Symbols and Abbreviations
Appendix E — Comparison of Elevation Plots of
Vertical-Polarized Ground-Mounted 160 Meter
Antenna Arrays

11
 FOREWORD
It’s probably fair to say that the topic of
antennas has generated more interest and
discussion among radio amateurs than any other.
Antenna installation or construction is often the
first do-it-yourself project an amateur
undertakes, and most of us have tried several —
or many! — different versions during our
amateur careers. Countless books and articles
have been written describing basic antenna
theory and “how-to” construction details, and
this is the largest category in the ARRL
Bookstore.
Antenna Physics: An Introduction is not another
one of those books. In this work, Bob Zavrel,
W7SX, strives to bridge the gap between basic
theory and graduate-level engineering texts. He
explains many of the underlying principles of
antennas and antenna physics and introduces the
reader to the mathematics behind these
principles. Although some competence in
mathematics is required, readers may follow
along and understand the concepts without
needing to solve the complex equations
presented. In later chapters, in the Applied
Antenna Physics section, Bob presents examples

12
to tie the concepts learned in earlier chapters to a
number of antenna types familiar to radio
amateurs.
We hope that you enjoy this approach and gain
a deeper understanding of how antennas work. It
might even inspire you to seek out some of the
more advanced texts listed in the References
appendix to delve deeper into this fascinating
and important subject.

David Sumner, K1ZZ

Chief Executive Officer
Newington, Connecticut
November 2015

13
 PREFACE
Why Write a Book About “Antenna Physics”?
After nearly 50 years of studying the subject
and designing, building, and using antennas for
both amateur and professional applications, I had
come to a conclusion. There are many very fine
books available on this fascinating subject, and
they usually fall within one of two categories: (1)
relatively simple treatments that provide “how-
to” material as well as basic theory to support
some level of fundamental understanding and (2)
highly advanced theoretical graduate-level
engineering texts.
There appeared to be a gap between these two
extremes. Thanks to the well-written and
comprehensive “how-to” publications of the
American Radio Relay League, I was able to
master much of the basic theory of antennas by
the time I graduated from high school. These
books provide enough theory to gain a sufficient
background to design, build, measure, and use
antennas. However these texts only instilled a
craving for deeper understanding of the
underlying physics of antennas. The leap from
these introductory texts to the masterpieces of
Kraus, Jasik, Balanis and others necessarily had

14
to wait for competence in advanced
mathematics, including the indispensable
calculus. After many years of using these books
as references and thus providing ever deeper
insights into the workings of antennas, I
concluded that many of the more advanced
treatments could be explained, at least in part,
using a simpler approach. This book is an
attempt, therefore, to “bridge the gap” between
the introductory and “how to” books and the
advanced texts. I hope some readers will use this
introductory text as a springboard to the more
advanced treatments of this fascinating subject,
indispensable to all forms of radio applications.
Although many of the antenna applications will
be most useful to Amateur Radio operators,
about two-thirds of the text should also find
more widespread use. I hope that it will serve
electrical engineering students and professional
radio engineers as well as the amateur
community.
Advanced mathematics is an essential tool for
solving complex antenna problems. However, it
is not necessary for understanding the deeper
principles. This is one of the key motivations and
challenges to writing this book: you can
understand the physics without being able to
solve a calculus problem, but it is of great

15
advantage to understand what the equations are
saying. Indeed, many of the complex-looking
equations that are found in this text are actually
only special forms of much simpler equations
which are second-nature to anyone with a
competence in electronics. In this text I often
point out the simpler forms as the calculus is
presented. For example, many of the complex
equations are merely special forms of well-
understood relationships like Ohm’s Law or the
power law.
Some competence in mathematics is required
for this text, particularly algebra. Also, being able
to deal with some trigonometry as well as
logarithms is essential to grasping the
fundamental concepts. When more advanced
equations are presented, explanations are often
offered to explain what the reader is looking at,
and what it means, particularly for equations
involving calculus and linear algebra. Of course a
mastery of calculus and other forms of advanced
mathematics will afford an even deeper insight
when combined with dedicated study of antenna
physics proper.
Physics is classified as one of the natural
sciences, together with mathematics, chemistry,
biology, and geology. Students of these
disciplines (purists) are only interested in

16
describing nature. In contrast, the engineering
sciences, concern themselves with designing and
building something. The engineering sciences are
completely dependent upon the natural sciences
for the laws and theories needed to do or build
something useful. On the other hand,
engineering that utilizes theories based in natural
science provides excellent vindication of the laws
and theories rooted in natural science. The
business of natural science is to provide
dependable, repeatable experiments with similar
results. Engineers need repeatable results to
design and build something that will be
manufactured. For example, the esoteric
theoretical subject of sub-atomic physics was
validated with the first atomic explosion. For the
pragmatic person, this is the ultimate test of a
“theory”: what are the useful results? Or, what
are the possible useful results armed with this
new theory? What can we do with this new
knowledge?
In this book we will explore how a
mathematician (Maxwell) first predicted that
radio waves exist. Then a physicist (Hertz)
actually built an experimental apparatus to prove
radio waves exist, and finally an engineer
(Marconi) actually developed radio as something
useful. Of course it wasn’t that straightforward.

17
Thousands of scientists and engineers have made
contributions, large and small to develop
“wireless” as we know it today, but the sequence
of theory, proof, and finally “let’s do something
with this” is clear. Therefore, an antenna
designer, builder, or even just a user can become
more capable if the basic underlying principles
(the physics) are understood.
After a lifetime of learning, experimenting,
mentoring and teaching antenna theory, I found
several topics to be commonly misunderstood.
Having to repeat the same ideas multiple times
over the years, extra attention is given to topics
that reflect these misunderstandings among not
only amateurs, but also professionals. These
topics, understandably are the more complex and
less intuitive. Some repetition of the basic “how
to” books is essential in writing such a text.
However, I attempted to minimize repetition
assuming the reader already understands these
basics or has access to basic texts such as the
ARRL Antenna Book.
“Applied” examples were chosen to gain
insights into common antenna types, which can
then lead to aids in engineering antennas. Also,
many of these applied examples, indeed
chapters, reflect my serious interest in Amateur
Radio and the publisher, the American Radio

18
Relay League (ARRL). Amateur Radio has been an
avocation since my childhood, and participation
in this wonderful hobby led me ultimately into a
lifelong career in electronic engineering with a
specialty in RF. The publications of the ARRL
have been and remain indispensable for the radio
amateur and many engineers and technicians
alike. In this vein I hope that this text provides
value to amateurs, students, and professionals.
Preface to the Text
The organization of this text is different from
both the more traditional introductory and
advanced texts. The beginning chapters weave
actual electricity and magnetism (E&M) material,
such as Maxwell’s equations taken from pure
classical physics, with how these concepts relate
directly to antennas. The reasoning is simple. In
order to develop a deeper understanding of
antennas, it is necessary to gain a deeper
understanding of the fields with which they
interplay and the characteristics of free space, the
medium in which they perform.
Many of the terms in common use regarding
antennas are redefined in relation to terms
typically quantified in the more advanced texts
but ignored or only briefly encountered in the
basic texts. Hopefully these discussions will shed

19
new light and deeper insights on these basic
terms used in discussions on antennas. This is
the essence of the first three chapters.
It is common practice for authors to simply
reference equations already presented in previous
chapters. I have always found this practice
awkward, having often to thumb back through
the chapters to find the applicable equation. I
would prefer readers to find the text a bit
repetitive rather than assuming equations and
other points made earlier are now embedded in
the reader’s memory. Also this book is written
somewhat in a “narrative” form. The intent is for
chapters to flow from one to the next. Hopefully,
however, the chapters will also stand by
themselves as references to the applicable topics.
Many important topics are presented in an
abbreviated form, especially when these topics
are very well covered by introductory and/or
specialized texts commonly available.
Part 1: Antenna Physics
Chapter 1: Development of Antenna Physics:
We first present an introduction of the
theoretical development of radio and then show
how antenna theory was co-developed as an
essential element to “radio.” Taken in a historical
context, it also offers an acknowledgement of

20
some of the great indispensable contributors to
the art.
Chapter 2: Fundamentals: Building upon the
introductions in Chapter 1, we begin by
introducing some of the basic physical
parameters of antennas. The approach quickly
diverges from the “how- to” books and
concentrates on developing standards of
measurements, the concept of aperture, analogy
to light, relationships among directivity, gain and
aperture, the concept of an isotropic antenna,
path loss, and radiation resistance.
Chapter 3: Radiation of Radio Waves: Again,
building from the previous chapter, we develop,
in detail, how radio waves “radiate” and how
antennas perform this task. An introduction is
made to Maxwell’s equations, the wave equation,
relationship between the magnetic and electric
fields, the impedance of free space, power,
energy, and fields, detailed definition of radiation
resistance, and how fields and antenna patterns
are formed. In order to clarify some concepts, the
dipole antenna is also introduced. The concept of
“PADL” is also introduced. This chapter, in
particular, diverges from most antenna texts in
that it roots antenna performance and
characteristics in electromagnetic physics.
Chapter 4: Transmission Lines: The thrust of

21
this chapter is to distinguish the “guided wave”
from the “radiated wave.” Rather than repeat
excellent texts available on transmission lines, we
introduce transmission lines in the context of
principles developed in the earlier chapters. For
example, the creation of a transmission line from
a 1⁄2 wavelength dipole is discussed in detail. Then
the two basic forms of transmission line are
discussed and developed from the physical
properties defined in earlier chapters.
Chapter 5: Antennas Using Multiple Sources:
We explain how directive antennas can be
constructed from multiple elements, including
multi​element driven and parasitic arrays.
Part 2: Applied Antenna Physics
Chapter 6: Dielectric Effects Upon Radio
Waves: Most antennas we encounter are located
close to the Earth’s surface. Therefore, no
antenna text (even a physics book) is complete
without a treatment of the physics of the Earth
— “geophysics” — as relating to radio waves.
Since the Earth is a dielectric, dielectrics are
treated first in a general discussion. Then the
Earth is discussed as a special example of
dielectrics.
Chapter 7: Vertical Antennas: Much of the
material presented in Chapter 6 becomes

22
particularly relevant to ground-mounted vertical
antennas. Amateur Radio experimenters may
find this chapter particularly interesting
stemming from the intense interest in the
amateur community regarding vertical antennas.
A mathematical format (scalar matrix) is
introduced for the quantification of vertical
antenna losses. Also, a simple comparison of
vertical configurations is also offered with their
relative major advantages and disadvantages.
Chapter 8: Yagi-Uda and Cubical Quad
Antennas: The Yagi-Uda configuration has
become one of the most commonly used antenna
forms. Although volumes have been written on
this antenna type, the purpose of this chapter is
to relate its performance to the physical
parameters already developed in this text.
Chapter 9: Specialized Antenna
Configurations: This chapter presents some of
the more important, but less-known antenna
configurations. Loop and helix antennas are
presented first, followed by “aperture” antennas,
traveling wave antennas, and finally a detailed
discussion on small antennas (relative to
wavelength)
Chapter 10: Noise, Temperature, and Signals:
This final chapter attempts several goals. The
primary importance of noise and noise

23
temperature are presented and are shown to be
intrinsically related. Then “noise temperature” is
shown to be an important antenna specification,
especially for radio astronomy and deep space
applications. Radio link budgets are presented,
showing how noise sets fundamental limits. We
also return to include some historical narrative
culminating in great scientific achievements. The
epic discovery of Penzias and Wilson is
presented as an outstanding research application
of an antenna.
The work of Claude Shannon and information
theory are introduced. Although not related
directly to antennas, the discussion of
information theory is presented since it links
noise to fundamental physical limiting factors for
both the radio link and on the limits to
communication of information. Thus the reader
is left with all the known physical limits (and
opportunities) for communicating information
using radio links, for which the antenna forms a
critical function. These are the same limits for
garage door openers, cell phones, WiFi, and even
establishing a hypothetical link with an
extraterrestrial civilization.
Acknowledgements
First, I would like to acknowledge the American

24
Radio Relay League publications that provided
me with the critical first level of understanding of
radio technology and to the point of this text,
antenna theory. Also, Mark Wilson, K1RO, who
guided this text through ARRL’s publication
process and Larry Wolfgang, WR1B, with whom I
have worked as a volunteer ARRL technical
advisor for many years. It is my hope that this
text will offer yet another useful volume to the
League’s library.
Some of my “elmers” dating from the 1960s also
deserve mention: Martin Hornak, W8MAE, and
Orlando O. Okleshen, W9RX, provided needed
guidance and help through those difficult teenage
years. Later my professors at the University of
Oregon Physics Department, especially Dr. K.J.
Park deserve mention. Although I never met
him, the famous text written by Dr. John Kraus,
W8JK, remains my “bible” for antenna
engineering. My friends and fellow authors Wes
Hayward, W7ZOI, and Cornell Drentea, KW7CD,
showed me through example how hard work and
dedication can result in the production of fine
technical books. The narrative style of Cornell’s
book provided the inspiration to present this text
in a similar vein.
Gary Breed, K9AY, provided an excellent
editorial review of the entire text and offered

25
numerous useful suggestions and corrections.
Gary also wrote the Introduction. Finally Dr.
Nuri Celik of Trimble Navigation performed a
detailed check of the most difficult math,
particularly from Chapter 3, and yes, he found
some errors! However, the errors were minor and
few, and I hope this will ring true for the entire
text.

Robert J. Zavrel, Jr, W7SX

Elmira, Oregon
September 28, 2015

26
 ABOUT THE AUTHOR
Robert J. Zavrel, Jr, W7SX, was born in Chicago
in 1951. His interest in radio developed very early,
having dissected his first radio at age four, much
to his mother’s horror seeing him on the floor
playing with an opened tube radio plugged in. He
was raised in Chicago Heights, Illinois, where he
passed his radio amateur Novice test when he
was 14 years old. He earned DXCC (an award for
contacting 100 different countries) when 17 years
old. After attending Roosevelt University in
Chicago, he worked as a radio navigator in the
Gulf of Mexico and as a radio broadcast engineer.
In 1979 he returned to school at the University of
Oregon, where he earned a BS in Physics in 1983.
After graduation he moved to Silicon Valley
where he worked for 12 years in RF
semiconductors for Siliconix, Philips, GEC
Plessey and several startups. In 1995 he moved to
RTP in North Carolina and worked with IBM’s
SiGe process and then to ATMEL in Colorado
Springs as Director of their SiGe process. In 2002
he started an RF engineering consulting company
that he ran for 10 years. In 2012 he joined Trimble
Navigation Ltd. In Corvallis, Oregon, where he
remains a Senior RF R&D engineer with a current

27
concentration in antenna design.
Bob has five patents, plus two additional recent
patent filings, and has published more than 60
technical papers in both amateur and
professional publications. He wrote the first
papers on software defined radio and using direct
digital synthesis (DDS)M for both FM and HDTV
applications, now all developed technologies and
industry standards. He patented the 8-circle
antenna array and has designed many antennas
for commercial and military applications. His
work has sent him to 35 countries on six
continents and has lived a total of three years in
Europe. He has earned Honor Roll status for
DXCC and CW DXCC using only tree-supported
wire antennas and has earned DXCC on all bands
160 – 10 meters. His amateur station consists of
an array of homebrew amplifiers, tuners, control
systems, and of course novel antenna designs. In
addition to Amateur Radio he grows Pinot noir
grapes and makes award-winning wine in his
boutique noncommercial winery.

28
 INTRODUCTION
I have known Bob Zavrel, W7SX, for 30 years.
We share an intense passion to know more about
antennas and RF circuits — well, actually, just
about everything scientific, with a bit of
philosophy thrown in for good measure! We
understand that many other radio amateurs want
to know more about antennas, but do not have
the formal education in physics and mathematics
necessary to delve into the theory of
electromagnetism. This book was written with
the goal of helping those Amateur Radio
colleagues follow their curiosity, with the
intention of providing a substantial step forward,
beyond the how-to articles and basic concepts
presented in magazines and handbooks.
When Bob asked me to “look over his shoulder”
as he wrote this book, I agreed, knowing that we
would probably disagree at times on things such
as choosing the best example to illustrate a
particular concept, or finding an appropriate
analogy to explain a complex concept. I’m
pleased to say that those times were few.
A mid-level book like this is a challenge for the
author! How much technical depth is
appropriate? What topics need the greatest

29
emphasis? Readers will likely experience the
same sort of response. They will be pleased with
some explanations, but may find that others just
don’t resonate. One reader may find the math
difficult, while another wants more detail. If this
book encourages them to pursue further study to
gain the additional background and find the
alternate descriptions they are seeking, we will
consider it a great success!

Gary Breed, K9AY

Mount Horeb, Wisconsin
September 2015

30
 ABOUT THE ARRL
The seed for Amateur Radio was planted in the 1890s, when
Guglielmo Marconi began his experiments in wireless
telegraphy. Soon he was joined by dozens, then hundreds, of ​-
others who were enthusiastic about sending and receiving
messages through the air—some with a commercial interest,
but others solely out of a love for this new communications ​-
medium. The United States government began licensing
Amateur Radio operators in 1912.
By 1914, there were thousands of Amateur Radio operators—
hams—in the United States. Hiram Percy Maxim, a leading
Hartford, Connecticut inventor and industrialist, saw the need
for an organization to band together this fledgling group of
radio experimenters. In May 1914 he founded the American
Radio Relay League (ARRL) to meet that need.
Today ARRL, with approximately 165,000 members, is the
largest organization of radio amateurs in the United States.
The ARRL is a not-for-profit organization that:
• promotes interest in Amateur Radio communications and
experimentation
• represents US radio amateurs in legislative matters, and
• maintains fraternalism and a high standard of conduct
among Amateur Radio operators.
At ARRL headquarters in the Hartford suburb of Newington,
the staff helps serve the needs of members. ARRL is also
International Secretariat for the International Amateur Radio
Union, which is made up of similar societies in 150 countries
around the world.
ARRL publishes the monthly journal QST, as well as
newsletters and many publications covering all aspects of
Amateur Radio. Its headquarters station, W1AW, transmits
bulletins of interest to radio amateurs and Morse code practice
sessions. The ARRL also coordinates an extensive field
organization, which includes volunteers who provide technical

31
information and other support services for radio amateurs as
well as communications for public-service activities. In
addition, ARRL represents US amateurs with the Federal
Communications Commission and other government agencies
in the US and abroad.
Membership in ARRL means much more than receiving QST
each month. In addition to the services already described,
ARRL offers membership services on a personal level, such as
the ARRL Volunteer Examiner Coordinator Program and a QSL
bureau.
Full ARRL membership (available only to licensed radio
amateurs) gives you a voice in how the affairs of the
organization are governed. ARRL policy is set by a Board of
Directors (one from each of 15 Divisions). Each year, one-third
of the ARRL Board of Directors stands for election by the full
members they represent. The day-to-day operation of ARRL
HQ is managed by a Chief Executive Officer.
No matter what aspect of Amateur Radio attracts you, ARRL
membership is relevant and important. There would be no
Amateur Radio as we know it today were it not for the ARRL.
We would be happy to welcome you as a member! (An Amateur
Radio license is not required for Associate Membership.) For
more information about ARRL and answers to any questions
you may have about Amateur Radio, write or call:

ARRL — the national association for Amateur Radio®

225 Main Street
Newington CT 06111-1494

32
Voice: 860-594-0200
Fax: 860-594-0259
E-mail: [email protected]
Internet:www.arrl.org/

Prospective new amateurs call (toll-free):

800-32-NEW HAM (800-326-3942)
You can also contact us via e-mail at [email protected]
or check out ARRLWeb at http://www.arrl.org/

33
 Part 1
Antenna Physics

34
 CHAPTER 1

Development of Antenna
Physics
Early History
“Electricity” had probably been observed since the beginning
of the human species in the form of lightning and perhaps
static electricity raising hairs on skin clothing worn by early
cave-dwelling experimenters. From ancient Greece, Aristotle
and Thales described the effects of magnetism. Much later, in
the 11th century, the Chinese scientist Shen Kuo was the
inventor of the compass and employed it for navigation.
Detailed rational explanations of electricity and magnetism
had to wait until the 19th century. The invention of radio and
thus antennas followed quickly after the fundamental
breakthroughs.
Modern History
During the Renaissance in Europe, the scientific method
began to replace mystical explanations for a wide variety of
unexplained phenomena. Careful, measured experiments were
undertaken and documented. Following is a chronological
record of some of the more important breakthroughs related to
electromagnetics:
• 1590s: William Gilbert (English) conducted the first formal
experiments in electricity and magnetism. His observation that
the Earth is actually a magnet directly led to refinements of the
compass.
• 1746: Jean le Rond d’Alembert (French) formulated the one-
dimensional wave equation describing the motion of a
vibrating string. Although his work had nothing to do with
electricity or magnetism, it would prove indispensable a
century later.
• 1750: Benjamin Franklin (American) developed the law of

35
conservation of charges, found there were two types of charges
(positive and negative), and invented the lightning rod.
• 1755: Charles Augustin de Coulomb (French) invented a very
sensitive balance and used it to relate mechanical forces due to
electric and magnetic fields.
• 1780: Karl Friedrich Gauss (German) developed the
divergence theorem, which later would become part of
Maxwell’s equations.
• 1800: Alessandro Volta (Italian) invented the voltaic cell
and the battery, thus providing, for the first time, a continuous
moving charge — an electric current.
• 1819: Hans Christian Oersted (Danish) observed that an
electric current flowing on a wire deflected a compass, thus
showing that electricity and magnetism were somehow
intrinsically related.
• 1820: Andre Marie Ampere (French) invented the solenoid
and thus produced a significantly higher magnetic force from
an electrical current. He also quantified the relationship
between the current and magnetic flux.
• 1821: Georg Simon Ohm (German) derived his famous law
relating voltage, current, and resistance
• 1831: Michael Faraday (English) discovered and quantified
the phenomena of a changing magnetic field produced an
electrical current, which is the reciprocal of Oersted’s findings.
This implied that electricity and magnetism were more closely
related more than previously thought.
• 1831: Joseph Henry (American) independently observed the
same phenomena as Faraday.
Maxwell, the Turning Point
• 1860s/70s: James Clerk Maxwell (Scottish, Figure 1.1) saw
that Faraday’s work was the final concept needed to postulate a
unified theory of electromagnetism. In effect, ​Maxwell
combined the equations of Faraday, Gauss and Ampere into
one set of relationships that precisely define the behavior of
electromagnetism, and more importantly for this text, the
behavior of electromagnetic waves.

36
 Maxwell also postulated (correctly) that light is an
electromagnetic wave by a brilliant observation. Coulomb and
Ampere, through static measurements, were able to determine
the permittivity (ε0) and permeability (μ0) of free space by
measuring the physical forces of magnets and electric fields
imparted on objects. Maxwell noticed that the expression

actually defined a velocity, where

(Equation 1.1)
and

(Equation 1.2)

37
or

(Equation 1.3)
therefore

(Equation 1.4)
or 299,863,400 meters/second.
This is obviously a velocity, but of what? In other words the
basic free space coefficients of the magnetic and electric fields
were related to a velocity. Through simple algebra, Maxwell
reduced these measured values (distance, mechanical force,
static charge and current) to a velocity. Once he had performed
these calculations, he postulated that light must be an
electromagnetic wave after noticing that this velocity was close
to the then-measured speed of light. Since these static (and
current flow) terms could be measured with good accuracy
even in the 1860s/70s, Maxwell’s calculation of the speed of
light was more accurate than the direct measurements of the
speed of light possible in the 1870s. His calculated speed is very
close to today’s known measured speed of light.
Sixty years before Maxwell, Thomas Young discovered that
light waves could interfere with one another, supporting the
idea that light was composed of waves, and the distance
between the ripples in the interfering pattern demonstrated
the wavelength of light directly. If light were an electro​-
magnetic wave, then certainly electromagnetic waves of other
wavelengths must also exist! Thus in the 1870s Maxwell not
only predicted the existence of radio waves, but also laid the
accurate theoretical basis for their study and — more

38
important — their application. This quantification of radio
waves, even before they were discovered, still forms the basis of
electromagnetic theory today. Maxwell’s equations define
precisely the relationships among the electric field, the
magnetic field, and the direction of propagation. They also
define precisely the ratio of the field strengths, and the
mechanisms for producing a radiated or a directed
electromagnetic wave.
The pivotal work of Maxwell was not universally accepted at
the time of its formulation. Today Maxwell’s equations remain
the basis of all theory pertaining to electromagnetic waves, and
thus also antennas. It would take decades before the validity of
Maxwell’s work would become difficult to refute.
The initial difficulty with Maxwell’s equations was their
complexity. His original formulation was 20 equations with 20
variables! In the early 1880s, Oliver Heaviside applied his
understanding of both electrical circuits and advanced
mathematics to “clean up” and simplify Maxwell’s equations to
just four equations in two variables. This work made it much
easier to grasp the concepts and fundamental relationships
developed by Maxwell, but we should remember that the actual
equations that are familiar to us are the work of Heaviside. We
will return to Maxwell’s equations with more detail in Chapter
3.
• 1887: Albert Michelson and Edward Morley (American)
performed the first experiment that negated the theoretical
existence of the “luminiferous ether.” Previously it was
accepted theory to assume that space was filled with an
invisible “ether.” Since light waves traveled through free space,
they required some type of matter to travel in. After all, ocean
waves need water, sound waves need air (or some other non-
vacuum medium), and so on. This assumption was in large
measure derived from Newton’s Laws constituting Classical
Mechanics. This experiment, together with Maxwell and a
growing body of knowledge from many scientists, eventually
led to Albert Einstein’s Theory of Relativity. Later we shall see
how Maxwell found that electromagnetic waves did not require
an “ether” as a propagation medium in effect, questioning the

39
then-assumed universality of Classical Physics.
• 1886: Heinrich Hertz (German, Figure 1.2) built the first
complete radio system consisting of a spark gap generator, a
transmit antenna, a receive antenna, and a spark gap receiver
(Figure 1.3 and Figure 1.4). Since radio waves cannot be
detected by any human sense, it was necessary to build a
complete system to actually prove radio waves exist! Thus
Hertz invented the antenna, the radio transmitter, and the
radio receiver, and in so doing “discovered” the radio waves
that Maxwell had predicted to exist.

40
 • 1895: Guglielmo Maria Marconi (Italian, Figure 1.5 and
Figure 1.6) invented practical radio. Hertz was a physicist and

41
was interested in the discovery and study of this newly-
invented apparatus. Marconi was the engineer who used
Hertz’s inventions and findings to build something useful. If
radio were to be useful, it would be necessary to significantly
extend the range to provide a new and useful application —
wireless communication. Successive experiments and
refinements extended the range, and soon it became apparent
that the antenna played a crucial role in increasing the range
of the wireless devices.

42
 In one of the experiments he greatly lengthened the wires
and configured both into ground mounted “verticals.” The
result was a range of 2.4 kilometers. Marconi was perhaps the
first to appreciate that the size of an effective antenna needed
to be comparable to the wavelength being used. This was
synonymous to raising the radiation resistance of an antenna, a
critical specification for antennas.
After moving to England, Marconi received needed funding.
After further refinements to his antennas and radios, in 1901 he
shocked the world by completing a wireless trans-Atlantic link
from England to Newfoundland. Modern wireless had been
born.
• 1897 – 1904: Hendrik Lorentz, George Fitzgerald, and Joseph
Larmor derived the Lorentz transformation, stating that
Maxwell’s equations were invariant.
• 1905: Albert Einstein used Maxwell’s equations as the basis
of Special Theory of Relativity.

43
1901 to Present
Since its infancy, radio technology has progressed through a
continuous stream of innovations. The basic theory was
derived in the 1860s and 1870s, but the technology is still
evolving quickly some 150 years later. This technical evolution
covers spark gap, the invention of the triode vacuum tube,
discovery of the piezoelectric effect, the transistor, integrated
circuits, super-miniature digital integrated circuits,
miniaturization of all the passive devices, and countless other
advances small and large.
The first vacuum tubes had severe speed limitations and thus
were usable only at relatively low frequencies. However,
vacuum tubes also were far more effective in building receivers
of much improved sensitivity and selectivity compared to the
coherer detector. In the early 20th century, the first radio
applications used the region of spectrum now used by the
standard AM broadcast band, hundreds of meters in
wavelength. Longer wavelengths were considered “low
frequency” and shorter wavelengths were considered “high
frequency.”
The relationship between wavelength (λ) and frequency in
Hz (f) is:

(Equation 1.5)
where c is the speed of light, λ is wavelength, and f is the
frequency in Hertz (Hz).
As technology improved, thus pushing the useable spectrum
higher in frequency, it was necessary to use progressively
stronger superlatives to define these new bands as they opened
up to practical use as shown in Table 1.1. Legacy trapped us
into defining 3 MHz as a “high frequency.” ​Perhaps frequencies
above 300 GHz should be defined as Unimaginably High
Frequencies!

44
 Today, the opening of the microwave spectrum
(approximately >1 GHz) has made possible low-cost, low-power
portable devices. Microwave wavelengths are comparable to
the physical dimensions of the human hand. Thus by using
these wavelengths (and shorter), efficient antennas could be
built into hand-held equipment. The results have been
profound. For example, the personal cell phone and its
numerous derivatives were made possible by a combination of
two fundamental technical innovations: 1) miniaturization of
microwave radio components to permit a two-way microwave
radio fit in a hand-held package and be battery operated; and
2) digital integrated circuits that could permit the new
miniature two-way microwave radio to behave like a
consumer-friendly wide-area wireless “telephone” and/or
computer link. More recently, advanced computers could also
be miniaturized and included in tablets such as Apple iPads
and similar hand held computer/wireless devices.
Dick Tracy had his “wrist radio” in the 1940s (Figure 1.7),
complete fantasy at the time. The concept of “cellular radio”
was conceived in the late 1940s at Bell Labs, but at that time a
“portable” cellular radio would have been the size of a
telephone booth. In contrast, antenna technology for portable
applications at microwave frequencies is almost as old as
antennas themselves. After all, for higher frequencies you
simply make the antenna smaller! Antenna engineers had to
wait for radio and digital circuitry to catch up. But catch up,
they did, as shown in Figure 1.8.

45
46
 We are all familiar with hand-held receivers that are quite
effective at much longer wavelengths (such as AM broadcast).

47
However, the receive antenna’s inefficiency (due to very small
size versus the wavelength) is overcome by the stationary
transmitters’ high power (kilowatts) and antennas that do
approach the physical dimensions of the applied wavelength.
Thus, poorly performing antennas are not out of the question
when designing radio systems, as we will explore in later
chapters.
The advent of personal computers such as the one shown in
Figure 1.9 was made possible by the miniaturization of the
microprocessor, memory, and I/O digital circuits. The
application that led to the home computer and its endless
applications and form factors was the result of the clever
innovators and pioneers of the personal computer. The first
microprocessor was developed to provide on-board navigation
for the Apollo Program! Although considered an off-shoot
development at the time, developers of this innovation went
on to form a new company called Intel. In turn, Intel and other
companies, academic researchers, and many more individuals
and organizations contributed to the birth of the modern
microelectronics and semiconductor industries. Although the
main focus of semiconductor development remains building
ever faster and smaller circuits for digital applications, the
implications of highly integrated, very small and very fast
components for micro-radio circuits has been equally dramatic.
In effect, the goals of fabricating ever smaller and faster active
devices (mainly bipolar and MOS transistors) were of equal
interest to digital (mainly computer) and radio designers (RF
engineers). “Smaller” implied transforming rack-mounted
equipment into hand held devices, and “faster” implied ever
higher frequencies of operation.

48
 There is another advantage to applying ever higher operating
frequencies. The amount of data or information that can be
carried by a channel of finite bandwidth has definite limits (as
we will explore in Chapter 10). Thus higher frequencies
inherently permit wider operational bandwidths. For example,
the entire AM broadcast band spans from 540 to 1700 kHz, a
bandwidth of only 1.16 MHz. In contrast, a comparable
bandwidth in the UHF region is a small fraction of the total
band.
Enter the Antenna
Antenna design has followed this upward trend in frequency
and thus smaller form factors for shorter wavelengths. Unlike
the radically different technologies that permitted this trend in
the radio proper, many antennas remain remarkably similar in
form. For example, the standard AM broadcast antenna (circa
1920s) is a 1⁄4 wavelength, ground mounted vertical. At 1 MHz,
this implies a steel tower about 75 meters high (Figure 1.10). In
contrast, 2.4 GHz WiFi access points typically also use a 1⁄4
wavelength (or longer) simple vertical antenna, only inches
high! The differences between an AM broadcast system and a
WiFi for radio technology is enormous, but the antenna design

49
is practically identical, at least the first iterations of these
antennas. There is also a continuous effort to integrate the
antenna(s) into the various hand-held form factors. The
external flexible “rubber duckie” antennas (Figure 1.11) have
been gradually replaced with antennas integrated inside the
unit. These internal antennas are in a constant redesign cycle
to approach an optimum performance limited only by physics.

50
51
 Almost all antennas, past and present, are configured of
metal. Even very complex antenna arrays are built upon
simpler long-known configurations but still consist of copper,
aluminum, and/or other metals and alloys. Although there has
been considerable progress in optimizing antenna designs for
various applications, basic antenna designs remain in
widespread use. More importantly for this text, they represent
the focal point for discussions on antenna physics, which forms
the basis for all antenna types.
In principle, virtually any type of antenna can be used at any
radio frequency, but their physical sizes will dramatically
change as a function of wavelength. Consequently, different
antenna designs may be more or less practical as the frequency
of interest changes. This is a key point in developing an
intuitive understanding of antenna design: The physics remains
constant, but the practical implementation and choices of

52
design typically vary widely with frequency.
This is not to say there has not been significant progress in
the development of antenna technology. Over the past several
decades, enormous progress has been made by optimizing
antenna parameters ever closer to physical limits. Computer
modeling has permitted such optimized designs not possible
only 10 or 15 years ago. Optimization of antenna design
involves changing the configurations of metallic structures and
often substrates upon which they are mounted.
In the early days of radio we began with Zepps, loops,
dipoles, and monopoles. Today many antennas are simply
manifestations of the simple dipole or monopole (as shown
above). Then multielement arrays (including the Yagi-Uda
design) using multiple dipoles or monopoles were designed to
create more directivity. The helix antenna was invented in the
1940s, and the parabolic reflector was simply borrowed from
optics exemplified by the reflector telescope. Patch antennas
are mainly two-dimensional flat structures, but 3D structures
are also in use. Diversity antennas of various forms use
multiple elements of the same simple structures to improve
information rate capacity and/or reduce path losses due to
changing propagation conditions. These innovations often
utilize sophisticated signal processing algorithms by comparing
and combining signals from two or more antennas. The market
wants products to be smaller, cheaper, and with better
performance!
We will explore in detail the reasons why antennas, in
general, require their minimum physical size to be close to the
wavelength of operation, or as close as possible. This is always
a critical factor in antenna design. However, this is only the
first “rule of thumb” to gain an intuitive understanding of the
workings of antennas. There are many fundamental terms to be
mastered, but again, the basic physics is constant for all
antennas. This is good news. By mastering the basics of
antenna physics, any antenna at any frequency is within grasp
of understanding. An antenna engineer is limited only by
his/her understanding of antenna physics.

53
 CHAPTER 2

Fundamentals
What is an Antenna?
The relevant Webster’s definition of antenna states, “a usually
metallic device (as a rod or wire) for radiating or receiving
radio waves.” This is indeed the most common application for
antennas. Kraus provides a more technical definition: “…may be
defined as the structure associated with the region of transition
between a guided wave and a free-space wave, or vice-versa.” In
other words, it is the device that connects the guided wave
(usually a one-dimensional transmission line) to three-
dimensional free space and vice-versa (Figure 2.1). Some also
refer to antennas as “transducers,” converting the guided wave
— represented by currents and voltages — to radiated waves
and vice-versa.

Most of the practical antenna discussions in this text will be

confined to linear antennas, which also constitute most of the
earliest antenna designs mentioned in Chapter 1. A linear
antenna is constructed from straight elements that are much
smaller in diameter than their length. For example, practical
linear antennas are typically constructed from wires or metal

54
tubing. We will also discuss some nonlinear configurations in
later chapters, but it is best to begin this introductory
discussion by restricting ourselves to the simpler linear
antenna configurations. In turn, more complex antennas build
upon the basic principles derived from simpler types, but the
calculations (particularly the integral calculus) become more
difficult. There is also the fictional “isotropic” antenna, which is
an essential reference point for all other antenna types,
practical and imaginary.
Receive antennas capture radio frequency power from free
space and concentrate it into a guided wave, and then, usually,
to a receiver. Transmit antennas perform the reciprocal,
receiving guided wave power from a transmitter (RF generator)
and radiating — or “coupling” — that power into free space.
Free space is defined as an empty three-dimensional area
devoid of any matter. For most of the radio spectrum, air can
also be considered free space for antenna work. While air (the
atmosphere) often has profound effects on radio propagation, it
typically has little effect on an antenna’s characteristics per se.
The essence of Antenna Physics deals with the physical
characteristics of the antenna and how it interfaces with these two
ports.
In electrical and electronic circuit theory, there are theorems
that can be of great help in solving difficult problems. For
example, if we have a two-port circuit (as in Figure 2.2)
composed of linear elements and apply a source voltage (and
resulting current) across Antenna 1, we will observe a ​resulting
current in Antenna 2. Now if we apply the same observed
current to Antenna 2 and remove the original source voltage
from Antenna 1, we will see the same voltage across Antenna 1
that we previously applied. This is called the Reciprocity
Theorem.

55
 Since most antennas can be considered linear structures, we
can assume that the Reciprocity Theorem also holds for an
antenna (also defined as a two-port circuit described above).
Notice from Figure 2.2 that both antennas reflect some
impedance value, and there is also an intermediate impedance
value of the medium — in this case, free space. These
impedance terms are of fundamental importance to antenna
physics and will be carefully developed later.
The reciprocity theorem as applied to antennas is of critical
importance whenever the antenna is to be used for both
transmitting and receiving (Figure 2.3).

56
 The Sphere as a Measurement Tool
Since free space is three-dimensional, 3D measurements are
of fundamental importance in describing antennas’
characteristics. We need some convenient construction that
can serve to represent and quantify what is happening in free
space relative to an antenna’s performance. We will use an
imaginary sphere for this measurement tool, superimposed
within free space with a test antenna at its center. For antenna
work, we are usually concerned with only the center of the
sphere and the surface of the sphere upon which we can define
power density over a portion of the entire sphere’s surface. The
volume of a sphere is usually only concerned with energy that is
not typically used for antenna characterization. It is usually
assumed that the space inside and outside the sphere is indeed

57
free space — no matter allowed except for the antenna!
It is often impractical to perform actual measurements in
actual free space. Antenna measurements are often taken on or
close to the surface of the Earth. This causes difficulties in that
matter (solids, liquids, and sometimes gases) can reflect and
absorb power, and worse, affect the performance of the
antenna proper. Reflections, in particular, are problematic in
that they can add or subtract from the direct path wave
between the antenna and a measurement device, thus
distorting the measurement. Such reflections can cause
multipath conditions, where a receiver receives a signal from a
straight line from the transmitter and also a reflected signal.
See Figure 2.4.

It is necessary to eliminate reflections that may distort

attempts to perform accurate antenna measurements — in
other words, simulate a free space test environment but not
actually have to become airborne to escape the effects of the
Earth. A practical solution to this problem is to use an anechoic
chamber (Figure 2.5). As its name implies, enough of the
radiated energy’s “echoes” or reflections are absorbed by the
floor, walls, and ceiling to an extent sufficient for good antenna
measurements.

58
 Since the sphere is often used in antenna measurements, a
review of spherical geometry is in order with emphasis on
terms used in this book. Figure 2.6 shows a section of an
imaginary sphere. The antenna is located at the center of the
sphere and is usually assumed to be a “point source” in that the
antenna’s physical dimensions are much smaller than the
radius of the sphere. We can see the traditional use of spherical
coordinates analogous to latitude (θ) and longitude (φ).
Therefore any unique point on the sphere’s surface can be
identified with a unique value of θ (0 – π) and φ (0 – 2π), where
the value is given in radians. We can convert the radian
measurement to degrees with this relationship: 2π radians =

59
360 degrees.

Of particular importance to antenna measurements is the

area of a sphere’s surface and also portions of that area. A
sphere’s surface area is 4πr2, where r is the radius. Similar to a
full circumference of a circle being 2π rads (360 degrees), the
entire surface of a sphere is defined as 4π Ω, where Ω is the
symbol for a solid angle measured in steradians. A hemisphere is
defined by 2π Ω, a quarter of a sphere π Ω, and so on. In this
book we use a standard bold font for (Ω) solid angle to
differentiate the same Ω symbol for ohms.
Since the surface area of a sphere is 4πr2 and the solid angle
representing a full sphere is 4π Ω, the actual area of a portion
of a sphere is simply Ωr2. Now we can use this “yardstick” to
develop some basic parameters of antennas.

60
Analogy Between Light and Radio Waves and
the Concept of Aperture
Let us place a 100 W light bulb at the center of our imaginary
transparent sphere. We assume that this light source is
emitting exactly 100 W of light power and ignore any heat
generated through inefficiencies. Our light bulb also has a
special characteristic: Its pattern of radiation is isotropic, which
means that the light power emitted by the bulb is radiated in
all directions equally. Therefore, the power radiated is
distributed perfectly evenly over the surface of the sphere. If
the interior of the sphere is a vacuum, then all 100 W of light
power will reach the surface of the sphere. If we could calculate
the total power passing through the sphere’s transparent
surface, it would also be 100 W evenly distributed over the
surface of the sphere.
Furthermore, we can define the power of the light bulb (or
any electromagnetic wave) as a power density, or flux over an
area, or W/m2 (watts per square meter). Another term often
used is the Poynting Vector S (θ,φ). S is the vector defined as
Wm–2, and (θ,φ) represents the spherical area over which the
power density is constrained, or more conveniently by a
steradian value. See Figures 2.7 and 2.8.

61
 On the surface of the 100 W light bulb’s surrounding sphere
we will place a solar panel. The surface area of the sphere is
taken to be 1000 m2 and our solar panel is 1 m2 and 100%
efficient. Therefore we can capture 1/1000 of the light power, or
1/10 W, or 100 mW from the light energy falling on our solar
panel. If we double the area of the solar panel we double the
power received. Here we introduce another very fundamental
relationship in antenna physics: the “area” of an antenna
(which is more appropriately defined as aperture). As we shall
soon describe, aperture and power gain are directly
proportional. Thus we can now express how much power is
received for a given distance and a given aperture of the
receiver:

62
(Equation 2.1)
where
Pr is the power received
Pt is the power transmitted in the direction of the receiver
Ar is the aperture of the receive antenna
4π r2 is the total surface area of the sphere with a radius of r.
Notice that the term

is simply the ratio of the antenna aperture to the area of the

entire sphere, which also represents the ratio of received
power to the total power transmitted.
As an aside, if we substitute r with d (distance) we get the
coefficient 1/d2 by:

Thus we have derived the well-known “inverse-square law”

from spherical trigonometry, holding for any form of
electromagnetic radiation. For a given receiver aperture the
power received is proportional to the inverse-square of the distance
from the transmitter.
Of course no solar panel is 100% efficient, but this becomes a
trivial calculation. If the solar panel from the above example is
only 20% efficient, the electric power generated is simply 20%
of the potential.
Pr = 100 mW × 0.2 = 20 mW
Yet another analogy is real solar panels located on the Earth.
In this case, the light source is the Sun, which is very close to
being an isotropic radiator. The measurement sphere is defined

63
by a radius, which is the distance between the Sun and the
Earth. The Earth’s atmosphere absorbs about 1⁄2 of the solar
light before it reaches the Earth, and a 100% efficient 1 m2 solar
panel will provide about 750 W. Thus we can work in reverse
and calculate the total power output of the Sun!
The Sun is about 150 billion meters from the Earth, or 15 × 109
m. Therefore the surface area of this imaginary sphere (the
radius being the Earth – Sun distance) is 4π(15 × 109 m)2, or
about 28 × 1021 square meters. Thus if the Sun’s power is 1500 W
per square meter at the distance of the Earth, then the total
light power output of the Sun is about 42.5 × 1024 W! Thus we
can calculate any unknown variable in Equation 2.1 if the other
terms are known. Also note that thus far we assume an
isotropic transmitter and a known aperture of the receiver.
Reconciling these two variables is a key task for any advanced
discussion of antenna physics. The following discussion will
quantify this essential relationship.
Defining the Critical Fundamental Terms
We use the examples of a light bulb or the Sun, since they are
easy to visualize. Now we can take the leap and state that there
is a direct correlation between our brief discussion about light
and radio waves. We know this is a correct correlation because
Maxwell predicted the existence of radio waves using the then-
known properties of permittivity and permeability of free space
and their relationship to the speed of light. He then correctly
postulated that both light and the yet-undiscovered radio
waves were both electromagnetic radiation, in effect having
identical properties except for their differing wavelengths.
After the actual discovery of radio waves, endless
measurements and experiments have proven the validity of his
early postulations.
Solar cells remain inefficient because we remain restricted to
the photoelectric effect that relies upon the inefficient process
of a light photon dislodging electrons within the photovoltaic
material, thus generating an electrical current. Building an
array of receive antennas for light is possible, but converting

64
the “guided” light wave directly into a dc current would require
a futuristic diode. Again wavelength is the critical limitation. It
is possible to build a dipole antenna for light and/or infrared
(IR) wavelengths using micro-processing techniques. To
convert the light power into an electrical current, however, we
would need to rectify the light power into direct current. No
such diode exists. If it could be developed, solar cells could be
replaced with arrays of tiny dipoles and futuristic diodes. So for
a practical comparison of light and radio waves we need some
slight modifications, but remember that the physics is the
same. Fortunately, antennas for radio frequencies often do
approach 100% efficiency.
A transmit antenna can be compared to the light source (a
light bulb, LED, or a torch) and the receive antenna compared
to the solar panel. Unlike radio antennas, however, these
particular choices of light sources and light “receivers” are not
reciprocal. We do not normally think of a solar cell as a
radiator or a light bulb as a light receiver. On the other hand,
radio antennas are reciprocal, therefore transmit antennas are
also characterized as having an aperture (the same aperture as
they have when receiving).
A “shrinking” aperture due to inefficiency is analogous to a
reduction in antenna gain due to power losses (usually as heat).
Also, as above, if we double the aperture, we double the power
gain (about 3 dB difference). This is illustrated by doubling the
aperture (two rather than one solar cell) for a doubling of
power capability (Figure 2.9). The effect of cutting an
antenna’s physical aperture in half is identical to cutting the
efficiency of two solar cells in half. Therefore, the power gain of
a solar cell or antenna follows the same rule. For example, a
hypothetical 100% efficient solar cell would convert all the light
power falling on it into electrical power. In a real solar cell, the
absorbed power not converted into electrical power is
converted into heat. This is precisely what happens in an
antenna, either for transmit or receive antennas. Power
applied to the antenna, but not radiated from the transmitter
or not delivered to the receiver from a receive antenna, is lost
as heat.

65
 One of the advantages of using aperture instead of gain in
radio link calculations is that we can directly calculate how
much power will be received by the antenna. Indeed, any
equation for calculating the power loss of a radio link must use
aperture for the proper calculation, or a derivation thereof. The
answer to the very important question posed in Figure 2.10
requires the detailed understanding of the aperture concept
which follows.

A non-isotropic antenna will present different gain for

different directions and therefore also present different
apertures for different directions. Again we can use the solar
cell as an example. When oriented perpendicular to the Sun
(toward the light transmitter) it will receive the maximum

66
amount of power. This is the direction of maximum gain and
thus also the direction of maximum aperture. If you begin to
rotate the panel away from the Sun’s direction the effective
aperture decreases, and so the power received decreases.
Finally, with the panel facing 90 degrees off the Sun’s direction
there is no sunlight striking the panel and the received power
drops to zero. An antenna’s gain and aperture also depends
literally on the antenna’s point of “view”! Of course, those
familiar with directional antennas know that as the antenna is
rotated toward the direction of a transmitter, the signal power
received increases and peaks where the forward gain is
maximum. This is also the direction that corresponds to the
antenna’s maximum aperture.
We also know that any non-isotropic antenna that is 100%
efficient will radiate all the power applied to it, but not in all
directions equally. Therefore in some direction(s) a receiver
will “see” the equivalent of more than 100 radiated watts and in
other direction(s) a receiver will “see” less than the equivalent
of 100 radiated watts. In this case the antenna has gain greater
than isotropic in some favored direction(s).
Yet another optic analogy is the automobile headlamp
(Figure 2.11). The typical headlamp bulb radiates about 10 W of
light power. However, placing a parabolic mirror in back of the
bulb results in a very high power beam with a small steradian
value defining its narrow radiation pattern. In effect, the
parabola “robs” all other directions on the sphere of light
power and concentrates it within the narrow beam. If we can
concentrate all the 10 W of power into a steradian of 1% of the
total sphere 0.01 (4π)Ω we simply multiply the output power by
the power gain (in this case 100) to derive an effective radiated
power relative to an isotropic or 1000 W EIRP. That is why a
misaligned headlight can blind an on-coming driver.

67
 Again, the power analogy between optics and radio physics is
exact. Also, here is another critical point: If we take the average
and thus total power radiated over the entire sphere, we find
that the value remains 10 W. In the case of the isotropic the
average and total power is also 10 W. However, the effective
radiated power (EIRP, compared to an isotropic) is 1000 W in
the main beam of the headlamp, but 0 W over 99% of the
sphere, thus the average is, again, 10 W.
The 1⁄2 Wavelength Dipole Antenna
Perhaps the simplest form of a practical linear radio antenna
is the 1⁄2 wavelength dipole, as seen in Figure 2.12. The dipole
is the shortest linear antenna that is resonant. Resonance, in
the case of antennas, implies that the source impedance (when
used as a receive antenna) or load impedance (when used as a
transmit antenna) is a pure resistance (no reactive part of the
impedance). In Figure 2.12 we use RL(load) since we are
assuming (for this example) a transmit antenna. Therefore an
antenna can be understood as a device that is represented by a
load to a signal generator that absorbs power by radiating it
into free space. Of course by reciprocity, the imaginary resistor
can also form a signal source with the same impedance.

68
 The isotropic and dipole antenna types form building blocks
for both the theoretical and practical basis of antenna study.
We will soon provide a detailed understanding of their
behavior and their interrelationships.
Relationship Between Aperture and Gain
Aperture is a straightforward concept in that an antenna can
be defined as a simple two-dimensional area. Gain (dB) is
another issue: gain relative to what? Usually antenna gain
specifications reference the antenna to a defined gain of an
isotropic (dBi) or a 1⁄2 wave dipole (dBd). Without a reference, a
specification such as the antenna gain is “6 dB” is meaningless:
4 times the power of what? Since an isotropic antenna would
have a power gain of 1 (0 dBi), this is a logical reference to use.
The decibel measure of power is 10log10(P1/P2), where the P
values are linear values of power. It is essential for P is be a
ratio of power values, such as a ratio of some gain value to the
gain value of a dipole or an isotropic.
Assuming we are comparing lossless antennas, the isotropic
antenna has the minimum possible gain of any possible
antenna. For this reason, antenna gain references are
ultimately based on the gain of an isotropic antenna (0 dBi). As
a result it is necessary to know the aperture of an isotropic
antenna in order to relate any antenna’s gain to aperture.
Without equating the two, it is impossible to calculate how
much power will be received by an antenna for a given power
at a given distance and thus answer the critical question posed
in Figure 2.10. For example, you know your receive antenna has
a gain of 9 dBi and that the power density is 10 microwatts per
square meter at the receive location. How much power will you
expect to receive? A value of 9 dBi is meaningless without

69
equating it (or any antenna gain specification) to an equivalent
area. However, we know that the aperture of a 9 dBi antenna
will be about eight times as large as an isotropic aperture. But
the trick question is: What is the aperture of an isotropic
antenna?
Empirical Calculation of the Aperture of an
Isotropic Antenna
We can use the above introduction (basic physics and some
simple mathematics), coupled with a set of empirical
measurements, to derive the aperture of an isotropic antenna.
We will also use relatively simple instruments to measure
power. This calculation is critical to antenna physics, antenna
engineering, and radio system engineering. But how do we
calculate the aperture of the hypothetical isotropic antenna
that both radiates and receives to and from all directions
equally?
To begin, let us review what we now know:
1) An isotropic antenna radiates and receives in all 3D
directions equally.
2) A non-isotropic antenna does not radiate in all directions
equally, but radiates and receives more power in some
directions and less in others (relative to the hypothetical
isotropic)
3) The total and average radiated power of any antenna will
be the same as an isotropic antenna (assuming 100%
efficiencies). If an antenna radiates all the power applied to it,
the power must be going somewhere!
4) The inverse-square law applies also to radio waves.
Now we are prepared to derive an experiment and
subsequent calculations to determine the aperture of an
isotropic antenna. In this experiment, we begin with two
dipoles. We use dipoles because we have no idea how to design
an isotropic antenna. At present we assume we do not know
the gain of these dipoles, nor do we know their apertures. We
make sure that the dipoles are near 100% efficient by carefully
measuring the temperature of the antenna before and after we

70
apply power to them. We find negligible difference in the
temperature, so we assume near 100% efficiency. We also verify
a perfect (lossless) match to our RF generator by taking a
simple VSWR measurement — it is 1:1. To begin, we use 300
MHz for our first test since relatively high frequencies are
needed for practical measurements inside anechoic chambers.
In the anechoic chamber we use a dipole as a source antenna
and a second dipole at 10 meter distance as the receive
antenna. By rotating the receive dipole in three dimensions, we
can record the directivity of the dipole over the dipole’s entire
imaginary external sphere. We don’t need a calibrated power
source since we are only plotting the relative gain. Since our
dipole is near 100% efficient, we assume the relative gain is the
same as the relative directivity, where
Gain = Directivity × Efficiency
(Equation 2.2)
If we plot the relative gain of the 1⁄2 λ dipole in free space, we
find a “doughnut” pattern (Figure 2.13). By our statement 3
above, we know that the total and average power response of
this non-isotropic dipole antenna must be the same as an
isotropic antenna

71
 We can now perform an empirical process to calculate the
gain difference between the dipole and the isotropic antennas.
We know that the gain of the dipole is constant for all points
perpendicular (broadside) to the line of the dipole because we
measured it. However, the gain difference can be plotted in two
dimensions (a plane) that bisects the doughnut tangent to the
doughnut’s “hole.” Since we only need to find an average of the

72
dipole power gain, we can ignore performing a three-
dimensional integration and use only a “slice” of the doughnut
and perform a two-dimensional integration. We find the total
area of the two sides of the slice, which we know corresponds to
the total power output of the dipole. Then we take this total
area and calculate the radius of a circle we construct with the
identical area, which is the equivalent pattern of an isotropic at
an identical power (circular). The radius of this circle is 0.61 of
the maximum length of the doughnut in its maximum gain
direction. This means the dipole-doughnut has a power gain of
about 1/0.61 = 1.637 over the theoretical isotropic, or 2.15 dBi.
Thus we have just measured the dBi gain of a 1⁄2 wave dipole.
Since the actual sphere is 3D, we can form a conceptual
understanding of the power integration in 3D by imagining
that we have a doughnut shaped 3D piece of putty. We measure
the radius of this doughnut to be 1 inch. Then we take the
doughnut and reshape it into a perfect sphere (like making a
snowball) and measure the radius of the new sphere. We find it
to have a radius of 0.61 inch which is about a 2.15 dB difference.
Assuming we were very careful in our measurement
techniques, Figure 2.14 shows a linear plot of the dipole’s
response in linear/linear coordinates as we rotate the dipole
360 degrees perpendicular to the source.

73
 We also know that the aperture of an antenna is directly
proportional to its gain in the same direction from the antenna
as in our solar cell example. Therefore we can measure the
received power from our dipole example and calculate directly
the aperture of the dipole and thus also the aperture of an
isotropic. This is the relationship that we need to break into
new territory in understanding antenna physics.
We now know that the power gain of our 1⁄2 wave dipole is
1.637 over an isotropic. Therefore if we apply 1 W to the
transmit dipole we will have an EIRP of 1.637 W in a direction
broadside to the dipole. When we place our receive dipole 30
meters away from the transmit dipole (also broadside to the
transmit dipole) we record a received power of just over 23.036
µW. Since we know that we have a total dBi power gain in the
system of 2.6797 (product of the two dipoles’ linear gain

74
broadside to each other), we know the equivalent isotropic
assumption would be just about 7.036 µW. Our actual
maximum measured broadside-to-broadside received power
was 18.854 mW, or 2.6797 times the calculated isotropic
received power. Now we have normalized the dipole to an
assumed isotropic radiator radiating 1 W at the center of a 30
meter radius sphere and a calculated received isotropic power
of 7.036 µW. We can now calculate the aperture of the isotropic
antenna.
The area of the 30 meter radius sphere is about 11,310 square
meters. With 1 W of isotropic radiated power at the center, the
calculated power density is 88.417 µW per square meter. Since
we would actually receive 7.036 µW, the aperture of our
isotropic antenna is

or about 0.0796 square meters. This looks like the answer, but
we have another observation/potential problem. We know that
1⁄2 wave dipoles have different lengths for different frequencies,
such as the antennas shown in Figures 2.15 and 2.16. Intuitively
we believe that a larger antenna will have a larger aperture,
well, because it’s bigger! For example, a 7 MHz dipole is
considerably larger than a 144 MHz dipole. Intuitively we think
that a much bigger structure will couple more power into free
space, or receive more power from free space. If our intuition is
correct we should arrive at different aperture values for
different frequencies.

75
 So we prepare an identical experiment, this time at 600 MHz
which is 1⁄2 the wavelength and thus also 1⁄2 the dipole length in
the 300 MHz experiment. If there is a difference in the
calculated isotropic aperture, we will know there is a
wavelength term in the isotropic aperture equation. We run

76
the experiment again and find that the isotropic value is about
0.0199 square meters, or exactly 1⁄4 the area of the 300 MHz
dipole. Additional experiments at other frequencies show that
the aperture of a dipole, and thus also an isotropic antenna,
increases at a rate equal to the square of wavelength, or the
square-root of the frequency.
However, we have some coefficient we must also solve to get
a true equation that defines the aperture of an isotropic at any
wavelength. If we return to the original 300 MHz results, we
know the wavelength is a nice normalized number of 1 meter,
and 1 squared is 1. Thus we see that the coefficient is about
0.07958. The reciprocal of this is 12.566. Because we’ve been
using this number so often with our spherical tool and with the
insight of a James Maxwell, we immediately recognize it as 4π.
We use this coefficient in our other measured results, and thus
converge upon the actual definition of the aperture of an
isotropic antenna as

(Equation 2.3)
where the antenna aperture (area) is in the same units of
wavelength (λ).
This is one of the most fundamental equations (and
relationships) in antenna theory in that it relates antenna gain
to aperture. Indeed, many misconceptions regarding antenna
performance can be resolved by understanding this
fundamental relationship. The most striking implication of this
equation is that the aperture of an isotropic antenna (and all
other antennas referenced to it) depends heavily upon the
wavelength of operation. Now we have the tools to solve the
problem illustrated in Figure 2.10.
Free Space Path Loss
Path loss simply defines the ratio of power lost between the
transmitter’s EIRP and the power received by the receiver’s

77
antenna. In free space, we use the simple spherical model as
above and assume an isotropic receive antenna with the
appropriate aperture. If we receive or “capture” 1/1,000,000 of
the power radiated, our path loss is 10log10(0.000001) or 60 dB.
Therefore the power received will be 60 dB below whatever
power is transmitted in the direction of the receiver. The
relationship, in most situations can be considered linear and
thus proportional.
Virtually all calculations for free space path loss are based
upon these simple concepts: power radiated (EIRP), distance,
and receive antenna aperture. These terms represent the
“starting point” to the operation of a garage door opener, an
interplanetary radio link, radar systems, cell phone coverage
area, satellite services, and so on. The great complexity (and
cost) of some radio coverage prediction software tools reflects
their ability to approximate the effects of material
environmental effects (such as a portion of the Earth’s surface)
upon the coverage. But the starting point always begins with
these fundamentals. It’s the refinement that is difficult. Free
space calculations are simple; it’s when “stuff” gets in between
the transmitter and receiver that estimations become difficult.
No matter how sophisticated a path loss predictor may be, the
result is always one single number (the path loss, and possibly
some statistical deviation around that number).
Returning to the basic situation of an isotropic antenna at
the center of our sphere, we know that the power will be
equally distributed over the sphere’s area, 4πr2. We also know
that the power received will be proportional to the aperture of
the receive antenna. If our receive antenna is 100% efficient,
the portion of the transmitted power received is simply the
ratio of the antenna aperture to the area of the sphere, or,

(Equation 2.4)
and the dB loss is simply:

78
(Equation 2.5)
Notice that Equation 2.4 is simply a ratio of two areas — the
aperture of the receive antenna to the total area of the sphere
on which the antenna is set.
Let’s assume that our receive antenna is isotropic. After the
above calculation and measurements, we know its aperture is

Substituting l2/4π for A2 we get

Notice the beautiful mathematical simplicity of this term.

This simplicity emerges because of the similarity in the
equation defining the area of a sphere and the isotropic
antenna, whose radiation pattern is spherical! Hopefully the
elegance of using spherical geometry in calculating and
defining antenna parameters is now apparent.
If we know that the transmit and receive antennas are
isotropic, we can now easily calculate how much power will be
received by the receive antenna.

(Equation 2.6)

79
 As implied above, we can adjust the aperture value if we
know the gain value of the antenna. For example, if we have a
6 dBi antenna, that represents about 4× power gain (calculated
through using the inverse-log term), so the aperture of a 6 dBi
antenna is simply

On the transmit side, if a 6 dBi antenna is used with a 100 W

transmitter, then the EIRP (effective radiated power relative to
an isotropic antenna) is 400 W (in the maximum gain
direction). Therefore by these simple adjustments you can
calculate any free-space path loss for any power, antenna gains,
frequency, and distance. The other radiated power convention
is ERP (effective radiated power relative to a dipole). The
difference between the two is always 2.15 dB, but caution
should be used to determine which is being applied for a given
application. Again, if the 6 dBi antenna has only a single lobe
with a “brick wall” pattern, then the pattern — defined by
steradians — is πΩ, or 1⁄4 of a spherical, isotropic pattern.
The path loss using antennas with non-isotropic gains is:

(Equation 2.7)
where G is the linear power gain coefficient related to an
isotropic gain for the transmit and receive antennas.
Since

Then

80
 By substitution we can now also use the aperture of both the
transmit (Aet) and receive (Aer) antennas in a given link to get
the path loss by substituting these terms for the isotropic in
Equation 2.7, so we get

(Equation 2.8)
This is the free-space Friis Transmission Formula, where
again r is the distance between the antennas, λ is the
wavelength, and Aet and Aer are the antenna apertures all using
the same units. Here we can see another result of the
reciprocity theorem. Notice that both the transmitter and
receiver aperture (or gain) terms appear in the numerator of
the equation. This implies that you can compensate one for the
other. In other words, you will receive the same amount of
power by swapping the antennas, or by changing their values
provided that the product is the same.
Perhaps no better example of the concept of antenna ​-
aperture can be found in the parabolic (dish) antenna (Figure
2.17). The area of the dish is simply equated to the aperture of
the antenna (assuming 100% aperture efficiency). Therefore the
dBi gain of a parabolic antenna increases with increasing
frequency while the aperture (in principle) remains constant.

81
 Equation 2.3 points to another issue relating to wavelength.
An often confused relationship is frequency (wavelength) and
path loss. If all electromagnetic waves conform to the inverse-
square law, why do lower frequencies seem to propagate
“farther”? Understanding the relationship between gain and
aperture solves this common source of confusion.
Refinements of the Aperture Concept
The above discussion provides for an intuitive understanding
of antenna aperture. However, this first approximation of this
critical antenna parameter needs some refinement for a deeper

82
understanding. Thus far we have only discussed effective
aperture (Ae) but have also hinted at the effect efficiency has in
reducing the effective aperture. As mentioned above, the gain
and thus effective aperture of an antenna, decreases when an
antenna becomes less efficient. Thus when an antenna has no
loss and is perfectly matched to a load, the aperture is the
maximum effective aperture or Aem. When power is lost to heat
due to inefficiencies, the result is a “loss” aperture or Al.
Therefore,
Ae = Aem – Al

(Equation 2.9)
Thus Al equates an equivalent area that captures power that is
lost to heat. This is comparable to a solar cell’s maximum
possible aperture being reduced by inefficiencies.
For a properly matched antenna to its load:

(Equation 2.10)
We have seen that power gain is proportional to area
(aperture) in calculating antenna parameters. Therefore we can
begin to formalize this relationship.
Let an incident wave produce a voltage (V) at the feed point
of an antenna, as in Figure 2.18. For convenience we will use a
dipole as an example. This voltage produces a current (I) in
both Rr and Rt. Rr is defined as the radiation resistance, which in
this case is also the feed point impedance value. In effect, Rr is
the source impedance and Rt the receiver’s input impedance
(usually 50 or 75 Ω). Rr, a fundamental term, will be dealt with
in detail later. In this discussion we will ignore any reactive
parts of the source and load impedances, since they are
routinely “tuned out” of the circuits. It is the real parts of these
impedances that determine real power terms. Assuming no

83
complex terms in the impedances (pure resistances) the power
delivered to the load is thus:

(Equation 2.11)
and

(Equation 2.12)
where V is the induced antenna voltage. Thus

(Equation 2.13)
Assuming a conjugate match, Rr = Rt. Therefore,

84
(Equation 2.14)
or

(Equation 2.15)
By our basic discussion of Ae above, we know that

(Equation 2.16)
where S is the Poynting vector, or power density per square
meter, and Ae is the aperture that captures power proportional
to its area.
Therefore, with a conjugate match and no loss

(Equation 2.17)
This discussion points to a possible contradiction. Figure 2.18
shows that 1⁄2 of the received power is “lost” in the antenna’s
radiation resistance. If we assume a 100% efficient antenna,
then the power is not “lost” to heat. Also, if we assume we only
deliver 50% of the received power to the antenna, then the
aperture and thus gain should only be 1⁄2 of what it “should” be.
The answer is that an antenna perfectly matched to its load
reradiates 1⁄2 of the power by virtue of yet another aperture, the
scattering aperture As.
In the case of the scattering aperture, the scattered power is
simply lost back (scattered) as re-radiation into free space. The

85
issue is comparable to a simple dc circuit consisting of a battery
with an internal resistance and a load represented by RL(load)
(Figure 2.19). Maximum power will be delivered to the load
when the source and load resistances are equal. In the battery
circuit, 1⁄2 the power is dissipated as heat. In the Figure 2.18
antenna circuit, 1⁄2 of the power is re-radiated back into space.

The transmitting case is also similar to the dc battery case.

The transmitter can be modeled as an RF voltage source with
an internal resistance (source impedance). The antenna retains
its internal resistance as defined by Rr. There is an important
difference in the transmitter case, however. In the receiving
case, the goal is usually to achieve a conjugate match, to make
Rt = Rr. This will achieve maximum power transfer into the
receiver resulting in a maximum signal-to-noise ratio (more on
this subject in Chapter 10). In practice, the receiver’s input
impedance may be tuned a bit off the exact complex conjugate
impedance, as minimum noise figure is seldom found at the
exact conjugate match.
In a transmission system we are usually most interested in
maximum system efficiency. For example, if the transmitter
were 100% efficient and the source and load resistances were
identical, the overall efficiency would be 50%. In such a
conjugate match condition, real transmit systems would be
very inefficient. For maximum efficiency the transmitter
impedance is typically tuned to a lower impedance than the
load, so a higher efficiency is established. Some transmitters

86
can achieve very low source impedances at their maximum
efficiency operating point, resulting in efficiencies of over 90%.
However, the antenna “sees” a 50 Ω source impedance but all
the power supplied by the line is radiated by the radiation
resistance. As remains present in the transmitting case, but
only applies to power incident on the antenna from free space.
Figure 2.20 shows the equivalent circuit of an antenna and a
transmitter. Transmitters are usually specified as “50 Ω output.”
The real definition of this specification is that the transmitter is
tuned for maximum efficiency assuming a 50 Ω load. Thus for a
given antenna source impedance, the receiver input is tuned at
or near the complex conjugate of the antenna impedance. In
contrast, for a given load impedance the transmitter is tuned to
some output impedance that will permit maximum efficiency of
the system.

The mathematics that describes the transmit condition of

maximum efficiency is rather complex and belongs more in a
text on circuit theory rather than antenna physics. The terms
involve the actual characteristics of the amplifier device proper
(tube or transistor) and the LC tuning unit between that
device(s), including its Q and the transmission line impedance.

87
However, the above is an adequate simplification for
conceptual understanding of these basic relationships.
As is often used to great advantage in antenna design and is
often misunderstood. In our basic discussion of the aperture
concept we found the maximum effective aperture (Aem) of an
isotropic and a dipole antenna. Now let’s revisit this analysis.
A properly terminated antenna will deliver 1⁄2 its power to
the load (usually a receiver) and re-radiates the other 1⁄2. The
“total” antenna aperture that includes both the Ae and As is
defined as an additional aperture, the collecting aperture (Ac).
The collecting aperture is simply the sum of the thus far
defined apertures:
Ac = Ae + Al + As

(Equation 2.18)
Therefore, in a 100% efficient antenna, Ae is about twice the
area of Ae since Ac = Ae + As and Ae = As.
As is an extremely useful property of antenna elements. In
particular, parasitic elements use this property to maximum
advantage. In a parasitic element, we set the load impedance to
zero (a short circuit), as in a director or reflector element.
When the terminal impedance of a resonant or near-
resonant dipole is zero (as in a parasitic element), As is by
Equation 2.13:

(Equation 2.19)
Since the terminal impedance is zero, then Ae = 0. Therefore all
Pr will be re-radiated, or Prr.

88
(Equation 2.20)
Thus the re-radiated power from a parasitic element is four
times that of a conjugate matched antenna! We will explore
this in more detail in Chapter 8.

(Equation 2.21)
where V2/Rr is the power term and S is the power flux/square
unit on the sphere, so the aperture is simply Prr/S.
Again, compared to As for a terminated antenna dipole, often
referred to as a “driven” element, Ae is four times as large for a
“parasitic” element compared to a “driven” element shown
again in Equation 2.22.

(Equation 2.22)
Thus, the scattering aperture (As) has a maximum area when
the dipole feed point is shorted, or no power delivered to the
load and all power re-radiated. On the other hand, maximum
Aem occurs when the dipole is matched into a load. The re-
radiation from the parasitic element can be adjusted in its
phase and amplitude response referenced to the driven dipole
to create very desirable antenna patterns. The most common
application of this principle is in creating a Yagi-Uda array
discussed in Chapter 8.
Now consider the opposite condition: an open center to the
dipole, where Rt = ∞. As becomes very small because the
antenna is no longer terminated. Mismatching and low Rr
prevent “capturing” incident wave power and also re-radiation
of the received power. This double-effect renders the antenna
to have negligible effect upon the incident wave. In effect, the

89
1⁄2 wave dipole is now two 1⁄4 wave dipoles closely spaced end-
to-end.
Now we are prepared to present a graph of how Ae, As, and Ac
change when the terminal resistance is changed. Note that Aem
appears only at one point, the maximum possible value for Ae.
Figure 2.21 plots Ae, As, and Ac against the normalized value
of Aem. When Rr = Rt then Ae = Aem and maximum power
transfer is accomplished.

Summary of the various types of antenna aperture:

Ae Effective Aperture: the “effective” equivalent of an
antenna’s area that delivers power to a receiver or power
coupled into free space by a transmitter.
Al Loss Aperture: the area of an antenna “lost” due to heat.
Aem Maximum Effective Aperture: the maximum possible
area of an antenna where Aem = Ae + Al when the antenna is

90
matched to a load.
As Scattering Aperture: the equivalent area of the antenna
that re-radiates power back into space.
Ac Collecting Aperture: the total equivalent area of an
antenna that represents power delivered to the load, lost to
heat, and re-radiated, where Ac = Ae + Al + As.
Ap Physical Aperture: There is also an additional aperture
term, the physical aperture. This term defines the actual
physical area of an antenna (two dimensional physical
aperture). It is applied typically to “aperture” antennas, such as
horns and parabolic “dish” antennas, where the physical
aperture indeed resembles the collecting aperture. For
example, if we consider a dish antenna 10 meters in diameter,
then Ap = π52 = 75.5 m2. If the aperture efficiency were 100%, this
would also be the Ae. If the efficiency were only 50%, Ae would
be 37.75 m2. In contrast, the “area” of a wire dipole antenna is
very small compared to its Ac , therefore the physical aperture
has little meaning for single-dimension antennas.

91
 CHAPTER 3

Radiation of Radio Waves

In Chapter 2, some of the basic parameters of antenna
physics were presented. In addition to the imaginary isotropic
antenna, the “real” dipole antenna was discussed, with its
radiation resistance, apertures, gain, and radiation pattern
quantified. However, these parameters were not explained in
detail. They were only presented as self-evident, empirical
parameters — measured, but not explained in a convincing
manner.
A deeper explanation requires more sophisticated tools.
Therefore we will investigate the nature of Maxwell’s equations
and attempt to provide some intuitive knowledge of them.
Modern Expression of Maxwell’s Equations
While it is possible to express most of the simple parameters
of antennas with relatively simple mathematics, it is almost
impossible to present more subtle definitions without using
Maxwell’s equations. However, an intuitive insight is possible
without a mastery of advanced mathematics. Although on first
sight Maxwell’s equations look somewhat formidable, the
relationships that these equations represent are relatively
simple to understand.
For those readers not familiar with calculus, Appendix A is
provided. In addition, Chapter 2 has already provided the
concept of an invisible sphere, with power flux projected
through the sphere as a portion of the sphere’s surface. In
Chapter 2 we made no attempt to explain what this power is.
We simply accepted that light or “RF power” was propagating
through the surface of the sphere. We will now consider a
similar concept, but rather than power we will consider the
flux to be either an electric or magnetic field that, together,
constitute the power of the wave.
Here we explain “Maxwell’s” equations, which actually

92
originate from Gauss, Faraday, and Ampere. These equations
are applicable not only to static electric and magnetic fields,
but also to electromagnetic radiated fields. First we will explain
the totality of Maxwell’s equations and then concentrate on the
equations and their forms that are most useful for a discussion
of antennas.
The modern expression of Maxwell’s equations represents
refinements made by Oliver Heaviside, who used vector
calculus and some other notation changes. However, the
essence of the relationships was initially developed by
Maxwell.

Maxwell’s First Equation, Gauss’s Law for

Electricity

The closed integral (sum) of an electrical field

over the closed surface area (dA) is the sum of positive and
negative electrical charges contained within the surface (q) and
inversely proportional to the permittivity (ε) of the “material,”
or lack thereof, inside the surface.
In mathematics, the simplest form for “summing” is the
addition sign (+). A shortcut can be used if the summed values
take on a sequence conforming to some defined change (Σ), the
Greek letter “sigma” for “summing.” If we need to sum over
some function (no discrete values), we use the integral calculus
sign

This is a “smoothed sum” over some function f. For example, if

we want to sum the area under a sine wave, we need a
smoothed sum and thus a precise answer requires calculus.

93
 An additional refinement

is simply the notation for a “smoothed sum” (in this case the
electric field fluxes) over the surface of the same imaginary
closed surface, usually a sphere as described in Chapter 2. The
little circle is written over the integral sign indicates we are
summing over a closed area. q is an electrical charge in
coulombs (can be either negative or positive; for example, an
electron has the minimum electrical negative charge q). We
begin with an electrical charge at the sphere’s center (Figure
3.1). We measure the total flux of the electric field through the
sphere by using the operator

This is very similar to the problem associated with an

94
isotropic antenna, which radiates equally in all directions. A
“point charge” is also isotropic in that the field flux will be
identical at all points on the sphere (assuming of course we are
observing it in free space and the charge is located at the
center of the sphere). However, the total flux will always be the
same over any size or shape of enclosure (here we are using a
sphere as the example). Also, the total flux will always be the
same no matter how we distribute the charge or charges within
the enclosure. We can intuitively accept this by noting that a
100 W light bulb will radiate a total of 100 W through the
sphere even if it is not located at the center. The flux will
simply be greater on the surface nearer to the light bulb.
We use a dot product, which is just a special form of
multiplication for vectors, in this case over the surface
surrounding the total charge(s). The dot product simply
accounts for the angle of an area that is not normal to the
source of the electrical charge (Figure 3.2). This is much like
our solar cell example from Chapter 2 being tilted somewhat
away from the Sun, where:

Aperture = area (m2) sin θ

95
(Equation 3.1)
Thus if the aperture is perpendicular to the charge (90
degrees), then the aperture is equal to the actual area of the
surface. In this case, the dot product simplifies to a simple
multiplication.
Finally, q/ε0 represents the amount of charge (in coulombs
divided by the permittivity of the medium), in this case free
space where ε = ε0 This is the essence of Maxwell’s first
equation: The total E field flux through a surface containing a
total of ±q charge will be equal to the total charge divided by
the permittivity of the material or free space inside the surface-
defined volume.
If we place a negative and equal positive electric charge
inside the sphere and integrate the total electric field flux, we
will find the total flux will be zero, since the two charges cancel
and will result in zero charge. Thus in this first equation, q
must account for the sum of all charges (negative and positive)
within the closed surface.

Maxwell’s Second Equation, Gauss’s Law for

Magnetism

The closed integral (sum) of a magnetic field

contained over a closed surface area (dA) is zero.

A critical difference between electric and magnetic fields is
that an electric field can result from a point charge. If the
universe were occupied by only one electron, the entire
universe would be filled with a very small negative electric
field. There is no such analogy with magnetic fields. There
must always be an opposite charge of reverse value,
exemplified by the north and south poles of a magnet. If we

96
place such a dipole magnet inside our sphere, we find that over
the entire surface of the sphere there will be zero net magnetic
field simply because the two field fluxes sum to zero (Figure
3.3).

Unlike the straight field lines emanating from a point electric

charge, the magnetic field lines form a concentric curvature
structure. The only “straight” magnetic field line is found along
the line directly connecting the opposite poles. So if we
imagine a sphere surrounding our bar magnet, we can easily
envision any magnetic field line that propagates “out” of the
sphere must also come back “in,” thus cancelling the net
magnetic flux through its surface. The equation always equals
zero. Consequently Maxwell’s second equation is always equal
to zero.
Now things start to get interesting!

Maxwell’s Third Equation, Faraday’s Law of

Induction
Maxwell’s third equation:

97
 Faraday’s Law states that an electric field E is created when
the magnetic flux ΦB changes with time over some length (dl).
In other words, a time-varying B field creates an E field. An
example is pushing a magnet in and out of a coil of wire
(inductor), creating an electric current (moving electrical
charge) in the inductor. It also explains how an electric current
is generated when rotating the inductor inside a magnetic field,
thus creating a generator (Figure 3.4).

Maxwell acknowledged that it was the work of Faraday that

led him to his revolutionary theory. Faraday was not an expert
mathematician, so he was unable to express the subtleties of
his insights in mathematical form. Maxwell was perhaps the
first to realize the significance of Faraday’s work. He
understood it and was able to express it with complex
mathematics, with some help from Heaviside.

98
Ampere’s Law

Ampere’s Law states that the magnetic field (in this case
generated along a loop conductor of length l) is proportional to
the current I in that loop and the permeability of the medium
inside the loop. An example of this is the electromagnet, or
solenoid, where a magnet is created by running a current
through a conductor (the effect is increased by configuring the
conductor into a solenoid). Notice that the third and fourth
equations are reciprocal: an electric field is formed by a moving
magnet and a magnetic field is formed by a moving electrical
charge, or current. A stationary magnetic field produces no
electric field and vice-versa.
It should now be apparent that there is some type of
reciprocity between the electric and magnetic fields. We can
generate an electric field by moving a magnet in and out of a
solenoid, but we can also create a magnetic field by running an
electric current through the same solenoid. An electromagnet
such as the one shown in Figure 3.5 is a practical example of
Ampere’s Law at work.

99
 Although Maxwell’s equations look formidable to anyone
without knowledge of advanced calculus, their meanings are
easily understood by anyone with a basic understanding of
electricity and magnetism. The development of these equations
into forms that explain electromagnetic fields does require
advanced mathematics. Indeed, most of the time a calculus

100
student spends in advanced courses deals with learning how to
develop and solve the bread and butter equations (solving for
derivatives and anti-derivatives of functions). The concepts are
far easier to understand intuitively. Therefore, when possible, I
will skip most of the advanced mathematics and simply state
the relevant results, but also attempt a simple explanation.
Maxwell had a problem with the fourth equation. In the case
of a capacitor inserted into a wire loop, there is zero loop
current in that a current does not flow around the complete
length of the loop. However, a current is necessary to charge or
discharge the capacitor (to displace charge); therefore Maxwell
derived and added an additional term to Ampere’s Law to
account for this displacement current.

(Equation 3.2)
Thus. if we add this to Gauss’s Law we derive Maxwell’s
fourth equation.
Maxwell’s Fourth Equation

Thus the displacement current can be defined as an electric

field flux ΦE that changes with time (dt indicating the time
derivative of ΦE). This is the reciprocal operation of Maxwell’s
third equation. The addition of the displacement current was
the critical term for defining all the relationships within an
electromagnetic wave. From the third and fourth equations we
find that a changing electric field creates a changing magnetic
field and vice-versa. This is precisely how electromagnetic
waves propagate through free space, a continuous sequence of
E-M-E-M… fields as described by the third and fourth
equations.

101
 Usually, we refine the word “change” to “accelerate” for a
more precise description of how a field must behave in order to
create a propagated wave. Usually we assume a radio wave to
be sinusoidal, and indeed a sine wave is nearly always in a state
of acceleration. (See Appendix A for a proper definition of
acceleration.)
These four equations define exactly how fields are formed
and how they behave. In this book we are concerned only with
the formation and propagation of radio waves. In free space
(vacuum) there is no matter and thus no possibility of a
charged particle and no possibility of a magnet since both
require some type of matter present (even an electron). Also
there is no current present in free space because there is no
conductor on which a charge (electrons) can travel. (Of course
electrons can flow through a vacuum, but since electrons are
matter, the vacuum is no longer a vacuum!) Therefore, the first
two of Maxwell’s equations, in free space, equate to zero. The
fourth equation would also equate to zero, except for Maxwell’s
displacement current term, and this is the key equation to
describe the radiation of radio waves.
Therefore, everything you need to know about an
electromagnetic wave can be condensed into two equations:

(Equation 3.3)

(Equation 3.4)
On an antenna the situation is different. In order to create
the fields in the first place, we must either have an accelerating
charge (current). As we will soon discover, a moving charge
with constant velocity is equivalent to a dc current and will
thus create only a static magnetic field around the conductor.

102
However, an accelerating charge (current changing amplitude
and/or direction) over time creates waves that radiate. An
antenna must be made of matter and most are constructed
from conducting material, such as metals. But how do we know
these propagating fields constitute a wave?

The Wave Equation

Imagine a string that is stretched along the X-axis (Figure
3.6). We “pluck” the string at one end thus create a wave that
propagates along the string at velocity v. The wave exists in the
X-Y plane, with the amplitude of the wave being a
displacement in Y direction and the direction of travel along
the string along the X-axis. More importantly we see that the
amplitude of the wave is defined in both space (X vs Y) and
time Y vs T) and the coefficient in the equation is velocity.

The wave equation had been known since the 18th century
(see Chapter 1) when it was derived to describe the harmonic
motion of a plucked string. Any phenomenon that takes the
form of a wave must conform to the wave equation, a second-
order differential equation of the form:

103
(Equation 3.5)
In order to actually solve this equation for a given set of
physical circumstances, we would need to be able to solve a
differential equation. However, we only need a brief
explanation to understand what this equation means. Also the
derivation of this equation requires some rather sophisticated
calculus. I will omit this derivation and simply refer the reader
to most advanced texts on electricity and magnetism (E&M) for
the formal derivation.
The wave equation defines exactly what is happening in the
Y-axis (where the instantaneous wave’s amplitude exists). The
first term

explains what the wave is doing relative to the X-axis, which is

the direction of propagation. The second term

is a function of the velocity of the wave on the string, and the

third term

informs what the wave is doing as a function of time. All other

parameters of any wave can be derived from this simple set of
relationships. This is the essence of the wave equation. Now let
us explore a physical example for additional clarity and
perform some important derivations.
The velocity of the wave along a string is dependent upon the

104
tension on the string and the mass of the string/unit length.
Anyone who has ever played a stringed instrument (Figure 3.7)
knows this relationship: If you tighten a string, you get a higher
frequency (note) where

105
 Many readers will recognize this equation as defining the
relationship between a radio wave’s wavelength and frequency
where the velocity is usually taken to be constant (the speed of
light, c). In the case of a violin, v is also constant and we adjust
the tone (frequency) by changing the λ by shortening the
vibrating string with our finger.
The other term that describes velocity is the mass of the
string (more precisely the mass per length). The relationship
between the “tightness” of the string, the mass per length of the
string, and the velocity is

(Equation 3.6)
where σ is the “tightness” of the string and µ is the mass/length
of the string.
Strings on musical instruments optimized for lower
frequencies (notes) will be bigger (higher mass per length). If
we convert our “tightness” term into “looseness” (tightness
being the reciprocal of looseness) then the equation becomes

(Equation 3.7)
Or,

(Equation 3.8)
Also if we define “looseness” as ε, we get

106
(Equation 3.9)
Now that we have a velocity term for the mechanical version
of the wave equation we can investigate the similarity (and
differences) to the electromagnetic wave equation.

Wave Equation for Electromagnetic Waves

Since we do not need the loop current term, we rewrite
Maxwell’s fourth equation without Ampere’s current term.

(Equation 3.10)
Again, the third equation is:

(Equation 3.11)
The derivation of the wave equation from Maxwell’s
equations is quite complicated. For completeness, we present
some of the important steps in the process. However, again, the
important point is to understand that the third and fourth
equations lead directly to a wave equation.
From the third equation we derive the following:

(Equation 3.12)
Here we start with Equation 3.11 and equate it to Equation
3.10 by using the reciprocal to the closed integral, or the

107
differential. The integral (or anti-derivative) and the
differential (or derivative) represent the two main divisions of
calculus that solve problems by working in different directions.
A rough comparison would be multiplication and division that
solve for different circumstances in opposite mathematical
“directions.”
Also if we take the second derivative of

and some advanced mathematics, we can derive an equation

with two partial differential E terms, one relating to the X-axis
and the other to time:

(Equation 3.13)
If we take the partial time differential

of another partial time differential

we get a second partial time differential where

(Equation 3.14)
Therefore, from Equations 3.13 and 3.14 using simple algebra
we derive

108
(Equation 3.15)
And finally from Equation 3.9 we get

(Equation 3.16)
This equation is exactly the form in Equation 3.5 and
therefore satisfies the wave equation. Thus the E field (usually
defined as a sine function) is a wave. Also, by Maxwell’s fourth
equation we know that an accelerating E field creates a B field
and so on. Thus electromagnetic radiation must be a wave. Again,
we skipped much of the complex mathematics to arrive at this
conclusion for simplicity.
In Chapter 1 we described how Maxwell derived the speed of
light in the form:

(Equation 3.17)
This form of the velocity term has now been shown to be a
direct analogy to the wave action on a string. However, rather
than describing μ and ε as a mass/length and as “looseness” of
the string’s physical condition, we now define space itself in
terms of different physical conditions: permeability and
permittivity. As Einstein later postulated, the presence of a
field “bends” space, so an oscillating field “oscillates” space
itself, in space-time. There is no matter in free space (including
the imaginary “ether” once thought to exist); therefore space
itself is the medium for electromagnetic waves.
As Maxwell rightly credited Faraday for the breakthrough

109
that led to Maxwell’s theory of electromagnetism, Einstein
rightly credited Maxwell with the foundation of his relativity
theory. Einstein, in turn, discovered the photoelectric effect,
which is the mechanism for the solar cells we discussed in
Chapter 2. Also, he showed that light has momentum, and
therefore light is also a particle! The particle of light (or any
other wavelength of electromagnetic radiation) is called a
photon. When a photon has enough momentum (energy) it will
dislodge an electron, creating the photoelectric effect. This
effect is much like a billiard ball striking another billiard ball,
but at the sub-atomic level. This leads to the peculiar fact that
electromagnetic waves exhibit a wave-particle duality. In
Chapter 10 we will explore this duality in some more detail.
However, we know it is a wave because it conforms to the wave
equation, and it is also a particle because solar cells would be
impossible without the photon.
Thus Einstein played a pivotal role in formulating the basis
for the two great subdivisions of modern physics: relativity and
quantum mechanics. You may be pleased that for antennas
operating in the RF portion of the spectrum, Maxwell is all that
is required for even an advanced intuitive and practical
understanding. Maxwell can be thought of the “bridge”
between Newtonian classical physics and modern physics
pioneered by Einstein (and of course many others).

Relationships Between the E and B Fields

The wave equation is expressed in differential form. The
differential was briefly shown to be the reciprocal of the
integral form. For the wave equation and some additional
explanations, it is easier to use the differential form of the
equations. We now know that there must be an E and also a B
field to constitute an electromagnetic wave. The relative
amplitudes of the two fields will be derived later and this ratio
uses relatively simple mathematics. However, we need
differential vector calculus to describe another important
relationship: the relative polarity between the fields.
Again, we are concerned only with electromagnetic radiation
in this text, so we can focus upon the equivalent differential

110
forms of only the third and fourth equations. Again after some
rather difficult calculus we can re-write the third and fourth
equations as:
Maxwell’s third equation

(Equation 3.18)
and Maxwell’s fourth equation

(Equation 3.19)
As before, we can eliminate the Ampere current term, here
represented by J. So the differential form of Maxwell’s fourth
equation simplifies to

(Equation 3.20)
This form of differential equation uses the (∇) operator also
known as the curl. A simple intuitive demonstration of the curl
is called the “right-hand rule.” If you shape your right hand as
shown in Figure 3.8, you can see clearly the curl result. The
thumb points in the direction of the accelerating charge
(function of time), for example from Equation 3.18

111
and ∇ E (the curl of E) is represented by the four fingers
encircling (curling) around the thumb line. These two
equations show that the opposite fields tend to wrap
themselves around a concentrated accelerating field. When a
field becomes a plane, as in a “plane wave,” the opposite field
cannot wrap itself around, and thus it also becomes a plane
wave.
There is another important result to the curl. Notice the
direction the thumb points relative to the direction the fingers
point. This implies, correctly, that the E and B fields are always
orthogonal, having a 90 degree difference — ie “vertical” and
“horizontal.” Again, the equations use very sophisticated
mathematics, but express a very simple concept.
For mathematical correctness, these equations are presented
using some different terms. For our purposes H is roughly
equivalent to B, D is roughly equivalent to E, and J is roughly

112
equivalent to µI. Since we are exploring only the conceptual
understanding of these integral and differential equations, we
may upset some advanced mathematicians by being sloppy!
See Figure 3.9. From Maxwell’s equations we find that
electromagnetic radiation occurs from accelerating charges. A
charge moving at a constant velocity (no acceleration) will
create a static magnetic and electric field, as from a dc current
in a solenoid: constant current, constant charge. From classical
physics we know that acceleration is the time derivative of
velocity or, A = dV/dt (Appendix A). For radiation to occur, the
velocity of the charge must change (usually at the frequency of
the RF signal). Charge that changes velocity (accelerating) at a
time-dependent value of a sine wave will produce a field
directly analogous to the same sine wave. Since current (I) is
moving charge, a changing rate of moving charge is the cause of
radiation. Thus antenna current is of fundamental importance
for determining the key parameters of how an antenna
behaves.

Notice that the acceleration of charges and fields occurs (as

in the string example) in the Y direction only, which is the
change in the rate of amplitude. In contrast, an electromagnetic
wave’s velocity (also in the string example) remains constant

113
(the speed of light) in the X direction, or the direction of
propagation. As a second step to the intuitive right-hand rule,
image that the fist throws a punch. The direction of the punch
(the direction of propagation) is orthogonal to the direction of
both the thumb and the fingers.
The fields that propagate outward from the charge’s location
become detached at their source, since polarity is changing at
the rate of twice the RF frequency (Figure 3.10). The result is
that we create “waves” of electric and magnetic fields, not a
constant static field. The electric field can be compared to a
soap bubble first appearing on the surface of soapy water from
a bubbling air source underneath. The bubble expands until it
finally detaches from the water, propagating away from its
source, “up,” only to be followed by the next bubble already
being formed. Eventually a series of bubbles is formed. Of
course bubbles tend to become small spheres, while the shape
of the field “bubbles” will conform to the directivity of the
antenna. As they propagate away from the antenna, they
converge into “plane waves” as the wavelength becomes much
greater than the distance propagated as along the expanding
sphere.

114
Impedance of Free Space
Maxwell’s equations provide the relationships (direction and

115
ratio) between the magnetic and electric fields associated with
an electromagnetic wave. The ratio of the two fields’
amplitudes remains constant and the two fields are always
orthogonal, at least in the far field. This is the condition for
plane waves. As we will show later, there are exceptions to this
general rule.
Since the ratio of these two fields is constant in a free-space
electromagnetic wave, those familiar with Ohm’s Law (Z = E/I)
will correctly postulate that some type of resistance is present,
but in free space?
In order for propagation of electromagnetic waves to occur in
any medium (including free space), there must exist a finite
intrinsic impedance. If the impedance is zero, or if it is infinite,
the magnetic and electric fields would have to be either infinite
or zero. If one or the other cannot exist, there can be no
electromagnetic radiation. In order to have electromagnetic
radiation in free space, free space must have some finite real
impedance, greater than zero but less than infinity. Indeed, the
universe would be a very different place without a finite
impedance of empty space!
We have already defined two physical characteristics of free
space, permeability and permittivity. That should give us our
first clue that there is more to free space than just these two
simple terms, but we begin with these terms, and use some
algebra and equations from fundamental classical physics.
Ignoring the coefficients for simplicity, we show how we
derive ohms (a pure resistance Ω) from the permittivity and
permeability of free space:

(Equation 3.21)
and thus

116
where F is Farads, H is Henrys, and J is Joules where
J = Nm

(Equation 3.22)
where N is Newtons (a physical force) and m is meters. Thus we
can actually relate linear distance and physical force to the
electromagnetic terms that describe free space.
From Chapter 1 we found that

reduces to a velocity (of light) in free space, now we find that

reduces to an impedance (of free space). Thus it is clear that the

two terms are intrinsically related where:

(Equation 3.23)
Then

117
(Equation 3.24)
This is the ratio of the values of the electric and magnetic
fields, again comparable to the ratio of voltage and current
equating to resistance.
We can formalize the relationship between the speed of light
in free space and the impedance of free space by:

(Equation 3.25)
Because

(Equation 3.26)
consequently

(Equation 3.27)
and if

(Equation 3.28)
then

118
(Equation 3.29)
must be Z0.
Plugging the coefficients of these terms back in, we can
calculate the actual impedance of free space:

(Equation 3.30)
The speed of light is related to the impedance of free space by
the simple term 1/μ0. Since 377 ≈ 120 π, Z0 often appears as
either term.
Notice the similarity to Ohm’s Law, in that H roughly
coincides with current and E roughly coincides with voltage.
Also, for any space occupied by matter rather than a vacuum,
the values of μ and ε are somewhat larger. Thus the speed of
light inside the material and a material’s intrinsic impedance
are calculated by the same simple equations. This will become
an important concept in later chapters.
For most practical purposes, air’s terms are sufficiently close
to a pure vacuum to be ignored. The exceptions are almost
always at frequencies above about 1 GHz, where the size of
raindrops approach the operating wavelength, or at
frequencies where air molecules (particularly oxygen) render
the air as a partial absorption medium. The atmosphere can
also affect the propagation of nearly all frequencies, but the
antenna itself is seldom directly affected by its operation in air.
Above, we derived the impedance of free space from a
derivation of Ohm’s Law. We now have an electromagnetic
specification for free space! Such a specification is necessary to
quantify how we couple RF power into free space and how we
capture RF power from free space. In effect, free space becomes
part of the entire radio system “circuit.” As such, this number is
critical to all aspects of antenna design and performance.
At this point we simply accept that “space” indeed has
physical characteristics, even when there is no matter present.

119
To explore this non-intuitive reality, classical physics fails and
the curious are forced into the non-intuitive and often bizarre
world of modern physics. People often feel a need to
understand fields and radiation. We cannot experience first-
hand electric, magnetic, or even gravitational fields. We can
only observe the effects they have upon matter. However,
science allows us to quantify the actual physical characteristics
of these non-sensual phenomena. The proof of our physical
understanding is using this understanding to duplicate
experiments and then actually utilize this physics by
engineering radio equipment, including antennas.

Fields, Power, and Energy

The field strength of an electromagnetic wave can be
measured either by the electric field (volts/meter) or the
magnetic field (amps/meter). Usually RF field strength is stated
only in volts/meter (the electric field) since it is easier
(technological limitations) to directly measure than the
magnetic field. Notice that these terms are measured in single-
dimensional space (meter). If we multiply the electric and
magnetic fields, we are left with watts/meter2, a two-
dimensional (area) function reminiscent of the surface of a
sphere and also defining a power term.

(Equation 3.31)
Above we see that the relationship between field strength
and power is really a special calculation of the power law, but
also involves a dimensional transformation. The direct
calculation of power density from a field strength requires
knowledge of the exact relationship between the field strength
and power. For this we need a combination of the power law
and Ohm’s Law. The form we base this calculation upon is

120
(Equation 3.32)
In this case Z0 is the impedance of free space (120πΩ), E is the
electric field strength, (volts/meter) and Sr is the power density
(watts/meter2) both at the point of measurement at distance r
from the source, on the sphere where the measurement takes
place. Again, this is simply a combination of the power law and
Ohm’s Law.

(Equation 3.33)
or,

(Equation 3.34)
For example, we can calculate the effective radiated power
from an isotropic (EIRP) antenna in the direction of the
measured point. If we can measure the field strength at a
distance r (Er) we find the power density by Sr = Er2/Z0, and
then multiply Sr by the total area 4πr2. Thus the total
transmitter power (EIRP) is:

(Equation 3.35)
Thus we can calculate through these simple equations the
relationships among power density, EIRP, field strength,
and/or the impedance of free space.
We have to know that the transmitter is using an isotropic
antenna to make the isotropic assumption. In any event, if we

121
know the distance from the transmitter and the field strength
at that distance we can directly calculate the EIRP at the point
(also defining the direction from the transmitter) where we
took the measurement despite the directivity of the antenna.
Furthermore, if we also know the power of the transmitter, we
can also calculate the gain of the transmit antenna, in dBi. So
this is just an extension of the intuitive reasoning from Chapter
2 to now include fields into our arsenal of terms.
Although the energy term is not often used in conjunction
with antenna problems, we can complete our explanations by
defining the term. Energy is defined as the time integral of
power. For example, you are billed for electric energy, not
electric power. Electric energy is defined in watt-hours, or
more conveniently kW-hours. Electromagnetic energy can only
exist as a radiating set of fields propagating at the speed of
light. If we have a 100 W isotropic light bulb burning for 1000
hours in free space, then the total radiated electromagnetic
energy will be 100 kWhr. This energy will be contained within
a sphere with a radius of 1000 light-hours, at least for that
particular instant in time. So electromagnetic energy is not only
a time integral, but also a space integral. Where the E and H
fields are measured in m, power is measured in m2 and energy
in m3, in this case the volume of an expanding sphere

Such measurements are sometimes useful in astronomical

research, such as finding residual energy from the Big Bang
(Chapter 10).

Radiation Resistance
We will encounter several terms in antenna physics whose
very title can imply an incorrect intuitive understanding. The
term radiation resistance is often a source of confusion. It is easy
upon first encountering this term to interpret it as a resistance
to radiation. In other words, the higher the resistance the more
difficult it becomes to radiate. This is an understandable yet

122
mistaken interpretation of the definition of radiation
resistance. Unfortunately the confusion does not stop with the
title. We introduced this term in Chapter 2. Now we will
expand upon it, and specific applications will be presented in
later chapters to further improve understanding.
If we place an RF voltage source on a conductor (antenna)
and measure its impedance, we will see the real part of the
impedance (a resistance), and, likely, a measureable reactance.
This is the case even for a theoretical lossless conductor,
perhaps constructed of a super-conducting material. How can a
zero resistance conductor register some real value of
impedance? The answer, of course, is that power is being “lost”
to free space by radiation. The resistive value of an antenna at
a specific frequency that is associated with radiated power is
called the radiation resistance (Rr) of the antenna.
The formal definition of radiation resistance is very often
misunderstood, even by professional radio engineers. Rr is a
critical term for calculating antenna efficiency which, of course,
directly affects effective aperture, gain and overall antenna
efficiency.
Antenna efficiency is defined as:

(Equation 3.36)
where (Rl) is loss resistance, or resistance resulting in RF power
conversion into heat. This is a simple, straightforward
equation, but the determination of Rr can be complicated and
therefore often mistaken. It is obvious that you must use the
correct antenna Rr value to get a proper calculation of
efficiency. But how do you determine the correct Rr value? In
Equation 3.36, how can you determine the difference between
Rr and Rl?
In the most basic model, Rr, being a resistive value, may be

123
calculated through the power law where

(Equation 3.37)
where Pr is the radiated power and I is the current at a point
along the antenna. However, this only provides the Rr at that
point on the antenna, not the antenna as a whole, which is
necessary for a proper calculation of efficiency. The first step
toward a proper definition of Rr is to measure I at a current
maximum point along the antenna. Second, to calculate Pr it is
necessary to integrate (sum) all the power being radiated
through the sphere. The process is relatively easy if the
antenna and wavelength are small. In such cases the far field
Poynting vector may be integrated inside an anechoic
chamber. For antennas and wavelengths too large for practical
chamber sizes, power integration is exceedingly difficult,
especially when the antenna is mounted close to the ground or
other material that causes power loss.
On the other hand, I is a straightforward measurement at the
point of current maximum. Thus this equation has three
variables, two of which are very difficult to directly measure.
Consequently it is difficult to set up a closed form equation that
correctly defines Rr using Equation 3.37. Furthermore, this
equation does not explain other factors that shape the value of
Rr. Therefore it cannot explain why Rr can change, even if Pr
remains constant but I changes. To resolve this problem, Kraus
gives a more refined definition for the general definition
(Equation 3.38) where an antenna’s all-important 3 dB
beamwidth (gain) is included in the definition. The idea is that
the total power radiated will usually closely approximate the
solid angle that is defined by the 3 dB beamwidth. In effect, the
assumption is that the power remaining outside the 3 dB
beamwidth will “fill in” the solid angle, as shown in Figure 2.8
in Chapter 2 instead of the “actual” pattern shown in Figure

124
3.11.

(Equation 3.38)
Equations 3.38 and 3.37 are approximately equal. The (max)
term in S(θ,φ)max designates a specific area on the sphere: the
steradian value (Ωa) of the 3 dB beamwidth of the maximum
gain lobe. In other words, S(θ,φ)max is the area on the sphere
that contains the main lobe of the antenna pattern. When the
antenna pattern is almost isotropic (nulls are less than 3 dB
difference than the peak), then W is the same as an isotropic or
the 4π solid angle defining the entire sphere’s surface. If the

125
gain pattern of the antenna is a distinct lobe (like the pattern
from a Yagi-Uda antenna), W would be defined as the solid
angle containing the maximum gain point and surrounding
area within 3 dB of the maximum gain point.
On the other hand, Equation 3.38 is precisely equal to
Equation 3.37 when the steradian contains precisely all
radiated power as in Figure 2.8 (in contrast to Figure 3.11). For
example, let us assume that the main lobe has a maximum gain
of 10 dB, which is also a linear power gain of 10. However, this
maximum gain location typically defines a point on the sphere.
Around this point we must define an area that is defined by the
steradian value. Let us assume that this area corresponds with
the gain number of the steradian value, or 1⁄10 the area of the
sphere, or

This model assumes that all the radiated power is contained in

that steradian constraint and that the area on the sphere
defined by that steradian and that the power density is uniform
over that area with zero power appearing over the balance of
the sphere. Therefore the power integration over the entire
sphere (average power) will be 1⁄10 the peak radiated power on
the sphere.
I2 is the RF current squared appearing on the antenna
element at a current maximum, as detailed below. Thus, in the
general Equation 3.38 above, Rr depends upon the steradian
(Ω) beamwidth of the maximum radiated power, which also
directly relates to gain and thus aperture as well. Therefore Rr
is a function of antenna current, free-space impedance, power
radiated and antenna gain. We will later develop this general
equation for Rr for specific antenna types. In many cases, Rr is
the critical term for designing and building efficient antennas.
Kraus specifies the general definition of radiation resistance
(Equation 3.38) because it represents a far more utilitarian
definition than Equation 3.37. Equation 3.37 cannot inform if

126
the terms are being influenced by antenna gain and/or loss,
whereas Equation 3.38 isolates these two terms.
If we know the total power radiated from a lossless antenna
and its current maximum, we can directly calculate Rr by
Equation 3.37. However, care must be taken in the computation
of this current value. For example, if we feed a folded dipole
(Figure 3.12) at a current maximum, we are only measuring 1⁄2
the current flowing at that “point” (compared to the
wavelength the “folded element” is essentially at the same
points along the antenna as the fed element). This will give a
false Rr value of about 292 Ω rather than the correct Rr for a 1⁄2
λ dipole in free space of about 73 Ω. Therefore, again, Rr is not
necessarily equal to a power law calculation taken at the antenna’s
feed point. For any arbitrary point on the antenna, the radiation
resistance at that point is Rrp.

Relationship Between Radiation Resistance

and Feed Point Impedance in Linear
Antenna Structures
The real portion of the feed point impedance equals the
radiation resistance plus losses of the antenna only for single-
conductor antennas fed at a current maximum along the

127
antenna. It is a very common mistake to automatically equate
the feed point impedance (real value) to the antenna radiation
resistance measured at a current maximum or worse, elsewhere
along the antenna.
Rr will coincide with the real part of the feed point
impedance at the center of an antenna that is less than or
equal to 1⁄2 electrical wavelength long, or an antenna that is an
odd number times 1⁄2 wavelength. For all other antenna types,
great care should be used when calculating or measuring
radiation resistance, since simply using power input and feed
point current does not necessarily yield Rr. Indeed, the
difference between the feed point impedance Rrp and Rr can be
very large. For example, a single wire 1⁄2 λ dipole in free space
has the same value of Rr no matter where the feed is connected.
The only point where the two values coincide is at the center of
the dipole, which is also the current maximum. Rrp = Rr only at
a current maximum on the antenna. Rrp is usually associated
with a feed point that is not at a current maximum.
The resonant center-fed 1⁄2 λ dipole has an input impedance
of about 73 Ω, which is also the Rr of the antenna. Now if we
move the feed point impedance to 25% the distance from one
end (off center), the feed point impedance at Rrp is now about
137 Ω, but the antenna Rr remains 73 Ω.
Another complexity: In antennas longer than 1⁄2 wavelength
the current distribution (and thus the radiation resistance) can
be changed by changing the feed point location. So, when
calculating and/or measuring the location(s) of current
maximum(s) along an antenna element, be careful that key
terms that define Rr are not changed by changing the feed
point. Thus if the current is to be measured at a current
maximum by moving the feed point, the current maximum may
be moved as a result of moving the feed point! See Figures 3.13
and 3.14.

128
129
 There are few applications where Rr becomes so critical as it
does in ground-mounted vertical antennas and antennas that
are very small compared to the wavelength of operation. This
will be treated with great detail in later chapters. For any
experimenter working with antennas generally and these types
of antennas in particular, it is essential to master the proper
definition or radiation resistance.
Now let us return to Equation 3.38. Assume that the 3 dB
three-dimensional beamwidth (equivalent area) is

In other words, the area is 1⁄8 the total surface area of the
sphere. Since we know our steradian value, we also know we
have a power gain of about 8, or about 9 dBi. We carefully
measure our feed point current as 1 A.

results in 200 W EIRP with a gain of 8. Thus we can calculate

(Equation 3.39)
Thus Rr is inversely proportional to the antenna gain and
therefore also approximately proportional to the inverse of the
3 dB beamwidth. Notice that we have isolated and precisely
defined these terms by including antenna gain (or aperture) in
the equation. Measured deviations from these predicted values
indicate loss and/or mistakes in the measurement. As such, it is
a far more useful general definition than Equation 3.37.
Regarding ground-mounted vertical antennas, a yet more

130
special form is presented later.
As a final thought, it is interesting to note that if we
extrapolate the characteristics of a 1⁄2 λ dipole with infinitely
thin wires in free space (gain and Rr) to a hypothetical isotropic
antenna, we would find the Rr = 119.9 Ω, or exactly Z0/π, where
Z0 is the impedance of free space. To derive this number for the
isotropic, we simply adjust the gain from 2.15 dBi to 0 dBi and
the consequence of this adjustment to the steradian value.
Once again we see an example of the beautiful symmetry of
antenna physics, particularly when we use proper terms that
lead directly to closed-form equations.

Antenna Pattern Formation

The accelerating charges along the length of an antenna
element represent the source of the radiated wave. But what do
the resulting patterns look like? Thus far we have shown
several antenna patterns. In Chapter 1 we presented some
specific antennas (isotropic and a 1⁄2 λ dipole) and their
respective patterns. We also just showed the pattern of the 10/8
λ linear antenna fed at two different points. The question is:
What is the mechanism for producing these patterns?
We begin by introducing an imaginary antenna segment
called a short dipole. A short dipole can be thought to have
some amount of accelerating charge or instantaneously
changing current traveling along in a very short length of the
antenna. Indeed, with the help of integral calculus we can
make the length essentially zero to create an infinitesimal dipole
(see Appendix A). The short dipole, although very short, is also
assumed to be straight. In other words, the charges on the
dipole can only flow in two directions, along the length of the
straight segment. We know that an accelerating charge
produces a radiated electric field parallel to direction of the
acceleration and a radiated magnetic field in an orthogonal
direction, and the direction of radiation is orthogonal to both
fields. Indeed, we can calculate the pattern of such a short
dipole.
If we place many short dipoles in series, we can model a

131
“real” antenna at an arbitrary length.

PADL
The phase and amplitude of these accelerating charges from
each of these short dipoles (antenna segments) contribute to
the total pattern of the entire antenna in the far field. There
are also two other critical parameters that define the resulting
far field: the direction of the accelerating charges and the 3D
location of these accelerating charges. Thus we need to know
four variables to determine all the parameters of an antenna’s
far field: Phase, Amplitude, Direction, and Location — PADL. The
relationships of the electric field, magnetic field, and direction
of propagation become deterministic depending upon the
PADL variables.
Returning to the infinitesimal dipole antenna, the phase is
simply the point along the RF wave defining that phase point
(0-2π radians) along the RF sine wave. The amplitude is the
instantaneous RF current amplitude. The instantaneous
direction of the current is always the same (defined by a point
on the antenna element) changing twice per RF cycle and thus
reversing the polarity twice per cycle. There are only two
directions the current can accelerate on a linear element: left
or right (as viewed from the side). The location of the
infinitesimal dipole is a point in free space. All calculations for
any antenna in free space simply sum the PADL terms (using
integral calculus) to form a far field pattern. But first we need
to determine the pattern of an infinitesimal dipole so we will
know what to sum in more complex antenna structures.
Ignoring the integral calculus to derive this important term,
we find that the linear (power) pattern of the infinitesimal
dipole follows a curve defined by a sin2 function. The pattern is
a doughnut pattern very similar to the 1⁄2 λ dipole, but has less
directivity. The directivity of an isotropic antenna is 1.
Therefore the dBi directivity of an antenna is simply:

(Equation 3.40)

132
where again,
Gain = Directivity × Efficiency (expressed as fraction)

(Equation 3.41)
See Table 3.1 and Figures 3.15 to 3.17.

133
 So, as we build a full-size 1⁄2 λ dipole from many short dipole
segments, we find we increase the directivity by about 0.42 dB.
The amplitude of the current along a 1⁄2 λ dipole is sinusoidal,
and the amplitude of the current in the short dipole is assumed
to be constant along its length (a small portion of a sine wave).
Therefore, the amplitudes of each of the long series of short
dipoles will have to be adjusted to the sinusoidal value at the

134
same point on the 1⁄2 λ dipole in order for the definite integral
to provide the correct result. For example, the current
amplitude value will be maximum at the segment at the center
of the 1⁄2 λ dipole. As we move away from the center, the
segments’ current amplitudes will decrease as a sin function,
and at the ends of the dipole, the segments will have close to a
zero value.
The concept of the short dipole segmentation of larger
antennas is very convenient for direct calculation of patterns
by hand (using calculus) or by advanced mathematical
modeling tools such as MATLAB. Antenna modeling tools such
as EZNEC also use segmentation for equivalent calculations and
as shown in Figure 3.18.

135
 Although finite-length segments used in antenna modeling,
by definition, offer only an approximation, surprisingly few
segments offer results very close to results obtained through an
exact solution from integral calculus. Thus we begin with a
formal definition using calculus, but then discover that
adequate accuracy can be derived from much simpler summing
techniques.
The total antenna pattern is simply the sum of these smaller
patterns at a far distance from the antenna (shown in Figure
3.19). The doughnuts also contain phase information that can
lead to either field reinforcement or cancellation in the far
field. Therefore, the complete antenna pattern is simply a
smoothed sum of the PADL parameters. Of course, the
resulting fields from each of the segmented currents are
assumed to propagate at the speed of light. All antenna
patterns are thus the result of the current distribution in
amplitude and phase at all points on the antenna and the
position of those points in 3D space. In the case of a 1⁄2 λ dipole,
none of the doughnuts are far enough apart to provide
complete cancellation. As the length is made longer than 10/8
wavelength, the phase shift along the antenna is enough to
rapidly increase nulling in the broadside direction.

Current and Voltage Distribution

The previous section emphasizes the primary importance of
current amplitude and phase along an antenna element for
producing the antenna’s directivity. The mechanism that results
in the current distribution is key to antenna performance.
It is important to think of antenna “voltage” and “current” in
terms of RMS values. The instantaneous values are constantly

136
changing in accordance with forward and reflected waves
traveling up and down the elements. For this discussion we will
only consider the RMS values.
Again, if we consider a simple center-fed 1⁄2 λ dipole, current
will always flow in the same direction on both sides of the
dipole. We assume that the transmission line is applying equal
but out-of-phase current to the antenna’s center. Since the
wires point from the feed point in opposite directions, the
instantaneous currents run in the same direction (Figure 3.20).

When the accelerating charges reach the ends of the dipole,

they are reflected and start back toward the center. Since there
is an equal amount of charge flowing “forward” and “backward,”
there is no RMS current at the end of the antenna and thus no
radiation from the end of the antenna. This is the mechanism
that results in the sinusoid distribution of current along the
dipole. It also explains why the center of a 1⁄2 λ dipole is the
point of maximum current and thus contributes the largest
amount of radiated power compared to other antenna
segments.
Measurement of RF current is relatively easy. We can simply
place a sample loop around the element and measure the
induced voltage (Faraday’s Law). If we move the sampling loop
up and down the element we discover the current distribution
shown in Figure 3.17, which is sinusoidal. In a practical linear
antenna, the current and voltage values are not quire

137
sinusoidal due to a finite diameter of the linear elements.
However, for our purposes, we will assume a sinusoidal
distribution.
Voltage is represented as “force” per charge (coulomb). As a
charge is accelerated down a linear antenna element, it
eventually comes to the end of the antenna element. Since the
charge cannot continue flowing, the density increases and this
creates a high voltage point. An analogy is the displacement
current, which is simply charging and discharging a capacitor.
If the “end” of the antenna is thought to be a capacitor (which
it is!) then accelerating charges stop, before being reflected at
the antenna end, and create a oscillating charge in the
capacitor where Vc = q/C, where Vc is the voltage a the antenna
element’s end, q is the total charge (large) and C is the
capacitance value (small) at the antenna element’s end. By this
equation, assuming the end of a thin dipole represents a small
capacitance and the amount of charge can be very large, it
becomes apparent that very high voltages will exist even for
modest transmit power.
When a charge exists, but there is no current, there must be a
capacitance involved. The instantaneous voltage at the end of
the dipole increases while current is flowing toward that end,
and reduces to near zero when the current flows in the
opposite direction. Thus, if we consider the ends of the dipole
to be capacitors, the voltages on these two capacitors charge
and discharge in an out-of-phase relationship. As we move
away from an end, RMS current increases and thus the voltage
decreases. The current flowing on a dipole can be considered to
be Maxwell’s displacement current as it never completes a
“loop” as in Gauss’s original equation.
Not surprisingly the center feed point resistance (in this case
also very close to Rr) is simply

where f indicates the feed point.

138
 A general assumption is that the maximum voltage appearing
on an antenna’s element appears at its end point. This
assumption is helpful in developing an intuitive understanding
of voltage and thus current distribution along a linear antenna
element. The sinusoid distribution of antenna I and V are
usually 180 degrees out-of-phase along a linear antenna
element (Figure 3.21). Thus you can estimate voltage and
current peaks and nulls by working back from and antenna’s
end point (if there is one!) following a sinusoidal distribution at
the assumed RF wavelength.

Antenna Polarization
Thus far we have assumed that our antenna is linear, an
antenna element in a straight line. Such an antenna is said to
have linear polarization. In a linear polarized antenna, the
current can only flow along that line, either one way or the
other. The resulting radiated fields’ directions, according to
Maxwell’s equations are exactly defined: The electric field (E) is
parallel to the direction of accelerating charge (parallel to the
wire). The magnetic field (H) is perpendicular to the line of the
wire. See Figure 3.22. In “outer” free space a linear polarized
wave’s direction is arbitrary because there is no reference for a
line in space. Near or on the Earth, the Earth’s surface defines a
plane. With terrestrial reference, on Earth if the E field is
vertical it is said to have vertical polarization. If the E field is

139
horizontal, it is said to have horizontal polarization.

The other type of polarization is circular polarization (CP). If

we take two dipoles, one horizontal and one vertical, and feed
them 90 degrees out-of-phase shown in Figure 3.23, the
amplitude of the instantaneous radiated fields will follow the
sinusoid acceleration of the current on the elements. Since
they are 90 degrees out-of-phase, the E field vector will rotate
(either clockwise or counter-clockwise) at the same rotation
rate as the radiated frequency. The “clockwise” and “counter-
clockwise” direction of field rotation is termed right-hand and
left-hand circular polarization.

140
 The “hand” polarization is determined by viewing the CP
antenna from its rear and thus the direction of radiation. From
this vantage point if the field is rotating clockwise, it is “right
polarized,” if it is rotating counter-clockwise, it is “left
polarized.”
PADL thus also will define the wave polarization, mainly
which direction the charges are accelerating: up/down,
right/left, and or in a circle. An important antenna type called
a helix is a “natural” CP antenna and will be discussed in
Chapter 9.

141
 CHAPTER 4

Transmission Lines
A comprehensive discussion of transmission line theory can
easily fill its own textbook. Rather than repeating volumes of
information that is readily available from multiple sources, we
present a short discussion of transmission lines as they relate
to physics presented in earlier chapters. The discussion then
evolves into a more traditional description. We have shown
that Maxwell’s equations form the basis of electromagnetic
physics. These equations can be used to solve not only antenna
and free space problems, but also problems involving the
transmission line.

The Guided Wave

In Chapter 2 we stated that an antenna “…may be defined as
the structure associated with the region of transition between a
guided wave and a free-space wave, or vice-versa.” Following
that definition, most of the discussion has been on the antenna
and its relationship to free space. We alluded to the existence
of the “guided wave” and “transmission line,” but did not deal
with this important topic in detail.
Most often the antenna is not co-located with the radio and
thus a physical connection must be provided. Whereas an
antenna’s function is to form the transition between guided
and radiated waves, it is the function of the transmission line
to deliver the guided wave to/from the antenna to/from the
radio. There are two basic objectives within the line: minimize
power loss to heat Rl and minimize power loss to radiation Rr.
Therefore the ideal line will exhibit zero for both values. Also,
a line that does not radiate will also not behave as a receive
antenna, due to the reciprocity theorem. This becomes an
important characteristic to minimize the introduction of
undesirable noise into a receiver while simultaneously losing
power received by the antenna through re-radiation. An

142
excellent antenna design can be ruined by loss and pattern
distortion from a poorly designed or installed line.
Therefore the design goal of a transmission line is to
configure the PADL parameters (Phase, Amplitude, Direction,
and Location — introduced in Chapter 3) so that no radiation
appears in the far field from the line. If two conductors are
spaced << λ and the RF current is out-of-phase and equal in
amplitude, then minimum radiation will occur. In effect, the
entire B and E fields will be canceled according to Maxwell’s
first two equations at some finite distance from the line. This is
the essence of the PADL distribution in a transmission line.
The actual location (L) is irrelevant since at any location in free
space the conditions of phase/amplitude cancellation will be
the same.
Therefore, no matter which direction we “view” the
transmission line (from the surface of the sphere), we observe ​-
almost completely canceled fields and thus no radiation. If we
measure the input impedance of a perfect unterminated line,
there will be no real part of the impedance since Rr is zero. The
reactive values along an unterminated (or shorted) line change
between inductive and capacitive as we gradually change the
physical length of the line or the wavelength. Therefore we can
also use a length of transmission line as a capacitor, inductor,
parallel resonant circuit, or series resonant circuit.
An antenna typically radiates or receives RF power from
three-dimensional free space, but the transmission line
concentrates the power into a one-dimensional structure and
thus, in principle, does not permit the RF power flux to
attenuate due to dispersion (Figure 4.1)

143
Radiation Loss Rr
In the previous chapters we developed the concept of
radiation resistance as the term that defines the coupling of an
RF wave into space. If the goal of a transmission line is to not
radiate, then the Rr of the line should be 0 Ω. We are now
familiar with the characteristics of a 1⁄2 λ center fed dipole
antenna. The Rr is about 73 Ω when the wires on either side of
the feed point are aligned in opposite directions, or θ = 180° as
in Figure 4.2 and Table 4.1. As we rotate the wires from 180
degrees to smaller angles, Rr begins to drop. At 0 degrees, Rr
falls to nearly 0 Ω which is exactly what we want in a
transmission line: no radiation, because Rr = 0 W. This is a
simple derivation of how we can create a very effective radiator

144
from two conductors, and then by reconfiguring the same
conductors, we create a transmission line with no radiation.
(Of course all practical lines will have some small value of Rr
and thus exhibit some finite amount of radiation.) The
transmission line in this case is referred to as a “twin lead” or
“open wire line.”

Characteristic ​Impedance
Let us assume we have an infinite length two-wire line with
zero resistance, constant spacing between the wires, and the
wire diameters do not change for this very long line (Figure
4.3). The two wires each have an inductance/unit length, and
the two wires also form a capacitance/unit length. The line is

145
also assumed to be lossless, in that the resistance of the wire is
zero and the resistance of the dielectric is infinite. More often
the dielectric is specified as a conductance rather than a
resistance and also a dielectric constant (treated in detail in
Chapter 6) which is a function of the permittivity of the
material or free space.

Intuitively we can imagine that two parallel wires will have a

capacitance between them much the way two parallel
conducting plates will form a capacitor. They also are
inductors. Since the line is infinite, we designate the
capacitance and inductance in terms of C and L per unit length.
Therefore, the electrical equivalent of the line is shown in
Figure 4.4.

If we connect a dc voltage source to one end of the line, we

might expect to measure a displacement current (Maxwell’s
fourth equation), as one capacitor after another is charged
through the endless series of Ls and parallel Cs. Indeed we will
measure a constant dc current value. When we apply a source
voltage V and it results in a current I, we can conclude that

146
there is some corresponding impedance to this line.

(Equation 4.1)
From Equation 3.30 we derived the equation for free space:

(Equation 4.2)
This term not only determines the characteristics of a
radiated wave, but also of a guided wave. If the E and B fields
are constrained between two conductors separated by free
space, then intuition suggests that the impedance of the line
will be a function of 120pΩ and the characteristics of the two
conductors.
There is a rather non-intuitive equation from Equations 4.2
and 3.21

(Equation 4.3)
Equation 4.3 is derived directly from Maxwell’s equations and
an important result is that the L and C distribution along a
uniform line is always reciprocal.
Note: there are two definitions for the term Z0 — the
impedance of free space and the characteristic impedance of a
transmission line. From Equation 4.4 through the rest of
Chapter 4, Z0 will represent the characteristic impedance of a
transmission line.
For transmission lines, the form for the equation:

147
(Equation 4.4)
is particularly attractive since the inductance and capacitance
per length are both well defined by the geometry of the line.
We will use this equation to find Z0 for the most common
transmission lines: two-wire open line and coaxial cable.
In Figure 4.3, if we switch off the source voltage (to 0 V), the
capacitance in the line will discharge at the same rate,
resulting in the same dc value for Z0. Behavior is also the same
for a continuously varying wave such as an RF sine wave, thus
transmission line characteristics are valid for all types of
applied voltages. This proves the line’s reciprocity.

Characteristic Impedance of Two-Wire

Transmission Line
Again, through some rather complex mathematics, we can
derive the inductance/length and capacitance/length for two
uniform parallel wires with constant spacing. See Figure 4.5.
The resulting equations are:

148
(Equation 4.5)
where S is the spacing between the two wires and r is the
radius of the wires (we assume both wires to have the same
radius). And

(Equation 4.6)
Substituting these equations into Equation 4.4 we get

(Equation 4.7)
Notice that this equation for the impedance of an air core twin
line cable contains the impedance for free space,

and the appropriate coefficient

that accounts for the geometry of the twin line.

The length terms cancel and

149
(Equation 4.8)
where

(Equation 4.9)
and

(Equation 4.10)
or

(Equation 4.11)
where 377 Ω is the impedance of free space and 2.3 is the
conversion factor from ln to log10.
The log10 term is often the more familiar form. Sometimes
there is some confusion when using this equation. Rather than
using the radius of the wires, occasionally the diameter d is
used where

(Equation 4.12)
Therefore

150
(Equation 4.13)
These are the equations for the characteristic impedance of a
two-wire open line.

Characteristic Impedance of Coaxial Cable

Using a similar process to the open wire example, we can
calculate the capacitance and inductance per unit length of a
coaxial cable. See Figure 4.6.

The capacitance/length of a coaxial cable is

(Equation 4.14)
And the inductance/length is

(Equation 4.15)
Again we substitute these L and C terms into Equation 4.4 to
derive:

151
(Equation 4.16)
This can be simplified to the approximation

(Equation 4.17)
Thus we also have

(Equation 4.18)
εr represents the relative permittivity of the medium
separating the coaxial conductors, which is also the same as the
dielectric constant of the material where kε = εr / e0.
ke = 1 for free space. For any other dielectric, ke will be some
value more than 1 since er is always greater than e0. Typical
values of plastic insulators used in coaxial cable have ke values
near 2. For most practical purposes we can assume µ to be the
free space value of µ0, thus typically we only need be concerned
with ke. In a two-plate capacitor the capacitance is directly
proportional to ke. For example, if we have a two plate
capacitor of 100 pF with an air dielectric (ke = 1) and insert a
sheet of plastic with (ke = 2), then the capacitance is now 200
pF.
Also, for open wire lines that use a dielectric other than free
space, we use the same term

152
to adjust the equation for Z0.
The electromagnetic fields along a properly operating coaxial
line are constrained between the conductors by virtue of its
geometry. This is of great advantage in practice over open wire
line whose fields propagate well beyond the immediate vicinity
of the line. Only in the far field do the fields of open wire line
cancel. Open wire line requires installation some “stand-off”
distance from other objects to avoid distortion of the fields in
the line’s close proximity. Furthermore, by the reciprocity
theorem, coaxial cable also has much better rejection of
undesired RF sources, making it superior in spurious-free
receiver applications. Mainly for these reasons coaxial cable
has become the most widely used type of transmission line.

Velocity Factor
When εr increases by replacing the free space dielectric with
a material, usually a plastic dielectric insulator, the impedance
decreases. There is another effect upon the characteristics of
the line: the velocity of the wave along the line. We again
recall that:

(Equation 4.19)
We assume that µ will be very close to µ0 in free space,
therefore, the velocity along the line will be:

(Equation 4.20)
where c is the free space speed of light. The velocity factor of a

153
line is thus

(Equation 4.21)
For example, if we use a dielectric with a ke of 4, the velocity
factor will be 0.5, or 1⁄2 the free-space speed of light. A common
value for ke in coaxial cables is 2.3, and thus the result is a VF of
about 0.66.
The velocity factor of the line is related to the characteristic
impedance of the line only by

and not the geometry of the line per se.

Types of Guided Waves

The circumstances discussed thus far have dealt with only
TEM waves, or transverse electromagnetic waves. The TEM
wave is defined as having the E and H waves both
perpendicular to the line of propagation as presented in
Chapter 3. This is the case in free space waves (as discussed
earlier), but also in transmission lines where the separation
between the conductors is <<λ.
There are other types of guided waves, however, where one
or both of the fields are not necessarily perpendicular to the
direction of propagation. Thus we can define in general terms
the different modes of waves. In addition, strict conformance to
Maxwell’s equations (assuming free space plane waves) is not
always found in the near-field of radiated waves, where the
ratio and vectors of the B and E fields are often different from
the classical model. In fact, the near-field is generally defined
as the region in proximity to the antenna, where an observer is
near enough to “see” both the radiated electromagnetic wave
and some portion of the originating displacement currents.

154
 For open wire lines, we made the assumption that the
distance between the two wires <<λ. As the wires become more
separated, the “line” begins to radiate and gradually becomes
an antenna rather than a transmission line. Within a coaxial
cable, as the frequency increases the field lines inside the cable
begin to distort as the frequency is increased and the
separation of the conductors increases.

Transverse Modes
Transverse electric (TE) modes: the E field is perpendicular
to the direction of propagation, but at least a portion of the B
field component points in the direction of propagation
Transverse magnetic (TM) modes: the B field is
perpendicular to the direction of propagation, but at least a
portion of the E field component points in the direction of
propagation
Transverse electromagnetic (TEM) modes: neither the E or
B field lines are parallel to the direction of propagation (as
assumed in Maxwell’s equations).
Hybrid modes: non-zero electric and magnetic fields in the
direction of propagation.
When we consider a coaxial cable, as with open wire line the
assumption is that the separation between the conductors is
<<λ. We have also assumed (up to now) that the characteristic
impedance of the line is independent of frequency. This is not
quite so, especially as the separation becomes an appreciable
fraction of a wavelength. In such cases, the fields are
constrained by the boundary conditions within the waveguide
(or shield of a coaxial cable) resulting in a “bending” of the field
vectors (E, H, or both) within the dielectric or the space inside
the waveguide.
The implication is that coaxial cable for higher and higher
frequencies must become smaller and smaller to use the TEM
mode. This has three effects:
1) The attenuation of smaller cables is higher.
2) The attenuation for higher frequencies becomes higher in
a given coaxial cable.
3) Smaller cable reduces the power handling capability of the

155
line.
Therefore, as operating frequencies climb above 1 GHz or so,
waveguide becomes more and more attractive. Furthermore,
waveguide also becomes smaller (cross-sectional area) as the
frequency increases. At lower frequencies, waveguide becomes
impractically large for most applications and coaxial cable is
the usual choice.
In a coaxial cable operating in the TEM mode, the E and H
fields are perpendicular to each other and both are
perpendicular to the direction of propagation as shown in
Figure 4.7. Notice the H field “curl” around the direction of the
current.

Figure 4.8 shows a waveguide operating in the TE mode. The

E field (vertical) is perpendicular (transverse) to the direction
of propagation (to the left). However, the H field forms a loop
within the waveguide with a portion of the field parallel to the
direction of propagation.

156
Unlike waveguides, coaxial transmission lines do not have a

157
low frequency cut-off. As our first example of a battery and an
infinite line implies, a transmission line will exhibit its
characteristic impedance even at dc.

Losses in Transmission Lines

Thus far we have assumed that the conductors in our
examples are perfect conductors and the dielectric material
separating them is a perfect insulator, or a conductance of 0.
This, of course is never the case. The first issue is the effect
upon the characteristic impedance of the line when conductors
with some resistance and a somewhat lossy dielectric is used.
The simplest form of this equation is

(Equation 4.22)
where R is the resistance of the conductors and G is the
conductance of the dielectric. ω is the angular frequency where
ω = 2πf (f in Hz).

Reflected Waves
Many students find the concept of reflected waves and the
resulting effects of these waves to be somewhat non-intuitive.
This confusion has been magnified by the plethora of published
misinformation on the topic. For this reason, we will look at
these very important parameters from several different
approaches.
In Figure 4.4 we showed how the characteristic impedance of
an infinite transmission line will set the V/I ratio of an applied
voltage of any frequency. Also if we terminate a line of finite
length with a load resistor with the same resistive value of the
line (characteristic impedance) we will see the same current for
the applied voltage. However, if the line is terminated by a
resistor of a different impedance, some of the accelerating
charges will be reflected back similar to the description of the
reflection of a wave on a dipole antenna in Chapter 3.

158
 The result is identical to a free space wave striking an object
with a different characteristic impedance of free space. (We
introduce Snell’s Law in Chapter 6.) For a different
characteristic impedance there will be some reflection. This is
the principle of many forms of radio propagation and radar.
Radio waves “bounce” off of metallic objects such as aircraft,
disclosing their range (distance) and their direction. The
combination of these two parameters yields the aircraft’s point
location. Now imagine an aircraft coated with a very special
paint. This paint’s electrical characteristics show that it has a
resistance of exactly 377 Ω per square meter, the identical
impedance of free space. In effect the radar signal would be
completely absorbed, providing no reflection and rendering the
aircraft “stealth.” In effect, when the free space impedance of
377 Ω is “terminated” by an equivalent square resistance, there
is no reflection.
Returning to the guided wave, the same principle applies. An
impedance discontinuity results in reflection. Complete
reflection occurs at the end of a dipole since there is no
shunting impedance at its end. A similar condition applies for
two extremes at the end of a transmission line: a “short” and an
“open” termination where all power is reflected back to the
source. Any line terminated in an impedance showing some
value of resistance will absorb some of the power by that
resistance. If the load is a pure resistance, that resistance will
define the load resistance. In the case of a reactive portion of
the load impedance, the voltage and current will no longer be
in phase, and some reflection will take place. Use of the Smith
Chart and its applicability to determining reflection on
transmission lines has been published extensively, so I will not
duplicate that discussion. However, the Smith Chart is a very
convenient tool for such calculations….and more.
Generally, reflection on a radio transmission line is to be
minimized, since reflected power will eventually be wasted as
re-radiation and/or heat loss, both for transmitting and
receiving applications.

VSWR and Standing Waves

159
 If a transmission line is terminated with the proper load
value, the RMS RF voltage will be equal at all places along that
line. Thus the ratio of the highest to lowest voltage is 1:1,
indicating a perfect “match” between the load and the line.
However, if the load impedance is not the conjugate match for
the line impedance, a portion of the traveling charges will be
reflected. The backward and forward-moving waves (measured
in RMS) will sum differently along the line. At some points they
will reinforce and at other points they will tend to cancel. In
this case, voltage maximums and minimums will appear along
the line, indicating a non-perfect match, or a VSWR of >1:1. If
the voltage maximum is twice the voltage minimum, the VSWR
is 2:1. If the terminating impedance is lower than the line
impedance, there will be a current maximum and thus voltage
minimum at the load and vice-versa.
Equation 2.24 in Chapter 2 is the applicable formal definition
of VSWR. However, we can form an intuitive understanding of
how the reflected wave creates a “standing wave” and also how
VSWR relates to reflected power. For example, let us assume
that the VSWR is 2:1 as is the case when a 50 Ω line is
terminated into a 100 Ω load. If the forward wave’s RMS voltage
is 50 V (50 W), then in order to achieve a 2:1 VSWR there must
exist a reflected voltage that will add to and subtract from 50 V
to form a ratio of two, but will be symmetrical around the 50 V
level, or 66.666 and 33.3333 V. The reflected voltage to create
these maximums and minimum must then be 16.66666 V.
Therefore the reflected power is

(Equation 4.23)
Therefore, with a 2:1 VSWR, 11% of the power is reflected. For a
transmit system the reflected power will be dissipated as heat
in the equivalent transmitter’s resistance, as in Chapter 2 in
Figure 2.22’s Rr.
Now we can summarize the critical terms:

160
 1) A transmitter sends a 50 W incident wave of (50 V at 1 A)
down a 50 Ω line terminated by a 100 Ω antenna feed point
impedance.
2) A reflected wave of 16.6666 V RMS is sent back toward the
transmitter.
3) At points spaced 1⁄2 λ along the line the reflected 16.6666 V
adds to the forward 50 V resulting in 66.666 V at these points
(maximum).
4) At other points displaced 1⁄4 λ from the maximum V points,
the reflected 16.6666 V subtracts from the 50 V incident voltage
thus creating 33.3333 V at these points (minimum).
5) The VSWR is simply the ratio of the maximum to
minimum values, or 2:1. It is also the ratio of the Z0 of the line
to the Z of the load.
6) The power reflected back is as in Equation 4.23, or 5.56 W.
7) The power delivered to the antenna feed point is simply 50
– 5.56 = 44.44 W.
In a “real” line, one that has a finite amount of loss, the total
loss of the line increases as the VSWR increases. There are
several factors that determine this additional loss. First,
increased losses due to higher voltage and current at points
along the line will increase dielectric and ohmic losses at the
corresponding points. Second, since a portion of the power is
reflected back to the source, this reflected power will be
attenuated at the same rate as the forward wave. If the
transmitter side of the line reflects power it will again travel
back to the antenna, experiencing yet more loss. The ARRL
Antenna Book presents detailed graphs of increased line losses
as a function of VSWR for a variety of real lines.
For this reason, the VSWR is best measured at the feed point
of the load (antenna). For example, let’s assume that a 432 MHz
antenna’s feed line has shorted at the feed point. This would
indicate an infinite VSWR and the antenna would not radiate
or receive any power. Now assume a very long coaxial line
feeding the antenna with 10 dB loss in a matched condition.
The reflected wave would be attenuated by a minimum of 20 dB
(up and back). Therefore, if we drove the line with 100 W we
would see only 1 reflected watt at the source. We would

161
measure a VSWR of 1.22:1, indicating a very acceptable match,
but the antenna is shorted! If we measured the VSWR at the
shorted feed point, we would see the true condition, a near-
infinite VSWR.
The ratio of the reflected voltage to the forward voltage is
called the reflection coefficient where:

(Equation 4.24)
where Γ takes into account any complex components to the
load. The calculation of VSWR only requires the magnitude of
Γ, usually designated as “ρ.” Many texts only use the Γ
designation, however.
The formal equation for the above explanation of VSWR is

162
(Equation 4.25)
Let’s take two extreme cases and apply them to Equation
4.25: a perfect match where the reflected voltage Vr = 0 and an
“open” termination where all the voltage is reflected or Vr = 1.
(Vf) is taken to be 1. In the case where Vr = 0, by Equation 4.25,
VSWR = 1, or 1:1. However, if Vr = 1, VSWR → ∞, or ∞:1. Also in
the case of an infinite VSWR the maximum voltage along the
line is only twice the matched value, but the minimum voltage
is 0, therefore the ratio goes to 2/0 or infinity. See Figure 4.10.

163
 CHAPTER 5

Antennas Using Multiple

Sources
Thus far we have considered only pattern formations from a
single antenna element. The technique for determining
antenna patterns is described in Chapter 3 using small
segments (or infinitesimal dipoles). When we move into more
complex antennas (here exemplified by multielement arrays),
the utility of the PADL technique (Phase, Amplitude, Direction,
and Location — introduced in Chapter 3) becomes
indispensable. We may extend the PADL principle to include a
rather sweeping statement: It does not matter if the PADL
parameters are distributed along a single antenna element or
multiple elements distributed in 3D space. They form the basis
of the antenna pattern. It also does not matter if these
segments are excited directly via a conducted wave
(transmission line or wire) or through a radiated wave coupled
by a mutual impedance. These two mechanisms define the two
basic forms of multielement arrays: “driven arrays” and
“parasitic arrays.” Sometimes both techniques are used in a
single antenna (intentionally or unintentionally). However,
unlike the single element antennas thus far discussed, the
actual values of the phase, amplitude and direction for close-
spaced elements become interdependent among all the
elements.

Driven Arrays
We will begin by using two simple examples that can form a
more intuitive understanding of how PADL para​meters create
antenna patterns. Both examples use two parallel 1⁄2 λ dipoles
in free space and are separated by 1⁄2 λ (Figure 5.1). Both
dipoles are center fed from the same RF generator
(transmitter). Since both of the elements that constitute the

164
array are directly fed from the signal source, we call this a
driven array. In this first example the currents being fed to the
two dipoles are identical but we will use two phase
relationships: in phase (0 degrees) and out-of-phase (180
degrees).

The pair of dipoles in Figure 5.2 are fed in-phase. If we

observe the array from one of the two broadside directions, we
will see two fields adding in-phase. Consequently radiation is
maximum “broadside” to the elements, in the directions normal
to the plane of the dipoles. Conversely, the pair of dipoles in
Figure 5.3 are fed out-of-phase. Therefore, observing them
from a broadside direction will reveal two waves cancelling.
Now imagine the field from the left dipole in Figure 5.1
propagating toward the right dipole. By the time it propagates
to the location of the right dipole, the phase of the right dipole
has reversed, since the sine wave of the generator has changed
180 degrees. Thus the fields cancel (off the ends), or in-line
with the plane of the dipoles in Figure 5.2. However, in Figure
5.3 by the time the field travels to the other dipole, the field has
changed 180 degrees, but the fields are now in-phase in the
plane of the dipoles. This is referred to as an “end fire”
configuration.

165
166
 The result is that we can provide switchable bidirectional
“figure 8” patterns by simply changing the phase relationship
between the two elements, again as shown in Figures 5.2 and
5.3.
The gain parameters in Figure 5.2 may appear a bit
conflicting upon careful investigation. The maximum gain is
5.94 dBi, or 3.82 dBi. This is counter-intuitive if we assume a
doubling of a 1⁄2 λ dipole aperture, the gain should be 5.15 dBi,
or 3 dB higher than that of the 1⁄2 λ dipole. Thus we must
explain: where did the extra gain and thus aperture come from?
Simple addition of apertures assumes completely
independent apertures and the added assumption that there is
no “duplication” of aperture coverage, like the shading of one
solar panel over another. If, however, antenna elements are
placed in close proximity, they become “coupled” and all their
independent characteristics change, including phase,
amplitude, and direction of the charge acceleration. Since these
terms directly relate to aperture, then aperture can change as a
result of the terms changing in a non-independent manner.
The intuitive process above serves as a general explanation of
how directivity is formed, but it ignores the critical subtlety
hinted above by inter-element “coupling” in multielement
arrays.

Mutual Impedance
Now we will consider this more complicated coupling
problem. Although multielement arrays can constitute driven
elements spaced at distances well into their far fields, we will
first consider near-field interactions. The far field and very
large aperture antennas will be considered in Chapter 9. In the
case of far-field coupling (through free space) there is no
measurable effect from one antenna to the other, only the
induced currents and voltages the transmitting antenna
imparts to the receiving antenna through free space. In effect, a
single antenna suspended in free space is immersed in a
medium defined by the intrinsic impedance of free space
(120pΩ). When we move the two antennas closer, approaching
the near fields of both, significant changes result in the new

167
“array.” In effect we have formed a new antenna with a new set
of parameters. The two elements become coupled not by a
conductor, but by virtue of their respective apertures.
In our example in Figure 5.1, some of the power radiated by
one element will be received by the other element; they are
not mutually independent. Because of the received power, the
element now has a modified voltage and current value: it is the
sum of the initial fed power and the received power from the
other element. This is called “mutual coupling” between the
two elements. Since the affected voltages and currents have
been affected in both elements, there must be some sort of
impedance involved.
From Figure 5.1 we now define the two antennas as antenna 1
and antenna 2 (it doesn’t matter which one is which, as they
are equal for this analysis). We now define the self-impedance,
which is also the feed point impedance, and also the Rr of both
the antennas independently. Therefore the self-impedance of
the two antennas is Z11 and Z22 respectively.
However, antenna 1 will induce power into antenna 2 and
vice-versa. The current, voltages and their phase relationships
define a mutual impedance as Z12 and Z21 respectively.
We can now determine the feed point impedances for both
antenna elements where
V1 = I1Z11 + I2Z21

(Equation 5.1)
In other words, the term V1 = I1Z11 is the voltage on antenna 1
as if it were the only element involved, and I2Z21 accounts for
the additional voltage term introduced by the close proximity
of fed element 2.
Since the two elements are identical and fed in-phase, we can
assume I1 = I2. Also, for the same reason, we can assume that Z11
= Z22 and Z12 = Z21.
Therefore the terminal impedance of element 1 is
Z1 = V1/I1 = Z11 + Z12

168
(Equation 5.2)
Also, because of the inherent symmetry of this simple array, Z1
= Z2.
Thus for this simple case, using two identical elements with
identical feeds (either 0 or 180 degree phase difference and
equal amplitude), we derive a simple equation for the new feed
point impedance, and, in this case, also the new Rr. We know
that the self-impedance (Z11) of a 1⁄2 λ dipole is 73 Ω. After some
rather rigorous trigonometry we derive Z12 at –13 + j14 Ω. Thus
Z1 = 73 – 13 + j14 = 60 + j14 Ω

(Equation 5.3)
Since there are two elements in parallel, the real part of the
feed point is 30 Ω where the two 60 Ω feeds are connected in
parallel. Again, for this simple case we can directly calculate Rr
as 30 Ω. We now provide a simple calculation that verifies this
value. We have already provided a general equation for Rr. In
Chapter 7 we will derive a more convenient form for
calculating the Rr of a ground-mounted vertical antenna.

for a vertical antenna. The equation for a dipole in free space is

of the same form. Of particular importance to this discussion is
that Rr ∝ 1/Ae. We simply take the ratio of the values of Rr for a
1⁄2 λ dipole and our new array, and the reciprocal of the
numeric gain numbers. Therefore, in this simple case,

(Equation 5.4)
Indeed, 2.433 ≅ 3.96/1.64 = 2.419, or 3.836 dBd, or 5.99 dBi.

169
This is very close to the modeled 5.94 dBi shown in Figure 5.2.
This is yet another indication of the reciprocal relationship of
gain (aperture) and Rr.
From this discussion there are two points which should be
made:
1) Often when modeling in EZNEC or other programs, exact
correlations do not occur with theory. This is mainly due to not
providing exact dimensions when modeling the antenna.
However,

is close enough!
2) Z12 is taken to be –13 + j14 Ω. This is not an error. Negative
resistances are commonly encountered when modeling
multielement arrays. It simply implies that more current is
flowing toward the source rather than flowing toward the
antenna due to the mutual coupling. Therefore in this case the
sum of the real part of the feed points is 60 Ω.
The array in Figure 5.3 uses a similar method of calculation.
As arrays become more complex, we can no longer make the
assumption of symmetry as in the simple case of Figure 5.1’s
array. Therefore the computation becomes more complicated,
but the principle shown here is the same. A natural progression
is to the quadrature array.
Note: The calculation of multielement arrays is usually more
complex than this simple analysis which is presented to show
conceptually how Rr relates to the other critical terms.

Quadrature Array
In the quadrature array (Figure 5.4), we use the same two
dipoles, as above. However, the element spacing is reduced to
1⁄4 λ (90 degrees).

170
 The two dipoles retain center feed points, but now we will
define the phase relationship as 90 degrees rather than 0 or 180
degrees, with the fed amplitude (current) remaining identical
in both feeds. We can follow the same intuition as in the
examples using 1⁄2 λ spacing. In this case, element 2’s phase is
leading element 1’s phase by 90 degrees. By the time element 2’s
field reaches element 1, the phase of the current delivered to
element 1’s feed point has advanced by the same 90 degrees.
Thus, in the negative X direction (Figure 5.5) the fields add
completely and we derive maximum gain.

171
 However, where element 1’s field propagates toward element
2, the phase of element 2’s current has advanced another 90
degrees, for a total phase shift of 180 degrees, and there is
complete cancellation in the positive X direction. Between
these directions of maximum and minimum radiation, there is
partial addition or subtraction of the fields, resulting in the
overall pattern shown in Figure 5.5.
Unlike the 1⁄2 λ spacing examples in Figures 5.1 to 5.3, this
array is non-symmetrical, in that Z12 ≠ Z21. Since the self-
impedances of elements 1 and 2 are identical at 73 Ω pure
resistive we can avoid very difficult computations by allowing
EZNEC to perform the necessary calculus by simply reading the
resulting Z1 and Z2 values, which are 120.3 + j76.95 and 52.85 –
j9.92 Ω respectively. Therefore we can now directly compute
the two different mutual impedances.

172
Z12 = (120.3 + j76.95) – 73 = 47.3 + j76.95 Ω

and
Z21 = (52.85 + j9.92) – 73 = –20.15 – j9.92 Ω

where Z11 = Z22 = 73 Ω.

We can see that the gain of this array is almost exactly 3 dBd
or 2× the power gain over a dipole. The two real parts of the
feed point impedances are 120.3 and 52.85 Ω. We calculate
these two parallel resistances the same way we calculate the
value of two resistors in parallel:

As expected, Rr for the array is almost exactly 1⁄2 Rr for the

single dipole, according to Equation 5.4
Note: For the quadrature array we must use a slightly more
complicated technique for the calculation of Rr. Once again, as
arrays become more complex, the equivalent calculations
become also far more complicated.

Parasitic Arrays
The usual configuration for a parasitic array (Figure 5.6) is to
employ one driven element and rely completely upon the
mutual impedance(s) between or among the driven element
and the parasitic element(s). In the case of multiple fed and
multiple parasitic elements forming one antenna, a
combination of these two analytical techniques becomes
necessary. The parasitic element usually employs the scattering
aperture, where the termination (center of a 1⁄2 λ linear
element) is a short, or 0 Ω.

173
 The parasitic array analysis becomes quite different from the
driven array analysis. The first step is understanding that there
is a phase reversal from the driven to parasitic element. Since
like charges repel, then we can assume that accelerating like
charges will repel. Therefore the induced voltage and current
in element 2 will be 180 degrees out-of-phase from the inducing
field striking it. With this in mind, we can begin to quantify the
terms
V1 = I1Z11 + I2Z12

In other words the voltage on the driven element will be the

sum of the fed current/self-impedance (as if it were alone) plus
the current/mutual impedance from the parasitic element.
Therefore the voltage and current on the driven element is also
a function of what the parasitic element “received” from and
retransmitted back to the driven element! Since the parasitic
element current is 180 degrees out-of-phase with its induced
current,
I2Z22 + I1Z12 = 0

In other words, the current flowing on the parasitic element

174
due to its own impedance is the negative value of the current
flowing on the driven element that is induced into the parasitic
element, so they sum to zero. Now we can describe the
fundamental equation for the action of a parasitic array, and its
important practical configurations including the Yagi-Uda.

In other words, the current on the parasitic element and its

phase is the negative value of the current incident upon it. The
amplitude and phase of the parasitic element is determined by
the complex coefficient of the mutual impedance and the self-
impedance of the parasitic element. However, the
instantaneous phase value of the parasitic element’s current will
also be a function of the time (and thus phase difference) due
to the spacing of the driven and parasitic element. For
example, if the spacing is 1⁄4 λ there will be a 90 degree phase
lag between the two elements. Therefore, the phase on the
parasitic element will be –90° – 180°, or –270° or +90° relative
to the driven element. Then the complex coefficient further
modifies the phase difference. Finally through PADL, we can
derive the parasitic antenna’s pattern. Of critical importance in
determining the phase of the parasitic element is Z22. A very
slight change in the complex value (slightly inductive to
slightly capacitive reactance) can “flip” the phase response
dramatically.
This idea should be familiar to anyone with a good
understanding of ac electronics: current leads in capacitive
reactive circuits and lags in inductive reactive circuits. If we
extend that intuition to a current flowing up and down an
antenna element it is easy to see how the phase of the current
for the entire element can change with a change in the reactive
impedance of the element. This, or course, leads to a change in
“P” in PADL and therefore affects the directivity of the array.
Figure 5.7 shows the array’s pattern.

175
 To reiterate, the basic PADL principle applies to both driven
and parasitic arrays. Driven arrays are inherently more stable
since the characteristics are largely driven by the self-
impedances of the elements. Parasitic arrays, in contrast,
depend completely upon mutual impedances for their
excitation and very careful tuning of the self-impedances for
proper phase response. Finally, greater gain can be obtained
from parasitic arrays because of the much larger scattering
apertures inherent to parasitic elements. This is in contrast to
the quadrature driven array where the antenna apertures
simply add, thus providing only about 3 dBd gain. An optimized
two-element Yagi-Uda provides about 5 dBd.
This completes the first section of this book. Chapters 7 and 8
of the second (and last) section will build upon this and all the
previous chapters to apply these principles to understanding
specific array forms. However, the first “applied” chapter will
deal with the idiosyncrasies of antennas’ inter-reactions with
our own planet.

176
 Part 2
Applied Antenna Physics

177
 CHAPTER 6

Dielectric Effects Upon Radio

Waves
In Part 1 we defined the most important physical properties
of electromagnetic waves and how antennas radiate and
receive them. These explanations assumed that the antennas
were located in free space. We can assume the free space model
is all we need if we are indeed using the antenna in free space,
for example satellite-to-satellite communication (without the
Earth in the way!). Obviously almost all antennas are located
on or near the Earth’s surface. The presence of the Earth has a
profound effect upon the performance of an antenna located on
or near its surface. We can divide these effects into two basic
types: near field and far field effects.
In effect, all material (desired or undesired) within the near
field of an antenna must be considered part of the antenna
proper. The usual goal is to design and mount an antenna so as
to minimize the effect support structures, vegetation, buildings,
and other materials have on the antenna’s intended
performance. Sometimes the ground is intentionally made part
of the antenna, such as with ground-mounted verticals. The
effect of the ground proper has a very great impact on ground-
mounted vertical antennas, and this will be covered in detail in
the next chapter.
In the far field, the antenna’s radiation is no longer affected
by the adjacent Earth and other materials. Only the
characteristics of the antenna described in free space will be
affected. The far field effects are due to the propagation of
waves — not due to the Earth becoming part of the antenna. In
practice, there is a “grey” area between the far and near fields.
Therefore, in many instances both effects are present, which
leads to complications for modeling the antenna — not to
mention forming an intuitive understanding of what is actually

178
happening. Although there are numerous physical mechanisms
for wave propagation due to the Earth, we will cover the two
most influential on forming the final gain pattern.

Reflection and Absorption

The bad news (for antenna calculations) is that the Earth’s
surface is composed of a very wide variety of materials both
natural and man-made, thus complicating predictive modeling
efforts. The distribution of these varied materials can be quite
uniform over a large area, such as sea water. It can also be very
non-uniform, such as when a line of exposed rock protrudes
from deep soil. In addition to surface discontinuities, the solid
ground is most often stratified — such as top soil over sub soil
over bedrock. Radio waves will penetrate ground, and thus
these lower stratified layers will affect both reflection and
absorption. Furthermore, most propagation modeling tools also
assume that the Earth is an “infinite” smooth plane, which, of
course, it is not. The real topography of the Earth will distort
smooth plane assumptions. The depth of ground penetration is
a function of frequency and the characteristics of the dielectric
ground. In general, lower frequencies will penetrate the ground
to greater depths. Therefore very lossy bedrock under a “good”
soil may become critical at medium or low frequencies, while at
HF the bedrock may have no effect whatsoever.
Propagation effects can even take place in the very far field.
For example, a multi-hop path to the other side of the world
will use ionospheric propagation, but it will also use reflections
from the Earth’s surface. If one of those reflection areas is on
sea water, the condition of the sea can determine if the path is
“open” or not (Figure 6.1). If the waves become an appreciable
fraction of a wavelength in height, the sea will represent a
dispersive reflection surface and scatter much more power than
a presumed smooth surface. Mountainous areas are known to
be problematic for the same reason, whereas flat land is a
better reflector. Although the mechanism is reflection, the
effect is attenuation due to dispersion.

179
 The good news is that these models often do provide fair to
excellent approximations of what a system may provide. They
are certainly better than using a free space model!
It is also fortunate that the actual geophysical terms relevant
to radio wave interaction can usually be reduced to two terms:
conductivity and permittivity. Therefore sufficiently accurate
models can be derived by reducing the Earth to these two
parameters. Therefore the subject of electromagnetic
geophysics (at least for radio waves and thus antenna
performance) is really the subject of how those waves interact
at the boundary between free space and the lossy dielectric of
the Earth as well as down to some depth.

Conductors and Dielectrics

We recall from Maxwell’s 4th equation

180
(Equation 6.1)
The term m0I is the conduction current. However, if there is
a resistance in the current path (1/G), we can rewrite this term
as GE. Also the term

represents the time-dependent displacement current (current

causing time-dependent charging and discharging of a
capacitor). If we represent this term by a sinusoidal E field
variation, it becomes (in complex notation) jωkdE. Thus we
have

(Equation 6.2)
In this chapter we will assume the µ term in Equation 6.1 will
be m0 because we will only consider materials that have µ
values identical to free space. Under most circumstances, µ of
the Earth, from the surface down to about 10 meters, can be
assumed to be the same as free space or very close to it.
Therefore we can now write a ratio of the two terms as

(Equation 6.3)
where DF is the dielectric dissipation factor, G is the
conductivity (frequency independent), and ωkd is the frequency
dependent dielectric constant of the material. Conductors will
have very high conductivity, thus DF is >>1. Good dielectrics
(insulators) will have very poor conductivity, thus DF becomes
<<1. Some readers may correctly see a correlation with the Q of
a capacitor. Indeed,

181
(Equation 6.4)
Therefore we can determine the theoretical maximum Q of a
capacitor if we know the DF of the dielectric. A “perfect”
dielectric will exhibit a kd that is constant for all frequencies.
There are no materials that have this property. Only free space
has such a frequency-independent characteristic. The DF of soil
and rock is very complex and variable with temperature and
moisture level.

Wave Propagation in Dielectrics

Let us assume first, a perfect dielectric with a dielectric
constant kd. The speed of light, and therefore of a radio wave
will be

(Equation 6.5)
where ed is the permittivity of the dielectric. We can simplify
the relationship by multiplying the free space speed of light (c)
by a simple coefficient to derive the speed of light inside a
dielectric material

(Equation 6.6)
This is the most important effect upon an electromagnetic
wave inside a perfect dielectric (zero conductivity). The
frequency remains constant, but the speed slows and therefore
the wavelength decreases.

182
Snell’s Law
Let us assume that a free space plane wave strikes a dielectric
surface with a kd = 4. The angle of incidence is qi. The reflected
wave will be the “mirror image” of the incident wave. However,
the transmitted wave, propagating into the dielectric, will be
traveling at 1⁄2 c by Equation 6.6. Figure 6.2 shows the effect
upon the angle qi through the dielectric. Since the wave
velocity is now 1⁄2 c, the wavelength inside the dielectric is now
half the wavelength in free space. This has the effect of
changing the angle of the transmitted wave. If you imagine the
incident wave striking the boundary in slow motion, and the
wave slows down inside the dielectric, it is easy to envision the
cause of this transformation of θ where:

(Equation 6.7)
In other words, the ratio of the sin values of the angles is
equal to the inverse-square ratios of the dielectric constants of
the two medium. In this case the two medium are free space

183
and a material with dielectric constant of 4.
Notice that when the incident wave is striking the boundary
from a perpendicular direction, in this case, from straight up qi
approaches 0 degrees, but qi also approaches 0 degrees. How
can this be, when

This is called a limit in mathematics. In this case, where the sin

values become infinitesimally small, the factor of 2 becomes
less and less significant. So, for practical purposes the angles
become equal.
The reflected wave always will “mirror” the incident wave,
while the transmitted wave’s direction will change at the
boundary. Again, if the dielectric material is “perfect,” i.e. zero
conductivity, there is no dissipation of power. The incident
wave’s power is simply divided into the transmitted wave and
the reflected wave, where
Pi = Pt + Pr

(Equation 6.8)
The ratio of Pt and Pr is a function of the ratio of the dielectric
constants and the values of θ.
Inside an imperfect dielectric (such as soil), the situation
becomes more complicated. A finite value of conductance
implies some finite value of resistance, which in turn will cause
attenuation of a wave inside the material. This will be true for
directed as well as radiated waves.

The Earth as a Dielectric

Any material comprising the “ground” will have a higher
value of ε and some non-zero value for conductivity. The
boundary between these two dielectrics forms a discontinuity
that has profound effects upon electromagnetic fields.
Again, the wave equation for free space is:

184
(Equation 6.9)
In a dielectric, ε0 has changed to εd, where εd is always > ε0.
Inside a non-free space dielectric, a wave is still a wave, but
with two profound modifications: the speed of light has slowed
down and the intrinsic impedance has decreased where

(Equation 6.10)
and

(Equation 6.11)
Also from Chapter 4

(Equation 6.12)
and

(Equation 6.13)
where V is velocity.

Effect on Wave Polarization

We have thus far only considered “a wave” incident upon the

185
dielectric. Vertical and horizontal waves interact with the
dielectric differently. This is particularly true of the
relationship between the phase of the incident and reflected
waves.

Horizontal Polarized Wave Reflection

When a wave’s E field is parallel to the surface of the
reflecting Earth dielectric, the wave is horizontal (Figure 6.3)
According to Snell’s Law, the reflected wave angle will mirror
the incident wave. However, some of the power will be
transmitted into the dielectric and thus lost as heat in the
dielectric’s resistance. Figure 6.4 shows how the direct wave
and the reflected wave will differ. Figure 6.5A shows the rela​-
tive magnitude of the reflected wave and Figure 6.5B shows the
relative phase of the reflected wave. For horizontal waves, the
phase reversal is nearly 180 degrees for most incident angles.
Figure 6.6 shows a simulation of a 1⁄2 λ horizontal dipole at 2 λ
over real ground.

186
187
Vertical Polarized Wave Reflection
A vertical polarized wave’s E-field is perpendicular to the
Earth’s dielectric surface, or “vertical.” But upon careful review
there can be no “true” vertical polarization that is being
reflected. Figure 6.7 shows why this is true. The reflection
angle resulting in a 90 degree phase shift corresponds to the
minimum magnitude of the reflected wave. This is defined as
the Brewster or pseudo-Brewster angle. Figure 6.8 shows the
relative magnitude and phase of the reflected wave for vertical
polarization.

188
 In the case of a 1⁄4 λ ground-mounted vertical, nulling due to
reflections is restricted to very low (grazing) angles. The point
source of radiation is simply too low to provide ground-
reflected waves at higher angles, so wave attenuation is
primarily due to ground attenuation and the requirement for
the values of the E-fields to remain constant along the
boundary of the dielectric and free space. Therefore, we do not
see ground reflections causing nulls and reinforcement in the
resulting pattern in Figure 6.9. Figure 6.10 illustrates vertical
polarized wave reflection.

189
 Unlike a ground-mounted vertical, with the elevated 1⁄2 λ
vertical dipole, the center of radiation is located at a sufficient
height to provide very significant reflected waves, even at high
angles. Figure 6.11 shows a simulation of a 1⁄2 λ vertical dipole
at 2 λ over real ground.

Assumption of an Infinite Plane

Antenna modeling software usually assumes that the antenna
is above an infinite plane representing the Earth. Of course the
Earth is not a plane — it’s a sphere (almost). For many practical
solutions, the infinite plane assumption is acceptable, but for
some modeling problems it is not.
Even at hundreds of wavelengths away from the antenna, the
maximum gain does not approach the 0 degree angle, since the
Earth’s curvature will not permit the convergence (Figure
6.12).

190
 Extrapolating this curve and using MF (medium frequency)
wavelengths, we can see the Earth’s curvature will serve as a
fundamental limit to how low the maximum gain elevation
angle can be (Figure 6.13). Figure 6.9 implies even more low-
angle attenuation over “average ground”. We will explore how
these terms relate to vertical antennas in the next chapter.

191
192
 CHAPTER 7

Vertical Antennas
In Chapter 6 we discussed the physics at a boundary between
free space and a dielectric for radio waves. We also discussed
the differences between horizontal and vertical polarized
waves at these boundaries. In this chapter we will focus upon
the characteristics of vertical antennas, return to complete the
discussion of their relationships to the Earth, and then
conclude with comparisons among various vertically polarized
configurations.
Vertical antennas may be placed above the surface of the
Earth or mounted on the Earth proper. Ground-mounted
verticals actually use the Earth as part of the antenna, or as the
“image” of the antenna. Verticals that are mounted above the
Earth may also include the Earth as part of the antenna,
depending upon how far above the Earth the “lowest” part of
the antenna is placed. For example, a 1⁄2 λ vertical dipole with
one end just above the Earth actually may be considered a
ground-mounted vertical, although a direct connection to the
Earth will significantly change the antenna’s characteristics.
An antenna connected directly to the ground is a “driven
ground” array. An antenna located near the ground, but not
connected to the ground proper, uses the ground (for better or
worse) as a parasitic surface, but with very complex properties.
As will be shown, the gain response of driven and parasitic
ground antennas mirrors the difference in gain between driven
and parasitic arrays, as in Chapter 5.

Ground-Mounted Vertical Antennas

Since the ground proper is actually part of a ground-mounted
vertical antenna, its relationship to the ground is of particular
importance. Of even greater concern is when compromised
ground systems are employed. In Amateur Radio publications,
there has been intense interest in how to optimize ground

193
systems for better overall antenna efficiency. For professional
AM broadcasters, very stringent ground requirements are
mandated so that minimum ground losses can be assumed. In
effect, the reality of compromised ground systems and
complete freedom of antenna design has resulted in much
more research in amateur than professional circles. In AM
broadcasting, antenna parameters have been established for
many decades and are highly accurate in practice. In AM
broadcast engineering, lossless ground is assumed and only
ground-coupled vertical elements are mandated. That makes
life much easier!
The general equation for radiation resistance, Rr, from
Chapter 3 is in an awkward form for easy use with vertical
antennas. Special attention must be given to Rr in ground-
mounted verticals, especially for amateur applications where
compromised ground systems are often used and/or the vertical
element is shorter than 1⁄4 λ.

Radiation Resistance of a Ground-Mounted

Base-Fed Vertical Antenna
The height (in wavelengths) of the vertical section can be
used to provide a good approximation of the radiation
resistance, Rr, and the idealized pattern of the antenna. Since
ground-mounted verticals radiate primarily “above ground,” we
characterize them over a hemisphere, not a sphere as we
previously assumed for free space antennas. We will begin to
quantify the performance characteristics of a ground-mounted
vertical by deriving its Rr. In Chapter 3 we defined the
following relationship:

(Equation 7.1)
There is an alternative and more useful method to determine
the equivalent of the power term in the general equations for

194
Rr in ground-mounted vertical antennas.

Effective Height
A critical term in our to-be-derived equation for Rr is he
which is the effective height of the antenna. he is another
potentially confusing term in that it is a function of the physical
height of a vertical antenna (including the ground image), but
should never be equated to it. It should also not be confused
with “electrical height.” Only in very specific circumstances
does he approach the actual physical height, hp. he is always less
than hp. The two terms can be equated using only the physical
height (without the image) in some specific case. The general
definition of he is

(Equation 7.2)
or
V = heE

(Equation 7.3)
The concept is quite simple. The voltage induced into an
antenna (of the same polarization) is proportional to the
incident E field and the “effective height.” The computation of
he is a bit non-intuitive for this simple definition. The
relationship between he and the physical height hp is simply
multiplying the average current along the antenna by hp and
then dividing by the maximum current along the same
antenna. For a linear vertical antenna whose he is greater than
about 1⁄8 λ, the current distribution will be sinusoidal.
Therefore, we need to calculate the average of a sine wave (or a
portion thereof) by integrating current over the physical
height. This integral simply computes the average value of a

195
portion of a sine wave. The portion of the sine wave to be
averaged is defined by two phase points along the sine wave.

(Equation 7.4)
In this equation we use two phase points α and β rather than
an arbitrary physical height. This is because the average
current can be different over the same physical length
depending upon which portion of the sine wave is present
along the length. We will show an example a bit later.
For computations of he it is convenient to take the maximum
current (I0) to be 1 A and the λ to be 1 meter to simplify the
equations. We then represent the physical height hp as the
appropriate fraction of λ, which we assume is 1. Therefore, the
term Iav/I0 is a dimensionless number which is less than 1. he
and Rr are independent of the power fed to the antenna, and
thus also independent of the actual current value or the actual
wavelength. We are only interested in the average current
referenced to the maximum current. This ratio is independent
of the amount of power applied to the antenna. Also he and Rr
are independent of λ as long as we scale the dimensions of the
antenna and thus the two phase values. For example, a 1⁄4 λ
vertical will show the same results no matter what the
wavelength, as long as the actual height is 1⁄4 λ at the actual
wavelength involved.
Using radian measure and integrating radian values for the
average current also permits us to use radian values as he For
example, a 1⁄4 λ vertical would be π/2 radians high.
A ground-mounted vertical antenna’s image is critical for
grasping a conceptual understanding of ground-mounted
verticals as well as setting up equations that define their
parameters. The image is considered to be an identical vertical
antenna that extends vertically underground, mirroring the

196
actual vertical element above ground. In the calculation of he
the image’s hp must be added to the hp of the vertical element,
or he of a ground-mounted vertical is twice the value of the
computed value for he of the vertical element taken alone.
There are some exceptions to this rule. In the case of
verticals higher than 1⁄4 λ, the image may actually be shorter
than the physical length. Also, for very lossy dielectrics (very
low conductivity and dielectric constant) that approach the
characteristics of free space, the image disappears and
significant radiation occurs into the dielectric. This radiation
must be taken into account when calculating Rr. The
disappearing image is due to the gradual decrease in ground
conductivity, which lowers the average ground current —
eventually rendering the image insignificant. Rl only accounts
for heat loss of RF current inside the dielectric, not radiated
power propagating inside the ground dielectric.
The power delivered to the load that is matched to a receive
antenna is:

(Equation 7.5)
From Chapter 2 we calculated that half of the power received
by an antenna is reradiated back into space and that only Ae is
significant for the transmit application. Now from Equations
7.3 and 7.5 we derive:

(Equation 7.6)
And from Equation 2.16

197
(Equation 7.7)
where

So,

(Equation 7.8)
Therefore

(Equation 7.9)
To confirm that this equation results in a resistance, we can
look at the dimensions within the equation, ignoring the
coefficients. Recalling Equation 7.4

therefore

(Equation 7.10)

198
 The m2 and I2 terms cancel, leaving R. Therefore, this special
form for the general equation for Rr provides a pure resistance.
It is critical to understand that in a ground-mounted vertical,
the physical height (hp) also includes the image. Notice also that
the current squared term in the denominator is the maximum
current term, identical to the other two equations thus
presented for Rr. All three equations are equal or very close to
it.

Example of the Effects of he on Rr

In this example we use two identical 1⁄4 λ ground-mounted
vertical antennas, except one uses top loading to place the
current maximum halfway up the vertical element. The other is
a simple vertical with no top loading, and thus the current
maximum appears at the base. The physical height hp is the
same in both cases, 1⁄4 λ.
Figure 7.1 shows two possible current distributions along a 1⁄4
λ vertical. The upper shows the current at zero at the top of the
vertical while the lower shows the current at the top to be
0.707 compared to the maximum value, at 1⁄8 λ height, which is
halfway up the vertical element. The usual technique involves
top loading to move the current maximum “up” the vertical
element. The average value of the top configuration is about
0.64, while the average current in the lower figure is about
0.92. Since the ground is actually providing the other 1⁄4 λ
section of the vertical, the total hp of the antenna is 0.5 λ.
Therefore, he for the unloaded vertical is 0.32 λ, and it is 0.46 λ
for the optimized loaded vertical. On the other hand, raising
the current maximum also increases the gain, in this case very
slightly, and thus the equivalent Ae.

199
 We can now use Equation 7.9 to actually calculate the values
of Rr for these two cases.

Simple 1⁄4 λ vertical: 36 Ω

Top loaded 1⁄4 λ vertical: 74 Ω

It is apparent that careful placement of the current

distribution along a vertical’s element can provide appreciable
efficiency gain for a fixed value of hp. After comparing the two
1⁄4 λ examples, it is also apparent that maximum Rr — and thus
maximum efficiency — will occur when the current maximum
is halfway up the vertical element, assuming that the ground

200
loss is the same.

Directivity of a Ground-Mounted Vertical

Antenna
From Equation 7.9, the other critical term for calculating Rr is
Ae. In Chapter 3 we showed the directivity of a 1⁄2 λ dipole in
free space. The 1⁄4 λ vertical is often described as a dipole, with
the ground (the image) representing the other half of the
antenna. The ground also divides free space into two
hemispheres with the radiation “forced” into the upper
hemisphere. Therefore, we might expect that the gain will be
higher from a ground-mounted vertical than a free space dipole
since the power in the lower hemisphere (in the free space
dipole) is not “added” to the upper hemisphere. Intuition
suggests that the gain should be doubled. The maximum gain of
the dipole is 2.15 dBi, while the maximum gain from the 1⁄4 λ
vertical is 5.15 dBi, or exactly 3 dB higher, or twice the power
gain of the dipole. Also, the directivity of the vertical is simply
the same pattern as the dipole, except that the dipole’s pattern
is cut into two (Figure 7.2). The same comparison can be made
between an extended double Zepp and a 5⁄8 λ vertical, as in
Figure 7.3. Of course, this assumption is for the idealized case
of an infinite perfectly conducting ground. For conditions of
“average” or “real” ground, the mechanisms were shown in
Chapter 6.

201
202
 Unfortunately, these very encouraging graphs become
somewhat disappointing when real ground is substituted for
perfect ground, as we shall see.

Ground Influences on Ground-Mounted

Verticals
In professional applications, such as AM radio broadcasting,
extensive ground radials systems are mandated by regulation.
The use of 120 buried 1⁄4 λ radials can usually be assumed to
reduce antenna ground losses (Rl) to a very small value, say 1 or
2 Ω (in calculating power they are assumed to be 0). In such
professional applications, Rl is taken to be zero and thus

203
(Equation 7.11)
where Pr is the licensed power output. Rrp is used since the
actual feed point may be a bit below the actual current
maximum. Often the broadcast tower is specified as physically
1⁄4 λ high (hp), which is about 5% higher than a resonant
vertical. If no losses can be assumed, however, the radiated
power can be calculated using the Rrp at any point and feeding
the appropriate current.
In contrast, for the average amateur installation, both hp and
the ground system will be limited by cost and/or the
availability of suitable space. Extensive empirical testing has
provided insightful approximations, especially when solidly
based upon physical principles. The usual intent of all this
work centers on the improvement of antenna efficiency.
Directivity can usually be assumed by standard models as there
is generally insignificant modification of directivity due to
antenna ground losses in the near field. However, gain
reduction directly results from these antenna ground losses. In
contrast, ground losses in the far field can have a dramatic
effect upon the final directivity and thus gain of a vertical.
These losses are a propagation effect in that they are not part of
the antenna proper. Therefore the directivity of the system is
determined in the far field, and thus little if anything can be
done to affect it, except move the antenna to a better location.
Thus far we have derived the specific equation for Rr of a
ground-mounted vertical antenna. This value is of primary
importance for understanding and designing vertical antennas.
Since such an antenna uses the Earth as part of the antenna,
the characteristics of the Earth in the vicinity of the vertical
have a profound effect upon the antenna’s performance. Of
particular interest is the ground’s contribution to the efficiency
of the vertical. Rr is usually taken to be an independent term in
the calculation of antenna parameters. The general equation
for determining the efficiency of a vertical antenna is

204
(Equation 7.12)
and

(Equation 7.13)
In Equation 7.12, Rl is the loss resistance of both the vertical
element and losses due to ground resistance (Figure 7.4). Some
authors also include Rl in matching losses. Although these
matching losses are not part of the antenna per se, they do
become important for system-level efficiency calculations.

This efficiency equation is deceptively simple. As we have

shown, the Rr term requires some rigorous computations for an
accurate derivation. The Rl term is exceedingly difficult to
predict in compromised ground systems. However, there are
only two terms — Rr and Rl. Therefore, at a first approximation,
we can increase efficiency by maximizing Rr or minimizing Rl
or a combination of both. Other terms come into play in the
overall efficiency, but this basic equation is of prime
importance.
As will be shown later, it is also critical to note that Rl
represents ground losses and not a fixed resistance. Somewhat
paradoxically, this “distributed” ground resistance actually

205
converges upon an equivalent lumped resistance shown as part
of the feed point impedance. This equivalent resistance must be
normalized to Rr to satisfy Equation 7.12. Also the resistance is
distributed over and under the ground. Simply stated, resistive
losses depend upon the current applied through them. With a
distributed resistance, the current distribution over that
distributed resistance will determine the conducted dielectric
losses. Therefore the solution demands a partial differential
that will be developed later. Fortunately, however, all
contributions to the distributed ground loss can be simplified
to a single resistor and applied to Equation 7.12.
The Rl of the vertical element proper can usually be ignored
except when the hp of the element is much less than 1⁄4 λ. In
such cases, Rr can fall to less than 1 Ω or so, and thus Rl can
become the dominant part of the power dissipation, and thus
efficiency will fall to very low values. Ground losses are a
concern in all cases, but it is important to consider both Rl and
Rr in the all-important efficiency equation. We will treat the
special case of short vertical antennas in later sections.
The directivity of the vertical will be mostly influenced by
the vertical’s height and the current distribution along that
physical height, as with any antenna. Lowering ground losses
increases the efficiency, and thus increases the gain of the
vertical. Lowering ground losses in the near-field ground plane
has an insignificant effect upon the directivity of the antenna,
but does increase the efficiency and thus the gain by Equation
7.13.
The ground-mounted vertical can be broken into three
fundamental basic components that directly affect the
directivity, efficiency, and gain of the overall system:
1) The vertical radiator
2) The near-field ground
3) The far-field ground
The height of the vertical radiator and the distribution of
current along that height largely determine the maximum gain
(assuming no ground losses), he, and thus also the Rr of the

206
vertical. The near-field ground plane is the main contributor to
Rl and thus, along with Rr, determines the antenna efficiency by
the simple Equation 7.13. The limited conductivity of the far-
field ground will attenuate the power in the far field
particularly at low elevation angles and thus further reduces
the system efficiency. The usual method of reducing near-field
losses is the use of ground radials in an attempt to provide a
low resistance return path for the antenna ground current.
The electrical characteristics of the soil, rock, water, organic
material, salts, and so on, have a profound effect upon the
operation of a ground-mounted vertical antenna. Since the
ground is part of such an antenna, it contributes directly to the
efficiency, directivity, and gain of the antenna.
There is also a secondary effect that becomes somewhat
ambiguous to the performance of the antenna proper. In the far
field, ground attenuation is usually significant and thus shapes
the final transmit (and receive) directivity of the antenna
(Figure 7.5). Because this is in the far field, we usually consider
these effects to be propagation effects rather than effects due to
the antenna proper.

207
 The attenuation mechanism of “real” ground can be
simplified into the one single term Rl assuming that the vertical
section’s resistance << Rl. If Rl could be determined easily, then
the basic characteristics of the antenna would be easily
modeled. Unfortunately, determining accurate ground
characteristics and how they actually combine with a radial
system to create Rl depends upon multiple variables that are
difficult to quantify.
Quantitatively, the difficulty of the problem can be
summarized by a differential equation. We will use this
equation as a step to formulating a matrix which will more
accurately quantify the interrelationships among the terms.
Each of these terms is difficult to quantify:

208
(Equation 7.14)
If these terms were mutually independent, we would have
this simple equation to define Rl. Unfortunately, these terms
are mutually dependent, thus Equation 7.14 can only be used as
a conceptual step.

This term defines the soil conductivity’s contribution to the

overall Rl and represents the most important term. If the “soil”
is a perfect infinite plane conductor, there would be no need
for radials, and radiation resistance would not matter since Rl is
zero.

This term defines the reduction in Rl as a function of the

number of ground radials used. Rl will decrease as the number
of radials increases. However, there is yet another complexity
in that the number of radials often forms a trade-off with the
length of the radials.

This term defines the reduction in Rl as a function of the length

of the radials with the above caveat. As we will imply later,
many short radials are often more advantageous to fewer long
radials.

This term defines the effect the feed point impedance (Zfeed) has

209
upon the overall Rl. Rl will not vary by changing Zfeed. However,
feeding a radial system with a higher voltage will result in
lower power losses, especially losses close-in to the vertical
element. An example is the use of a 1⁄2 λ vertical which
significantly reduces ground losses when compared to a 1⁄4 λ
vertical over the same ground system. Power loss is Ploss = I2R
and current will be lower for a higher feed point impedance.
Equation 7.14 is provided as a first approximation to the
actual quantification of the problem. The solution for Rl is
actually more complex than solving a simple differential
equation. All the variables listed above are mutually dependent
upon each other. When each term is partially dependent upon
all the other terms, that is, they are all mutually
interdependent, we can use a matrix to quantify this particular
type of problem. For example, if we want to know the
interrelationship between the number of radials and the length
of the radials as it affects Rl, we look at the terms

Now we will redefine Equation 7.14 into a more accurate and

formal relationship. If we define Rl as a scalar dependent upon
a system of mutually interdependent partial differentials, we
can set up a matrix that, in principle, will capture the most
important variables that define a vertical’s Rl.
Matricies are very useful tools in an area of mathematics
called linear algebra. A matrix can be used to simplify
simultaneous equations. Let’s say we want to solve to R, and R
is a function of two other variables (X and Y), and the system of
equations is:
3R + 2X – Y = 1
2R – 2X + 4Y = –2
–R + 1/2X – Y = 0

210
 Notice that all values of the variables are mutually dependent
upon each other’s values. The solution is R = 1 and X = –2 and Y
= –2. “1” could easily be an equivalent resistive loss. I will not
present the method of solution. When a system of equations
becomes very complicated, suffice it to say that use of a matrix
becomes a very convenient “shortcut” to a solution. If we use
partial differentials as the elements, the form is as follows:

(Equation 7.15)
Here we use X to identify the matrix. We use four variables
which we defined above as x1-4 where
x1 is the ground conductivity (g)
x2 is the number of radials (radial#)
x3 is the length of the radials (radiall)
x4 is the base feed point impedance (Rfeed)
The 16 partial differential terms that comprise this matrix
represent all the important interrelationships that define losses
associated with a ground-mounted, base-fed vertical. For
example, in free space a ground plane antenna will have
infinite “ground” resistance, but

is zero since there is no effect upon the radial’s contribution by

the ground, so Rl approaches zero. As the same ground plane
antenna is lowered onto a ground dielectric,

211
becomes significant and Rl becomes higher. Finally, when the
radials come in contact with the dielectric, Rl is maximum for
the given antenna. Ground plane antennas whose radial
systems are very close to the Earth(<< 1⁄4 λ) have been a subject
of intense study in the amateur community, mostly to
empirically determine these differential terms.
Thus far there are two key variables that have not been
treated in the ground-mounted vertical discussion: the
frequency of operation and the height of the vertical above
ground. The main difference realized when the frequency is
changed is the characteristics of the ground and the ground’s
interaction with the other variables. Therefore, the ground
coefficients of the matrix will differ for different frequencies.
The only situation where the outcome is independent of the
ground characteristics is when the ground is assumed to be an
infinite perfect-conducting plane. The height of the verticals in
the above discussion have been zero, since they are all ground-
mounted. A following section will deal with “raised” vertical
antennas, or ground plane antennas.

Experimental Method to Derive Rl

Equation 7.15 is formidable. It is offered as an approach to
quantify the variables and interdependencies of the problem.
Empirical tests remain the best method to approach some
maximum efficiency for a given set of constraints. The
extensive empirical work that has been done has reflected one
or more of the differential terms expressed in Equation 7.15.
The “holy grail” would involve combining the results of this
work into a unifying equation like 7.15.
Having quantified the overall problem, let’s return to the
other extreme at problem-solving: simple empirical
measurements. There is a non-ambiguous method to determine
Rl for a given location for a vertical 1⁄4 λ or shorter. Install a 1⁄4 λ

212
resonant vertical over the ground plane. The base feed point
impedance should show as little reactive value as possible (a
pure resistance, but some small value is acceptable). We are
only interested in the real portion of the impedance. The value
for Rr will be very close to 36 Ω. Therefore the efficiency will
simply be
Rl = Rfeed – 36 Ω

(Equation 7.16)
As you add radials, change their lengths, and so on, the feed
point should change accordingly. There are many empirical
studies in print. These can form the basis of an approximation
for a general case. However, the multiple differentials involved
will be different for any given location. There is no substitute
for empirical testing at a given location. As you experiment
simply use Equations 7.16 and 7.13 to maximize efficiency. If you
attempt to use a vertical shorter than 1⁄4 λ, then you must
calculate the Rr and adjust Equation 7.16 accordingly. For
verticals higher than 1⁄4 λ, Equation 7.16 will not provide an
accurate answer but can provide relative data.
The results of this exercise will provide a very good
approximation for efficiency of a 1⁄4 λ vertical at a given
location. It will also provide a first approximation for efficiency
of verticals of different heights and loading. Equation 7.15 is a
method for quantifying all the important variables involved.
There is also a possibility of another term that implies Rr may
itself be influenced by Rl. This is certainly the case for very low
conductive dielectrics. For example, if the “ground” under a
vertical/radial system is gradually made more lossy, it will
begin to approach the characteristics of free space. Thus the
“ground-mounted” vertical has morphed into a ground plane
with significant radiation into the ground. This will clearly
affect both he and Ae and thus modify Rr (see Appendix B).

Feed point R vs Maximum Rr

213
 We have assumed that maximizing Rr will, by definition
maximize efficiency. This remains a correct assumption
provided that the term

does not become more significant in reducing Rl. We have

quantified and shown a simple method to maximize efficiency
by maximizing Rr. We have also found that the effective losses
can be minimized by placing the current minimum at the feed
point impedance (also called voltage fed), again through top
loading. Figure 7.6 shows three configurations for a 1⁄4 λ
ground-mounted vertical.

For top loading there is yet another important term, R losses

in the capacitive hat. Loading inductors are often placed in
series with the capacitive element(s) to increase their effect.
These inductors can cause losses due to resistance in the
windings. Quantitatively, the effect of these loss resistances can
simply be added to Rl. Therefore, in an attempt to keep raising
the current maximum farther up the vertical, the top loading
structure eventually reaches a point of diminishing returns, in
that the increased top losses begin to approach the diminishing
ground losses. Another problem with large top loading

214
structures is unwanted radiation. As the loading conductors
increase in length they begin to radiate. This radiation is
usually highly undesirable since it will be horizontally
polarized and at a high elevation angle maximum. The usual
desired directivity from a vertical antenna is a low elevation
angle maximum.

Verticals Higher than 1⁄4 l

We have thus far focused on the 1⁄4 l vertical element. We
have also shown how placing top loading on this antenna can
significantly increase efficiency by redistributing the current
and increasing the effective height of the vertical. Over a
perfect ground, low angle directivity increases as the vertical is
made higher. However, when a linear vertical antenna hp is
raised above 5⁄8 λ, the low-angle directivity begins to decrease.
With further increase in height the directivity eventually
converges in a very high angle, or “straight up.” For most
practical purposes, there is no reason to build a vertical higher
than 5⁄8 λ hp. Figure 7.7 shows a comparison of vertical
antennas’ fundamental relationships up to 5⁄8 λ hp.

215
Raised Vertical Antennas
As we raise the vertical antenna together with its radial
system above ground, we gradually leave the characteristics of a
ground-mounted vertical and begin to see characteristics of a
ground plane antenna (Figure 7.8). Since we have explored the
ground-mounted vertical with some depth, we will now
consider the opposite extreme, a ground plane antenna in free
space. The ground plane antenna in its simplest form is a 1⁄4 λ
vertical section fed against two or more radials. The usual
configuration is three or four radials.

216
 A ground-mounted vertical antenna can be assumed to have
a complete image that mirrors the vertical element below
ground. Since there is no “ground” under a free space ground
plane, it is interesting to investigate what effect the radials
have upon substituting for “ground.”
We have already quantified he, Rr, and directivity for ground-
mounted verticals with an assumed “full” image. Since the
image directly effects he and thus Rr, we are in a position to
explore the free space ground plane.
The word “counterpoise” can reflect several definitions. In
this text we will define it as a raised ground system for a
vertical antenna. A counterpoise may consist of a single wire,
multiple wires, a flat conductive plane, or other configurations.
Unlike a large dielectric (such as the Earth) under the vertical,
a counterpoise does not create an image under the vertical
unless the size of the counterpoise becomes very large
compared the wavelength.
Step 1: 1⁄4 λ vertical with no radials in free space, simulating
no counterpoise (Figure 7.9). In both cases, the directivity is
about 1.82 dBi but Rr is about 25 Ω for the end-fed and about 12
Ω for the center fed. This result is very close to the expected
ratio using 0.64 he for the end fed and about 0.5 for the center
fed, everything else being about equal. Also notice that the

217
directivity is just a bit better than the infinitesimal dipole from
Chapter 3, which was 1.76 dBi. The directivity pattern, as might
be expected is a doughnut.

Step 2: 1⁄4 λ vertical with 1 radial, often referred to as an

inverted-V configuration, also in free space (Figure 7.10).The
Step 2 example is identical to the development of a 1⁄4 λ
transmission line from a dipole in Chapter 4. There is now a
counterpoise relative to the vertical element. As we know, if
this were a vertical 1⁄2 λ dipole, the gain would be higher, the he
would reflect an hp of 1⁄2 λ and Rr would 73 Ω.

218
 The feed point has risen from 22 Ω to about 42 Ω. In the case
of a single wire counterpoise, the counterpoise becomes part of
the antenna. With a multiple wire radial counterpoise, the
fields cancel and do not contribute to he, and therefore gain.
Step 3: 1⁄4 λ vertical with a four-radial configuration, also in
free space. Now we return to the typical ground radial shown
in Figure 7.8. The results (Figure 7.11) are counter-intuitive: the
directivity has dropped again slightly to 1.25 dBi, the directivity
pattern is nearly identical to the doughnut pattern in Step 1
(symmetrical in the upper and lower hemispheres), and Rr is
also very close to Step 1, about 22 Ω.

219
 The only advantage the radials provide with a 1⁄4 λ ground
plane is to create a counterpoise for the vertical element. In a
practical sense, complications of end-feeding a 1⁄4 λ element are
greatly reduced by including a counterpoise. This is a very good
reason to use radials. The radials accomplish little if anything
else; again, from a practical sense, there is no effect on gain or
Rr

Raised Vertical Antennas Over Earth

Step 4: The presence of the Earth under a ground plane has a
very profound effect upon its final directivity. At ele​vations
down to about 1⁄2 λ, Rr remains relatively constant at about the
free space 22 Ω value. Below this height some anomalies
appear until Rr returns to about 36 Ω when the vertical again is
ground-mounted. At the higher elevations, propagation effects
from the ground dominate the formation of the final directivity
pattern, but the antenna proper is largely unaffected. But, as

220
the vertical is lowered to ground level, the image gradually
returns — thus increasing antenna gain, he and Rr. In effect, as
the vertical is lowered closer to the ground, the ground itself
gradually becomes part of the antenna.
Much empirical work has been performed, mainly by
amateurs, to determine the trade-offs between ground-
mounted verticals and slightly elevated radials. In effect, this
work attempts to resolve several more partial differentials
involving the scalar Rl. These include the same terms that
appear in the Rl matrix above, but the addition of radial height
as an additional partial variable. The trade-off is the decreased
image in the ground vs. an effective decrease in ground losses
as described in the above paragraph.
Step 5: 1⁄2 λ dipole lowered from a high elevation to ground-
level. In free space the vertical dipole will have an antenna gain
of 2.15 dBi (power gain of about 1.64) and Rr of about 73 Ω. With
one end nearly touching the ground, the idealized gain
becomes 6.94 dBi (power gain of 4.94). In effect, the dipole
operating just above or connected to the ground behaves like a
2-element collinear antenna with the image providing the
other 1⁄2 λ. When lowered close to the Earth, there appears an
image, also 1⁄2 λ. Therefore, he doubles and he2 quadruples.
Aperture increases by 4.94/2.64 = 2.97, and therefore Rr
increases to 4/2.97 (73 Ω) or about 97 Ω.

Short Vertical Antennas

For very short vertical antennas (hp < 1⁄8 λ), the small portion
of a sine curve can be approximated by a straight line.
Therefore, the average current will be close to 1⁄2 the maximum
current value. See Figure 7.12.

221
Data for Various Ground-Mounted Unloaded
Vertical Physical Heights
Table 7.1 shows calculated data for various vertical physical
heights with no top loading. For the three shortest lengths, the
current distribution is assumed to be linear, therefore the
average current is simply 50% of maximum (we don’t need to
solve the integral in Equation 7.4). For about 1⁄8 λ and higher
the sinusoid distribution is assumed and Equation 7.4 must be
used.

222
 In Table 7.1 are the physical heights in wavelengths of
various vertical elements. heλ is the effective height stated as a
fraction of the wavelength; Giso is simply the numeric gain
value; and Aeλ is Gisol2/4π, where λ is normalized to 1. And
finally, Rr is from Equation 7.9.
As also shown in Figure 7.7, notice that the apertures (gain)
for very short verticals (less than about 1⁄8 λ) remain nearly
constant as we change hp. For very short verticals (from 0 to
about 1⁄50 λ), the dBi value derived from EZNEC modeling was
exactly 4.76 dBi (a power gain of exactly 3). Also, since the
gains and thus apertures do not change with antenna height
for very short heights, we can rearrange terms by substituting
3l2/4π for Ae in Equation 7.14, and simply multiply the isotropic
aperture by 3:

(Equation 7.17)
This is the simplified equation commonly published for short
vertical antennas. With no top loading, the average current is
about 50%. If appropriate top loading is installed, then the

223
current average approaches 100% and he is now nearly twice hp.
For an optimally top loaded short vertical, the average current
is 100%, thus Rr quadruples. This equation is simpler than a first
glance implies. For short verticals with no loading (a simple
vertical whip antenna), he actually equals hp since the average
current is 0.5, and we double this including the image, which
results in a coefficient of 1 between the two terms. Thus you can
simply substitute the actual physical length (without the
image) of the whip for he. A specific example of very short
verticals is presented at the end of this chapter.
Table 7.1 summarizes the three critical issues regarding very
short vertical antennas:
1) The directivity, and therefore potential gain, does not
significantly decrease with shrinking height.
2) The major problem is a rapid decrease in Rr, which
requires very careful design of the antenna and matching unit
to maximize efficiency. For example, a 1⁄32 λ vertical would
require only 0.4 Ω of loss in order to achieve only 50%
efficiency!
3) With optimum top loading, Rr quadruples, thus offering
significant efficiency improvement.

Ground Proximity to Current Maximums

We have already discussed the terms critical for maximizing
a ground-mounted vertical antenna’s efficiency. In addition to
maximizing Rr, the real part of the feed point impedance is the
other most critical term. For example, a base-fed 1⁄2 λ vertical’s
feed point impedance is very high, and since it is base-fed, the
current minimum also appears at ground level. Since the actual
value of ground resistance (at a specific frequency) remains
constant, power loss through ground current will be less at
high voltage points along the ground. This is similar to why
high voltages are used for long-distance transmission lines: for
a fixed R value of the line’s wires, higher voltage will exhibit
lower loss. Furthermore, as the distance increases from the
base of the vertical, the total current increases but is distributed

224
over a larger area of the ground surface. (The current is not
confined to the surface, but does penetrate into the ground. For
the sake of this argument, the ground is assumed to be a two-
dimensional plane.) In the case of a 1⁄4 λ vertical, the ground
current is maximum at and very near to the antenna’s base, and
therefore all the ground current is forced to traverse through a
small portion of ground.
The ground can be considered to be a two-dimensional
“square resistance.” However, the feed point becomes a point
source for current on the square resistance. We can
approximate this idea with a simple equivalent schematic
diagram.
Figure 7.13 shows how the current values decrease through
each equivalent resistor as we move away from the vertical’s
base. Assuming a constant total current along a radial angle, we
can calculate the total power dissipated as a function of
distance from the feed point. Let us assume the value of all
resistors is 1 Ω. Therefore, the power dissipated by the first
ground ring (closest to the feed point) is 1 W (12 R). The second
ring is 0.5 W (2(0.52 R)), and the third ring is 0.25 W (4(0.0252
R)).

225
 Of course the current is not constant as a function of distance
from the feed point, but this simple model does serve to
illustrate the point: for any current distribution ground
attenuation decreases as the distance from the feed point
increases. When the voltage maximum is at the feed point, it is
easy to see why 1⁄2 λ base-fed verticals have significantly lower
ground loss than 1⁄4 λ verticals for similar ground systems.
The actual current distribution as a function of distance
results from yet another partial differential equation involving
the sinusoidal function, dissipation due to ground attenuation,
and dielectric effects upon the wave velocity in the ground and
on the ground’s surface.
This simple model also shows why wire radial ground
systems are most effective. They inherently provide for the
highest density of conductivity where it is most needed: closest
to the feed point. They also provide this conductivity in the
most optimum direction(s): normal to the feed point. It also
points to the importance of the

226
term (the number of radials being used).
It should now be clear why avoiding maximum current at
ground level inherently increases efficiency in a compromised
ground system. The 1⁄2 λ vertical is one example, but other
vertically polarized configurations can accomplish this same
goal. We will show later that efficiency is only part of the
system performance. We will now explore particular vertical
polarized configurations and then compare their performance.

The Near-Ground Inverted-V Antenna

The near-ground inverted-V in its simplest form is a 1⁄2 λ
dipole supported by a single high support, and the two legs
form a 90 degree angle with their ends near ground level as
shown in Figure 7.14. Although seldom considered as a
“vertical” antenna, it should be considered as such.

Since the inverted-V legs form 45 degree angles with the

ground’s surface, this antenna has both a horizontal and
vertical polarization response. Broadside to the antenna (into
and out of the page in Figure 7.14), the response is
predominantly horizontal, while the end-fire response (to the

227
right and left of the figure) is predominantly vertically
polarized. If we compare the azimuth patterns of a single
vertical element and that of a near-ground inverted-V, we can
see that the low-angle gains are nearly identical (Figures 7.15
and 7.16).

The inverted-V, like its dipole cousin, has its current

minimum at its ends. The ground, in effect, does form part of
the antenna, as with the ground-mounted vertical. In addition,
high-voltage coupling to the ground is by means of capacitive

228
coupling. The capacitive values are very low, although the
voltages are very high. Again, for total losses in such cases, a
partial differential equation is required to solve the problem.
The ground losses due to capacitive coupling from the high
voltage ends is usually far less than the advantage of getting
the current maximum(s) as far away from the ground as
possible.

Near-Ground Phased Inverted-Vs

We can take additional advantage of the vertical polarized
end-fire low-angle directivity of the inverted-V by phasing two
together. See Figures 7.17 through 7.19. As a practical matter,
the feed point impedances of multielement, near-ground
inverted V arrays drop to a few ohms. A simple technique to
raise the feed point impedances is to use a folded V, similar to
the folded dipole in Figure 3.12.

229
Broadside Bobtail Curtain
The Bobtail Curtain (Figure 7.20) is an example of a

230
broadside, voltage-fed vertical array that automatically
provides for voltage maximums at the base of all three
elements. The array is base-fed at the center vertical element.
A minimum of radials are required only at the center-fed
element, in practice to help stabilize the high feed point
impedance. The three vertical elements are 1⁄4 λ and the two
horizontal wires are 1⁄2 λ making the array 1 λ long. These are
the dimensions for optimum gain. However, “mini Bobtails”
can also provide excellent results using shorter vertical
elements and/or narrower width as long as the phases and
amplitudes along the three elements are re-optimized for best
results. Figures 7.21 and 7.22 show the patterns.

231
Single-Element Vertical (Monopole Antenna)
A single ground-mounted vertical element has become the
“standard” vertical antenna for a very wide range of
applications. The 1⁄2 λ dipole is the only antenna that is more

232
common. In this text, the first mention of these simplest of
antenna configurations was in Chapter 1. The monopole’s main
advantages are its mechanical simplicity and its
omnidirectional azimuth pattern (Figures 7.23 and 7.24).
Figure 7.25 shows a 5⁄8 λ vertical for comparison.

233
Multielement Vertical Arrays
As we discussed in Chapter 5, the addition of elements allows
configuration of directional arrays. In AM broadcasting, nulls
are created in antenna patterns to “protect” other AM station
coverage areas, particularly at night. For amateur applications,
the usual goal is to maximize gain in a particular direction,
hopefully with many directions easily selectable.
In this section we will explore three multi-element

234
configurations: 2, 4, and 8 element configurations. The plots
shown in Figures 7.26 through 7.31 are taken at 1.8 MHz, but
the gain can be corrected for other frequencies with a single
coefficient for a good approximation. For 75/80 meter
applications, subtract 0.9 dB from these numbers. For 40
meters, subtract 1.5 dB (based on EZNEC modeling results).

235
236
237
 In 1994 while I was working up the first known 8-circle
design, I noticed that with 1⁄4 λ spacing of the elements around
the circle, the resulting spacing of the two cardioid pairs would
be very close to 5⁄8 λ, which is optimum for a broadside array!
This results in a very effective configuration by virtue of the
array’s geometry. See Figures 7.32 to 7.34.

238
239
Gain Comparison of Vertical Arrays
We are now prepared to summarize the relative advantages
and disadvantages of the various configuration presented in the
following tables. The arrays detailed in Table 7.2 are arranged

240
by increasing gain.

When placed over “real ground,” the frequency of operation

does have an effect upon the absolute gain numbers. As the
operating frequency rises above “average ground,” the gain for
the equivalent array decreases. The gain values given in Table
7.3 are for 1.8 MHz. For all these antenna configurations,
simply subtract 1.5 dB for 7 MHz, The actual range of deviation
is about 1.4 to 1.6 dB. For 3.6 MHz subtract 0.9 dB. Also the
elevation angle for maximum gain rises with frequency, but not
more than 2 or 3 degrees. Since the 1⁄4 λ ground plane in free
space has no ground influences, its gain remains constant over
all frequencies.

241
 These EZNEC models show some interesting differences
between the perfect and real ground systems. Although the 1⁄4 λ
vertical has a small gain advantage over the infinitesimal
vertical in the idealized perfect ground circumstance, it
actually shows less gain over real ground. The 1⁄4 λ vertical
begins to be affected by ground reflection cancellations,
although it remains minimal. Unlike the 1⁄4 λ vertical, the
infinitesimal vertical far field ground effects are limited to only
tangential dielectric absorption since there can be no ray
cancelations. However, from a practical viewpoint, the Rr of the
infinitesimal (or very short) vertical will be very low and thus
the efficiency will also likely be very low. Also, from a practical
sense, losses in the matching network will also add to
inefficiencies. Thus the slight advantage of a very short vertical
over a 1⁄4 λ vertical will in most circumstances will be more
than offset by efficiency concerns. Appendix E shows plots for
various vertical arrays, providing a graphic comparison of
expected gain responses.
The top-loaded (capacitive hat) 1⁄4 λ vertical shows a very
slight compromise in gain due to the raising of the current
maximum up the vertical height. However, the Rr has increased
significantly thus increasing efficiency significantly —
especially over a compromised ground radial system.
With the 5⁄8 λ vertical, the effects from reflection nulling

242
become dramatic. The current maximum on the 5⁄8 λ vertical is
3⁄8 λ above ground, thus providing enough height to generate
multiple nulling reflections in the far field. The inverted-V
configurations suffer comparatively little over real ground
because the current maximum is at the antenna’s apex, and the
vertical polarized source is distributed over both the X- and Z-
axis (sloping elements), thus “fuzzing” the reflection nulling.
The 5⁄8 λ gain maximum is at a very low 12 degree elevation
angle. Shorter verticals will have significantly less gain at these
angles. Therefore, the take-off angle desired for a given
frequency and target area should be considered when
comparing antennas. Usually for long-range F2 paths, higher
frequencies typically require lower take-off angles. Thus 13
degrees may be too low for many 160 meter paths, but may be
excellent at 40 meters. Again, from a practical sense, the slight
advantages of the 5⁄8 λ ground-mounted vertical are usually
offset by the fact it is more than twice as high as the 1⁄4 λ
option.
Also notice that the 2, 4 and 8 element phased arrays have
nearly identical maximum gain elevation angles, near 21
degrees for 160 meters. This is also nearly identical to the
monopole. Thus for a given vertical height, operating
frequency, and ground conditions, there is a conservation of
maximum elevation angle gain irrespective of the number of
elements. This also holds true for the higher vertical
configuration. Thus, for example, an array of 5⁄8 λ verticals
would exhibit similar azimuth patterns to the 1⁄4 λ arrays, but
the maximum elevation gain angles would be about half, or
about 11 degrees.
These real ground models in Table 7.2 and Table 7.3 assume a
100% efficiency ground radial system. Therefore this
comparison must be tempered with compromised ground
systems. For example, it is very unlikely that a practical
comparison between the infinitesimal and 1⁄4 λ vertical will
show an advantage to the infinitesimal antenna, especially
over compromised ground systems. Also a 1⁄2 λ or 5⁄8 λ vertical
will have higher Rr than the 1⁄4 λ vertical and higher feed point
impedance values, thus increasing the overall efficiency in

243
compromised radial systems.
The 2, 4, and 8 element directive arrays provide a coinciding
possible number of switchable azimuth directions. For
example, the 2 element array inherently provides two
switchable direction. The 4 element array provides 4
directions, and the 8 element provides 8 directions. It is also
fortunate that in all cases these directional settings provide
sufficient over-lap of azimuth patterns that result in only about
2 dB nulls for full azimuth coverage.

Conclusion
When weighing the relative advantages and disadvantages of
vertical arrays, Table 7.3 can represent a good “starting point”
in that it compares:
1) Physical dimensions and necessary area for the antenna.
2) Gain that can be expected using an “adequate” radial
system under the antenna when needed.
3) Comparative optimum take-off angles for maximum gain.
4) Relative radial requirements.
5) Feed point location.
In the case of a compromised radial system (when needed)
the quoted gain will diminish as Equation 7.15. Since this
equation is not closed form, for practical information the
volumes of empirical data from amateur experiments can
provide good approximations for real implications of Equation
7.15. However, Equation 7.15 is an attempt to actually quantify
all the important terms into one unifying equation.

EIRP Calculation for Ground-Mounted Low

Frequency Vertical Antennas
Vertical antenna efficiency problems become particularly
acute at very long wavelengths. Recent amateur bands below
the AM broadcast band (the 630 meter band, 472 – 479 kHz)
present new challenges for system design. Although the same
issues apply for these frequencies as outlined above, added
attention is required. EIRP is a power output specification
where:

244
EIRP = Pi (Gdbi)

(Equation 7.18)
where Pi is the power input to the antenna.
For example, if the input to an antenna is 5 W and the
antenna has a gain of 1.78 dBi, the EIRP in the direction of the
4.78 dBi gain is 15 W. A short vertical antenna (<1⁄8 wavelength)
has a maximum gain of 4.78 dBi (linear gain of 3) with a perfect
ground system and a perfectly conducting infinite ground. This
gain figure is independent of the vertical’s height for heights
below about 1⁄8 wavelength. Unless ground losses in the far
field are taken into account for EIRP calculations, this will be
the coefficient in calculating EIRP values. For example if we are
limited to 5 W EIRP and use a vertical shorter than 1⁄8
wavelength, our maximum power input to the antenna will be
5/3 or 1.6666 W.
At these frequencies, the typical amateur transmitting
antenna will likely be very inefficient. A full size 1⁄4 wave
vertical is about 500 feet high! The efficiency will also be
complicated by an almost certain compromised ground system.
AM broadcast stations use 120 ground radials each 1⁄4
wavelength long. The equivalent system at 475 kHz would
require a circle of radial wires about 1000 feet in diameter
(almost 20 acres).
The three primary efficiency concerns are as following for
very short ground-mounted verticals:
1) Ground losses.
2) Vertical element losses.
3) Matching losses.

Ground Losses
The short height and compromised radial system results in a
very inefficient transmit antenna system. However, the FCC
will be limiting the maximum EIRP to very low values, ≤5 W.
Assuming 1.6666 W antenna input, this implies that an antenna
only 1.6% efficient would require an antenna input power of 100
W. Wasting 98.3333 of 100 W is much more cost effective than

245
the above extravagant solution. This is a huge advantage of
regulating power in terms of radiated power rather than
transmitter output power.
If we want to calculate the required antenna power input Pi,
we use:

(Equation 7.19)
where EIRP is the desired effective radiated power.
Now we need to find the proper values of Rr and Rl.

(Equation 7.20)
where hp is the physical height of the vertical (without the
image) and λ is the wavelength, in this case about 2052 feet, or
632 meters. For the special case of verticals less than about 1⁄8 λ
high, he ≅ hp making the calculation easier. For example, a 50
foot high vertical ground-mounted antenna for 632 meters will
have an antenna Rr of 0.23 Ω. There is little advantage in
attempting top loading since the loading structure would likely
be very large, especially for verticals under about 100 feet.
The calculation of Rl requires a direct measurement of the
real portion of the feed point impedance.
Rl = Rfeedpoint – Rr

(Equation 7.21)
where again, Rr is calculated by Equation 7.20. Now we have all
the terms to calculate the needed antenna input power for a
given EIRP (Equation 7.19). It is very likely that the measured
Rfeedpoint will be much higher than the calculated Rr indicating

246
the losses will be very high and thus efficiency very low.

Vertical Element Losses

For such very low values of Rr it is also very important to
lower the resistive value of the vertical itself in addition to
ground losses. The losses along the vertical length will not be a
linear function of Rr and Rlv. EZNEC can estimate the losses for
you, by applying the correct element diameter(s) and
material(s). Using EZNEC, you can simply perform two models:
one using “zero” for “wire loss” and the second using the
appropriate metal and element diameter. The maximum gain
difference will yield the dB loss due to the vertical element loss.
Some may think that vertical element loss will also affect Rr
which, in general, is an incorrect conclusion. The effective
height will be reduced because there will be a reduction in
average current along the vertical height. But also there will be
a reduction in gain, and thus also aperture. These terms usually
balance, and thus the calculated value of Rr remains constant.
The exception is when the loss resistance is “lumped” and of a
high value. In effect this shortens the physical height of the
vertical and simultaneously distorts the current distribution
from its assumed linear function. However, when typical values
of loss resistance are distributed over the length (especially the
lower half of the vertical) and are less than about 1 Ω, the
above assumption can be made. EZNEC makes short work of
this calculation by the above technique.
The current distribution along the vertical element will be
linear along its length with the maximum at the base.
Therefore, the greatest effort should be to reduce resistance at
the base of the vertical. For obvious reasons this is good news
— heavier conductors are best placed closer to the ground.

Matching Losses
Thus far we have limited the discussion to the antenna
proper. However, in this case, short lossy verticals will exhibit
relatively low real values at the feed point, perhaps tens of
ohms. However, the imaginary part of the feed point

247
impedance will be a high negative value (high capacitive
reactance). This calls for the use of high Q matching,
particularly the inductor. Typical inductive values for the tuner
will be about 1 millihenry. An air core inductor with 1⁄4 inch
copper tubing would be about 6 feet long, 18 inches in diameter
and 100 turns. It may be possible to build a comparable
inductor out of ferrite cores, but this very high inductor value
must also handle about 100 W.

Summing the Losses and Determining

Required Transmitter Power Output
Now we have all the terms we need to determine what input
power is necessary for a given EIRP for a given system. Stated
in dB losses and dBW power levels it is

(Equation 7.22)
where 0 dBW= 1 W and Pi (dBW) is the power output from the
transmitter. A detailed description of actual power levels using
logarithmic notation is presented in Chapter 10.
This represents the basic equation for determining the
proper transmit output power for a desired EIRP. The system
design will be challenging if the FCC limits transmitter output
power. For example, if the FCC restricts transmitter output
power to 100 W and the maximum EIRP permitted is 5 W,
implying an antenna input of 1.66666 W, a total of only 17.8 dB
loss will be required to achieve maximum EIRP output.
Many of these complications will hopefully be settled with
clear power rules. Aside from the EIRP calculation, will the
calculation for Rr require computer modeling? Will transmitter
output power consider matching losses? Answers to these and
other questions may change the coefficient values in Equation
7.22, or even change the equation itself.

248
 CHAPTER 8

Yagi-Uda and Cubical Quad

Antennas
No antenna text can be considered complete without a
discussion of parasitic antennas generally and Yagi-Uda
antennas in particular. During the 1930s, Shintaro Uda (Figure
8.1) was a student of Professor Hidetsugu Yagi (Figure 8.2). Mr.
Yagi acknowledged the Mr. Uda was indeed the inventor of the
Yagi-Uda array. It continues to carry both names.

249
 The Yagi-Uda has been the subject of intense investigation
for many decades, resulting in highly optimized designs. There
are many fine texts that detail Yagi-Uda and quad type arrays.
This chapter will only review the basic computational methods
used for optimized designs and link previous discussions in this
book to those methods. Therefore, as one of the final chapters,
the parasitic antenna can be considered as a specific
culmination of the previous chapters.

Gain and Aperture of a Yagi-Uda Antenna

Most technical publications about the Yagi-Uda antenna
stress F/B (front-to-back ratios) and forward gain as functions
of boom length, thickness of the elements, element spacings,
and lengths of the parasitic elements. I will not repeat this
readily available and extensive information. Rather, we will
explore the relationship of the Yagi-Uda to its effective
aperture and continue the general discussion of parasitic arrays
from Chapter 5.
The Yagi-Uda is a derivative of the simple dipole. There are
two ways to increase the aperture (gain) of a 1⁄2 λ dipole

250
antenna: increase the length, and/or create a multielement
array (either by driving multiple elements, or by adding
parasitic elements, or a combination of all these).
First we will explore the increased length dipole. We have
already derived the broadside power gain of a 1⁄2 λ dipole as 1.64
(2.15 dBi). We have also defined broadside gain as a function of
length (Figure 3.17 in Chapter 3).
Again, the aperture of a 1⁄2 λ dipole is

(Equation 8.1)
If we double the aperture, the gain will also double and we
achieve about 3 dBd gain. A 2-element collinear antenna is
twice the length of a 1⁄2 λ dipole, but only achieves about 2 dBd
gain. If we extend the length to 10/8 λ, we create an extended
double Zepp. This is the length of a center-fed linear element
that achieves the maximum broadside gain (about 3 dBd).
The boundaries shown in Figure 8.3 are approximate. The
actual boundary lines are a bit “fuzzy” but this figure serves as a
good starting point to conceptualize what the aperture areas
actually look like around linear antennas.

251
 The second method of increasing the gain of a single dipole is
to add elements. In Chapter 7 we showed many multielement
vertical driven arrays and their relative gains and patterns. The
same basic techniques can be used for free space dipoles with
similar results in gain. Thus free space driven arrays will not be
covered here in detail to avoid repetition.

252
 We begin the discussion with a simple two-element Yagi-Uda
parasitic array (Figure 8.4). By making the parasitic element
slightly longer than a resonant dipole and placed in back of it,
the backward pattern can be nearly completely eliminated and
aperture and gain increases in the opposite (forward) direction.
By making the element slightly shorter than resonant, a
different phase response is achieved, in that additional gain is
realized in the forward direction when the director is placed in
front of the driven element.

The very high scattering aperture (discussed in Chapter 2) of

a shorted dipole makes possible the high gain characteristics of
the Yagi-Uda antenna. For example, a typical 2-element Yagi
has about 5 dBd of gain (5 dB higher than a simple dipole).
Notice in Figure 2.23 in Chapter 2 that for a shorted 1⁄2 λ dipole
the scattering aperture is 4× the maximum possible effective

253
aperture. On the other hand, a 2-element quadrature driven
array has a maximum gain of about 3 dBd, which is what would
be expected by a simple doubling of a single dipole’s aperture.
See Figure 8.5.

By adding directors in the desired maximum gain direction,

multielement Yagi-Uda arrays can be built. Antennas with 20
or 30 elements are practical for some applications. In the
majority of configurations, there is one driven element and one
reflector. Additional elements take the form of directors. This is
the optimized design configuration because the reflector is
more effective than directors at maximizing the front-to-back

254
ratio. The addition of more and more directors is about
proportional to the log of the gain. Therefore, adding more and
more directors provides less and less gain addition per element.
This is because, again, the additional directors begin to
duplicate the available aperture around the antenna. On the
other hand, the inherent advantages (electrical and
mechanical) often outweigh this diminishing return.
Now let us assume a 3-element Yagi-Uda array. The addition
of a single director is taken at a typical value of 2 dB more than
a 2-element array. Therefore, we assume the Yagi-Uda has a
gain of 7 dBd, or about 9 dBi, or a numerical power gain of 7.94
(in free space). Therefore the aperture of our 3-element Yagi-
Uda is

(Equation 8.2)
In Figure 8.3 the shape of the apertures is elliptical. This is
due to the horizontal current distribution along the wires. As
the gain increases with a Yagi-Uda, the array proper begins to
resemble a point source, and thus the shape of the aperture
becomes progressively more “circular” as shown in Figure 8.6.
In turn we can define the radius of this aperture circle by

255
(Equation 8.3)
Stacking Yagi-Uda Antennas
We may increase the gain of a single Yagi-Uda by simply
adding a second identical antenna. In principle this should
double the aperture and thus double the gain (+3 dB). If yet
more gain is needed, doubling the aperture again will yield (+3
dB) more (using four antennas). Another 3 dB can be realized
by doubling the aperture again (now eight antennas), and so
forth. In Figure 8.3 we showed that if two dipoles are spaced
very closely (end-to-end), some of the potential aperture is
covered by both antennas. This is analogous to our discussion

256
of solar cells in Chapter 2. If one solar cell partially covers
another (shading), the amount of aperture (gain) is reduced.
The correct spacing among “stacked” Yagi-Uda antennas can be
determined with fair accuracy by calculating the equivalent
aperture (as in Figure 8.6) and then make sure the antennas
are sufficiently separated to provide minimum “shading” of the
other as shown in Figure 8.7 for antennas stacked side-by-side.

For maximum gain there is no reason to increase the spacing

further. Therefore, as we increase the number of elements of
the Yagi-Udas, we increase the gain, increase the aperture, and
thus the spacing among the elements must increase to avoid
wasting the gain by duplicating the aperture coverage.
Furthermore, if the spacing is made progressively larger,
sidelobes begin to appear.
Figure 8.8 shows the azimuth pattern of a simple 2-element
21 MHz Yagi in free space. We can stack two of these antennas
horizontally as shown in Figure 8.9, and Figure 8.10 is the
resulting azimuth pattern for the stack. Figure 8.10 shows that
the aperture of the array of Figure 8.9 is “distributed” more
along the horizontal axis than the vertical axis. This results in a
narrower azimuth response than the single antenna, but has
little effect upon the vertical beamwidth.

257
258
 We can also “stack” the two antennas vertically. The
difference is how this will affect the directivity of the array.
Vertical stacking of Yagi-Udas over real ground provides
diversity of the maximum elevation angle of radiation. Low
horizontal antennas will radiate at higher elevation angles
than higher horizontal antennas (assuming the maximum
gain).
Figure 8.11 shows an array of eight Yagi-Uda antennas for
432 MHz using both horizontal and vertical stacking.

259
Design and Optimization of a Yagi-Uda Array
The fundamental theory of parasitic arrays was presented in
Chapter 5. We discussed the fundamental importance of the
mutual impedances of the array. However, we only dealt with a
2-element array. Now we will present numerical methods for
multielement arrays. In Chapter 7 we presented the concept of
using the matrix as a method of linking multiple
interrelationships of multiple functions. The modeling of a
multielement Yagi-Uda antenna is most easily performed using

260
such a matrix. Unlike the Chapter 7 discussion where we
resorted to differential equations for the terms, with a Yagi-
Uda we are only dealing with (in principle) relatively simple
algebraic functions, in particular solving for Ohm’s Law with
complex impedances.
Consequently, matrix algebra is used to calculate the relative
amplitudes and phases on each element that, in turn,
determine the final directivity of the antenna. We begin by
expanding the 2-element discussion in Chapter 5 to a 3-
element Yagi-Uda. The first step is to compute the voltages that
appear on each element as a result of all the self and mutual
impedances:

(Equation 8.4)
The resulting voltages on each element are the sum of their
corresponding currents and the complex impedances
determined by the self and mutual complex impedances. It is
these complex impedances that alter the relative phases of the
currents on each element. Once again the phases, amplitudes,
directions, and locations (PADL) of these currents, determine
the final directivity of the antenna.
The matrix in Equation 8.4 actually represents three linear
equations, with the currents being unknown. The mutual
impedances can be computed analytically by knowing the
spacings and electrical lengths of the elements. Thus if we
apply a voltage to the driven element (element 2), we can
derive the resulting voltages by this set of linear equations
contained within the matrix.
The final step is to derive the currents by inverting the
complex impedance matrix to yield

261
(Equation 8.5)
which, of course is simply a special form of Ohm’s Law

(Equation 8.6)
Once we know the complex currents in each element
(amplitude and phase) and the spacing between the various
elements, we can directly calculate the far-field of the array,
again by the PADL method. This is the basic technique used by
antenna modeling software such as EZNEC.
As with most antenna configurations, today computers are
used to solve for these matrices and then calculate the resulting
pattern. Furthermore, optimization for forward gain, front-to-
back ratios, and other parameters involve hundreds if not
thousands of calculations through trial-and-error algorithms to
derive an optimized design. Again, today these computations
require only seconds on a laptop computer.

The W8JK Array

Although not a parasitic array, the W8JK array deserves
review and may be compared to the performance of a Yagi-Uda.
In 1937, John Kraus, W8JK, built a closely spaced driven array
that resembled a Yagi-Uda. The array consisted of two 1⁄2 λ
dipoles fed 180 degrees out-of-phase at 14 MHz. The result is a
maximum bidirectional gain of 5.88 dBi, or about 3.74 dBd. This
is about 0.74 dB higher than a mono-directional pair of dipoles,
but about 1 dB lower than the mono-directional 2-element
Yagi-Uda. The W8JK uses the 3 dB advantages of a doubling of
aperture, but also both driven elements provide additional gain
by virtue of their scattering apertures. This results in a slightly
higher gain than a simple aperture doubling.
The W8JK can be thought of as yet another derived design
from Chapter 5. See Figure 8.12. Rather than two dipoles
spaced 1⁄2 λ and fed 180 degrees out of phase, we now space
them about 1⁄8 to 1⁄4 λ. The flexibility of spacing is added to the

262
flexibility of element length (both as a function of wavelength).
Thus, the W8JK has the inherent advantage of providing
broadside gain over one octave or more of usable spectrum (not
mentioning the required matching for different frequencies!). If
the element lengths are made 5⁄8 λ long, and use 1⁄2 λ spacing
(let’s say for 28 MHz), then appreciable gain and tolerable feed
point impedances result as low as about 10 MHz (about 3⁄32 λ
spacing). At the highest frequency, the two dipoles are
extended double Zepps, and the bidirectional gain is comparable
to a 3-element Yagi-Uda (in both directions simultaneously).
The pattern is shown in Figure 8.13.

263
 For all its simplicity, as the operating frequency changes, the
feed point impedance moves “all over the Smith chart.” Thus a
practical broad-band W8JK should use open wire or ladder line
to a tuner and balun for matching.

Multiband Yagi-Uda Arrays

While the W8JK offers a very simple configuration for
multiband operation, the Yagi-Uda does not. Yagi-Uda
antennas are very effective when optimized for one narrow
band of operation. There are several popular methods for
designing a multiband Yagi-Uda.
1) “Traps” may be installed along the elements to electrically
set the equivalent physical length of the elements. The traps
are actually parallel resonant tank circuits placed in line with
the element. At the resonant frequency, the trap effectively
switches off the element length beyond its position. At lower

264
frequencies the trap appears as a series inductor, thus making
the element electrically shorter at lower frequencies. Good trap
design can result in low Rl loss, but there is always some loss.
The shorter element lengths also reduce gain and decrease the
VSWR bandwidth.
2) “Interlacing” elements simply places elements for different
frequencies along the same boom. The main problem is inter-
band interaction through mutual impedances of the various
elements. Again, very careful design is needed to minimize
these effects while maintaining acceptable performance. An
interlaced Yagi-Uda array may have separate driven elements
for each band, each with its own feed line, or it may use a
broadband or multiband driven element with a single feed line.
These may be a small log periodic cell or a set of closely-spaced
elements that exhibits resonance at multiple frequencies
(typically referred to as “coupled-resonator” or “open-sleeve”).
3) Remote control of element length. This is an excellent
technique used to change the length of the elements. Designs
have included using air pressure to set the lengths against two
stops for a dual-band configuration. Currently the most popular
method utilizes stepper motors that can set the element
lengths with high precision. The one disadvantage is shared
with trap antennas: the element separation is fixed. However
the design is inherently broad-banded in that it can be
absolutely optimized for any frequency given the spacing
constraint. Unlike the W8JK matching problem, the length of
the driven element can be adjusted to resonance and a simple
broadband balun-transformer can easily match to the popular
50 Ω coaxial cable feed line at any of the frequencies covered
by the antenna.
4) Linear loading can involve many different configurations
of multiple conductors running parallel to a “main” conductor
along portion(s) of the element’s length. Parallel conductors
can provide parallel resonant dipoles or sections of open or
shorted transmission lines that in effect, create multiple
resonances at desired frequencies without the need for LC
traps. This technique eliminates trap losses, but complicates
the mechanical design.

265
The Cubical Quad
The Yagi-Uda is a derivative of the linear antenna form: the
dipole. Large loops (≅ 1 λ circumference) can be configured in a
very similar manner to the Yagi-Uda by feeding only one loop
(the driven element) and using parasitic elements slightly
longer (reflector) and slightly shorter (director) to create more
gain and/or front-to-back ratios. This is the basis of the cubical
quad antenna such as the one shown in Figure 8.14. The
calculations are identical to the Yagi-Uda.

Figure 8.15 shows the directivity of a single 1 λ square loop,

which has gain about 1.15 dB higher than a 1⁄2 λ dipole. The
initial gain advantage of the quad over a dipole begins to
diminish as more elements are added. Most models and
empirical studies show a 2-element quad outperforms a 2-

266
element Yagi-Uda (for forward gain) by about 1.8 dB, but as the
boom length is extended for more elements, the Yagi-Uda
shows at least a parity of gain with the quad. This advantage,
coupled with the simpler mechanical structure of the Yagi-Uda,
has made the Yagi-Uda far more popular. However, the quad
has two big advantages in addition to the slight gain advantage
of 2-element performance. First, it is much easier to build
multifrequency quads. Smaller loops for higher frequencies are
simply added to the same spreaders. Inter-band mutual
coupling is also less with a quad, and there is no need for traps
or very careful design of “interlacing” elements. Second, the
quad loops inherently are lower in Q, thus providing a good
match over a greater bandwidth than a dipole/Yagi Uda design.

One final note: Good quality traps are expensive to buy or

267
even build. High quality high voltage “doorknob” capacitors are
usually required. Also, quality aluminum tubing is expensive.
After the support, rotator system and boom, the quad requires
only wire and bamboo poles, versus telescoping aluminum,
expensive traps, and very low error tolerance in cutting and
setting the aluminum. I built my first quad when I was 16 years
old, I could not beat the performance for the required
investment. Today I use Yagi-Udas, but would still consider a
quad if my requirements changed.

Quad Parasitic Elements in Vertical Arrays

We can also apply the advantages of scattering apertures to
vertical polarization. Using ground mounted vertical elements
as parasitic arrays is possible in principle. However, such
parasitic arrays are also prone to the same Rl losses as the
driven element. Also, the fine tuning required to establish a
parasitic element’s parameters is difficult to maintain with
changes in soil moisture and temperature particularly. This is,
again, particularly true over compromised ground systems.
Recalling our discussion from Chapter 5 on parasitic arrays, a
parasitic element relies completely upon its mutual impedance
with the driven element(s) to establish the necessary phase
relationship for optimum gain in the desired azimuth
direction(s). In contrast, a driven array’s element phase
relationships are primarily determined by the driven phase
relationships, which are much easier to stabilize in an unstable
environment (such as a typical ground-mounted vertical array).
An alternative is to use a quad loop parasitic element (Figure
8.16). In principle, a 1⁄2 λ parasitic element could be employed,
but the element would be considerably higher. We can also
bend the parasitic element, perhaps at the top and the bottom,
to maintain the high current portion of the element in a
vertical orientation. In effect, this forms half of the full quad
loop! By completing the loop we add little to the required
support, add additional gain, and provide for greater electrical
stability (lower Q). Figure 8.17 shows the gain pattern of a 1⁄4 λ
monopole for 7 MHz with a cubical quad loop director.

268
269
 There are several important notes to this type of
configuration.
1) The actual shape of the loop is not critical. We want a
response to vertical polarization, so the vertical wires should
be kept as long and “vertical” as possible.
2) We can change the loop from director to reflector by
simply adding inductor(s) for the reflector. However, the
inductor cannot be placed at the high voltage point (centers of
the horizontal wires). They will have no effect. Rather, for a
square loop, two inductors should be used, one each at the two
bottom corners.
3) The lowest wire (wire 2 in Figure 8.16) will exhibit very
high voltages and therefore should be kept well out of reach of
people and animals (as should all antennas used for
transmission).
4) The optimum length (circumference) for the parasitic
element will be greatly affected by the overall height and the
shape of the loop. Modeling will be necessary, but fine tuning
can be accomplished by breaking the loop at a corner and
resonating to the appropriate frequency above (for a director)

270
or below (for a reflector). However, despite this sensitivity it is
more inherently stable and efficient than a ground mounted
vertical used as a parasitic element.
5) An inherent problem with any parasitic element is that it
remains a parasitic element even when its associated antenna
is not being used. For example, the presence of a loop reflector
will also dramatically affect a receive antenna located nearby.
This undesirable affect can be eliminated by making the loop’s
total wire length shorter than a full size reflector, about 0.9 λ
or less (Figure 8.18). The loop is activated by switching in an
inductance and closing the loop. The switch that breaks the
loop is located adjacent to an inductor, so undesirable loading
is avoided. Figure 8.19 shows one method of creating a
variable inductor that can be used to tune the loop.

271
272
 CHAPTER 9

Specialized Antenna
Configurations
Antenna design has evolved into many different
configurations since the inception of radio technology. Thus far
we have considered center-fed linear elements and derivatives,
of which the 1⁄2 λ dipole is the simplest form. In this chapter we
will investigate some of the other important configurations for
professional and amateur applications alike.

Small Loop Antennas

In Chapter 8 we described the cubical quad antenna. The
quad is based upon elements about 1 λ circumference in size,
making the quad a “large loop” antenna. Here we will describe
the characteristics of “small loop” antennas. The shape of the
small loop may be circular, square, triangular or any other
shape as shown in Figure 9.1. It may also consist of one turn or
more, and can range in size (circumference) from about 1⁄2 λ to
a small fraction of a wavelength.

A loop’s characteristics will change significantly as the

273
diameter is reduced from a large loop to a small loop. Here we
will treat the small loop in some detail, but will compare it to
the large loop to gain insight into how the two configurations
create their respective patterns.
Let us begin with a very small loop. If we reduce the loop size
to a very small size compared to the wavelength, we create an
antenna that is “almost” a shorted transmission line. As
expected, this shorting section will exhibit very high current
and radiate very little power (very low Rr). As we make the loop
larger, but still very small compared to the wavelength, the
current begins to fall, and more power is radiated. However,
since the circumference of the loop remains much shorter than
the wavelength, the current value along the loop’s conductor
will remain essentially uniform as will the phase of the current.
We can once again employ the PADL principle introduced in
Chapter 3 to gain an intuitive understanding of how a small
loop’s pattern is formed. In Figure 9.2, since the circumference
of the loop is much smaller than the wavelength, both the
phase and the amplitude can be assumed to be constant around
the loop. However, the direction of the current (accelerating
charges) is equal in all directions. If we view the loop broadside
to the plane of the loop as in Figure 9.2, the distance to all
sections of the loop is equal from any point in line with the
loop’s axis. Therefore, there is perfect cancellation in the
broadside direction.

274
 In contrast, if we view the loop in Figure 9.2 from its side
(from the right, left, up or down from the loop), one critical
PADL parameter has changed. Again if we look at Figure 9.2,
but this time from a distant point to the “left” or “right” of the
loop, there is a slight difference from the distant point source
to the two sides of the loop (the distance of the loop diameter).
This difference is enough to provide a slight difference in the
phase, and thus provides a slight non-cancellation of the fields,
and thus some radiation maximum around the radius of the
loop — but again, complete cancellation broadside to the loop.
The very slight phase response normal to the loop’s axis results
in small loop efficiencies that are very low. Figure 9.3 shows
the current direction on small loop antennas versus large loop
antennas.

275
 Figure 9.4 shows the resulting figure 8 pattern with the null
broadside to the loop. This is completely opposite the pattern
of a large loop 1 λ in circumference, where the nulls are in the
plane of the loop and maximum directivity is broadside to the
loop. Figure 9.5 compares the E and B fields of a small loop and
a small dipole.

276
 As we might expect from any antenna that is small relative to
the operating wavelength, small loops tend to be rather
inefficient. Indeed, the common expression for antenna
efficiency also applies to loop antennas where

(Equation 9.1)
Loop antenna Rr is calculated from the general form, Equation
3.37 in Chapter 3:

(Equation 9.2)
Again where S(θ,φ)maxr2ΩA represents power concentrated in

277
the beamwidth defined as ΩA, and I2 is the current. So, again,
this is only a special form of the power law linking resistance to
power and current. By integrating the power over the pattern
(identical to the small dipole) over the sphere surrounding the
loop, we find

(Equation 9.3)
where A is the cross-sectional area of the single-turn loop.
Often the circumference of the loop is a more convenient
measurement. If we use the circumference rather than the
circular area of the loop, then the term C/λ becomes a
coefficient that is a simple ratio applicable at any frequency,
where C is the circumference. Thus

(Equation 9.4)
It is strikingly apparent that as the circumference of the loop
is made smaller, Rr drops at the fourth power of the ratio! For
example, a loop with a circumference of 0.1 λ will have an Rr of
about 20 milliohms. Cλ4 is often used rather than C/λ to make
clear that this is now a coefficient. By increasing the
circumference only to 0.15 λ, Rr jumps to about 100 mΩ, a five-
fold increase.
Furthermore, the dc resistance of the loop conductor (as in
any application) is not adequate for computing Rl in the
efficiency equation. Skin effect increases the effective RF
resistance of the conductor and must be taken into account
when calculating loop efficiency, even with 20 milliohms of
loop Rr.
Aside from using larger diameter wire, there are two

278
techniques for raising the Rr of a loop antenna: use more turns
in the loop (as in Figure 9.6) and/or wind the loop on a
material with higher permeability. These are precisely the
same techniques used to increase the inductance of a coil. Rr
will increase as the square of the number of turns, so we simply
add that coefficient to Equation 9.4.

(Equation 9.5)
The other technique is to use a high permeability material,
usually ferrite or iron core rods on which the coil is wound.
Thus we add one last term to the equation

(Equation 9.6)
Consequently, for small loop antennas, we can increase Rr and

279
thus efficiency by winding many turns on a ferrite rod core, as
shown in Figure 9.7.

Adding a ferrite core also increases the inductance, which

must be tuned out for an effective impedance match. Therefore
there is a limit on how big an inductor can be used in a real
design. This becomes a practical limit for the number of turns
(and efficiency) of small loops. Equation 9.6 makes clear the
primary importance of the size of the loop relative to the
wavelength.
There are also other issues with using a core material, such as
core loss. Notice that we use μer rather than the usual
specification of μr for ferrite rods. In practice μer will be
somewhat smaller than μr due to demagnetization effects.
The loop antenna is sometimes referred to as a “magnetic
dipole,” while the infinitesimal dipole is referred to as an
“electric dipole.” Small loop antennas are seldom used for
transmitting due to their inherent low efficiency. However,
under many circumstances they are very effective receive
antennas. Small loops exhibit very deep nulls, making them
useful for direction finding by establishing an azimuth bearing
from the antenna. They also provide excellent nulling of

280
undesirable signals and/or interfering noise.

(Equation 9.7)
We are now prepared to calculate the efficiency and thus the
gain and effective aperture of a small loop antenna. Rr is a
relatively straightforward calculation using Equation 9.6. On
the other hand, Rl is a function of several variables, all of which
are difficult to predict. However, Rr + Rl is a relatively simple
measurement. If we tune the loop to series resonance with a
capacitor we will measure a nonreactive total impedance equal
to Rr + Rl. Then

(Equation 9.8)
The directivity of a small loop is 1.5 or 1.76 dBi. If the loop had
no losses, its aperture would be

Therefore we can calculate the efficiency and thus also gain

and aperture by using Equations 9.6 and 9.7.
Finally, if we want to know how efficiency will change with
more or fewer turns, we simply use Equation 9.9. We begin
with the calculated and measured values of Rr and Rl. These
nominal values will be designated as Rrn and Rln. We know that
the value of Rr will change at a rate of n2 where n = number of
turns, and Rl will increase linearly by n. Thus

281
(Equation 9.9)
For example, we have a loop with 10 turns and Rrn = 0.0001 Ω
and Rln = 1 Ω. The efficiency of this antenna will be about
0.0001%, or 40 dB below 100% efficient (–38.2 dBi). Now we
want to know the efficiency if we replace the 10 turns with 100
turns, thus Dn = 10. The result is Rr is now 0.01 Ω and the new
Rl is 10 Ω. Thus the new efficiency is about 0.001%, or 30 dB
below 100% efficient (–28.2 dBi). Therefore, in the case of very
low efficiencies where Rl >> Rr, efficiency increases nearly
linearly with the number of windings.

Larger Small Loops

The coefficient

from Equation 9.6 is a great incentive to build larger loops for

greater efficiency. As we will see, efficiency for weak signal
reception is critical but other factors come into play. As
operating frequencies decrease, the usual trend is for both
natural and man-made noise to increase. On the other hand,
the desired signal strength is often considerably higher than
even the high noise levels at MF frequencies. Because of the
high noise levels at these frequencies, the most important
parameter becomes the received signal-to-noise ratio (SNR).
Many types of receiving loops have evolved over the years.
The K9AY loop shown in Figure 9.8 provides excellent over-all
performance for its physical size.

282
 The directivities of the Beverage as well as a short monopole
are shown here for easy comparison to the K9AY loop (Figures
9.9 to 9.11). Together with the plots in Chapter 7, a
comprehensive set of patterns for MF and low HF are
presented. The interpretation of these plots requires some
knowledge of noise in systems, and we will explore noise in
Chapter 10. In addition, we will present a discussion on
alternative methods of defining antenna parameters at a
system level. These techniques will be shown to be critical for
understanding the applicability of small loop antennas to
frequencies about ≤ 5 MHz.

283
284
Helix Antennas
As a small multiturn loop antenna’s size is made larger
(relative to the wavelength), other directivity characteristics
emerge, creating a helix antenna. In 1946, John Kraus was the
first to discover the remarkable characteristics of the helix
(Figure 9.12). The gain of the helix increased as the length
(number of turns) increased. It is inherently broadbanded and
is relatively easy to match to common 50 or 75 Ω feed lines. It
is also circularly polarized, making it preferable for a wide
range of application. It’s particularly useful for space
applications, where “horizontal” and “vertical” polarizations
become meaningless.

The helix is an excellent antenna with which to visualize the

subtleties of circular polarization. If we imagine the helix like
threads of a screw, we find the “left hand” and “right hand”

285
circular polarization depends on the direction of the
“threading.”
For example, when we look down on a standard thread
screw, we rotate it clockwise to tighten and counterclockwise
to loosen. If we have a threaded rod, we can turn the rod to the
other end and the same directions apply to tightening and
loosening. Helix antennas are similar in that no matter which
direction you view them from, they are either “left” or “right”
handed. A reverse-threaded screw will tighten when turned in
a counterclockwise direction, and thus has reverse threading. It
is a common mistake to assume a right-hand helix will become
a left-hand helix if it is turned over. The “threading” does not
change by flipping it over.

Small Helix Antenna

A small helix antenna may be considered identical to a small
loop antenna of multiple turns. When the loop circumference
is less than about 0.5 λ, the resulting antenna responds as a
small loop, the maximum directivity being normal to the axis of
the loop. This is defined as a “normal mode” helix. As the loop
is made larger, it begins to show a broadside pattern to the
loop, as does the 1 λ circumference quad. However, the
characteristics of the antenna become very interesting if we
treat the one-element quad as the first winding of a helix, and
begin to add more windings. This is the basis of the helix
architecture. Figure 9.13 shows the different modes of helix
antennas.

286
 As Figure 9.14 shows, the gain increases and the beamwidth
decreases with more helix turns. There is also a decrease in
bandwidth as the array becomes longer.

287
 When a circular polarized (CP) wave is reflected, its “hand”
changes. For example, if you use a right-handed helix to drive a
parabolic surface, the transmitted wave off the parabola will be
left-handed. Similarly, a left hand CP wave reflected off the
Moon’s surface will return as a right hand polarized wave.
Sometimes this is a desirable effect. For example, GNSS
navigation signals use right hand polarization. A major
problem with GNSS accuracy is ground reflections that distort
the time of arrival to the receiver from the satellite. CP
antenna inherently reject opposite polarized signals, so they
will attenuate ground reflections greatly. However, for ultra-
high performance GNSS, even the slightest ground reflected
signal can cause location errors. No “real” antenna is perfect,
and therefore a right-hand CP antenna will never have
“infinite” rejection of a left-hand signal.

288
Aperture Antennas
The parabolic antenna and corner reflector antennas are
often termed “aperture antennas.” The reflecting surfaces of
such antennas are not strictly “part” of the antenna, but rather
form passive reflection rather than taking advantage of mutual
coupling as in a true parasitic element.
The parabolic antenna derives its name from the fact that the
surface is shaped into a parabolic curve. Parabolic surfaces will
focus a plane wave onto a “focal point” where a true “antenna”
is mounted to receive or transmit power from or toward the
parabolic surface (Figure 9.15).

The antenna mounted at the focal point ideally will have a

directivity that results in a radiation pattern that uniformly
coincides with the surface of the parabolic reflector. Too much
gain will “waste” some of the parabola’s surface area (aperture),
while too little gain will expose the antenna’s response to
directions outside the parabola’s response. See Figure 9.16.

289
 The aperture of a parabolic antenna is perhaps the simplest
aperture calculation of all antenna forms. For a perfect
parabolic antenna Ae = cross-sectional area. For example, if the
diameter of a circular “dish” antenna is 10 meters, then Ae =
78.5 m2. At 1 meter wavelength, such an antenna will provide a
gain of 986.5 or about 30 dBi. A unique characteristic of the
parabolic “dish” is that the aperture is constant for all
wavelengths. However, since gain is a function of both aperture
and wavelength, the gain (dBi) of a parabolic antenna changes
as the square of the wavelength. Figure 9.17 shows the
Cassegrain antenna, a variation of the parabolic dish.

290
 The corner reflector configuration (Figure 9.18) has found
wide applications, particularly at VHF and UHF. Construction
is easier than a parabolic structure in that it is easier to build a
planar reflecting surface rather than one formed as a parabola.
Its advantages are also significant for linear polarization as the
reflector need not be “solid” but can be constructed of linear
elements in the plane of desired polarization.

291
 The feed antenna of a corner reflector is not as critical as the
feed antenna for a parabolic antenna. The corner reflector’s
driven element is typically located “inside” the reflecting
surface, thus a simple dipole can be used without the concern
illustrated in Figure 9.16.
On the other hand, the maximum gain available from a
corner reflector is limited to about 13 dBd, as shown in Figure
9.19. This is due to the fact that there is no “focal point” in a
corner reflector antenna, setting a limit on the potential
aperture (Figure 9.20). Therefore if more gain is required, then
a parabola antenna is necessary where the aperture, in
principle, can be expanded indefinitely

292
293
 The corner reflector can be modeled using the mutual
impedance methodology introduced in Chapter 8.

Small Antennas
In Chapter 1 we discussed the general trend in wireless
communication toward higher frequencies and smaller form
factors (size of device). Design problems arise when the
physical dimensions of the form factor become small with
respect to the wavelength of the wireless device(s). We also
discussed how antenna efficiency relates to antenna gain and
how both relate to the physical size of the antenna. Therefore,
a fundamental question arises: How small can an antenna be

294
made and still be effective?
In Chapters 2 and 3, some points we made hint at some
physical limitations on antenna size. In particular, as an
antenna size shrinks relative to the wavelength, Rr also begins
to drop. One may recall from ac circuit theory that efficient LC
matching circuits require greater Q to match higher ratios of
impedances. This also holds for matching from the typical 50 or
75 Ω transmission lines to the resistive portion of a small
antenna’s feed point impedance which is typically very low. A
simplified equation for the necessary Q of an LC matching
circuit is:

(Equation 9.10)
where R is the real component of the two impedances to be
matched.
In this equation, either the source or load real value can be
either the higher or lower value. If the load or source has a
reactive value, the equation becomes more complicated.
However, Equation 9.10 shows how the bandwidth of an
antenna becomes narrower as its load resistive value becomes
lower.
As the reactive-to-resistive portion of the antenna proper is
made larger, the Q of the antenna also rises. This shrinks the
bandwidth but also increases the antenna’s stored power versus
the radiated power. In Chapter 4 we described the basic modes
of electromagnetic waves as they relate to transmission lines.
These modes also begin to play a significant role in the
behavior of small antennas, particularly in the near field.
Therefore, one effect of making an antenna small is that
higher Qs become necessary and thus the bandwidth of the
antenna is reduced. The second major problem is that very low
values of Rr tend to drop the antenna efficiency as in Equation
9.1. See Figure 9.21.

295
 A “small” antenna can be defined as an antenna whose
maximum dimension(s) are confined to a spherical volume
which has a radius of
kr < 1

(Equation 9.11)
where

(Equation 9.12)
and r is the radius of the sphere, where

(Equation 9.13)
McLean quantified the minimum antenna Q (assuming zero
matching loss) for both linear and circular polarized antennas
as:

296
(Equation 9.14)
for a linear polarized antenna, and

(Equation 9.15)
for circular polarized antennas.
For example, when λ = 2 m, then k = π. If we set kr = 1
(maximum size of a “small” antenna) then r = 1/π. Therefore a
maximum size small antenna for the 2 meter band must be
confined within a sphere of radius 1/π meters or less. The
maximum size dipole that could fit inside this sphere size
would be 2/π meter (the diameter of the sphere), or 0.64 meter
long, while a full-size 1⁄2 λ dipole would be 1 meter long. Since
kr = 1, then by Equation 9.15 the minimum Q of a dipole of this
length is 2.
These limitations are derived from the assumption that equal
power is excited into the TM and TE modes (Chapter 4) in the
immediate vicinity of the antenna. A detailed analysis of these
“higher modes” requires an advanced mathematical treatment
which can be found in some of the texts referenced in
Appendix C.
As the size of the antenna decreases, maintaining efficiency
and a reasonable bandwidth becomes increasingly more
difficult. As a practical matter, impedance matching also
becomes more problematic, which adds to the overall system
efficiency. Figure 9.22 dramatizes the challenges facing
antenna designers with ever-smaller devices. However, as the
operating frequencies have increased (λ decreases) hand-held
devices can accommodate reasonably efficient antennas.
An intuitive understanding of the small antenna problem can
be taken from our discussion on vertical antennas (Chapter 7).
As the size of the antenna is reduced (compared to the

297
wavelength), the effective height he of the antenna decreases.
Again we can see this effect in

(Equation 9.16)
which is the same as Equation 7.9.
In effect, a longer length (and therefore the size) of an
antenna distributes the current over a larger distance and thus
requires less current to create the required field. As he shrinks,
larger current values (lower Rr) are required to generate the
same field strength. The same principle applies for a linear
vertical, a planar antenna, and a 3D structure.
Finally the Q of the antenna is a function of its self-
impedance. As antennas are made smaller, Rr decreases and
thus decreases the damping factor on the increasing capacitive
reactance. See Figure 9.22. Similar to circuit theory this results
in a higher antenna Q, including an increase in stored energy
around the antenna.

298
Traveling Wave Antennas
As we discussed in Chapters 6 and 7, as a vertically polarized
wave travels along the Earth’s surface, the velocity of the wave
nearer the ground is slower. The source of these waves can be a
ground-mounted vertical antenna with its wave propagating
over the ground (ground wave), or it can be a sky wave arriving
from a distant transmitter. This effect results in a tilting of the
wave “forward.” This forward tilting effect can be used to
advantage in creating a traveling wave antenna very close to
the Earth. A wire (about 1 – 4 λ in length) placed near the
ground surface and laid out in a straight line will receive waves
from the direction of the wire. Such an antenna is called a
Beverage, after its inventor Harold Beverage, who developed it

299
in the 1920s. See Figure 9.23. There are cases where a
combination of directivity, along with better efficiency is
needed, especially for very weak signal detection in the MF and
low shortwave bands. The Beverage will exhibit better
efficiency than small loops, but at the expense of laying out a
very long wire in the desired direction(s).

The traveling wave antenna differs from most other antenna

types in that the waves travel in only one direction along the
wire due to the presence of the terminating resistor. In effect,
received signals from the unwanted direction are simply
dissipated as heat in the terminating resistor. There is no
reflection of the signal as at the element ends of dipoles and
verticals. Therefore, the RMS voltage level will be constant at
all points along the Beverage.
Another important type of traveling wave antenna is the
rhombic. The rhombic is based upon directive patterns
obtained from wires about 1 λ or longer. We have already
described the patterns of shorter wires, including the extended

300
double Zepp, where broadside gain is maximum. As the wire is
made longer, the azimuth pattern quickly changes from a
“figure 8” to a “4-leaf clover” pattern (Figure 9.24 and Table
9.1).

The rhombic is constructed from four wires, each longer than

about 1 λ at the antenna’s lowest operating frequency (Figure
9.25). Opposite the feed point is a terminating resistor that
performs the identical function as the terminating resistor in
the Beverage. Without this terminating resistor, the rhombic
would be bidirectional. With the resistor, half the power (from
the rear direction) is dissipated in the resistor. Unlike the
Beverage, the rhombic efficiency is typically very high.

301
Therefore, if used as a transmitting antenna, the resistor must
be able to dissipate half the transmitter’s power output. This is
a major disadvantage of the transmitting rhombic.

Antennas for Navigation and Location

Determination
The sharp nulling effects of a small loop can also be used as
effective direction-finding and thus navigation aids. By nulling
a radio “beacon” of known location, an azimuth line may be
drawn through the point location of the beacon and the user’s
unknown location. With two beacons of sufficient directional
difference, two azimuth lines are drawn. The intersection of
the two lines yields the unknown user’s location, as shown in
Figure 9.26.

302
 Prior to the advent of GNSS (global navigation satellite
system), medium wave (1.6 – 2.0 MHz) and LF wavelengths
were used. These wavelengths have the advantages of excellent
ground-wave propagation and also small portable loop
antennas can provide the sharp nulls necessary to provide a
high precision azimuth.
More advanced navigation systems such as LORAN utilized
multiple beacons with time (and thus phase) synchronization.
These types of systems provided excellent 3D location for the
users. These older MF wave systems have been largely replaced
by GNSS (global navigation satellite system) which includes
GPS (US), GLONASS (Russia), Galileo (European), and others
that use microwave signals from satellites. The instantaneous
satellite locations are known with high precision (despite their
high velocity and altitude). Very precise location determination

303
(within centimeters) is possible using sophisticated algorithms
for error correction. However, even these very sophisticated
satellite systems rely upon simple position and timing for these
calculations.

Patch Antennas
With the nearly universal use of PC boards in radio
construction and the general trend toward higher operating
frequencies, it was only inevitable that antennas would soon be
fabricated onto the PC board. Also, low profile antennas
suitable for working off a ground plane were obvious choices
for aircraft and space applications. In response, a wide variety
of antenna forms have evolved. The simplest PC board uses two
sides of thin copper separated by a dielectric of fiberglass. Since
PC boards use at least two metal layers, transmission lines,
matching networks as well as the antenna proper can be
combined into one planar structure.
At frequencies above about 1 GHz, fiberglass (fire-retardant
FR4 material used in PC board construction) becomes a
comparatively high loss material. Therefore, for higher
performance RF circuits, other substrate materials are often
used. However, some compromised performance is often
accepted to keep manufacturing prices down.
At first, simple dipoles were etched onto the PC board,
followed by more complex geometric forms including spirals,
planar “patches,” and even fractal shapes. The various form
factors may either be placed above a ground plane or emulate
“free space” by keeping other metal traces away from the
antenna proper, or some combination of both.
An alternative approach has been developed that permits the
use of fiberglass PC board by creating a separate antenna
structure using much higher performance ceramic as the
antenna’s substrate (Figure 9.27). The ground plane is simply
included on the PC board’s top layer, thus making the
shortcomings of fiberglass mute. A rectangular, circular or
other shape metal plane with small spacing to the ground
plane forms most of configurations in use. The above discussion
on small antennas is directly applicable to patch antennas

304
since they are often quite small relative to the wavelength in
use.

PADL parameters become very complex since the patch uses

two-dimensional planes rather than linear wires or thin
cylinders for elements. Both transverse electrical and magnetic
fields appear between the patch and the ground planes as well
as outside the dielectric.
The patch antenna has gained wide acceptance for a variety
of applications from UHF well into the microwave spectrum.
They are typically mounted on a conducting ground plane and
thus form an approximation of a hemispherical pattern.
Depending upon the feed point within the structure, either
linear or circular polarization may be achieved. Because of
their small size and relatively good efficiency, they find wide
application in handheld equipment, particularly for GNSS and
UHF RFID applications.

Phased Array
In its simplest form, a phased array consists of multiple

305
antenna elements (sometime hundreds of individual antennas)
distributed over a plane (Figure 9.28). The resulting array may
be steered to any direction above the plane by changing the
phase and amplitude relationships among the elements.

If all elements are fed in phase, the pattern is broadside to

the array or “straight up.” If we progressively delay the phase to
elements on the left as in this figure, we can establish
directions that are dependent upon these phase relationships.
Changing the phase relationships among all the elements to
any desired phase delay/advance is difficult as the number of
elements grows. With the introduction of digital signal
processing, an alternate method of steering becomes possible.
We use separate receivers for each of the elements, as shown in
Figure 9.29. The receivers use the same local oscillator so that
the signals from each receiver will be phase-locked (identical),
but the location (as in PADL) is different. After analog-to-
digital conversion we feed the digitized signals into digital
phase shifters. Therefore the beam steering is accomplished at
the digitized IF frequency. In addition, multiple directions may
be received by simply increasing the number of digital phase
shifters. In Figure 9.29 three simultaneous directions may be
realized.

306
Aperture Distribution
No matter what the antenna configuration, sidelobes begin
to appear in the main pattern as the aperture becomes greater
than about 1.5 λ. See Figures 9.30 to 9.32 for the following
discussion.

307
308
309
 The simplest example is to plot the resulting pattern along
one axis as a result of the aperture distribution along that axis.
The aperture can consist of multiple individual antennas (as
with a phased array) or a large single aperture (as with a
parabolic antenna). Again, the calculation uses PADL for its
solution, but with a rather sophisticated pair of integral
functions.
The calculation of the sidelobe response from the aperture
distribution utilizes a Fourier transform. The Fourier transform
is a special form of integral that is extremely useful for a wide
variety of engineering problems. In electronics it can provide
the spectrum (frequency domain) of a signal captured in the
time domain. Time domain displays are most often shown on
oscilloscopes, while frequency domain displays are shown on
spectrum analyzers. If you have time domain data (for example

310
a voltage versus time) the spectrum of that signal (frequency
versus voltage or power) can be calculated using the Fourier
transform. On the other hand, if you have spectrum data you
can use the inverse Fourier transform to derive the time
domain signal. The simplest example is the Fourier transform
of a pure sine wave to a single spectrum line, and vice-versa.
Readers who are familiar with transitions between time and
frequency domains may recognize the similarity of the
relationships shown in Figure 9.30. If the uniform aperture
distribution shown in Figure 9.30A, were a “square wave” on an
oscilloscope, we would expect to see Fourier series of the
fundamental frequency and a series of odd-order harmonics
(3rd, 5th, and so on) decreasing at a

rate. This is exactly the decreasing rate in the sidelobe power

shown in Figure 9.30A.
A second example is the Gaussian distribution, which is the
unity function for the Fourier Transform. For example, if we
have a time domain signal that traces a Gaussian curve on the
oscilloscope, the frequency distribution of that signal will be
the same Gaussian spectrum. Thus the Fourier and inverse
Fourier transforms of a Gaussian function are the same
Gaussian function relationship as shown in Figure 9.30E.
The trace in Figure 9.30F shows the sidelobe response of an
inverse taper in the aperture. Such aperture distortion occurs
when a large focal point structure is placed over a parabolic
antenna. Thus careful design of the aperture distribution must
be taken if sidelobe response is to be minimized.

Polarization Diversity
In principle, an infinitely thin perfectly horizontal antenna
will have zero response from an infinitely thin, perfectly
vertical wire in free space. Since linear wire or tubular antenna
elements have non-zero diameters, there is some response

311
from one such antenna to the other. Furthermore, reflections
from the Earth and/or from the ionosphere often change the
polarization characteristics of the radiated wave.
This difference in response between polarizations can be put
to use in transmitting one channel of information using
vertical polarization and a second channel using horizontal
polarization but using the same frequency band
simultaneously. This technique has the potential of doubling
the capacity of the frequency channel. Left hand circular
combined with right hand circular can provide the same effect.
In such systems additional nulling can take place in the
receiver to offer better polarization isolation since both
polarization signals are received as shown in Figure 9.33. Since
the error signals bleed-over from the other polarization, we
can simply sample the undesired signal, correct for amplitude
and phase and provide additional nulling.

MIMO Antennas
“MIMO” is an abbreviation for “multiple input, multiple
output.” Consider the system is Figure 9.34. The transmit
power is divided between two antennas of some distance apart.
The receiver receives these two signals on two receive
antennas also separated by some distance. Therefore there are
now four possible paths for the signal to travel thus greatly

312
increasing the probability that the system can avoid multipath
problems. A second advantage is that since there are different
time delays among the four separate paths, it is possible to
transmit different information along the separate links, or a
combination of the two.

Other Configurations
The above examples of specific antenna configurations are
but a small sample of current development work in antennas.
The last few examples point to a progressively closer
relationship between the antenna and the receiver. The
development of digital signal processing and software defined
radios (SDR) has opened new frontiers for improving antenna
performance. In particular the difficulty of dealing with
expensive passive components for phase and amplitude
adjustments can now be performed using inexpensive yet very
powerful digital processors. In effect, critical adjustments to
antenna parameters are now being performed in the digital
domain. However, the physics of the antenna proper remains
constant.

313
 CHAPTER 10

Noise, Temperature, and

Signals
Thus far this text has presented some of the key concepts of
antenna physics. The physics of antennas sets physical and
theoretical limitations (and opportunities) for radio links. The
purpose of this final chapter is to link antenna performance to
the overall theoretical limitations of the radio link.
Of critical importance is the omnipresence of noise. The first
part of this chapter deals with the origins of noise and how
these mechanisms relate to the RF spectrum. Second, we
discuss how noise affects the radio link. Third, we introduce a
new parameter for antennas: antenna temperature. Aside from
the electromagnetic noise limitations to a radio link, there is
one other critical physical parameter that, together with the
electromagnetics, completes the physical limitations of a radio
communication system: information science. This will
complete an introductory treatment of all the theoretical
factors that place physical limits upon a radio communication
link. A discussion of medium-frequency (MF) atmospheric
noise serves as a practical application of many concepts treated
earlier. As an aside, we will show how an antenna was the
indispensable sensor for the first direct detection of the
beginning of the universe, the Big Bang.

Boltzmann’s and Planck’s Constants

Noise is a type of waveform present in all communication
systems. It also sets fundamental limitations upon all
communication links. Therefore an understanding of what
causes noise and how it imposes limitations can set the stage
for dealing with it most effectively.
For radio systems, noise has numerous sources, and each can
become the limitation for a given application. We begin the

314
discussion by looking at thermal or “Boltzmann” noise. We will
see that thermal “radio noise” is closely related to temperature,
as its name implies, but this relationship is often confusing,
particularly when its effects upon communication systems are
considered.
Thermal noise originates from the fact that any atom or
molecule with a temperature above 0 Kelvins or 0 K (absolute
zero) will vibrate. The classical definition of absolute zero (0 K)
is the point where all atomic vibration stops. Some
experiments as of this writing (2014) have achieved
temperatures only a small fraction of a Kelvin higher than
absolute zero. (Note that unlike the Fahrenheit and Celsius
scales, Kelvins is the unit of measure for this scale — without
“degrees.” So while we would normally write 50 °F or 50 °C,
when using the Kelvin scale, 50 K or 50 Kelvins is the correct
notation.)
The absolute position of a vibrating atom is impossible to
predict at any given instant. However, the distance from a
“norm” position is a matter of probability. This physical
vibration of atoms also has another effect: it produces a
random electrical voltage. Since the voltage is random, we call
it “noise.” If a voltage is “random” it must therefore also be
“alternating.” Since this is a random alternating voltage, we can
also deduce that it is broadbanded and (to the point of this text)
extends into radio frequencies. Furthermore, we recall from
Chapter 3 that radiation can only occur when an electrical
charge is accelerated. Atoms contain electrons (charges) that
therefore also vibrate, which in turn is a random acceleration
of charges, which creates a random electromagnetic radiation.
A random RF voltage is produced within a resistor at some
temperature above absolute zero. This random voltage will be
conducted and thus appear across the terminals of the resistor.
The source of this RF voltage will also radiate, thus causing
thermal radiated noise according to the dynamics of the
antenna PADL (Phase, Amplitude, Direction, and Location —
introduced in Chapter 3) of the resistor.
As we might suspect, the random movement of atoms will
increase with increasing temperature. This is indeed the case

315
and was first quantified by Max Planck (Figure 10.1). In 1900
Planck introduced the concept of black body ​radiation. This was
follow-on work of Ludwig Boltzmann (Figure 10.2), who, 23
years earlier had linked entropy (from theory of gasses) and
probability. The mathematics dealing with probability
functions is of primary importance in the study of quantum
mechanics. In turn, black body radiation derives its theoretical
basis in quantum mechanics and is also described in terms of
probability functions.

316
 As temperatures increase, the total power emitted from a
black body increases very quickly, proportional to T4! The
spectral distribution of this power is not linear across the
electromagnetic spectrum, but limited to RF and microwave
frequencies a linear relationship can be assumed. The peak
power typically occurs well into infrared wavelengths and
shorter. Even for very low temperatures near absolute zero, the
noise temperatures at RF frequencies change linearly with
temperature defined by Boltzmann’s constant. For example, the
peak spectral power of a very cold 3 Kelvin black body is
centered at about 160.2 GHz. Although this can be considered a
very short microwave wavelength, technically it is still in the
RF spectrum. So for radio noise considerations, the Planck
constant need only be considered at superlatively very high
(RF) frequencies and measuring at very low temperatures. The
general equation across the entire electromagnetic spectrum
for thermal noise is:

(Equation 10.1)
where h is Planck’s constant, 6.626 × 10-34 J-s (Joule-seconds)
and k is Boltzmann’s constant (given in Equation 10.3). It was
Planck who first calculated (within a few percent) the now-
known value of Boltzmann’s constant.
Fortunately for radio engineers, when f is made
comparatively very low (RF frequencies), this equation
simplifies to a much simpler linear equation using Boltzmann’s
constant. When an object is very hot, the peak frequency
becomes very high. For example, the Sun’s highest power
spectral output occurs in the visible light spectrum. The Sun is
about 5000 K. If it were made hotter, the peak power would
increase, and the frequency of maximum output would also
increase.
Figure 10.3 shows thermal noise power plotted at four
different temperatures. Note that Figure 10.3 plots black body

317
radiation. The Sun, for example emits higher power on the
spectral lines of hydrogen and helium, which make up the bulk
of the Sun’s mass. Few “real” objects emit a “pure” black body
spectrum. If we look closely at Figure 10.3, we see the noise
power (Y-axis) converges into a mostly linear function at lower
frequencies (RF). Therefore, for the majority of considerations
concerning RF, the complicated general equation converges to
a much simpler equation:

(Equation 10.2)
for RF frequencies, where T is temperature (Kelvins), B is
bandwidth, and k is Boltzmann’s constant:

318
(Equation 10.3)
for 1 Hz of bandwidth at a rate of 1 second, where J is energy in
Joules, and K is the Kelvin temperature. It is interesting to note
that k is an expression of energy/temperature. How do we
derive a power term from kTB?
“Frequency” was defined by “cycles per second” (cy/sec.) until
the ISU (International System of Units) redefined “frequency”
as “Hertz” where today

(Equation 10.4)
The old “cycles” are now dimensionless and therefore simply
become the coefficient for Hertz — and thus in a noise
calculation, also power. So if

(Equation 10.5)
and J = W × sec (energy = watts × seconds), T = absolute
temperature, and B (bandwidth) = X/sec, therefore,

(Equation 10.6)
Thus kTB is in watts, with temperature (T) and bandwidth
(X/sec). Also, W × sec = energy.

Boltzmann Noise in Circuits

During the 1920s, John Johnson at Bell Labs measured a
“noise voltage” across a resistor that was not connected to any
power source. Harry Nyquist (Figure 10.4), who was also
working at Bell Labs, was able to explain the mechanism of the
source of this thermal noise by linking it to Planck’s constant.
Suitably, “resistor noise” is today referred to as Johnson-

319
Nyquist noise. Like black body radiation, noise voltage is
directly related to the absolute temperature (Kelvin) of the
resistor and the value of the resistor, where:

(Equation 10.7)
This is the noise voltage value in 1 Hz of bandwidth produced
by a resistor (R) at an absolute temperature (T). To derive the
noise power we can use the simple equation:

(Equation 10.8)
or

(Equation 10.9)

320
or

(Equation 10.10)
for power in 1 Hz of bandwidth.
Adding the bandwidth term, we get

(Equation 10.11)
There are two important points to be made regarding this
equation for thermal noise:
1) In the noise power equation the R term cancels, indicating
that the noise power generated in a resistance is independent of
the resistor value, whereas noise voltage and current values are
dependent upon the actual R value.
2) The coefficient 4 is a common source of confusion. This is
the voltage seen across the thermal resistor source. However, if
we want to transfer the power to an external matched load as
in Figure 10.5 the power becomes:

(Equation 10.12)
By the following derivation (from Equation 10.7):

321
Then the voltage found across R2 is

(Equation 10.13)
or,

(Equation 10.14)
And, simultaneously adding the bandwidth correction

(Equation 10.15)
This is the derivation of Boltzmann’s noise.
A final point which is very important when considering noise
in circuits and antennas is that although the power generated
in a resistor is independent of the resistor value, the noise
source will have an impedance equal to the value of the source
resistance. Figure 10.5 shows a Thevenin equivalent circuit
which is a good approximation for source impedances of finite
value. However, in the case of near-zero or zero ohm source
impedances in thermal noise modeling, the signal noise source
is actually a short circuit. Thus the very low resistance noise
source provides a “short” across the input impedance of the
load, and no power is transferred. In effect, the random
thermal noise power generated in a non-connected resistor
simply re-dissipates itself as heat. Therefore we have a noise-
voltage/current-noise cycle inside all resistors. So, a simple
carbon resistor sitting on your bench contains a frenzy of
electromagnetic activity! Extracting a portion of that noise
power is another matter. As we will see, for antenna noise
analysis, this source impedance is simply equated to Rl for

322
efficiency and noise calculations. For circuit noise calculations,
including amplifiers, the mathematics becomes significantly
more complicated.

Random Processes
Equation 10.12 provides a simple definition of thermal noise,
either radiated or conducted. At first glance we might expect to
measure a constant voltage or power due to Johnson-Nyquist
noise. This is not the case since thermal noise is created by a
random process and thus the instantaneous power value must be
governed by the laws of probability. Equation 10.12 defines a
mean value. The instantaneous value of a black body radiator is
defined by a Gaussian distribution.
Note that the theoretical probability curve never quite reaches
zero even for an infinite power level! Using the Boltzmann
constant we can compute an actual mean power level at any
temperature. This is the reason that Boltzmann’s constant is
represented as an energy term. The energy term, by virtue of its
time integration, also finds an average power over time. In this
case it represents the average of the Gaussian function over
time, or the peak of the Gaussian probability curve as in Figure
10.6.

323
 At room temperature (290 K), the power in 1 Hz bandwidth is

(Equation 10.16)
in 1 Hz of bandwidth.
This is about 204 dB below 1 W. We have found that antennas
are usually specified in dBi or dBd. These are related to power
gain and express ratios, not absolute power levels. When we get
into radio systems engineering, we need to specify real power
levels. In Chapter 7 we introduced such an absolute power term;
now we will formalize it relative to RF system engineering. The
usual convention in radio engineering is the use of dBm, which
uses the reference:

324
0 dBm = 1 milliwatt (mW)

(Equation 10.17)
Therefore, Boltzmann noise at 290 Kelvins, also +16.85 °C
and +62.3 °F (or “room temperature”) is –174 dBm/Hz. The
noise power will be directly proportional to the bandwidth,
therefore we can simply add the term
10 log (B)

(Equation 10.18)
to derive
Pnoise = –174 dBm + 10 log (B)

(Equation 10.19)
at 290 K, where B is the bandwidth in Hertz.

Radio Receiver Sensitivity

There are two more terms needed to derive the critical
specification of receiver sensitivity. Any radio receiver will
generate its own noise. The process of amplification,
attenuation and mixing all contribute to the receiver noise
figure usually expressed as NF (dB). The noise figure (specified
in dB) is the log of the noise factor (F), which in turn is the ratio
of the input to output signal-to-noise ratios (SNR). Therefore the
noise factor is a ratio of two ratios. The noise factor of a circuit
(active or passive) is simply defined as the degradation of the
signal to noise ratio when a signal passes through the device.
Therefore,

(Equation 10.20)

325
for actual power.
Noise figure is defined in dB. Therefore, for the NF expressed
as power,
NF (dB) = 10 log (F)
(Equation 10.21)
For example, a “perfect amplifier” will not degrade the SNR of
the amplified signal, therefore, NF = 0 while the F is 1. When we
refer to a “device” we are referring to amplifiers, mixers, filters,
and other components of a radio. Other sources also must be
considered, including noise power that is received by the
antenna. Consequently, we can divide “noise” into two basic
sources: external to the receiver, and internal to the receiver.
Both must be considered in the calculation of the quality of a
communication link. However, the first step is defining the
receiver noise figure.
Finally, the minimum performance level of a receiver requires
some level of SNR for audio, video, or digital bit error ratio
performance. In effect, the quality of a link is directly related to
the ratio of signal power (or energy) to noise power (or
energy). In analog systems, we usually refer to the SNR, which
implies a power ratio. However, in digital systems it is more
appropriate to specify an energy ratio since the symbol rate of a
digital communication system implies that the symbol value
(amplitude, frequency, and/or phase) stays constant for the
time that symbol is transmitted.

(Equation 10.27)

(Equation 10.28)
where rs is the symbol rate and rbit is the bit rate. Thus, a

326
receiver performs a time integration (Tsymbol), which results in
an energy value in contrast to an analog signal’s power value
SNR.

(Equation 10.29)
where Eb is the “energy per bit” and No is the noise energy.
Therefore, at room temperature receiver sensitivity is defined
as

(Equation 10.30)
The conversion factor between SNR and Eb/No is simply:

(Equation 10.31)
where Rb is the channel data rate (bits/second) and B is the
channel bandwidth.
To summarize, Equation 10.30 simply states that at 290 K, we
first calculate thermal noise power per Hertz, then correct for
the bandwidth of the channel, degradation from the receiver
NF, and finally the required ratio of signal to noise (power or
energy) to maintain a predefined quality of the link. This is the
definition of receiver sensitivity.
For example, let us assume that we have a system that
requires 1 MHz of bandwidth, the noise figure of the receiver is
5 dB, and the required SNR is 10 dB. What is the sensitivity of
the receiver?
Sensitivity = –174 dBm + 60 dB + 5 dB + 10 dB = –99 dBm

327
(Equation 10.32)
This receiver requires a –99 dBm minimum of signal power
assuming that only thermal noise is present at the receiver
input.

Calculating a Minimum Antenna Gain

Specification
In Chapter 2 we described the physics of a radio link. Now we
can use that physics coupled with a radio sensitivity
specification to calculate how much antenna gain will be
required to span a free-space distance using some restricted
transmitter power. See Figure 10.7 for the following example.

Let us assume that we need –99 dBm of power to provide the

minimum signal level at the receiver input. Let us also assume
we have a transmitter that provides 1 W output (+30 dBm).
Through a simple calculation (–99 – 30 dBm = –131 dBm) we also
know that we can lose 131 dB between the transmitter output
and the receiver input. This is termed the total link budget and
like all budgets, we can scrimp and save some places and
splurge in others. However, at the end of the day we have to
stay within our budget.
Now we also know that we are using a frequency of 1200
MHz, or a wavelength of 0.25 meters. We need to link two
mountaintop sites 100 km apart, but we have a free-space link.

328
We also assume that we have 5 dB of transmission line loss at
both sites, for a total of 10 dB loss. How much total antenna
gain will we need to provide minimum quality to the link?
From Equation 2.6 in Chapter 2 we know that the antenna-
to-antenna link budget (assuming isotropic antennas at both
ends) to be:

(Equation 10.33)
Now we can plug in the numbers and calculate the power
received:

(Equation 10.34)
After we subtract 10 dB from the transmission line losses, we
find –114 dBm at the receiver, or 15 dB lower than we require.
We can gain this 15 dB by adding a 7.5 dBi gain antenna at each
end; or a 10 dBi antenna at one end and a 5 dBi antenna at the
other; or use lower loss transmission line with lower gain
antennas. Any of these changes will result in an increase of 15
dB at the receiver. We could also increase the power of the
transmitter, or use any combination of these approaches. This
rather complex problem simply is a matter of adding and
subtracting decibels to the actual power (dBm) levels of
Boltzmann noise, transmit power, and necessary received
power. All the other terms are simply dB loss or gain, but we
begin with dBm (transmit power) and end with dBm (signal
power) at the receiver input.
Now suppose we have transceivers at both ends of the line,
and we assume both are identical to the transmitter and
receiver specifications stated above. Since links are reciprocal
we will find the power to the receivers at both ends to be

329
identical.

Noise Temperature
In the above example, we assumed that the noise at the input
to the receiver was –174 dBm/Hz since the assumption was a
system temperature of 290 K. Therefore NF is a temperature-
dependent value related to the reference of 290 K. Suppose by
Equations 10.20 and 10.21 we measure and calculate a NF of 3
dB, the F is 2 and let’s define the gain as 10 dB. Let us drive the
amplifier with a 1 W signal with an accompanying 1⁄4 W noise.
The output will contain 10 W of signal, but 5 W of noise. We
can see the input SNR is 4/1 and the output SNR has been
reduced to 2/1, therefore F = 2 and NF = 3 dB. We have assumed
an input noise temperature of 290 K and the amplifier has
contributed an equal amount of noise (before amplification).
Therefore the noise temperature of the amplifier must also be
290 K.
For NF or F, we can calculate the equivalent noise
temperature of any device by
Te = T0 (F – 1)

(Equation 10.35)
where
Te is the equivalent noise temperature
T0 is 290 K (since F is a function of 290 K)
F is the noise factor
Good quality low-noise amplifiers can exhibit noise figures
approaching 1 dB, even operating at 290 K. This corresponds to
a noise factor of 1.26 and by Equation 10.35 an equivalent noise
temperature of about 75 K. Thus the noise temperature of an
amplifier may be considerably below the ambient temperature
at which it is operating. This very beneficial but somewhat
paradoxical feature is explained by relationships in
semiconductor physics.
Since NF is taken as a log function of F, we must use an
inverse-log function to derive the NF.

330
(Equation 10.36)
We can also lower the Te and thus also the F of an amplifier
by physically cooling it down. The relationship is linear where:

(Equation 10.37)
Here we will begin with the same amplifier as above with a NF
of 1 dB, or a Te of 75 K. If we cool the amplifier down to the
liquid helium temperature of 4.22 K, we derive a new Te for the
device of 1.09 K. In Equation 10.37, our Te290 is 75 K. (The Te of
the amplifier at 290 K.) Thus the Te4.22 is a very low 1.09 K.
The effort of cooling an amplifier to such extreme low
temperatures is worthwhile is some specific applications which
we will soon address. With this background information we are
ready to describe an antenna specification critical to many
system-level design concerns.

Antenna Temperature
A resistor can be used to measure temperature by simply
measuring its noise voltage by Equation 10.7. However, a
resistor cannot measure temperatures below its own physical
temperature, while it is possible for an antenna to do so. In a
“perfect” antenna the “resistance” is radiation resistance (Rr),
which cannot be specified by an ambient temperature. The
concept is quite simple. If we measure the power level (dBm)
being received by an antenna, then we can calculate an
equivalent temperature using Boltzmann’s constant as the
relating term. However, now the noise power source is the Rr
of the antenna instead of a “real” resistor generating Johnson-
Nyquist noise. Rr cannot have a physical temperature.
Therefore in a “perfect” antenna there can be no Johnson-

331
Nyquist noise produced by the antenna itself as with a resistor.
Antenna noise temperature is only a function of the RF noise
power received by it. And the noise power is only a function of
the equivalent temperature of the matter in the antenna’s
steradian gain response. This “temperature” is coupled to the
antenna by the antenna’s Rr. The “matter” might be the ground,
the moon, or a distant galaxy. Here again, the voltage received
in 1 Hz bandwidth will be:

(Equation 10.38)
However, in this case, T is the equivalent temperature of the
noise/signal being received by virtue of the antenna’s Rr, not
the physical temperature of Rr. Therefore,

V2 = kTRr

(Equation 10.39)
and since
V2 = PRr

(Equation 10.40)
then

(Equation 10.41)
and

(Equation 10.42)

332
 Thus the noise power (dBm) received by an antenna is
directly proportional to the antenna temperature in Kelvins.
Note: this assumes that the antenna is 100% efficient (lossless).
In the case where the antenna is not 100% efficient, we must
consider the relationship between the radiation resistance and
the loss resistance Rl (actual temperature of the antenna’s loss
resistance). This is similar to the efficiency calculation of
ground-mounted vertical antennas in Chapter 7.
See Figure 10.8. We have already explained the mechanism
for a simple noise measurement of a resistor (Figure 10.8A). If
we place an antenna whose radiation resistance is the same as
the resistor in Figure 10.8A in an anechoic chamber at the same
temperature, we will measure an identical voltage and thus
power level (Figure 10.8B). Finally, in Figure 10.8C, we point the
antenna at a part of the sky with an equivalent noise
temperature and again, we read the same noise power.

In the third example above, we can see how an antenna can

become a remote temperature sensor. This is somewhat of a
peculiar circumstance. The distance to the “sky” temperature is
cancelled by the antenna aperture, thus neither affect the antenna
temperature! This is provided that the portion of the sky
(measured in steradian solid angle) has uniform thermal
temperature over the beamwidth of the antenna. Until this
point in this text we have assumed that a distant signal source
is a point source. With noise as the source, we must now
consider it as a source with some steradian “area” of the sphere.

333
 For example, if we use a 20 dBi gain antenna, we are looking
at 1⁄100 the area of the full spherical sky and measure X Kelvins.
Now we replace the 20 dBi antenna with a 23 dBi antenna, so
we are only viewing 1/200 of the spherical sky. Therefore we
might conclude that we are only receiving 1⁄2 of the thermal
power, therefore the temperature should be 1⁄2 X. However, we
just increased the antenna gain by 3 dB so the smaller area’s
power is now increased by 2. This leaves us with a temperature
again of X Kelvins (an identical noise power)! Also, the distance
(even to the edge of the universe) does not matter by the same
cancellation. Of course the sky temperature is not constant, so
if within the beamwidth of our antenna there are different
source temperatures, we need to adjust our measurement
technique. However, in many directions, sky temperature is
constant, even for lower-gain antennas.
Now let us assume that the antenna is not 100% efficient. If
the antenna is receiving the equivalent of 50 K of thermal noise
and the antenna is 99% efficient operating at 290 K, we can
assume that this resistive loss (due to antenna inefficiency) will
contribute thermal noise, and thus raise the antenna
temperature. Thus in this example antenna noise would be:

(Equation 10.43)
or, about 53 K.
Therefore we suffer a 3 K rise in noise temperature due to the
antenna’s inefficiency. The equivalent would be to place a
resistor with a value of 1⁄2 Ω at 290 K in series with a radiation
resistance of 50 Ω at 50 K. Unfortunately, if we point the
antenna toward the Earth, we will see an antenna temperature
close to 290 K, the ambient temperature of the Earth. However,
if we point the antenna at the sky, we must also minimize all
thermal radiation from the Earth and keep our system noise
temperature very low. Only then can our antenna become an
extremely sensitive heat detector for man-made and natural

334
objects. Even a small sidelobe response to the 290 K Earth will
overwhelm an attempt to measure very weak extraterrestrial
thermal noise sources.
However, if we know Rr and Rl of the antenna (usually an
easy calculation), we can subtract the contribution from Rl and
determine the temperature of a remote object, or even space
itself. But if we point our antenna to “empty” sky, it is
reasonable to assume the antenna should record zero power
since there is no matter emitting thermal radiation from
“empty sky.” This would be true in a “static” universe, but our
universe not static, so what will empty sky provide?
Figure 10.9 shows the first radio map detailing the
temperature of the sky. This map was made by Grote Reber,
W9GFZ (Figure 10.10), who used the home built antenna
shown in Figure 10.11 for his early radio astronomy
experiments.

335
336
A Nobel Prize
Over a period between 1922 – 1923, Edmund Hubble observed
that several celestial objects were actually outside the Milky
Way, eventually overturning the previously held scientific
orthodox that the universe was constant and limited to the
Milky Way. Further observations by Hubble and others showed
that the universe consisted of countless numbers of galaxies all
moving away from each other, some at near-light speed. This
observation led to the theory of the Big Bang. If all objects are
moving away from each other, then they must have had a
common point of origin. And, if that much matter was in one
“place” at one “time,” then the universe must have begun with a
Big Bang, much like a bursting fireworks explosion expels its
fragments away from the point of explosion in three-

337
dimensional space. Observing the rate of expansion and the
size of the universe, the age of the universe was/is calculated to
be about 13.7 billion years.
One of the implications of such a violent event was that
residual radiation should remain in the universe in the form of
black body electromagnetic radiation (including RF). Through
the mid-20th century, many predictions and spirited debates
took place regarding this topic. Most of the physicists
concluded that this residual noise would have a temperature in
the range of 2 – 5 K with the most likely range being 2 – 3 K. If
this residual temperature of the universe could be found it
would all but prove the Big Bang theory to be valid.
Then in 1964, two engineers, Arno Penzias and Robert
Wilson (Figure 10.12) were working on a project at Bell Labs to
reduce noise in microwave radio links. They became puzzled
by a minimum 3 K noise temperature no matter which way
they pointed their 6.2 meter horn-reflector antenna at 4 GHz.
They would later find that they had discovered the residual
energy from the Big Bang and received the Nobel Prize for
their discovery. We now know that the actual temperature is
2.7 K and the noise is in near-perfect conformance to the
theoretical black body frequency distribution predicted by
Planck. Thus an antenna was the indispensable tool for
observing the first direct evidence of the beginning of the
universe: microwave radiation from the Big Bang.

338
 Penzias and Wilson found that the absolute minimum noise
an antenna can detect is 2.7 K, which i s about –194 dBm/Hz.
This will be true of any antenna located anywhere in the
universe and pointing in any direction! It also sets a
fundamental limit to background noise.
This measurement may seem a bit puzzling. Since the horn
antenna used by Penzias and Wilson was Earth-bound, the
antenna was pointed at the sky through the atmosphere. Since
the atmosphere is considerably warmer than absolute zero, one
might expect that they would have observed a much higher
value of thermal noise. In fact, the atmosphere is a “thin gas”
that doesn’t radiate (or absorb) black body radiation in the
same manner as “true” black bodies. Once again, the answer to
this dilemma lies in the laws of quantum mechanics. For our
purposes, the atmosphere does not emit black body radiation
over the frequencies investigated by Penzias and Wilson,
allowing the above measurements to be made. In contrast the
atmosphere has an extremely high black body temperature at
much lower frequencies.

Atmospheric and Galactic Noise

339
 There are countless objects in the observable sky that emit
much higher temperatures. If the antenna is pointed at these
objects, higher noise powers will result. Figure 10.13 is
provided to illustrate Earth-bound sky noise as a function of
frequency. By using Boltzmann’s constant the actual dBm/Hz
values can be easily computed.

As shown in Figure 10.13, atmospheric and galactic noise pose

a basic limitation to communication systems at frequencies

340
below about 1 GHz. The air molecules in the atmosphere are
constantly being bombarded by electromagnetic radiation and
particles from the Sun. The effects are twofold: ionization of
the atmosphere (particularly at higher atmospheric elevations)
and electrical charging. Furthermore, rapid vertical movement
of liquid particles (particularly water droplets) in the
atmosphere adds to electrical charging. The most dramatic
electromagnetic noise generation is lightning caused by sudden
discharging of highly charged clouds.
The Milky Way Galaxy (our home) consists of billions of stars
and a myriad of other fascinating celestial structures. Many of
these structures emit radiation at radio frequencies as well.
Detailed study of these sources of radiation began after WWII
and became known as radio astronomy, which has clearly added
very significant data to the understanding of a wide variety of
celestial objects as well as fundamental discoveries in
cosmology.
Figure 10.13 also shows a minimum of natural noise occurring
between about 1 and 10 GHz. This is the range of frequencies
where most SETI (search for extraterrestrial intelligence)
activity has been conducted, since this band provides the
minimum sky noise for Earth-based observations. Therefore if
“they” were interested in interstellar communication they
might use these same wavelengths below about 10 GHz.

Photons in Radio Waves

We have already stated that electromagnetic waves are both
“waves” and “particles.” The particle is called a photon. When
the photon has enough momentum, dislodging electrons
becomes possible — thus creating an electric current as in a
photoelectric or “solar cell.” Although defining the photon
relative to radio waves is usually not of practical value, a brief
description is presented for completeness. Again Planck’s
constant is used to calculate both the momentum and the
energy of the photon.
The energy (in Joules) of a photon is linearly proportional to
the frequency of the wave.

341
(Equation 10.44)
where h is Planck’s constant as in Equation 10.1 and f is the
frequency in Hertz. This is the amount of energy contained by
one photon. If we should ever need to know how many
photons we are emitting per second, we simply divide our
output power by the power/photon. Keep in mind 1 Joule = 1
watt-second (1 J = 1 W-s), and to keep terms in order, remember
that h = Js and f = 1/s, therefore we end up with Joules, an
energy term.

(Equation 10.45)
For example if we are transmitting 100 W at 144 MHz, we are
transmitting about 1027 photons/second. Notice that as we raise
the frequency but keep the power constant, we transmit fewer
photons.
When we consider the momentum of a photon, classical
Newtonian physics once again breaks down. “Momentum” in
classical physics must have mass, but a photon has no mass.
However, a photon must travel at the speed of light or it loses
its energy. Any “thing” that has no mass and no energy is a
“nothing.” The value of a photon’s momentum is the same
relationship for its energy (hf). However, since momentum is
usually stated as a direction as well as a function of energy, it is
usually expressed in vector notation.
This result is also a bit peculiar in that the power of a radio
wave can be calculated from its field strength or by the number
of particles/second. The answer, in watts, will be identical. This
is the essence of Einstein’s wave-particle duality of
electromagnetic radiation. The power of electromagnetic
radiation is not “shared” between the “wave” and the “particle,”
but coexists in both simultaneously.

342
Atmospheric Noise Considerations at MF
and Low HF
Under most circumstances, RF power at these relatively long
wavelengths does not penetrate the atmosphere. This is true
for both extraterrestrial galactic noise intersecting the Earth
and terrestrial noise propagating toward the sky. In both cases
the power is either absorbed and/or reflected. On the Earth we
observe this noise as thermal noise coming from a huge
equivalent noisy resistor (the atmosphere) under a huge
reflector (the ionosphere). However, the noise problem at
these frequencies is considerably worse than the Earth’s 290 K
thermal noise.
Figure 10.13 dramatizes the noise problem at these
frequencies. Figure 10.14 provides a more detailed plot of
atmospheric noise at MF and HF frequencies with the “noise
floor” on the graph being the noise temperature at 290 K.
Table 10.1 shows the relationships among noise power levels,
equivalent noise temperature, and the S unit equivalent.

343
 The solid black line in Figure 10.14 shows the maximum
natural atmospheric noise. The natural atmospheric noise level
dominates the over-all noise power when it is at its maximum
and may be taken as the total contributor to the noise level.
Lightning is the major contributor, the equivalent of hundreds
of ultra-high-power spark gap transmitters keying at hundreds
of times per second. When propagation permits, this massive
power will appear at a receiver hundreds or thousands of
kilometers away. In effect, the actual nighttime noise level at a
given location at MF will be the grand sum of lightning
occurring over the Earth’s dark hemisphere and the quality of
propagation to these areas. This is the mechanism that creates
a constant “static” from the sum of thousands of strikes/second.
The dotted line shows the minimum natural noise level.
Before the spread of electrical grids worldwide, this would have
been the lower threshold for noise. Notice the huge variation
of natural noise possible at 1 MHz, about 100 dB. Again,
lightning is the main source of atmospheric noise at these
frequencies and varies widely depending upon the season, time
of day, and geographic location.
The line with alternating dots and dashes shows the galactic
noise. In a quiet location where man-made noise is low and

344
during periods of simultaneous low natural noise, galactic noise
may predominate the total noise power at frequencies above
about 5 MHz. The reason for the low frequency roll-off is the
fact that the source is extraterrestrial and thus the ionosphere
will reflect these frequencies back into space and not affect
terrestrial receivers. The actual roll-off frequency is determined
by the MUF and D-layer absorption.
The line with dashes shows the mean man-made noise level
at a quiet rural location. When natural noise falls to low levels,
man-made noise becomes the dominating factor. Like natural
noise, nighttime noise is typically much higher than daytime
noise because of propagation effects. The usual areas producing
the most noise are highly populated regions such as China,
Japan, North America and Europe. Consequently, pointing a
directional antenna in these directions will result in a higher
noise response, if, of course the band is “open” to these areas.
Statistically we can add to this noise level to adjust for
additional local noise. For these general locations simply add
the following expected power (dB) to this line:
1) “Normal” rural area, +11 dB
2) Residential area, +18 dB
3) Business/commercial area, +24 dB
Of course these are average difference values. Actual
additional noise due to proximity to populated areas can vary
widely. Marconi certainly had a much better noise
environment when he completed the first transatlantic radio
contact! Also, man-made noise will almost certainly increase in
the future. This will result in ever increasing difficulty for
working DX on the low bands.
The line with two short dashes and a long dash alternating
shows the “best” case of man-made noise at a quiet location.
Here, “quiet” location might mean Antarctica, a remote island,
or Greenland — but only when propagation to noisy areas is
not good.

Receive Antenna Parameters for MF and Low

HF

345
 In the previous section we described the various noise
sources in this frequency range. The problem at these
frequencies is similar to the problem of radio astronomy in one
important way. Both applications present the challenge of
detecting a point source on the hemispheric sky with that
hemisphere filled with noise. In contrast, the problem for MF
and low HF receivers is dealing with very high noise
temperatures, while Penzias and Wilson faced measuring
temperatures near absolute zero. The general idea of point
source versus hemispheric source points to some clear
objectives for MF and low HF receive antenna design.
If we assume that the sky has a uniform noise temperature
over its entire hemisphere (which it usually does not have!)
and we increase the gain of our receive antenna, the noise
power we receive will be the same as the hemispheric pattern
antenna, as in the radio astronomy explanation. However, if
we increase the gain of our receive antenna, a desired signal
will increase as a function of the higher gain since it is a point
source. Suppose we use an isotropic receive antenna with the
resulting total received power including both noise and the
desired signal,
Pr(iso) = Pn + Ps

(Equation 10.46)
Now if we assume the noise to be constant over the
hemispheric sky, and add gain (3 dB for this example) and,
more importantly directivity, then
Pr(3 dBi) = Pn + 2Ps

(Equation 10.47)
Notice that the noise power has remained constant but the
signal power has doubled. Thus we have increased our SNR by
a factor of two. Adding more directivity will further increase
the SNR linearly as the numerical directivity increases.
SNR ∝ directivity

346
(Equation 10.48)
The receive antenna efficiency can be relatively low since we
are mainly concerned with the received SNR and the received
noise is likely to be very high. Again the relationship is
Gain = Directivity × Efficiency

(Equation 10.49)
If our receive antenna is 100% efficient, then our antenna
noise temperature will be equal to the noise temperature of the
sky. For example, by Figure 10.14, a very quiet condition for 1.8
MHz implies an antenna noise temperature of about 290 × 104
K, which places the antenna thermal noise floor at about –134
dBm/Hz or –104 dBm for a 1 kHz bandwidth. Therefore the
received desired signal strength in dBm is used to compute the
SNR against –104 dBm noise.
Now, what happens when we use a small receive antenna
such as a 20 foot vertical, or a small loop? How much
inefficiency can we tolerate? For this calculation we use the
same equation for radio astronomy equivalent noise
temperatures, Equation 10.43. For example, let us assume we
are using a K9AY loop. From Figure 9.9 in Chapter 9 we see the
maximum gain is –26.6 dBi, which is very good for such a small
loop. We will first calculate the noise powers assuming that the
loop is isotropic in the hemisphere (+3 dBi). The efficiency of
such a loop is about 0.2%.
Tant = 290 × 104 (0.002) = 6344 K

(Equation 10.50)
We will also assume a weak signal condition, where the SNR
= 1. Assume 99.8% of the noise source for this antenna is
attributed to the actual temperature of the antenna Rl, and
0.2% will originate in the antenna’s Rr. We do not care what
the actual values of Rl and Rr are, since our given efficiency
defines the ratio. Rl is simply the ohmic value of the antenna,
and thus the noise power will be –174 dB/Hz. Notice that the

347
equivalent sky noise temperature is only about 13 dB higher
than the thermal noise emanating from the antenna’s thermal
resistance. Again, this is due to the antenna’s inefficiency, thus
lowering the “received” noise power.
The sensitivity equation (Equation 10.32) further lowers the
SNR due to the receiver’s noise figure. A typical noise figure for
MF receivers is 15 dB, placing the receiver noise temperature at
9170 K. Often the receiver’s attenuator will be used in an effort
to lower intermodulation and other nonlinear responses in the
receiver proper. These nonlinear receiver responses have the
effect of raising the equivalent noise response and/or creating
spurious signals in the receiver. In-band intermodulation
products are reduced a minimum of 3 dB for every 1 dB of
attenuation added. Therefore finding the “sweet spot” is a
balance of improving the intermodulation performance versus
not affecting the sensitivity limit. The “sweet spot” will change
for different noise power (and signal) inputs.
Thus for the given efficiency, a hemispherical response, and a
very quiet atmospheric noise condition, our weak signal will
not be detected. However, the K9AY loop also has directivity,
and by Equation 10.46 we gain back some of the SNR. It is
critical that the desired point source is aligned with the
maximum directivity of the antenna, or as close as possible.
The object is to maximize directivity in the direction of the
desired point source. Sky noise can be assumed to be arriving
from all elevation angles, while some distant weak desired
signal will be arriving at a relatively low angle. A small vertical
antenna will naturally reject the noise arriving from higher
angles, where a small loop will not. A low dipole will be a very
poor receive antenna under these circumstances since its
directivity maximum is “straight up.” The loop will respond to
low-angle vertical polarized waves, but it will also respond to
noise from “straight up.” Consequently the small vertical has an
inherent SNR advantage over the loop. Small loops that also
offer azimuth gain, such as the K9AY loop, will also benefit
from some high-angle rejection but not as well as a vertical.
Starting with the vertical and then adding elements results in
the additional advantage of azimuth as well as elevation gain.

348
The Beverage also has a low angle response and gain in the
desired direction. The Beverage also exhibits higher efficiency
than smaller loops, allowing more flexibility with receiver
controls. This explains why Beverages, vertical 4-squares and 8-
circle antennas are the usual preference for low frequency
receiver antennas, not to mention a quiet antenna location.
Multiple plots of these antenna configurations are presented in
Chapters 7 and 9. When considering the all-important SNR
figure of merit, the total steradian pattern must be used to
compare expected SNRs, especially high angle responses.
Unfortunately, these very effective receive antenna arrays tend
to be very large, compared with small loops. Therefore the
small loop might be the best (or only) choice when limited by
real estate.

Shannon’s Limit
There is one more expression to define the absolute physical
limits to any communication link, including radio
communication. Although not usually considered in an
antenna text, the Shannon Limit is the fundamental limitation
of any communication link, including all radio links. Since it is
a physical limit that also involves noise, we include at least a
basic discussion in Antenna Physics. Also by reviewing the
Shannon Limit, we find that noise becomes a fundamental
theoretical limitation to any radio link regarding the rate of
information conveyed as well as a fundamental theoretical
limitation to the dynamics of the radio link. Therefore,
antennas play a crucial role in determining the fundamental
theoretical limits of any radio communication link from both
physical limitations.
The receiver must interpret the received waveform as
containing some desired information. To this end the
waveform, or signal, must contain information that is unknown
to the receiver or there is no point in transmitting the
information. The format of the signal must be known by the
receiver, but the information carried by the signal is usually
unknown. For example, when we tune an SSB receiver we
assume a certain type of waveform, but the information we

349
receive may be unpredictable (for better or worse).
Claude Shannon (Figure 10.15) was working at Bell Labs in
the late 1940s when he provided the world with a remarkable
and profound insight: Information, in whatever form, can be
directly equated with a ratio of a desired “signal” containing
power or energy to that of the noise power or energy in a
“channel.” We usually think of “information communication” as
being some abstract phenomena. Shannon actually successfully
equated it to physical terms. His work directly led to the
formation of an entirely new discipline, Information Science.

Now we return to the concept of SNR, but now include it in

an equation of fundamental importance:
The Shannon Limit

350
(Equation 10.51)
where
C is the capacity in information bits/second
B is the bandwidth of the “channel” in Hertz
S/N is the signal-to-noise ratio
In this equation the capacity (C) term needs some
explanation. We usually think of a digital communication
system specified in bits per second. This says nothing about the
information per bit. “Information,” in turn, is only conveyed
when the receiver learns something from the transmitter it did
not know before the transmission.
A simple analogy is when someone tells you something that
you already know. No new information has been sent and
received. In effect the energy and time taken to transmit the
data was wasted. Now imagine the speaker knows exactly what
you know and don’t know. The speaker then cleverly crafts what
he/she says so that each word and inflection conveys the
maximum possible amount of new information without
repeating anything that you already know. The speaker also
knows exactly how much information you are capable of
receiving, the characteristics of your hearing, and so on. This is
an example of maximizing the information contained for a
limited acoustic energy expended, constrained by the audio
bandwidth in the presence of undesirable noise, and “noise”
within the listener’s hearing, nervous system, neural receptors
and whatever else affects the reception of the desired
information.
This example is closely analogous to a digital communication
system, including radio links, which, of course, include
antennas. Now imagine a digital communication system
sending some text written in English. If the transmitter sends a
“q” then there is no information conveyed by also following the
“q” with a “u,” assuming, of course, all English words that use a

351
“q” are always followed by a “u.” In effect, the time and energy
spent sending the “u” are wasted since the receiver already
knows English.
The next leap is to imagine that some code is developed that
permits each bit of information conveyed to contain the
maximum amount of information. Both the transmitter and
receiver know English, and both also have the ability to encode
(transmitter) and decode (receiver) the 100% efficient coded
message. Therefore, one of the implications of Shannon’s Limit
is to convey the maximum amount of information per bit
possible. This has resulted is a multi-billion dollar industry
developing and implementing codes, all to overcome the effects
of noise and the resulting bit errors.
The type of modulation used in a radio system must also be
optimized to provide maximum spectral efficiency. Of course
this topic also fills volumes. However, we must consider a
theoretically perfect modulation scheme together with some
ideal code to reach the Shannon Limit.
There is another intriguing implication of the Shannon Limit
equation. Notice that the equation does not have an error
term. The “limit” is quite literal. If the information is coded
with theoretical precision, there will be zero errors if the
equation is satisfied. Below the limit, no information can be
communicated. Above the limit the error rate is zero (again
assuming a “perfect code”).
As an aside, the equations that define the amount of
information in a bit are quite similar to the equations defining
entropy in thermodynamic systems. Despite the fact that there
is no relationship between information theory and classical
“entropy” (except the equations look similar), the amount of
information in a bit is termed its “entropy”! See Figure 10.16.

352
 Finally, let us assume we could build a communication
system with zero noise. Using Equation 10.51 we find that the
channel capacity becomes infinite even for an arbitrarily narrow
bandwidth. With zero noise, we could run error-free Gb/sec
links through 1 Hz bandwidth channels. Also, any radio link
could have infinite range if all noise (and interference) could be
eliminated. Indeed the universe would be a very different place
without the existence of random radio noise being present
everywhere.
Even today many people, including many engineers, are not
aware of this incredible and counterintuitive understanding of
information and noise. The reader is encouraged to pursue
Shannon and information science further. However, from a
physical standpoint, Shannon’s Limit represents the starting-
point for defining limitations on any communication system. If
you know the noise level and bandwidth, you can quickly
calculate a physical limitation to the channel’s information
capacity. Finally, because this limit is a physical limit, it applies
to all information systems, radio, fiber, and even biological
systems, including human beings. Shannon’s Limit has found
direct applications in diverse areas as neuroscience and

353
electronic communication systems.

Extreme System Design

In this section we will apply fundamental limits from
antenna physics and information science to define a system
limited only by these physical constraints. This analysis will
use the concepts shown in Figure 10.7 combined with the
Shannon Limit.
Sometime in the future we may wish to communicate with
another civilization in a solar system in the direction of an
otherwise empty sky. We can use a very slow data rate since it
may take many years to reach the target receiver. We have
already received signals from this civilization thus we know
what frequency and coding to use. Our parameters are an
equivalent 1 Hz bandwidth and a minimum SNR of 1/1. What is
the maximum capacity of this system? By Equation 10.49 we
find the capacity is 0.69 bits/second. Now what is the
sensitivity of the receiver assuming a background noise of 2.7 K
and a receiver noise temperature of 1 K?
From Equation 10.32 we rewrite the appropriate terms
Sensitivity = –194 dBm + 10 log10(1) + NF (–198 dBm) + 0(NF)
where –194 dBm/Hz is the noise power from the Big Bang
(since we are using only 1 Hz bandwidth there is no cor​rection
for bandwidth), the receiver noise temperature is 4 dB below
the antenna noise temperature, and the F = 1 which equals 0
dB. Therefore the sensitivity is 1.5 dB above the antenna
temperature, or –192.5 dBm. Since the assumed SNR is 1, then
the desired signal must be at least –192.5 dBm.
We assume equal gain antennas at each end of the link and a
carrier frequency of 6 GHz (λ = 0.05 m). What is the EIRP
necessary to deliver this power at a distance of 100 light years
(LY)?
1 light year ≅ 9.4715 meters

(Equation 10.52)

354
 We recall from Equation 2.5 the isotropic loss is

In this case

Based on Equation 2.7 in Chapter 2, we can now set up the

necessary equation:

(Equation 10.53)
or

(Equation 10.54)
or

(Equation 10.55)
This is the amount of EIRP and receiver antenna gain to
satisfy the minimum link budget. Now we can derive the
necessary antenna gains and transmit power.

355
 At 6 GHz the aperture of an isotropic antenna is

(Equation 10.56)
Now let us assume that at each end are 200 meter diameter
parabolic antennas that are 100% efficient (or very close to it).
Therefore the antenna aperture at both ends is 31.4 × 103 m2 =
Aparabola. This represents a power gain of 15.7 × 106 or about 72
dBi for each antenna. Therefore from Equation 10.52 we derive

(Equation 10.57)
The value of the linear power gains of the two antennas
(here equal) in the link budget analysis are multiplied. This is
identical to adding the dB gain numbers. The necessary
transmitter output power is 1.45 million watts assuming zero
transmission line losses to communicate a maximum of 0.69
bits/second of information a distance of 100 light years. Thus
the equations defined by the physics of the antenna, radio
performance, and information theory apply to wireless garage
door openers the same way they apply to interstellar
communication links.

Physics Defines the Ultimate Limits to Radio

Communication
In this chapter we have defined the fundamental physical
limitations to a radio communication link. First, even if we
point our antenna with a background of “quiet” sky we will still
have an antenna noise temperature of 2.7 K (assuming a 100%
efficient antenna with no sidelobe response). Second, even if
we achieve the Shannon Limit by maximizing the information
sent with each bit, we will still be limited by how much

356
information we can send per unit time since we will have some
finite noise from the Big Bang. Third, we are forced to conform
to the physical realities of link budgets, the inverse-square law,
antenna aperture, receiver sensitivity, and transmit power.
These limitations are due to the fundamental physical
characteristics of free space itself.
In addition to these limitations defined by this physics, there
are further engineering limitations mainly constrained by the
state-of-the-art in all relevant areas: transmission lines,
amplifiers, mixers, data converters, and, of course, even
antennas are never perfect. They all have losses and a host of
other undesirable effects. Also, the myriad of matter’s physical
characteristics and the resulting effects upon antennas and
radio links must be considered.
From these daunting realities, we actually described a system
that is limited only by physics, not current engineering state-
of-the-art. This text attempts to touch on all the basic variables
pertaining to the antenna and the antenna’s place in a system.
Further advances in theoretical limits will have to wait for the
inevitable evolution of human knowledge and the resulting
upgrading of our understanding of the limits of physics.

357
 APPENDIX A

You Already Know Calculus

Back when I was a student studying mathematics, “calculus”
seemed like a daunting and mysterious subject. My teachers
never explained to me what it is or why it is so valuable. I was
only told to master algebra, geometry, trigonometry, and then
“pre-calculus.” Had I simply known what it was — and more
importantly, its great usefulness — I would have been much
more excited to prepare for it. Why not simply tell students
what calculus is and, more important, how powerful a
technique it is?
The intention of this Appendix is not to teach calculus. The
purpose is to explain what it is, and convince you that you
already understand the basic principles. Everyone who has ever
driven a car understands the relationships among acceleration,
velocity, and distance driven. Looking at this relationship in a
slightly different manner will lead to the first realization of
what calculus is.

A Simple Example
See Figures A.1 and A.2 for the following discussion. Imagine
a motionless car at a starting line. The car is not accelerating,
has no velocity and is at point 0 (starting line, X = 0) along a
long straight road. Now let us consider these three variables:

358
 1) Acceleration
2) Velocity
3) Position (along the road (X-axis))
These three variables will be plotted against time. At time = 0
seconds, the car begins to move, and therefore has

359
acceleration. Acceleration is defined with two functions of time,
for example 5 miles/hour/second. In other words, the car has a
velocity of 5 miles/hour after 1 second, 10 miles/hour after 2
seconds, and so on. Also notice that velocity is defined with
only one term of time (miles/hour or mi/h). Position, however is
defined with no function of time. A mile is simply a mile along
the X-axis; it does not require time to define it.
Now, if we accelerate 5 miles/hour/second at a constant rate,
it is a simple matter to multiply this acceleration rate to
determine the velocity. For example, if we accelerate at this
rate for 10 seconds, our velocity will be 50 miles/hour. At any
point in time, we know exactly how fast we are going.
In this simple example, all we need to do to calculate velocity
is multiply acceleration by time where

(Equation A.1)
The seconds cancel and we are left with mi/h (or mph).
But what about distance traveled (X)? If our velocity is a
constant 50 mi/h, then the calculation is similarly trivial, we
multiply velocity by time to get miles.

(Equation A.2)
However, in our above example, velocity is changing for the
first 10 seconds, so we cannot simply multiply by time to get
distance traveled. Arithmetic breaks down and we need
something more sophisticated to “add up” the velocities over
time to get distance traveled. For this we need to “integrate”
the velocity over time.
For this operation it is more convenient to use feet/second as
the velocity, since miles will be a small fraction after only 10
seconds of acceleration. So let’s redefine acceleration as 5

360
feet/second/second.
We can now define the velocity as 5t. That is, velocity
increases at a rate of 5 feet/second/second; in this case we’ll use
time (t). So after, say, 4 seconds, our velocity will be 20 mi/h.
If we want to find how far we have traveled during the first 10
seconds, we need to set up an integral from t = 0 to t = 10. The
equation we use is a definite integral with the start and end
points, and we define the velocity as a function of time (5t)
since the velocity is simply 5 feet/second/second (the
acceleration) multiplied by time. Now to determine the
distance traveled in the same 10 seconds, we cannot simply
multiply the velocity × time because the velocity is gradually
changing with time. We need to use integral calculus to
calculate the function of velocity over time. Again the distance
traveled (X) is equal to the function of velocity integrated over
the proper amount of time (10 seconds).

(Equation A.3)
The actual calculation of the time to solve this problem does
involve knowledge of integral calculus. However,
understanding that it is the method of determining the
distance traveled at the end of 10 seconds is a giant leap to
understanding what integral calculus is. We will cheat here and
present the actual solution:

(Equation A.4)
Now we take two points, where t = 0 and t = 10. Finding the
difference will provide the answer. When t = 10, the result is
250 feet. When t = 0 the result is 0 feet. Therefore
X = 250 – 0 = 250 feet

361
(Equation A.5)
Therefore, starting from zero velocity, with an acceleration
of 5 feet/second/second, after 10 seconds the car will have
traveled 250 feet. Similarly, we can calculate the total distance
traveled for any time. After 5 seconds, the car has traveled 62.5
feet. Without calculus an exact calculation is impossible.

Integral and Differential Calculus

We can think of integral calculus as a very powerful tool for
“summing up” functions, including functions that cannot be
“summed” by simpler mathematical tools (addition,
multiplication, and so on). Distance traveled is simply a sum of
all the velocities (including changing velocities) over a portion
of time. Velocity is simply a sum of all accelerations (and
changes of accelerations) over time. Thus if you know the
function that defines acceleration over time, it is easy (using
calculus) to find the velocity, and also the distance traveled.
But what about calculating velocity from distance traveled? If
we know the distance traveled in 1 hour (for example, 50
miles), we can easily calculate the average velocity (50 mi/h).
However, we cannot calculate the velocity at any given point in
time. We might have driven 100 mi/h in a rural area, and then
slowed down through a town. However, if we know the
function of the distance traveled over time, then we can
calculate the velocity at any given time by taking the derivative
of the function.
For example, if the function of distance traveled is
X = 50t miles
Then our velocity has indeed been constant and the calculation
is simple. However, if our distance traveled is 5/2t2 feet/second,
then we can substitute any value of seconds and calculate
velocity by differentiating 5/2t2. The differential is written as

362
meaning we are finding the rate of change of x as t changes.
Again we will cheat and present the answer for the
derivative. This is simply finding the rate for change in velocity
over time. In this case, as above, the velocity is increasing for
the duration of the 10 seconds (as in a drag race). Thus the
velocity is simply 5 mi/h/sec (t) or

You now know the two basic divisions of calculus: integral

and differential.
Integral calculus is “summing up” even difficult functions to
provide “the integral of” function.
1) The (time) integral of acceleration is velocity.
2) The (time) integral of velocity is distance traveled since we
are integrating (summing) over time.
Differential calculus is “finding the rate of change,” or the
derivative of a function.
1) The time derivative of distance traveled is velocity.
2) The time derivative of velocity is acceleration.
If acceleration is zero, velocity is constant. If acceleration is
positive (going faster), velocity increases. If acceleration is
negative (braking), velocity slows. If velocity is changing, there
is some non-zero value for acceleration. The derivative of a
constant function (not changing) is always zero.
We can also say that the second derivative of distance traveled
is acceleration, since we have to take the derivative of a
derivative. The form is:

and the distance traveled (X) is the integral of the integral of

acceleration, the form being

363
It’s really that simple!
Now we can explore other examples. Energy is the time
integral of power. For example, if we run a 1 kW machine for 1
hour, we have used 1 kW-hour of energy, another simple
calculation. However, if we are constantly changing the power
being used according to “smooth” functions, we will need
calculus. For example, suppose we have a transistor amplifier
that exponentially increases its power until there is a large
surge of power, the transistor is destroyed and all power
consumption suddenly stops. If you know the function that
defines the destruction sequence you can calculate the energy
consumed by the sequence. In this way, your power meter is
performing a power integration to tell the power company how
much your bill should be. It has integrated all the power you
have used, no matter how nonlinear the consumption has
been, to yield the integral of the power over time: the energy
used.

Not Limited to Time

Time is a very frequently used variable for a wide variety of
real-world calculations with or without calculus. However, it’s
not the only possible variable. Many dielectrics’ constants will
change as a function of frequency. Such a differential term
would take the form:

This equation signals we are defining how the dielectric

constant of a material (εr) is changing with changing frequency
(ω). In this text we also use a partial derivative. Let’s assume that
our material’s (εr) dielectric constant changes not over just
frequency, but also over temperature. Thus εr is partially
dependent upon frequency and partially dependent upon
temperature. Therefore

364
 This is known as a partial differential equation. This equation
informs us how the dielectric constant of a material changes
with frequency and with temperature. Of course we need some
starting point value here represented by a reference function.
By plugging in the actual numbers, we can perform an actual
real calculation. The above equation only illustrates the form.
Again, solving for an actual solution usually requires several
years of training in advanced mathematics, but understanding
what it means does not.
The applications are endless: the rate of growth of trees over
different temperatures, humidity, soil types, sunlight, and other
factors. Or we could analyze the rate of weight gain in humans
as a function of exercise, carbohydrate intake, genetics, toxic
exposures, and so on.
In Chapter 3 we presented Maxwell’s equations using a
special form of an integral

This simply informs us that we are summing (in Maxwell’s case

magnetic or electric fields, here simplified as “x”) over a closed
surface, usually a sphere. This looks like a difficult operator, but
in principle it is quite simple to understand.
We all know how to find the average of a number of values:
add up the values (sum) and divide by the number. In Chapter 7
we used an integral of a sin function to find the sum of the area
under a sin curve, and then divided by the maximum value.
This is a necessary step to determine the radiation resistance
along an antenna element with a sinusoidal current
distribution, easily accomplished with integral calculus,
impossible otherwise.
As a final example, in Chapter 8 we set up a matrix of
differentials to describe how five differential variables are all
interdependent in order to set up an equation that defines
antenna loss due to soil resistance. This equation uses
techniques at least one year past calculus, but again, it’s just
one more step in complexity for precisely describing multiple

365
variables all interdependent with respect to a final desired
term, in this case loss resistance.
We live calculus functions all the time. They define almost
everything that changes as a result of something else changing.
We solve them all the time without knowing that we are doing
so. Many relationships (such as driving a car) are second nature
and become instinctive. Understanding calculus permits
insights into a very wide breath of problems impossible
without it.
To conclude, many if not most of the complicated looking
equations presented in advanced science and engineering texts
are just a more sophisticated way of defining something that
simpler techniques cannot. That’s what calculus is all about,
but you already knew that.

366
 APPENDIX B

Summary of Rr Terms ​-
Described by Kraus
Most authors of advanced antenna texts discuss radiation
resistance. Of all the authors, Kraus offers the most
comprehensive treatment and for the most part is in agreement
with the other authors. Reference: John D. Kraus, Antennas,
2nd ed, McGraw-Hill, 1988.
In particular, the discussion of Rr for ground-mounted
vertical antennas is a common source of confusion. Kraus
provides several approaches to solve for Rr under different
circumstances. I have introduced a new term (Rrp) to
distinguish between the Rr of the antenna versus the Rr at a
particular point along the antenna (Rrp). Following is a
summary of these terms.
1) (from Kraus, page 216)

This conceptual definition has several weaknesses. The term

Pr is difficult to measure directly in the best of circumstances.
In the case of ground-mounted verticals it is nearly impossible.
Furthermore, attempts to use computer modeling become
extremely problematic. In the free space hemisphere above
ground, a far field power integration includes the effects of
power loss and losses due to propagation effects. In the
dielectric hemisphere below ground, any specific surface of
integration will yield a false value, since you are only
integrating power at a specific distance from the feed point and
inherently ignore power loss beyond that distance. Worst of
all, integration of power in the ground usually cannot
distinguish between power lost to heat and power actually
radiated underground (which also ends up being dissipated as

367
heat!).
Using computer modeling techniques, power integration is
typically performed over a surface, such as a sphere. Within a
lossy dielectric power “radiated” into the dielectric and “lost” to
heat are both distributed through the three-dimensional
dielectric volume. Consequently an accurate differentiation
between radiated and absorbed power is nearly impossible. We
rightly consider underground radiation to be “lost,” but a
calculation of Rr must account for all radiated power, “desired”
or not.
The portion of power that can be directly attributed to
antenna loss and not propagation loss is best calculated by a
direct measurement of Rr at a point of maximum current and
then compared with a modeled antenna over lossless ground.
Only loss in the return current can provide the proper term to
calculate losses in ground-mounted verticals. This value simply
appears as an equivalent resistance at the feed point.
Theoretically it is possible to actually perform a power
integration that will correctly differentiate radiated and lost
power, but this would likely be an exceedingly difficult model.
For direct measurements we must somehow know that the
antenna is lossless, since there is no compensation for loss in
this equation. This equation also does not provide a term for
the Ae or the he of the antenna, which additionally limits its
usefulness. Although this equation does automatically
compensate for a change in Rr for Ae, it does not provide the
necessary terms to tell us why.
The reader should be wary of any attempt to directly
calculate Rr using this equation, particularly for ground-
mounted verticals.
2) (from Kraus, page 847)

This is Kraus’ general definition of Rr. It represents the

simplest closed form equation that takes the two critical terms
(Ae and he) into account. The numerator contains the

368
equivalent of the power input term from the simpler equation.
However, the radiated power term is defined in terms of gain,
which in turn, can also be easily converted to aperture, or in
this case, to a power density concentrated into a steradian
value. The actual definition of this general equation is the 3 dB
beamwidth, again stated in steradian value. A 3 dB beamwidth
corresponding to a steradian value approximately contains all of
the power radiated. However, if we hypothetically create an
area on the sphere that contains all the power (defined by a
“solid” steradian angle) normalized to the maximum gain, this
equation becomes precise.
From a practical view, the 3 dB beamwidth is usually easier to
measure than the Poynting vector over the entire sphere, so
this equation is not only closed form but is also much more
useful. Since we now know the gain, we also know what the
lossless feed impedance will be, so all the terms become
deterministic, including loss.
3) (from Kraus, page 42)

It is instructive to note that this equation is the first

definition Kraus offers for Rr in the introductory Chapter 2. It is
also instructive to note that this equation can be directly
derived from Equation 2 above. As Kraus further notes, this
equation’s form is particularly applicable to ground mounted
verticals and small antennas (relative to wavelength). Indeed,
in these applications, Rr becomes a much more critical term
than in most other antenna configurations. The reason for the
great utility of this equation is that it takes into account both
antenna loss and antenna gain. Most importantly, the radiated
power term is calculated by a distributed and maximum current
on the antenna and not a far more difficult radiated power
calculation.
Antenna loss results in both a reduction in the antenna’s
average current and maximum current (he) and gain
(represented by Ae). As long as these terms remain linearly

369
proportional there can be no change in Rr in the presence of
loss, unlike the implication from Equation 1. Also, this equation
provides the accurate effect gain has on Rr, an inverse
proportion. This equation represents the proper definition for
the ground mounted vertical since it includes all relevant
variables in a closed-form.
In the case of very low conductivity dielectrics, it is
exceedingly difficult to separate radiation into the dielectric
and actual heat loss in the dielectric. Eventually the radiated
power will indeed be lost to heat, but in the far field where it
can be considered as a propagation loss, not an antenna loss.
From a practical sense, any power lost in the dielectric is loss,
radiated or lost to heat in the near field since it is not radiated
into free space. However, the value of Rr “doesn’t care” it only
responds to power lost in the antenna proper (including
through mutual impedances) or power reflected back to the
antenna. Again, extreme care must be taken in the actual
calculations of “loss” and “radiation.”
Using relatively simple antenna modeling programs such as
NEC, it is possible to set up a model that can force return losses
into an equivalent single resistor and eliminate the effects of
far field loss simultaneously. The equivalent loss resistor is
placed just above the ground, with the feed point located just
above it, both very close to the ground. This simple technique
will be suitably accurate for most “real” grounds. Antenna loss
proper will converge upon a simple single resistance. The
antenna is modeled using a perfect ground to eliminate
propagation losses. A decrease in he and thus efficiency will
track perfectly with a reduction in Ae (gain) just as Equation 3
states.

4)

This equation provides a direct and abbreviated method to

calculate the radiation resistance of a short vertical: Kraus does

370
not present this special equation, but does derive a similar
expression for very short dipoles. This equation is published by
many sources, and can be seen to be a direct derivation of
Equation 3. For ground mounted verticals less than about 1⁄8 λ,
the gain versus physical height becomes nearly constant at 4.77
dBi, or a power gain of 3. Thus the aperture simplifies to

for any vertical height below about 1⁄8 λ. The other

simplification involves he2. For short verticals, the current
distribution can be assumed to be linear (a small portion of a
sine curve). Therefore, a linear short vertical’s he value will be
1⁄2 its physical height multiplied by 2 (to include the image), or
about 6 Ω for a 1⁄8 λ physical height. For a perfectly top loaded
1⁄8 λ vertical, he becomes 1× its physical height × 2 (to include
the image), or about 24 Ω. Therefore a properly top-loaded
short vertical has a fourfold increase in Rr when compared to
an unloaded simple vertical of the same physical height.
5) Radiation Resistance at a Point which is Not a Current
Maximum (Kraus, page 227)
This section make two simple points: First, with linear
antennas greater than 1⁄2 λ, the choice of feed point location
will change the current distribution along the antenna, and
thus the location of the current maximum and directivity.
Second, for a lossless antenna, the real portion of the feed
point impedance can be taken to be the radiation resistance at
that point. Therefore for radiated power calculations, the
simple formula I2Rfeed = Pr applies. An example of this is in AM
broadcast antennas, where it assumed that the antenna is
lossless. That is usually close enough for radiated power
calculations, but never precise. In effect, Kraus is describing
here what I have defined as Rrp for clarification.

371
 APPENDIX C

References
The following articles and texts were valuable in the course
of preparing this book.

International Telecommunication Union Recommendation

ITU-R P.372-6, Radio Noise, www.itu.int/dms_pubrec/itu-
r/rec/p/R-REC-P.372-6-199408-S!!PDF-E.pdf
“Project Cyclops: A Design Study of a System for Detecting
Extraterrestrial Intelligent Life,” NASA, Ames Research
Center, 1971.
Bianchi, Cesidio, and Meloni, Antonio, “Natural and Man-made
Terrestrial Electromagnetic Noise: An Outlook,” Instituto
Nazionale di Geofisica e Vulcanologia, Roma, Italy, Annals of
Geophysics, Vol 50, No 3, June 2007, p 435.
Balanis, Constantine A., Antenna Theory, 3rd Edition (Wiley-
Interscience, 2005).
Devoldere, John, ON4UN’s Low-Band DXing, 5th Edition
(ARRL, 2010).
Drentea, Cornell, Modern Communications Receiver Design and
Technology (Artech House, 2010).
Fujimoto, K., Henderson, A., Hirasawa, K. and James, J. R., Small
Antennas (Research Studies Press, 1987)
Johnson, Richard C. and Jasik, Henry, Antenna Engineering
Handbook, 2nd Edition (McGraw-Hill, 1984).
Jordan, Edward C., Electromagnetic Waves and Radiating
Systems (Prentice-Hall, 1950; also in paperback by Pearson
Education, 2003).
Kraus, John D., Antennas, 2nd Edition (McGraw-Hill, 1988).
Lawson, Dr. James L., Yagi Antenna Design (ARRL, 1986).
Maxwell, James Clerk, “A Dynamical Theory of the
Electromagnetic Field,” Philosophical Transactions of the Royal
Society of London, Vol 155, Jan 1, 1865, pp 459 – 512.
McLean, James S., “A Re-Examination of the Fundamental

372
 Limits on the Radiation Q of Electrically Small Antennas,”
IEEE Transactions on Antennas and Propagation, Vol 44, No. 5,
May 1996, pp 672 – 675.
Silver, H. Ward, Ed., The ARRL Antenna Book, 23rd Edition
(ARRL, 2015).
Taub, Herbert and Schilling, Donald, Principles of
Communication Systems, 2nd Edition (McGraw-Hill, 1986).
Uda, S., “On the Wireless Beam of Short Electric Waves,”
Institute of Electrical Engineers of (JIEE), Mar 1926
Wangsness, Roald K., Electromagnetic Fields (Wiley & Sons,
1979).
Wheeler, H. A., “Fundamental Limitations of Small Antennas,”
Proceedings of the I.R.E., Vol 35, No. 12, Dec 1947, pp 1479 –
1484.
Wheeler, H. A., “The Radiansphere Around a Small Antenna,”
Proceedings of the I.R.E., Vol 47, No. 8, Aug 1959, pp 1325 – 1331.
Yagi, H., “Beam Transmission of Ultra Short Waves,”
Proceedings of the I.R.E., Vol 16, No. 6, Jun 1928, pp 715 – 740.
Zavrel Jr., Robert J., “How Antenna Aperture Relates to Gain
and Directivity,” QEX, May/Jun 2004, pp 35 – 39.

373
 APPENDIX D

Symbols and Abbreviations

A

area (vector notation)

A

area
A

ampere
Ac

collecting aperture
Ae

effective aperture
Aem

maximum effective aperture

Aer

effective aperture of the receive antenna

Aet

effective aperture of the transmit antenna

374
Al

loss aperture
Ap

physical aperture
As

scattering aperture

bandwidth
B

magnetic flux (vector)

BER

bit-error ratio

velocity of electromagnetic radiation in free space

C

Coulomb
C

capacity bits/second

375
C

capacitance (farads)
C

circumference of a loop antenna

directivity
D

electric flux density (vector)

DF

dielectric dissipation factor

dB

decibel
dBi

antenna gain relative to an isotropic antenna

dBil

linear polarized gain relative to a linear polarized

isotropic antenna
dBic

circular polarized gain relative to a circular

376
polarized isotropic antenna
dBd

antenna gain relative to a 1⁄2 wave dipole (0 dBd =

2.15 dBi)
dBm

power expressed referenced to 1 milliwatt (1 mW)

dBW

power referenced to 1 watt (1 W)

energy
Eb

energy per bit

E

electric field (vector)

EIRP

effective radiated power relative to an isotropic

antenna
ERP

effective radiated power relative to a 1⁄2 wave

dipole

377
f

frequency, Hz
F

noise factor

gain
Gr

gain of the receiver antenna

Gt

gain of the transmitter antenna

magnetic field (vector)

h

Plank’s constant
he

effective height
hp

378
physical height

current
Iave

average current along an antenna element

I0

maximum current point on an antenna

Joules
j

designates a complex function

k

Boltzmann’s constant
ke

dielectric constant

length
L

379
inductance (Henrys)

meters

Newton
NF

noise figure
N0

noise energy

power
PADL

phase, amplitude, direction, location

Pi

power input to an antenna

Pn

noise power

380
Pr

radiated power
Pr

received power
Prr

re-radiated power
Pt

transmitted power

quality factor
q

electrical charge

resistance in ohms
Rb

bit rate
Rl

381
loss resistance
Rload

resistive part of a load impedance

Rr

radiation resistance of an antenna

Rrp

radiation resistance at a point on an antenna

Rt

terminal resistance
r

radius

Poynting vector, power flux in W/m2

S

power flux density

s

second
SNR

382
signal-to-noise ratio

temperature in Kelvins
Tant

antenna noise temperature

Te

effective noise temperature

T0

290 Kelvins

voltage
V

velocity
Vn

noise voltage
VF

velocity factor

383
impedance
Zfeed

impedance at an antenna’s feed point

Z0

impedance of free space

Z0

characteristic impedance of a transmission line

reflection coefficient
ε0

permittivity of free space

μ0

permeability of free space

ω

angular frequency (2pf)

λ

wavelength
θ

384
a degree or radian value, corresponding to
latitude in spherical coordinates
φ

a degree or radian value, corresponding to

longitude in spherical coordinates
Ω

ohms
Ω

steradian solid angle

Ωa

beam area

385
 APPENDIX E

Comparison of Elevation
Plots of Vertical-Polarized
Ground-Mounted 160 Meter
Antenna Arrays
These three plots (Figures E.1 to E.3) are presented as a
simple visual comparison among various configurations using
the same gain scales. All data assumes lossless ground systems
over average earth. The relative shapes of these plots are valid
for higher frequencies as well, so this may serve as a basic
comparison for higher frequencies. However, the absolute gain
numbers will change with different frequencies. ARRL plotting
is used to offer more detailed graphing for small differences
among the gain figures. All three graphs include a 1⁄4 λ vertical
for reference comparisons.

386
387
388
389
390