Cubical Quad Vol 2
Cubical Quad Vol 2
Volume 2
Rethinking the Quad Beam
L. B. Cebik, W4RNL
Published by
antenneX Online Magazine
http://www.antennex.com/
POB 72022
Corpus Christi, Texas 78472 USA
Copyright © 2001 by L. B. Cebik jointly with antenneX Online Magazine. All rights
reserved. No part of this book may be reproduced or transmitted in any form, by any
means (electronic, photocopying, recording, or otherwise) without the prior written
permission of the author and publisher jointly.
ISBN: 1-877992-10-0
Table of Contents
Dedication ........................................................................................................................... 5
Preface ................................................................................................................................ 6
Chapter 1: Introduction and Reference Data ...................................................................... 8
Chapter 2: Calculating the Length of a Resonant Square Quad Loop .............................. 19
Chapter 3: 2-Element Quads as a Function of Wire Diameter: Understanding Some Quad
Properties ................................................................................................................... 37
Chapter 4: 2-Element Quads as a Function of Wire Diameter: Automating the Design
Process ....................................................................................................................... 51
Chapter 5: 2-Element Quads as a Function of Wire Diameter: Fatter Elements from “Mere
Wire” ......................................................................................................................... 64
Chapter 6: Automating the Design of 3-Element Monoband Quad Beams: A Wide-Band
Model ........................................................................................................................ 83
Chapter 7: Automating the Design of 3-Element Monoband Quad Beams: A High-Gain
Model ...................................................................................................................... 105
Chapter 8: 40-Meter Wide-Band 3-Element Quad Designs ........................................... 120
Chapter 9: 4-Element Monoband Quad Design .............................................................. 137
Chapter 10: Some Notes on Long-Boom Quads ............................................................ 154
Chapter 11 A Place for Narrow-Band Quad Designs ...................................................... 172
Chapter 12: Some Notes on Size and Height .................................................................. 185
Chapter 13: Some Notes on VHF Quad Design ............................................................. 197
Chapter 14: Quad Horizons ............................................................................................ 229
Other Publications .......................................................................................................... 231
Cubical Quad Notes
Dedication
Dedication
Preface
Preface to Volume 2
Rethinking the Quad Beam
2. The typical quad array is a wire affair, and wire has a very small diameter
relative to the length of an RF wave. If the typical 20-meter Yagi-Uda array uses
elements that average 1" in diameter, the typical #14 AWG copper wire quad uses
an element under 6.5% as large. Element diameter does make a difference in quad
design and performance, a sizable difference. What remains to be developed is
just how large a difference element diameter does make and whether that differ-
ence is sporadic or systematic.
quad designs when working with more than two elements. What ramifications these
outcomes may have for quad design is yet to be seen.
If we add together all of the questions so far raised, how do they combine to
yield more adequate quad designs? In the combination there may be further sur-
prises not revealed by the individual factors, and they remain to be discovered.
This volume attempts to rethink the quad, taking into account the role of
element diameter, optimal element spacing for maximum performance across de-
sired bandwidths, and the range of feedpoint reactance across a desired passband.
In the course of our exploration, we shall discover that once certain critical factors
are fully accounted for, the remaining elements of monoband quad design can be
systematized into simple computer, spread sheet, or antenna modeling programs.
Unlike past formula-laden methods of quad design, both element diameter and de-
sign frequency will have to be entered for a workable design to emerge.
Although the designs that emerge from these further studies are in one
sense optimized, they do not constitute a set of ultimate quad designs. Rather, the
designs represent the best composite of guiding parameters that I have been able
to come up with so far. Not only are these design directions being investigated by
others, but as well, there are many branch directions of design along the path that I
have followed--too many directions for one person to fully explore in a short period
of time. So at most, these notes present only the best that I have so far found.
Much is left to be done, and the field of quad analysis and design remains wide open
for the efforts of serious students of antenna properties.
Chapter 1
Introduction and Reference Data
Rethinking the design of quad beams covers a vast territory--much more than
any single individual can hope to cover in a single study. Therefore, we shall have to
confine the scope of the study to a workable unit. As well, as in any study, the
methods of analysis should be made plain so that anyone can replicate them--or go
beyond them. Finally, there is some common data used throughout the study. Plac-
ing it here at the beginning seems most fitting and perhaps more useful than devel-
oping an appendix.
The chapters ahead will look into only monoband quad design for a very basic
reason: there is so much to learn about quad behavior with respect to element
diameter and optimized spacing that trying to include multi-band arrays would be
self-defeating. Virtually all multi-band quads involve compromise dimensions and
compromise performance figures. What this volume hopes to accomplish is to
discover what a quad might be able to do when it is designed without a compromise.
Fig. 1 shows the range of quad that we shall work with in the course of the
study. Essentially, any monoband quad between 1 and 5 elements will be our sub-
ject matter. Even within those limitations, we shall discover that the larger the num-
ber of elements, the more tentative the conclusions that we may draw about opti-
mized quad behavior.
All quads will use a standard configuration. For an element count of 2 or more,
there will be 1 reflector and a driven element, with the remaining elements (if any)
being directors. Some design variants have included multiple reflectors. Others
have attempted to use a phasing line between the driver and the reflector. However,
these notes will stick to the more common design. In addition, each array will be fed
at the bottom of the driver at a single, center location. All quad loops will be squares.
though rear quadrants will be examined to discover if there are any problems in the
quartering side lobes that are natural to quad radiation patterns.
Fig. 1-2 shows a typical 2-element quad free space azimuth pattern. At the
design frequency, the array has been optimized for a 180° front-to-back value of
nearly 54 dB. However, the quartering side lobes are down by less than 17 dB. This
latter figure is the worst-case front-to-back ratio and must be accounted for in evalu-
ating the overall performance of an array.
10 meters has two entries. One is for the entire band. However, since so many
antennas are designed to cover only the first MHz of this wide HF band, entry “10A”
is included. Clearly, for most of the work to be discussed, the non-harmonic (or
WARC) HF bands are not of central interest, since almost any narrow-band array
will more than adequately handle them. Of primary interest are the harmonically
related HF bands and the two lowest VHF bands.
Second, the arrays will be designed for near resonance at the design frequency.
Near resonance will be defined as a feedpoint impedance with under j1.0 Ω of reac-
tance. In many generalized designs, the basic feedpoint impedance will show a
larger variation in reactance as the frequency is varied significantly from the basic
30-MHz baseline. Even with perfect scaling of all element factors (circumference,
spacing, and diameter), there will be some degree of variation due to skin effect.
Skin effect does not vary at the same rate as the other factors involved in scaling an
array from one frequency to another. In addition, the feedpoint reactance will vary
according to the material chosen for the elements.
Table 1-2 provides a short list of common materials often used in antenna con-
struction. Some of the materials may be better known than others to hams who
build their own antennas. For example, most hams buy antennas made from stain-
less steel rather than using the material for antenna construction. A few of the
entries may be of predominantly historical interest, such as brass and phosphor
bronze, although we still find them used in some home brew and commercial prod-
ucts. The value for “solder” on the list is an approximation for the value that might
be found in a galvanized steel tower. Of the values in the list, copper and aluminum
(especially 6063-T832 and 6061-T6) will be the most relevant to our efforts, al-
though the remaining values provide a general context for understanding the merits
of the most common antenna metals. Values are given for both resistivity--in Ohms/
meter--and conductivity--in Siemens/meter (where the Siemen is the current name
for what used to be called the “Mho”). The two values are, of course, simple recip-
rocals of each other. The listings are in “engineering” notation. 1.59E-08 equals
0.0000000159, while 6.2893E7 equals 62,893,000.
In the range of materials from a perfect or lossless conductor through the alumi-
num entries in the table, the losses in the antenna due to the material can make a
3. Element Diameter: One the most fundamental variables in quad design will
turn out to be element diameter. To a large degree, element diameter will determine
for any size quad the required circumference for each loop and the spacing be-
tween the loops of multi-element arrays. In developing some automated design
programs, the only two input values required will be the design frequency and the
element diameter.
Table 1-3. Common AWG Wire Size Diameters in Inches and Millimeters
AWG Diameter Diameter AWG Diameter Diameter
Gauge# Inches Millimeters Gauge# Inches Millimeters
1 0.2893 7.348 21 0.0285 0.723
2 0.2576 6.544 22 0.0253 0.644
3 0.2294 5.827 23 0.0226 0.573
4 0.2043 5.189 24 0.0201 0.511
5 0.1819 4.621 25 0.0179 0.455
6 0.1620 4.115 26 0.0159 0.405
7 0.1443 3.665 27 0.0142 0.361
8 0.1285 3.264 28 0.0126 0.321
9 0.1144 2.906 29 0.0113 0.286
10 0.1019 2.588 30 0.0100 0.255
11 0.0907 2.305 31 0.0089 0.227
There are equations for determining the diameter of any particular wire gauge-
-or the gauge from a given diameter. However, for most uses, the table is most apt.
Note that in the AWG system, the wire diameter doubles with each 6-gauge de-
crease. Hence #12 AWG is twice the diameter of #18 AWG. Memorizing a few
common gauge-diameter relationships is a useful exercise.
Table 1-4. Common BSWG Wire Size Diameters in Inches and Millimeters
BSWG Diameter Diameter BSWG Diameter Diameter
Gauge # Inches Millimeters Gauge # Inches Millimeters
1 0.3000 7.620 21 0.0320 0.813
2 0.2760 7.010 22 0.0280 0.711
3 0.2520 6.401 23 0.0240 0.610
4 0.2320 5.893 24 0.0220 0.559
5 0.2120 5.385 25 0.0200 0.508
6 0.1920 4.877 26 0.0180 0.457
7 0.1760 4.470 27 0.0164 0.417
8 0.1600 4.064 28 0.0148 0.376
9 0.1440 3.658 29 0.0136 0.345
10 0.1280 3.251 30 0.0124 0.315
11 0.1160 2.946 31 0.0116 0.295
12 0.1040 2.642 32 0.0108 0.274
13 0.0920 2.337 33 0.0100 0.254
14 0.0800 2.032 34 0.0092 0.234
15 0.0720 1.829 35 0.0084 0.213
16 0.0640 1.626 36 0.0076 0.193
17 0.0560 1.422 37 0.0068 0.173
18 0.0480 1.219 38 0.0060 0.152
19 0.0400 1.016 39 0.0052 0.132
20 0.0360 0.914 40 0.0048 0.122
In the BSWG system, wire diameter approximately triples for every 12-gauge
decrease in number.
In order to familiarize yourself with how large the diameter of a certain element
is at a given frequency, Table 1-5 provides some sample values of wire diameter in
wavelengths and the corresponding element diameter in inches. In parentheses,
where applicable, is the nearest AWG wire gauge to the listed diameter.
For the largest diameters listed, there are unlikely to be corresponding quad
arrays. However, as we shall see, it is possible to simulate very large diameter
elements using multiple strands of thinner wire. So using a 0.01 wavelength ele-
ment diameter at 80 meters is not impossible, however improbable. Even less
probable is the use of a 0.00001 (1E-5) wavelength diameter element at 2 meters.
However, a 0.9" diameter element (or something close) is certainly feasible at 144
MHz.
The entries in Table 1-5 are divided in increments such that the corresponding
common logarithms of the wire size progress from -5.0 through -2.0. The entries
containing “316” represent wire diameter values whose common logarithm is X.5.
(The more precise value of wire diameter would use “31622777,” but that fussiness
would yield spurious precision in most cases.) It will turn out that most of the dimen-
sions and performance figures for quads, when varied by element diameter, yield
more usable curves when related to the common logarithms of the various element
diameters. Hence, every wire diameter--whether given in inches, in millimeters, or
as a wire gauge--will end up converted to the corresponding wire size as a fraction
of a wavelength and from there to the appropriate common logarithm.
Development Methods
In developing a precise quad model that meets the criteria set for the design
and that uses progressively thicker (or thinner) element diameters (with consequential
changes in loop circumferences and element spacings), it is easier to set up the
basic model so that the coordinates of the wires for each loop and the wire diameter
are variables. The numeric values of each variable can be changed with a single
entry which will result in several changes to the array dimensions.
For any given wire size, when the array is declared optimized, the resulting
values may be recorded or saved. The progression of value sets can then be ma-
nipulated to arrive at equations for the resulting curves. Regression analysis will be
the main tool for translating the curves into equations of an order (4th order equa-
tions will be most commonly used) that provides suitable design guidance.
Several of the quad designs shown in succeeding chapters will demonstrate the
process in various levels of detail. However, Fig. 1-3 will provide a sample of the
process. In this case--the reflector circumference for a 4-element quad--the data
from optimized samples is entered. The smooth curve represents the continuous
output values that emerge from the 4th order equation having the form noted in the
inset. The precise values for a, b, c, d, and e will be given in Chapter 9. The only
difference among these equations in the many applications of them in this study will
be the numerical values generated for the constants.
Regression analysis equations are simple curve fitting exercises and have no
foundation in an electronics or antenna theory. They are useful calculating tools
within the context of this study. No attempt will be made to integrate these equa-
tions with any facet of antenna theory. They will have served their purpose if they
allow us to design the dimensions of a given quad array type and to have some
relevant indications of its performance.
The GW Basic versions of the design equations are available in the suite of
utility programs called HAMCALC. The suite of well over 200 programs comes on a
CD-ROM that includes a copy of GW Basic. It is available for $7 (US) from George
Murphy, VE3ERP, 77 McKenzie Street, Orillia, ON L3V 6A6, Canada. The fee cov-
ers the cost of the disk and air mail postage (anywhere in the world), with excess
proceeds over costs being donated to the amateur radio program of the Canadian
National Institute for the Blind.
In all cases, the design results are offered tentatively. They represent the best
that I have so far been able to develop. Further developmental work in the future by
me or by others may well result in superior designs as the dynamics of quad opera-
tion gain a better and more systematic understanding. My only reason for reporting
on these interim results is that, even if imperfect, they represent some interesting
and useful directions in monoband quad array design. Moreover, the volume of
such results has reached the level of forming a relatively coherent progressive treat-
ment.
However, we shall not know how far short of the goal these notes end up if we
do not get started. The proper place to start the investigation of quads is with the
single quad loop.
Chapter 2
Calculating the Length of a Resonant Square Quad Loop
I have received numerous inquiries about calculating the length of wire needed
to form the circumference of a single resonant square quad loop at various frequen-
cies. Everybody seems to “know” that such a loop in the HF range requires use of
the old formula
1005
L ft =
f MHz
Folks also seem to know that this equation does not work at VHF.
In fact, this old formula does not work at HF either. It has been, is, and always
will be wrong for common sizes of bare wire. I have heard by the unreliable grape-
vine that it is the formula to use at HF for insulated wire, where we have a shortening
effect or velocity factor. However, being over 4% short, the formula exceeds the
velocity factor effects of most wire insulations with which I am familiar.
Fig. 2-1 shows us the basic ingredients for what we need. A single quad loop in
free space has only two directly related dimensions: wire diameter and circumfer-
The only reason we also need to know the frequency is so that we can translate
the input wire diameter from a common unit of measure, such as inches or millime-
ters, into a fraction of a wavelength. If we use feet and/or inches, then a wave-
length, L, becomes
983.5592 11802.71
L ft = and Lin =
f MHz f MHz
To find the relationship between the wire diameter and the perimeter length, I
modeled a large number of quad loops and brought them to resonance, where
resonance is defined as a feedpoint impedance with less than +/-0.1 Ohm of reac-
tance. The process was considerably eased by using the model-by-equation facili-
ties of NEC-Win Plus. Once I had placed variables for the wire dimensions and the
wire diameter, a new model required only that I change two values on the equations
page. Using this facility is how I came to realize that no matter what frequency I
plugged into the model, the impedance did not change if I defined the variables in
terms of a wavelength.
The restriction on this result is the use of lossless or perfect wire and a free
space model. However, changing any model’s material into copper or aluminum
yields no significant adjustments to the result. The effects of material losses will be
greater with very thin wire (as a fraction of a wavelength) than for think wire. The
resistive component of the impedance grew to reflect the wire losses (an Ohm or
two), and the remnant reactance near resonance remained less than +/-1 Ohm.
Nevertheless, ground effects will vary from application to application and must be
the responsibility of the builder and user of the antenna.
Fig. 2-2 shows the relationship between the wire diameter and the perimeter
length, when both are expressed in terms of a wavelength. Note that the wire diam-
eter axis is a log scale, and the “3162” steps are points at which the common log has
a value of x.5. The graph is limited to wires sizes from 1.0E-05 to 1.0E-2, which
covers the range of #18 on 80 meters to small tubing at VHF. As is evident, the
relationship is not at all linear.
While doodling with the values for wire diameter and loop circumference, I evolved
following equation is a first order approximation of the curve in the graph.
2
QL wl = 1.0413 + ( log ( d wl * 1E5 )* 0.0128 )
where QL is the perimeter length of the quad loop and d is the wire diameter, both in
wavelengths.
The approximation is satisfactory for most amateur building projects, since the
maximum error is about 2%, relative to the NEC models from which the algorithm
was generated. Nevertheless, I was not satisfied with the results. Therefore, I
turned to regression analysis to see if a more adequate fit might be obtained be-
tween the modeled data and the results of a calculation. If one has enough data
points that vary along relatively smooth curves, regression analysis can provide
equations that permit very accurate interpolations between the data points. The
sample curve in Chapter 1 is typical of the results. Programs such as Datafit auto-
mate the process, while allowing the user to select the degree of equation best
suited to the job. For the loop circumference and for the feedpoint impedance at
resonance, 4th order equations proved sufficient.
I shall not parade the equations at this point for reasons of economy of space
and because they appear later in a utility program. The general form of the equation
will suffice:
y = a x 4+b x 3+c x 2+d x+e
where x is the common logarithm of the wire size expressed as a fraction of a
wavelength and y is the parameter being related to x, in this case, either the circum-
ference length (as a fraction of a wavelength) or the resonant feedpoint resistance.
(The form of the equation, of course, will change if one selects other than a 4th
order version.) The circumference dimension can be the full circumference, the
length of one side, or half the length of one side, since all three of these measures
are related to each other by simple multipliers or dividers. The modeler tends to
prefer 1/8 the circumference, which is half the side length, a convenient number to
plug into a model that is centered along one of the axes.
The constants a through e emerge from the regression analysis and are gener-
ally carried out to the available space in the cell in which they are recorded. Be-
cause they are constants, my own preference is to use them to the full level of
precision given, saving all rounding operations until the last move in a calculation.
For hand calculator work, this procedure is fraught with problems, but when placing
such values into a calculating program on a spreadsheet or in Basic, they work very
well. They also relieve one from the task of determining if a deviation from the
baseline data is a function of the equation itself or of some one or more rounding
steps along the way.
The regression analysis provided equations that much more closely fit the data
than the original approximation, even though the approximation might well suffice
for amateur construction. Models based on the calculated outputs from both schemes
were created and then checked for the amount of feedpoint reactance. The devia-
tion from precise results is shown in Table 2-1. The table lists a span of wire sizes
(in terms of a fraction of a wavelength) and then lists the reactance of quad loops
modeled on the results of each algorithm.
For the initial approximation, the maximum error occurs at a wire size of about
0.001 wl. For larger and smaller wire sizes, the error in the algorithm is far less than
1%. For the equation yielded by regression analysis, the maximum error occurs
with the next-to-largest wire size listed and is the only figure to exceed 2 Ohms. The
modeled resonant resistance of the baseline models ranges from 121.6 Ohms for
the thinnest wire model (0.000001 wavelengths) to 141.4 Ohms for the largest di-
ameter element (0.01 wavelengths). A reactance of 20 Ohms would not produce an
objectional SWR value.
The superiority of the regression equation over the initial approximation is per-
haps even more graphic in Fig. 2-3. The graph plots the original baseline modeled
dimensions, which appear as the circumference and the element diameter as frac-
tions of a wavelength. As well, there are curves that show the results of using the
original approximation and the regression equation. The original approximation
was calibrated to a minimum wire diameter of 1E-5, so the results approaching 1E-
6 move rapidly away from the baseline data. In contrast, the regression equation
shows a close overlap with the baseline, with the exception of the next-to-largest
wire diameter for which the data is correlated.
The end values for the correlations were selected for quite different reasons. At
the smallest diameter end of the scale, values less than 1E-6 wavelengths are im-
practical at any frequency for which radio amateurs would build a quad loop. The
cut-off at wire sizes approaching 0.01 wavelength was a function of the segment
length-to-diameter ratio of the element. Larger diameters would have required a
shift to fewer segments per quad loop side, which in turn would have disrupted the
smoothness of the baseline curve.
A Utility in GW Basic
To save you the trouble of doing the transformation and algorithm calculations
for each possible quad you may wish to build, I have put the basics of the discussion
above into a simple utility program. The listing of the GW Basic program follows.
The program let’s you enter the wire diameter in either inches or millimeters.
Anyone so inclined can add an AWG wire table module for direct entry of wire gauges.
However, the table of diameters in Chapter 1 should satisfy most needs.
Lines 190 through 230 do most of the calculating, with each algorithm broken
into steps. One line lists the constants, while the next calculates the result. Two
outputs are calculated: the perimeter of the loop and the anticipated resonant im-
pedance.
You can repackage the equations into almost any form and calculating medium.
The program was written in GW Basic, which only recognizes log-base e (natural
logs). But then, GW Basic was a product of the 1980s (my edition is dated 1987).
Hence, the log functions contain a conversion factor to change the natural logs into
log-base 10 (common logs). You can expand the 0.4343 multiplier indefinitely for
greater precision of calculation (say, to 0.4342945), but the common conversion
factor is precise enough for virtually all applications. If you translate the program
into another medium that recognizes both common (LOGx) and natural (LNx) logs,
you should remove the conversion factor if you retain the “log” notation. Current
spreadsheets—a common substitute for Basic—tend to have both functions.
Outputs (to screen only) include the input data, the wire size in terms of a wave-
length, and the resulting perimeter length, also in terms of a wavelength. In addi-
tion, the outputs include the length of a wave in feet and in meters, along with the
perimeter lengths in these units. At the bottom is the resonant impedance. I shall
assume that conversion to inches, millimeters, or centimeters is a routine hand
calculator job. However, you can feel free to doctor the utility program to include
these outputs as well.
I tend to prefer working up utilities in GW Basic, since the listing makes all of the
arithmetic transparent, both for error detection and for transferring the information
to other media. For example, all of the constants required for the regression equa-
tions are clearly apparent in the listing.
Let’s run a few examples through the calculating utility and see what they might
reveal about quad loop properties in general. First, we shall use #12 wire on 20, 10,
6, and 2 meters, as shown in Table 2-2.
The series of #12 examples shows the “growth” in the wire diameter as a func-
tion of a wavelength. We can also see that as the loop circumference grows, so too
does the feedpoint impedance. However, as the wire diameter grows as a function
of a wavelength, the losses reduce, as evidenced by the smaller differential be-
tween the resistive components of the perfect-wire and the copper wire impedances.
span of calibrated wire sizes from 1E-6 to 1E-2 for perfect, copper, and 6063-
T832 aluminum.
For practical purposes, by the time the element reaches a diameter of 1E-4
wavelengths, the differential between a perfect wire and the most common materi-
als used in quad loop construction becomes insignificant. At a diameter of 1E-3
wavelengths, the differentials are negligible. As a practical example, we might have
added to the list in Table 2-3 the following quad loop for 146 MHz using a 0.25"
element.
Note that the element diameter is about mid-range between the 1E-3 and the
1E-2 entries in Table 2-3. So, too, is the impedance, which shows negligible change
in the transition from perfect to copper wire.
To see the effects of real ground on our calculated loop sizes, let’s model the
#12 28.5 MHz copper wire loop at various heights above average ground (conduc-
tivity 0.005 S/m; dielectric constant 13). Table 2-4 provides some interesting data.
The lesson contained in this table is that calculating the single quad loop via the
algorithm will result in usable dimensions for any common antenna height. The
antenna impedance will slowly stabilize near its free space value as we further in-
crease the height. The range of variation in feedpoint impedance for heights from
0.5 to 1.0 wavelengths is somewhat less than we find for linear dipoles. Thus, the
quad loop acquires a reasonably justified reputation for being relatively immune to
the effects of nearby objects, including the ground.
While we are cataloging general quad loop properties as we change the ele-
ment diameter, we should pay heed to the changes in gain from a loop. The quad
loop (in free space) presents a figure-8 pattern broadside to the plane of the loop.
As we move from very thin wire (1E-6 wavelengths) to very fat wire (1E-2 wave-
lengths), even with perfect wire, we shall see considerable change in the free-space
gain of the quad loop.
The total change in gain runs from about 3.23 dBi for the thinnest element to
about 3.57 dBi for the fattest element. Although the differential would not be opera-
tionally significant for a single quad loop, the effects of this phenomenon are cumu-
lative as we add loops to form complex quad arrays. As shown in Fig. 2-4, the
increase in gain (not to be confused with the gain itself) nearly doubles for each
additional order of magnitude in the element diameter.
Single quad loops also display the classic SWR curves that are commonplace
with more complex quad arrays. As shown in Fig. 2-5, Each curve in the set rises
more steeply below the design frequency than above it. In this example, element
diameters from 1E-6 through 1E-2 wavelengths were used to create quad loops at
30 MHz. Each loop was referenced to its own resonant impedance for VSWR
curves from 28 to 32 MHz.
As one might well expect, the versions with the largest diameter elements showed
the shallowest SWR curves, while the steepest curve belongs to the thinnest ele-
ment. However, regardless of the element diameter, the VSWR ratio rises more
rapidly below the design frequency. As we shall see in future chapters, adding
elements to form a complex array sometimes makes the differential more dramatic.
A Design Model
As earlier noted, the dimension equation for the single square quad loop can be
placed into a NEC-2 model, so long as the modeling software has a design-by-
equation facility. NEC-Win Plus has such a facility, so I wrote a model that automati-
cally sets the dimensions for a quad loop at resonance. The user need only enter
the design frequency and the wire diameter in the units to which the program is
currently set.
However, the process requires that the wires and geometric coordinates be set
up as variables. Fig. 2-6 illustrates the process (although the black and white illus-
tration blanks out a bit of the detail, for example, the dividing lines of the spread
sheet cell structure).
The second step is to set up the equations page. For this design project, Fig. 2-
7 shows the various parts of one method of accomplishing this task. Column D
contains the constants (a through e) produced by the regression analysis. Cell D8 is
highlighted so that the 4th order equation appears in the working space above the
cells. Since the user can add any number of equations, whether or not used in
creating the quad dimensions on the wires page, I have added cell D7. Note that the
The user specifies two variables in this spreadsheet. Variable A in cell B4 is the
wire diameter. Note the required entry format (=0.3934. . .). As well, the user
specifies the design frequency in cell B9 as variable G. Cell D9 converts this fre-
quency to its corresponding wavelength value. This step is necessary so that the
antenna can be stepped from frequencies below the design frequency to frequen-
cies above it. (If the automatically produced variables F and W had been used, the
model would be dimensionally defined by the “Start” frequency in the upper left
corner of Fig. 2-7).
Each side of the quad loop is twice the value shown in each cell of the X and Z
columns. At this point, the modeler can run the program as is or alter other model
parameters, such as the frequency sweep or the conductivity of the model wires. All
normal tabular and graphical outputs will be available for analysis.
The comparative data shown in the graphs in this chapter were produced by
running the design model through each wire diameter increment. The exception, of
course, is data for the baseline models, which consisted of individual models opti-
mized for resonance at selected wire sizes. The model results and the GW Basic
program results, of course, agree, since they both use the same equations to deter-
mine the dimensions, given any frequency and element diameter entry.
One of the conveniences of using the design model is that one can select any
desired frequency sweep parameters. The curves shown earlier used the span
from 28 to 32 MHz in 0.5 MHz steps. However, for most amateur design projects,
the frequency span is likely to be an HF or VHF ham band, with the steps selected
so that the sweep starts and stops at the desired band edges. For single designs, it
is convenient to use the built-in SWR plot button. As well, we shall find occasion to
use the Z button to plot the change in resistance and reactance across a desired
passband. The change in reactance will be, in some cases, of special interest.
The set-up shown for NEC-Win Plus will apply in general terms to any other
program having a design-by-equation facility. However, there may be differences of
detail related to the order of the set-up and the manner in which variables and equa-
tions are entered.
Conclusions
It is certainly legitimate to ask why I have gone to such fussy levels of explana-
tion and correlation with NEC models in the presentation of the simple BASIC pro-
gram to calculate quad loop circumferences. The answer has two parts.
1. We began with a traditional wire cutting formula from sources that actually
were intended only for use with thin wire multi-element quad arrays. That formula
has always simply been presented as if it were correct, and it has become embed-
ded in the minds of antenna builders. I did not wish this presentation merely to
present an alternative “formula” in equally simple terms, trusting only to the belief
potential of readers.
Instead, my aim has been to show both the method and the rationale for gener-
ating a usable design program that is adequate to virtually all applications for single
quad loops. Correlating the program to NEC-2 models of quads loops--which are
themselves very accurate relative to such antennas in the real world--provides a
degree of confidence in the adequacy of the result.
2. The circumference of a single resonant quad loop turns out not to be ame-
nable to a simple cutting formula at all. The required perimeter length is a direct
consequence of the wire diameter selected for the loop. Of equal or greater impor-
tance than the design program is an appreciation of the fundamental properties of
the quad loop, especially as it forms the basis for increasingly complex quad arrays.
We have been able to survey those properties rapidly partly because of the ease
with which the design program provides us with resonant quad loops of any wire
diameter and at any desired frequency.
It will be useful to keep in mind the general properties and patterns relating to
quad loop gain, SWR, and feedpoint impedance as we move from the single quad
loop to multi-element arrays. One of the questions that we shall keep in the back of
our mind is this: which properties of quad arrays emerge from the quad loop itself
and which arise from the parasitic assembly that we create by combining quad
loops? Just as understanding a Yagi is impossible without an understanding of the
dipole, so, too, understanding a quad beam requires a preliminary understanding of
the simple 1-wavelength loop.
The design program that we have developed, of course, even within the limits of
its accuracy, applies only to square and diamond loops that are individually reso-
nant. The required dimensions of the loops within a quad beam using 2 or more
elements will differ from the di-
mensions of a single quad loop,
just as do the linear elements of
a Yagi relative to a single resonant
dipole.
Chapter 3
2-Element Quads as a Function of Wire Diameter:
Understanding Some Quad Properties
There are some very old formulas for cutting the lengths of 2-element (and
larger) quad beam elements, formulas that have persisted since at least the 70s, if
not before. The driven element should be 1005/f(MHz) feet long, while the reflector
should be 1030/f(MHz) feet long. Spacing should be between 0.14 and 0.2 wave-
lengths. (See, for example, The ARRL Antenna Book, p. 12-1, in most recent edi-
tions.) Moreover, the quad has been called a very “low-Q” antenna, meaning that it
is wide-banded compared to other beam antennas, presumably including Yagis.
Unfortunately, using the formulas will result in a relatively poor 2-element quad
at any spacing. As well, we should not be too attracted by the so-called low Q of the
quad, because that feature has very restricted application to SWR. A quad, as we
saw in Volume 1, is not a low-Q antenna when it comes to other operating band-
width considerations
In fact, we should very likely start all over again, beginning with the one piece of
information that early quad builders thought was too insignificant to notice: the wire
size. In this and the next two chapters, we shall examine the properties of 2-ele-
ment quad beams using a driver and reflector based on the wire size we select for
the elements. In this chapter, we shall look at quad properties based on very careful
modeling with NEC. In the next chapter, we shall provide a way of automating the
design process. In the final part of our work with 2-element monoband quads, we
shall look at a way to improve the operating performance bandwidth of wire quads.
There are many operating specifications that we might emphasize for the de-
sign frequency we choose. For this exercise, I shall pick two, letting the others
become what they will in the designs. First, the driven element will be resonant at
the design frequency. Before we are done, we shall show how and why to vary that
parameter without significantly affecting the reflector. Second, we shall select a
spacing between elements and a reflector length that provides maximum 180-de-
gree front-to-back ratio at the design frequency. These choices are consistent with
fundamental work done on VHF models by Dan Handelsman, N2DT, and David
Jefferies, G6GPR, and generally provide the widest operating bandwidth for most of
the quad beam’s other parameters.
The range of wire sizes that we shall examine will run from 1E-5 to 1E-2 wave-
lengths. Refer to charts in Chapter 1 for detailed correlations with standard wire
gauges. By way of review, 1E-5 (0.00001) wavelengths is about #20 AWG at 80
meters and successively smaller in physical diameter as the frequency increases.
1E-3 (0.001) wavelengths is about #12 at 2 meters and increasingly fatter physically
as the frequency goes down. Although the single quad loop exercises in Chapter 2
used wire sizes down to 1E-6, such wire sizes are impractical at any HF or VHF
frequency and have been omitted. As well, as we shall discover, these wires sizes
result in impractical arrays with respect to performance.
We shall follow the same procedure that we used for the 1-element quad loop in
the preceding chapter and check not only the rounded values (for example, 1E-4)
but as well intermediate values of an order that yields -x.5 as common logarithms.
Not only for single element quads, but for virtually all quad designs, we may more
easily track dimension and performance curves using the common logarithm of the
wire diameter expressed in wavelengths than by using the wire diameter itself.
Models for each wire diameter were developed to obtain element lengths and
spacing to resonate with driver within +/-j1 Ohm of remnant reactance. As well, the
front-to-back peak value had to exceed 50 dB. NEC-2 or -4 are equally adept at this
task, since the quad presents no pressure on either core’s limitations until the wire
size exceeds 0.01 wavelength. The models used 21 segments per side to ensure
good convergence. 11 segments per side would have yielded equally valid model-
ing results. However, in examining a sequence of models from which curves and
other generalized data are to be drawn, it is important to retain a single segmenta-
tion if it is satisfactory. Changes in segmentation will result is operationally insignifi-
cant differences in output values, but the resulting curves will be displaced a notice-
able amount numerically at the point of a significant change in segmentation.
The models were calibrated for the most common material used in quad beam
construction: copper. Test aluminum models showed little change in characteris-
tics relative to the baseline copper models, but the slightly higher wire losses can
add roughly an Ohm to the feedpoint impedance—less than 1% of the actual value.
The test frequency was 28.5 MHz, about the geometric mean for the combined HF
and VHF frequency range.
Since copper wire material losses vary with frequency, but at a rate that differs
from the changes in wire diameter, there will be slight variations from the results to
be shown at the extremes of the frequency range. At very low HF frequencies, the
gain will be higher, but only by a maximum of a few tenths of a dB. Since the gain
increase results from lower material losses, the source impedance will be lower—
perhaps as much a 5 Ohms at 80 meters. Conversely, gain at VHF for a given wire
size will be very slightly lower than at 28.5 MHz, while the feedpoint impedance will
be correspondingly higher. However, these slight variations occasion no significant
changes in the physical dimensions of the quad as a function of a wavelength.
Before examining the properties of 2-element quad beams in detail, let’s take a
summary view of the modeling results in the form of Table 3-1. All sizes and lengths
are in wavelengths. We shall use our shorthand notation for wire sizes. Gain is the
free-space value.
The key element in this table is the value set for the thinnest wire. Notice the
gain associated with this wire size. As the wire size increases—given our design
criteria of resonance and peak front-to-back value—the frequency at which peak
gain occurs decreases. For a wire size of about 3.16E-5, the maximum front-to-
back value and the peak gain value occur at about the same frequency. For all wire
sizes greater than 3.16E-5, the peak gain frequency is below the design frequency
and grows more distant from it with increasing wire size.
In contrast, for wire sizes below 3.16E-5, the peak gain value occurs at a fre-
quency higher than the design frequency. At the design frequency the gain value
may be quite low, since the gain decreases more rapidly at frequencies below that
at which peak gain occurs. The result is that 2-element quad beams with any per-
formance potential at all and designed to the criteria used in this exercise are not
feasible in wire sizes below 3.16E-5. As a result, model results for the thinnest wire
in the table have been removed from the graphs to follow.
The first significant feature appears in Fig. 3-2, which plots the required element
spacing for maximum front-to-back ratio against the wire sizes. The thinner the
wire, the closer that spacing must be to achieve maximum front-to-back values at
the design frequency. Notice that the curve becomes nearly linear with wire sizes
above 1E-4, which is thinner than #14 at 10 meters. The shallowness of the curve
above this value indicates that for most practical monoband beam designs, 0.17
wavelength spacing represents a limit. However, since tuning a 2-element quad for
peak front-to-back also tends to result in the widest operating bandwidth for most
parameters, it becomes advisable to use the spacing that is correct for the selected
wire size.
Fig. 3-3 presents the driver and reflector element circumference lengths. Of
special note is the fact that the reflector length increases at a faster rate than the
driver length. In practical terms, the fatter the wire, the more the length of the
reflector must exceed the length of the driver for peak performance.
Combining the results for spacing and element length yields the conclusion that
inter-element coupling increases with wire size. Arriving at the correct value for a
given wire size involves both element length and spacing so that the parasitic ele-
ment currents have the correct magnitude and phase for maximum front-to-back
performance. This performance will only be achieved if both the upper and lower
wires of the reflector have close to optimal currents at their centers—and thus an
optimal current distribution along their length. Part of the reflector sizing curve is a
result of this function.
In Fig. 3-4 we have the resonant feedpoint impedances of the modeled quads.
The general increase in value is apparent. Part of the stair-step nature of the curve
results from the fact that close to resonance, the resistive component of the imped-
ance can show significant changes as we change the remnant reactance from close
to j-1 Ohm to +j1 Ohm. In the value range of the graph (125-155 Ohms), the stair-
step result is visually apparent, but not operationally significant. However, the over-
all difference between the thinnest and the fattest wire values may be significant for
the method chosen to match an optimized quad beam to a given main feed line.
The change of gain with respect to wire size, shown in Fig. 3-5, must be read
with great care. For wire sizes greater than 1E-4, the gain increase with additional
wire thickness would not itself justify using a larger wire size. However, remember
that as the wire thickness increases, the frequency of maximum gain becomes
increasing lower than the design frequency. One consequence of this fact is that for
thicker wires, the amount of gain change across a given amateur band will be less
than for thinner wires. For example, consider a 2-element quad at 10 meters with a
28.5 MHz design frequency. Using #14 wire, the gain at 29 MHz will drop to under
6.5 dB, despite the 7+ dB value at the design frequency. With 0.5" elements, the
gain at 29 MHz will be above 6.7 dB. In general, fatter elements provide both better
and smoother performance across the ham bands.
There are other dimensions to operating bandwidth that are worth noticing. Two
are included in Fig. 3-6. The graph shows clearly the VSWR bandwidth as a per-
centage of the design frequency. Even the thinnest wire on the graph shows a 6.2%
operating bandwidth, while the span from 28 to 29 MHz is only about 3.5% of the
design frequency. It is the wide SWR range that has given the 2-element quad
beam the illusion of being a low-Q antenna.
However, notice the second curve on the graph, which provides the front-to-
back operating bandwidth. Because a 20-dB front-to-back ratio can be commonly
achieved with Yagi design across a relatively wide frequency range—that is, a range
covering all or almost all of a ham band—the operating bandwidth was defined as
the percentage of design frequency over which the model had a front-to-back ratio
in excess of 20 dB. These percentages are very low and do not reach the 3.5%
level until we have a wire size of about 3.16E-3. This value represents a wire over 1"
in diameter at 10 meters. Since quads are rarely made from wires this thick at HF,
the quad user will generally have to be content with front-to-back performance at
the band edges that is inferior to that which a well-designed Yagi may provide.
(Note: this comparison only suggests that the rate of change of a 2-element
driver-reflector Yagi optimally spaced for maximum front-to-back ratio will be less
than for a 2-element quad. However, such a Yagi may have a peak front-to-back
ratio of only about 12 dB, with lesser values away from the design frequency. As
well, the 2-element Yagi will have a lower gain—with a considerable variance across
a band as wide as 10 meters. Hence, the 2-element quad will generally outperform
a 2-element Yagi when both are designed to the same criteria. The point of these
notes about quad performance is not to assess the overall superiority of one an-
tenna type over the other. Instead, these notes call attention to the specific proper-
ties of the quad that must be taken into account when designing one or when reach-
ing an overall evaluation of the quad’s performance—especially, when considering
options within quad design factors.)
When we initially examined the gain of quads relative to wire size, we noted the
frequency position of the gain peak and its influence on the rate of change of gain
across an intended band of operation. Fig. 3-7 formalizes those notes by showing
the rate of gain change per 1% of design frequency in terms of dB. Let us use the
1E-4 wire size as a marker, since it is the first data point where peak gain is signifi-
cantly lower in frequency that the maximum front-to-back frequency. The curve is
nearly linear from that point through 1E-2 wire sizes. In fact, for each decrease of
wire size by a factor of 10, the rate of gain change across a working passband
increases by nearly 1/2 dB. Wherever stable gain is required across a passband,
the fattest element diameter feasible is the order of the day.
Although the graphs based on element wire size have much to teach us about
the performance of 2-element quads, they cannot teach everything. For example,
once one has designed a 2-element quad for a chosen wire size, it is usually a good
procedure to reresonate the driver from the mid-band design frequency to a lower
frequency—somewhere between 1/4 and 1/3 the way up from the lower limit of the
passband. The obvious questions is this: how do we arrive at such a recommenda-
tion?
The design frequency values can be easily identified. Let’s begin with the VSWR
curve. Note that the SWR value increases much more rapidly below the design
frequency than above it. If we wish to have roughly equal SWR values at the band
edges of a chosen operating passband, then it will be necessary to reduce the
resonant frequency of the driver. In fact, the small change in element size needed
to effect this change will have negligible effect on the performance of the reflector or
the frequency of maximum front-to-back. For example, reducing the resonant driver
frequency to about 28.3 MHz will move the front-to-back ratio peak by only a few
kHz.
Such a move will likely be beneficial (although minimally so) to the overall front-
to-back performance within the passband—for example 28 to 29 MHz. Similarly to
the SWR performance, but to a lesser degree, the front-to-back value changes
more slowly above the design frequency than below it. From the graph, you may
determine that 0.3 MHz above design frequency, the front-to-back ratio is above 20
dB, while 0.3 MHz below design frequency, the value is only about 18.5 dB. In fact,
using the span from 28 to 29 MHz as a passband definition, there is a 3 dB differ-
ence in the front-to-back ratios at the passband edges.
It is possible from the graph in Fig. 3-8 to create an illusion—namely, that there
is good performance to be had from the antenna above 29 MHz. However, remem-
ber that across the passband, the antenna gain is steadily decreasing, and it contin-
ues to decrease as we raise the frequency further. By the frequency at which the
SWR approaches 2:1 on the high side of the design frequency, the gain has dropped
to about 5.5 dBi, and the front-to-back ratio is only about 7.4 dB. At 29.7 MHz, the
gain is down to 6 dB, with a front-to-back ratio of about 10.6 dB. Although usable for
some purposes, these figures are down considerably from the design frequency
values.
The point of the exercise has been to demonstrate the changes in 2-element
quad performance as we change wire size. I noted early on that one would have to
make some adjustments as we change frequency, since the graphs are calibrated
to copper wire. For a better sense of the degree of change, let’s sample our 3.16E-
4 wavelength diameter wire quad at a number of frequencies. Note that 28.5 MHz is
the original design frequency. All of the quads use the same length elements and
spacing in terms of fractions of a wavelength.
For thinner wire sizes, the variance will be greater, while fatter wires will show
less variance.
The small range of the variance should tell us two things. First, the general
properties of 2-element quads are indeed largely a function of the diameter of the
wire we use for the elements. In general, performance—especially the operating
bandwidth of most essential performance specifications—improves with the use of
larger diameter elements. Unfortunately, this fact is at odds with the conventional
ways in which we construct quads. The supporting structure for quad elements will
handle thinner wire but certainly not aluminum tubing. However, there are alterna-
tive ways of simulating fat elements that make use of wire. Remember that the
inter-element coupling is the key contribution of thicker elements. Material losses
are secondary. Hence, if we can increase the coupling through the use of light-
weight simulated fat elements, we may generally ignore their slightly higher material
losses. We shall defer that task to the third part of the chapters on 2-element
quads.
Chapter 4
2-Element Quads as a Function of Wire Diameter:
Automating the Design Process
In Chapter 3, we examined
some of the properties of the 2-
element monoband quad beam
as they emerged from the wire
size. For that exercise, we used
a series of graphs based on se-
lected models. The models used
wire diameters from 3.16E-5
through 1E-2 wavelengths in di-
ameter so that the X-axis of each
graph would follow a linear pat-
tern according to the common
logarithm. The Y axes of the
graphs explored various proper-
ties of the beam. The design
specification for each model was that the driver should be resonant within +/-1 Ohm
of remnant reactance and that the wire dimensions and element spacing should
produce a maximum value of 180-degree front-to-back ratio. Fig. 4-1 shows the
salient quad dimensions.
The task of finding suitable beam dimensions was considerably eased by the
use of “modeling by equation” techniques. The program used was NEC-Win Plus,
but any other NEC-2 or NEC-4 program with a similar facility would work as well. All
models used copper wire and were in free space.
Fig. 4-2 shows the equations page of the model used to explore quad proper-
ties. The standard design frequency (variable G) is 28.5 MHz, which is roughly the
geometric mean between the lowest HF frequency for which a 2-element quad might
be built and the high VHF frequency to be used. By moving the design frequency
down from a more perfect 30 MHz, the resulting model beams could be compared
more easily to physical beams built for the 10-meter amateur band.
For each wire size selected for variable H (multiplied by the wavelength in cell
D3 to provide physical dimensions for the program), it became necessary only to
vary three values. A is 1/8 of the driver circumference, while B is 1/8 of the reflector
circumference. D is the spacing between elements. As a matter of course, I exam-
ined spacing in 0.005 wavelength increments to find highest front-to-back peak and
then refined the spacing.
Fig. 4-3 shows the equations version of the wires page. The wires correspond-
ing to the variables are clear enough. Each model was swept in 0.01 MHz (10 kHz)
increments to track the pattern of both resonance and the maximum front-to-back
ratio. The closer to resonance, the smaller the change to the driver length variable
A. The closer to maximum front-to-back ratio, the smaller the changes to the reflec-
tor length variable B. Then the spacing would be changed, with a further zeroing-in
exercise until no further improvements could be made.
Note the Radiation Pattern entry. The gain and front-to-back ratio were obtained
from a tabular readout. Since I needed only the 90-degree mark for gain and the
270-degree mark for rearward gain, I set the parameters for the radiation pattern to
produce only 4 values. This set-up will not yield any usable graphical patterns, but
it does simplify scanning the output tables.
From this preliminary work, which one might replicate either in greater detail or
for other antennas, I obtained the data points for the graphs in Chapter 3. These
graphs are simple “connect-the-dot” constructions that are useful for seeing pat-
terns. However, they do not provide any basis for calculating 2-element quads for
other design frequencies.
I subjected the data obtained from the systematic modeling exercise to regres-
sion analysis. At one time, such analysis was painstakingly slow. However, pro-
grams like DataFit automate the process, providing both tabular and graphical out-
puts to test the equations that result. For the 2-element quad, 4th-order equations
yielded results that fell within about 0.5% of optimized models throughout the physi-
cal wire size limits and the frequency limits of the project, with the exception of very
thin wires. Where the common log of the wire diameter in wavelengths was -4 or
less, the error rate increases slightly. In all test cases, the resulting model had its
resonant frequency and its maximum front-to-back frequency within about 10 kHz
of the requested design frequency. These results were judged to be well within the
construction variables for most antenna-building situations.
There is a limit to how well regression analysis can track data points. Higher-
order equations tend usually (but not always) to provide a better track than lower-
order equations. 4th-order equations were the maximum possible with the limited
number of data points used. Some data point sequences make easy curve fits for
regression analysis. Fig. 4-4 shows the reflector circumference curve produced by
the analysis along with the original data points. A more exacting fit is hard to imag-
ine. However, Fig. 4-5 shows a more difficult fit between the curve and the data
points for the rate of gain change. Although the curve comes close to the data
points, the reversal of direction for the thinnest wire makes the shape of the curve
peak more open to question. Remember that with the thinnest wire used, the 2-
element quad reaches a coincidence between the frequency of maximum gain and
the frequency of maximum front-to-back ratio. Hence, the rate of change is less
than the next thinnest wire sized used. From 0.0001 wavelength wire diameters
onward, the frequency of maximum gain is always lower than the frequency of maxi-
mum front-to-back ratio. For the purposes of the advisory approximation, the curve
and the regression equation is perfectly adequate.
The equations produced by regression analysis are perfectly adequate for cal-
culating all of the data we examined in Part 1 for all points between the listed data
points. However, the equations have no inherent theoretic import for electronics or
antennas beyond their ability to calculate.
The production of a set of calculating equations does have the merit of allowing
one to create a small program in any number of media to automate the design
process for 2-element quads that meet the basic specifications (resonance and
maximum front-to-back ratio on the design frequency). By specifying the wire size
and the design frequency, we can let the program generate the remaining data.
Therefore, I produced the following little GW Basic program to do just this task.
300 SW=(AW*(D1^4))+(BW*(D1^3))+(CW*(D1^2))+(DW*D1)+EW
310 AF=-.00266666667#:BF=.388#:CF=4.790666667#:DF=19.55485714#:
EF=28.76628571#
320 FB=(AF*(D1^4))+(BF*(D1^3))+(CF*(D1^2))+(DF*D1)+EF
330 AN=-.08333333333#:BN=-.9462962963#:CN=-3.943055556#: DN=-
7.582671958#:EN=-5.23234127#
340 DG=(AN*(D1^4))+(BN*(D1^3))+(CN*(D1^2))+(DN*D1)+EN
350 WL=299.7925/F:PRINT “Wavelength in Meters =”;WL
360 WF=983.5592/F:PRINT “Wavelength in Feet =”;WF
370 PRINT “Quad Dimensions in Wavelengths, Feet, and Meters:”
380 PRINT “Driver Side =”;(DE/4);” WL or”;(DE/4)*WF;”Feet or”;(DE/
4)*WL;”Meters”
390 PRINT “Driver Circumference =”;DE;” WL or”;DE*WF;”Feet
or”;DE*WL;”Meters”
400 PRINT “Reflector Side =”;(RE/4);” WL or”;(RE/4)*WF;”Feet
or”;(RE/4)*WL;”Meters”
410 PRINT “Reflector Circumference =”;RE;” WL or”;RE*WF;”Feet
or”;RE*WL;”Meters”
420 PRINT “Reflector-Driver Space =”;SP;” WL or”;SP*WF;”Feet
or”;SP*WL;”Meters”
430 PRINT “Approximate Resonant Feedpoint Impedance =”;ZR;”Ohms”
440 PRINT “Approximate Free-Space Gain =”;GN;”dBi”
450 PRINT “Approximate 2:1 VSWR Bandwidth =”;SW;”% of Design
Frequency”
460 PRINT “Approximate >20 dB F-B Ratio Bandwidth =”;FB;”% of
Design Frequency”
470 PRINT “Approximate Rate of Gain Change =”;DG;”dB per 1% of
Design Frequency”
480 INPUT “Another Value = 1, Stop = 2: “;P
490 IF P=1 THEN 10 ELSE 500
500 END
The program does not contain a module to convert AWG wires gauges into
physical diameters, so Table 4-1 may be useful as a set of checkpoints. See Chap-
ter 1 for a full set of values for wires gauges from 1 to 40.
Fig. 4-6 provides a truncated view of the screen data produced by the pro-
gram. The test case is a 28.5 MHz quad using wire just slightly larger than #18
AWG. Conveniently, the selected wire size is 0.0001 wavelength in diameter. The
remaining entries show the calculated data.
The following table parallels the data from the program and from the 28.5 MHz
test model that served as the 0.0001 data point. (“DF” means “design fre-
quency.”)
There are some cautions to be observed in using the program or its equations.
At the frequency extremities of the program, that is, at low HF or middle to upper
VHF, certain systematic variations will appear between the calculations and actual
models of the antenna. They are best illustrated by reference to Table 4-3 . The
28.5 MHz reference frequency corresponds to the program design frequency and
was used in Table 4-2 above.
The rate of change of material losses in real materials does not occur at the
same rate as the change in inter-element coupling. NEC calculates efficiency solely
on the basis of material losses, which the table shows to increase with frequency if
the wire diameter is held constant as a function of a wavelength. Increased material
or resistive losses also appear as increases in the source impedance, not to men-
tion small reductions in the antenna gain. Conversely, well below the design fre-
quency, efficiency increases, gain increases, and the source impedance decreases.
Had the program been calibrated for lossless wire, there would have been no
difference in the results at any of these frequencies. However, calibrating the pro-
gram in terms of copper wire provides a more realistic basis for planning. The
differences between copper and aluminum will be minimal. It is probably useful to
note also that few amateur quads for 80 meters will be constructed from 3.37" diam-
Chapter 4 ~ 2-Element Quads as a Function of Wire Diameter:
Automating the Design Process
60 Cubical Quad Notes
eter wires, and equally few built for 2 meters will use 0.0082" wire (about #32 AWG).
Nonetheless, the advice given in Part 1 to use the fattest element diameter pos-
sible—or a simulation of a fat wire--remains valid.
For those using a NEC program with a “model by equation” facility, such as
NEC-Win Plus, the equations in the GW Basic program can be entered directly into
the model itself.
Fig. 4-7 shows the equations page for such a model. Columns E, F, and G
contain the constants and equations for determining variables A, B, and C. Since
each final equation (line 7) produces a value for the driver, reflector, and spacing in
wavelengths, line 8 converts these values to physical values using the current di-
mensional units (inches in the figure). Of course, the values for the driver and
reflector are for the total circumference, so the required values for A and B are 1/8 of
the line 8 numbers.
Although this model has only the dimensional equations entered, there is no
reason why one cannot also enter the other equations. The spread sheet is fully
functional and has sufficient columns (out of sight to the right) to handle the supple-
mental calculations. One need only enter the design frequency and the wire size (in
current dimensional units) to obtain results for both the model and the supplemental
information. (Note that the spreadsheet here does know a LOG from a LN, so the
conversion factor has been omitted in column D, line 4.)
Of course, one also needs to attend to the frequency or frequencies for the
model, as noted in the upper left corner of the figure. In this case, a sweep of 0.01
MHz each side of the design frequency provides sufficient data to determine the
resonant frequency and the frequency of peak front-to-back ratio. The radiation
pattern can be modified for a full azimuth pattern, if one needs it.
The model itself appears in dimensional form in Fig. 4-8. The dimensions cor-
respond precisely to the output from the Basic program, once one moves from feet
to inches or back the other way.
The model in the figure has been set in order to correspond to our initial design
case. However, let’s survey a few designs using material that may be more likely at
the chosen frequency than the constant 0.0001 wavelength wire we have so far
used. In Table 4-4 below, C means calculated by the program and M means mod-
eled results. In all cases, the reference models on which the program is based
show a front-to-back ratio that is greater than 50 dB. Hence, the Front-to-Back
column may be used as an indication of program accuracy, understanding that val-
ues change very rapidly near the peak. Hence, values of about 40 dB or so indicate
a peak within about 10 kHz of the design frequency.
The limitations noted earlier have appeared in full force. Thin wire models tend
to show up to 1 to 1.5% errors in some data, while fat wire models come very close
to calculated values. The break point falls at about the wire size where the log of the
diameter in wavelengths reaches -4. However, the actual frequencies of peak front-
to-back ratio are within 10 to 15 kHz of the calculated value. Thin wire models tend
to show more rapid decreases in front-to-back ratio relative to the peak value for
smaller changes in frequency. Consequently, the calculated values for antenna
dimensions would easily fall within construction variables.
Before we leave our 2-element bandwidth-optimized quads, let’s tackle one more
question: can we improve the operating bandwidth of 2-element quads (especially
the front-to-back ratio bandwidth) and avoid large, heavy, tubular elements, espe-
cially in the HF range? That will be our task in Chapter 5.
Chapter 5
2-Element Quads as a Function of Wire Diameter:
Fatter Elements from “Mere Wire”
In this chapter, let’s re-examine the reasons for wanting to use fat elements in
our quads. Then let’s perform a modeling test to see if we can get the simulation to
work well.
The simplest way to get a handle on why fatter elements are beneficial to 2-
element quads is to examine the performance of a couple of optimized designs.
The design frequency will be 28.5 MHz. The design passband will be 28-29 MHz:
we shall be interested in performance across that span, and not just at the design
frequency.
The element diameters chosen are 0.0641" and 0.5". The thinner wire corre-
sponds to #14 AWG copper wire, perhaps the most popular quad element material
used in the U.S. The half-inch size is arbitrary, but sufficiently larger to show major
performance differences—differences that can make a difference in operation across
the designated passband.
Each beam was designed using the automated program and model presented
in Chapter 4. The #14 version uses a spacing of 0.158 wavelength. The driver
circumference is 1.012 wavelength long, while the reflector length is 1.071 wave-
length. The design frequency resonant impedance is 136.1 Ohms. The 0.5" diam-
eter version requires a spacing of 0.164 wavelength. The driver is 1.020 wave-
length long, while the reflector is 1.103 wavelength long. The resonant impedance
at the design frequency is 141.1 Ohms.
Fig. 5-1 shows the gain curves for the two quads. As we have noted, the thinner
the wire size for a 2-element quad, the closer in frequency are the maximum front-
to-back and the maximum gain points. Hence, the #14 quad starts at a higher gain,
being closer to the gain peak. However, the gain of the #14 version decreases more
rapidly across the passband of interest in this exercise. In contrast, the 0.5" version
has a lower rate of change across the band, which tends to even up gain perfor-
mance between band edges.
If gain were the only parameter in question, then there would likely be no reason
to work with fatter elements. However, we should also look carefully at the front-to-
back curve, shown in Fig. 5-2.
It is tempting to focus on the front-to-back peak, which is the same for both
antennas. However, of much greater importance is the front-to-back ratio toward
the band edges. The front-to-back ratio of the #14 antenna is barely 12 dB at 28
MHz and 16 dB at 29 MHz. 20 dB front-to-back ratio is a common amateur stan-
dard. For the #14 version, we achieve this goal only between 28.26 and 28.78 MHz.
In contrast, the 0.5" version of the antenna shows better than 20 dB front-to-back
ratio between 28.15 and 28.94 MHz, a 50% improvement in bandwidth between the
20-dB markers. At 28 MHz, the 0.5" antenna improves the front-to-back ratio by 4
dB over the #14 version, while at 29 MHz, the improvement is 5 dB.
To complete the record, Fig. 5-3 shows the comparative SWR curves of the two
antennas, each referenced to its own resonant feedpoint impedance. Once more,
we see the more rapid increase in SWR below design frequency than above it.
Since the SWR never reaches 2:1, we might be satisfied with either curve. How-
ever, for very long coax runs, the 0.5" curve is superior in reducing line losses.
For a “fat” element, we may construct an equivalent element from 2 wires spaced
by a distance that is not at all arbitrary. When using relatively thin wire to simulate
very thick elements, a three wire scheme may be required, but the transition from
#14 to 0.5" on 10 meters is far from needing the added wire.
For linear elements, such as used in Yagis, the technique is simple. We may
take each element and find its resonant frequency. Then we construct a wire pair
(shorted at the ends and the center) of the same length as the tubular element. We
adjust the spacing until we arrive at the same resonant frequency. For most simple
arrays of linear elements, the wire spacing used for one element will generally suf-
fice for all of them.
The dual wire element will have a higher material loss than the original tubular
element, since its surface area is smaller. However, the performance of a parasitic
array depends more upon the inter-element coupling than on material loss (within
limits, of course). The dual wire element restores the level of coupling that is re-
duced in the move from a fat to a thin element. Hence, the modeled performance of
Yagis using thick tubular elements and their dual wire equivalents is generally within
0.1 dB gain and indistinguishable with respect to the front-to-back ratio and the
impedance curve. These factors apply not only to the design frequency, but as well
across the passband.
For 2-element quads, with their closed element geometry, the construction of a
2-wire equivalent requires a different procedure—more of a trial and error tech-
nique. Constructing a 2-wire element for each of the quad’s elements requires a
trial spacing of the wires and then adjustment of the overall loop sizes to bring the
antenna to resonance and to maximum front-to-back ratio at the design frequency.
The feedpoint impedance at resonance is a good indicator that the wire spacing is
correct: it should be about the same as the fat-element antenna being simulated.
There are two main ways to construct dual-wire elements, shown in Fig. 5-4.
One method, shown on the left, involves placing a cross-piece at the loop sup-
port point—a sort of Tee-configuration. Then, for each element, identical loops are
built, spaced by the desired amount. At each corner (minimally), bridge wires be-
tween loops are required to ensure that each loop in the element has the same
current distribution. The mid-point between loops in each element—where the sup-
port arm is—represents the point for measuring the spacing between elements.
The second method is to construct loops in the same plane. This method is
illustrated on the right in Fig. 5-4. The two loops for each element will have different
circumferences, one larger and one smaller than the reference length used to cal-
culate them. Once more, bridge wires are necessary at the corners (at least) to
ensure equal current distribution along the wires of each loop. In this planar con-
struction method, each element is a constant distance from the other.
For our test antenna, I finally settled on a wire spacing of 5" for both the Tee and
the planar models. To see what is necessary in quad element adjustment for the
simulated fat elements, Table 5-1 may be useful.
For the Tee configuration, the actual spacing is +/- 2.5" for each of the loops.
For the planar configuration, the actual loop lengths are 4 x +/- 2.5" for each loop.
From the perspective of resonance and maximum front-to-back ratio at the design
frequency, it makes no difference whether one uses the Tee or the planar configura-
tion. The results are the same to 4 significant figures, with differences only in the
5th digit for numerical fussiness on my part.
More interesting at this point is the fact that each of the dual-wire antennas
requires a smaller driver and a larger reflector than the 0.5" antenna which they
simulate. In fact, the dual-wire driver is even smaller than that required for the more
closely spaced single #14 antenna. Whether these loop size adjustments will affect
performance remains to be seen.
Modeling dual wire elements involves great care. The 5" spacing at 10 meters
presses the limits of NEC-2. To ensure equal current distribution in the driver, one
must model a single wire of at least three segments, with the source placed at the
center. Then, 1-segment wires move at right angles to the center feed wire to the
required spacing limit. For 5" spacing, these wires were each 2.5" long. Since NEC
prefers that wires meeting at angles have similar segment lengths, 2.5" became the
standard segment length for the entire model. The planar model required different
levels of segmentation for the inner and outer loops in order to keep the segment
junctions parallel with each other. In the end, the models required between 645 and
670 segments. This level of segmentation still provided a large segment length-to-
wire diameter ratio, also desirable for accuracy.
Even with such care, the NEC-2 models pressed the core limits for the close
spacing of wires. Initial NEC-2 models showed a systematic 0.3 dB gain deficit
relative to the 0.5" model. Applying the average gain test in NEC-Win Plus to all of
the models produced values of 1.002 for both of the single wire models. However,
the dual-wire models yield values 0.969. Values close to 1.000 indicate a precise
model within the limits of the test (which is a necessary, but not a sufficient condition
of model adequacy). For many general purposes, values as low as 0.96 and as
high as 1.04 are considered very good. However, for correlating models, especially
when one design is a proposed substitute for the other, the average gain test values
were considered inadequate.
The models were reconstructed in NEC-4, which is, while not perfect, consider-
ably better than NEC-2 with respect to close wire situations. Unfortunately, the
average gain test was not available in the version of NEC-4 used. However, the
gain deficit dropped to 0.1 dB. I suspect, but cannot prove, that the remaining deficit
is a function of the core and not a real difference between antennas. (Proof will
have to await the next generation of modeling cores.) Part of my suspicion arises
from the fact that the dual wire models virtually eliminate material loss as a source
of reduced gain. The single #14 wire quad has an efficiency of 97.9% as a function
configuration (where “Tee” refers to the unseen supporting structure). The struc-
ture, including the feedpoint modeling, is apparent from the sketch.
Fig. 5-6 presents the gain curves across the first MHz of 10 meters for the 0.5"
antenna and its counterpart. Of note is the NEC-4 deficit of about 0.1 dB average
that we have previously noted. More important however is the shallower curve for
the dual wire antenna. The lower rate of change in gain tends to indicate that the 5"
spacing between wires is actually simulating a wire somewhat fatter than 0.5".
The impression given by the gain curve in Fig. 5-6 is confirmed in the front-to-
back curve in Fig. 5-7. The dual-wire version of the antenna actually increases the
passband for better than 20 dB front-to-back ratio from 28.1 to 29.0 MHz. This
operating bandwidth is not a function of the very slightly lower peak front-to-back
ratio at the design frequency. That difference only indicates that the 0.5" curve is
better centered at 28.5 MHz.
With respect to gain and front-to-back ratio, then, there is little to choose be-
tween the dual thin-wire and the single fat wire models.
The model construction of the planar configuration of the dual-wire quad ap-
pears in Fig. 5-8. The construction of an actual planar antenna using the dual wire
The front-to-back curve in Fig. 5-10 shows the same performance spread as
the equivalent curve for the Tee—a 28.1 to 29 MHz passband for better than 20 dB
front-to-back ratio. By now, it should be apparent why I have presented the Tee and
planar curves separately: placed on the same graph, we could not see one through
the other.
To illustrate this point, Fig. 5-11 presents the SWR curves for both dual-wire
antennas, along with the curve for the 0.5" model. The Tee and planar curves
overlie each other so closely that they are indistinguishable. In fact, for every oper-
ating parameter about which we might have concerns, the two versions of the dual-
wire antenna are indistinguishable. Moreover, the curves also suggest one more
time that the dual-wire antennas have a broader operating bandwidth in every im-
portant way than the 0.5" model for which they are a substitute.
Finding the exact wire spacing to be a precise substitute for the 0.5" element
model would have been an exercise in unwarranted fussiness. 5" is a nice round
number and convenient for modeling. #12 AWG wire (.0808" diameter) would have
yielded a different dual-wire spacing for the same equivalence to a 0.5" single ele-
ment. However, the more likely course of further experimentation should be to find
the spacing (in easy-to-handle numbers) that yields a true minimum front-to-back
ratio of 20 dB across the entire 1 MHz span of 10 meters.
The present exercise has been aimed at establishing the principles of dual-wire
simulation of fat-wire elements in the optimized 2-element quad. I suspect that the
wider bandwidth of performance with the 5" spacing suggests that a slightly wider
element spacing might show further gains. However, that increase would be of the
order of 0.001 wavelength for a total spacing of 0.165 wl at the design frequency.
Such differences would likely be lost in the variables of actual antenna construction.
since it requires fewer parts. There is no need for fixing the Tee support to the main
support arm. The absence of any significant difference between the Tee and planar
model performance suggests that the planar model would perform equally well on
either of the two main types of quad construction: the use of a spider hub and
slanting support arms or the use of a boom with flat-plane support arm structures.
The only possible deficit for the planar model is the need to have loops for each
element that have different circumferences.
designs, with an equal increase in surface area. Hence, before tackling such a
task, one would do well to assess the adequacy of the support structure.
In addition, the design of a multi-band 2-element quad using dual wire elements
for some bands requires a total redesign of the elements relative to conventional
values. Not only must the wire spacing be selected so that the coverage on each
band meets design specifications, but as well, the driver and loop sizes must be
refigured for the element spacing selected—along with the spacing between ele-
ments.
The design task is far from simple. However, it is feasible and may be one way
to bring quad design from the 1960s into the new millennium. There is no good
reason why quads should not enjoy the same high front-to-back ratios across the
ham bands as the Yagis with which they compete.
The matter of the gain deficit that appeared in NEC-2 models of the dual-wire
wider-band 2-element quads continued to disturb me. Even though the final deficit
was small in NEC-4 models, I still wondered if there was a way of rebuilding the
dual-wire models to eliminate any question of whether the deficit was real or a prod-
uct of core limitations.
I rebuilt the planar model in the following way: I eliminated the driven element
section for the feedpoint and ran both wires continuously from corner to corner. I
then placed a source at the center of each lower driver wire to simulate a parallel
feed of the two loops. The resulting model could use fewer segments, since the
minimum segment size was now about 5" to keep the segment junctions parallel
between the inner and outer loops of each element. Slight adjustments of driver
and reflector brought the model to the optimal conditions of resonance and maxi-
mum front-to-back value. The Average Gain Test in NEC-2 registered 0.9996, a
value considered to indicate a highly accurate model.
The resulting frequency sweep with the model on both NEC-2 and NEC-4 yielded
an average gain differential of 0.025 dB between cores. At 28.5 MHz, the NEC-2
dual wire model gain is now within 0.06 dB of the 0.5" single wire model. There are
no significant differences in the front-to-back and the SWR curves.
Fig. 5-12 shows the wire geometry of the final model. Wires 1-8 and 13-20 are
the element loops, with wires 9-12 and 21-24 being the corner links. These links are
crucial to maintaining symmetrical currents in the elements, both in the model and in
a physical quad built to these specifications. In fact, additional interloop links would
be in order for a physical quad.
The source can be handled in one of two ways. One may place two sources on
the model, one at the center segment of each bottom driver loop wire. The imped-
ance for the array then becomes the solution to a parallel impedance problem. Treat-
ing the resistive and reactive components separately, one may take the reciprocal of
each resistive value and add them, finally taking the reciprocal of the result. Do the
same for the reactive components.
Alternatively, you may connect a very short (for example, 1") transmission line
with a characteristic impedance close to the value on one of the parallel sources (for
example, 300 Ohms in this case) between the two parallel source segments. Then,
eliminate one of the original parallel sources. The single reported impedance will be
within 0.5% of the value calculated from the dual sources.
The Table 5-2 summarizes the differences between the original and the alterna-
tive models at the design frequency. NEC-2 performance figures are shown.
Of course, the alternative model tends to confirm the earlier noted suspicion that
there is no significant gain difference—by even minuscule amounts—between the
dual-wire quads and the 0.5" single element quad.
Scaling a 2-wire design follows the same rules as scaling a design for a 1-wire
design. One cannot simply apply the 5" wire spacing used at 10 meters to a 20-
meter design and expect it to replicate a scaled version of the 0.5" 10-meter quad.
First on the list of problems is that a correctly scaled 20-meter version of the half-
inch 10-meter quad would have 1.0" diameter elements.
A correctly scaled 2-wire version of the 10-meter quad would end up with (roughly)
double the element lengths, double the spacing between elements, double the spacing
between wires, and double the wire diameter. The resulting wire diameter (0.1282")
is #8 AWG, a wire size not in general use in quad construction. It weighs about 4
times as much as the more common #14, although much of the weight differential
can be eliminated by the use of aluminum wire.
A trial 20-meter quad perfectly scaled for 14.25 MHz covered 14.0 to 14.5 MHz
with the same properties as its 10-meter original. The natural tendency is then to
see if the same quad can be developed for something like #14 wire, perhaps by
spreading the wires further apart. If success includes a peak front-to-back ratio that
approaches 50 dB (one of the markers in developing a sequence of maximum band-
width 2-element quads), then success will be elusive.
Trial quads using decreasing wires sizes from #10 through #14 showed two
tendencies. Initially, a #10 version of the antenna using 13" spacing showed a
performance peak at the design frequency (14.25 MHz). However, the maximum
attainable front-to-back ratio was only 35.5 dB. The passband edges showed de-
creases in gain relative to the #8 scaled model. Further decreases in wire size
showed additional performance reductions and a decrease in the frequency of peak
performance--down to 14.1 MHz for #14 AWG wire, even with optimized spacings
that peaked at about 16".
wire surface area. The inter-element coupling becomes insufficient to simulate the
original 1" modeled elements. When the designer reaches this point, he has two
general options. One is to return to a fatter wire. The other is to model a potential 3-
wire substitute for the tubular element.
Despite such limitations, the 2-wire simulation of much fatter and heavier single
elements shows some potential for developing quads that have wider operating
passbands than traditional single-wire models. Since gain and operating bandwidth
are both functions of element diameter in 2-element quads, and since quads are
mechanically designed for thin-wire elements, obtaining the highest gain and the
widest bandwidth may call for such techniques.
However, these notes have only scratched the surface of the work needed to
make such designs routine. Much remains to be done, especially in the area of
developing substitute elements using 2, 3, and possibly 4 wires for the mid-to lower
HF region. As well, codifying the limitations of each substitute in terms of usable
wire size relative to the frequency and basic element diameter of interest remains a
task for the future. It is a field for the combined efforts of both experimenters and
modelers.
Chapter 6
Automating the Design of 3-Element Monoband Quad Beams:
A Wide-Band Model
The exercise of automating the design of 2-element quads raised the question
of whether a similar technique might be applied to 3-element quads. One answer is
in this set of notes. A second answer appears in the next chapter.
The director was sized and spaced to yield a good gain with a resonant feedpoint
impedance between 70 and 80 Ohms. In general, this procedure does not yield the
very highest possible gain or the shortest possible boom length. However, it does
produce a very good gain (as judged in quad terms) with the widest possible operat-
ing bandwidth. These results are consistent with the conclusions reached in Vol-
ume 1 as we explored spot designs of quads and attempted to optimize various
operating characteristics. For the design sequence we shall explore in this chapter,
the required driver-to-director spacing is nearly double that of the reflector-to-driver
spacing. Fig. 6-1 illustrates the relationships among the elements.
The following GW Basic utility program requires only the entry of the wire size
and the design frequency to set the calculations in motion. Refer to Chapter 1 for a
correlation between wire diameter and various wire gauges.
Besides the usual dimensional outputs, the program will also display the wire
diameter as a function of a wavelength. The performance data includes the ap-
proximate gain at the design frequency, the feedpoint impedance, the 2:1 SWR
bandwidth as a percentage of the design frequency, the >20 dB front-to-back band-
width as a percentage of the design frequency, and the rate of change of gain over
a span of 1% of the design frequency. Remember that the line with the “LOG” entry
is, for GW Basic, a natural log and requires a correction factor to create a common
log. If you translate the program to another medium, you can drop the conversion
factor if the medium recognizes common logs.
The dimensional portion of the program can also be placed in a model within a
program having model-by-equation facilities. Fig. 6-2 shows the equations screen
for a NEC-Win Plus model of the program, set for a design and test frequency of
146 MHz and a wire diameter of 0.0625". In this spreadsheet arrangement, LOG
already means a common logarithm and so the correction factor required by GW
Basic is omitted.
Only the design frequency and the wire size need be entered for each design
revision. No regression-derived equations for the supplemental data on operating
bandwidth is included, although a modeler might easily place those equations into
the spreadsheet. The added data in the GW Basic program can be directly derived
for any particular design simply by executing the NEC core run of the program. Use
the design frequency or a frequency sweep to derive the desired curves.
The effects of wire size on gain are as vivid for a 3-element quad as for a 2-
element quad, as shown in Fig. 6-3. In this figure, wire size is listed in wavelengths,
using values that translate into a linear progression of the logarithms of the wire
sizes. There is well over a dB difference in the gain of arrays using the thinnest wire
size and arrays using the fattest wire size. Moreover, the increase in gain over the
corresponding 2-element quad also increases with wire size. The thinnest wire size
3-element quad shows a 1.3 dB improvement in gain over a 2-element quad using
the same wire, whereas the fattest wire 3-element quad shows a gain improvement
of nearly 2 dB over its corresponding 2-element array.
Fig. 6-4 shows the change of maximum front-to-back ratio with increasing wire
size. Theoretically, the curve should be smooth and almost linear across the scale.
Since the checkpoint models were hand optimized, allowing the maximum front-to-
back ratio to occur as little as 10-15 kHz from the design frequency yields the flat
portion of the curve. However, in practice, this slightly less than optimal design
curve makes no practical difference, since constructing a quad so that its front-to-
back ratio maximum is precisely at the design frequency is more hope than reality.
Nevertheless, the increase of both gain and front-to-back ratio with wire diameter
demonstrates the importance that wire size has in effecting maximum mutual cou-
pling between quad elements. Thin wire quads of the sort we generally construct at
HF with #14 or #12 wire simply are not capable of achieving all of the performance
that a quad can provide.
The feedpoint impedances as a function of wire size appear in Fig. 6-5. Here,
the curve is very real and not a function of optimizing variance. With the thinnest
wire, the gain peak and the front-to-back ratio peak are very close together, yielding
less than a peak feedpoint impedance value. As the wire size increases, the gain
peak occurs well below the design frequency so that the front-to-back maximum
value dominates the production of the feedpoint impedance. As a general rule,
attempting to construct a 3-element quad with wire sizes less than 0.0001 (1E-4)
wavelengths is unwise if one has any hope of obtaining any operating bandwidth
from the array. Conversely, those services which operate on specific frequencies
are not limited by bandwidth constraints for antenna design. Indeed, in many in-
stances, the narrower the bandwidth of the array, the better the performance in
terms of immunity from coupling to nearby arrays that are relatively close in fre-
quency.
driver size. However, the required reflector circumference is shorter in the 3-ele-
ment quad than in the 2-element quad for any given wire size.
Interestingly, the director circumference does not follow the pattern for the other
two elements. As the wire size increases, the required director size decreases. If
we were to normalize the driver circumference so that it graphs as a straight line
across the page, the director line would move down at almost the same rate as the
reflector line moves up.
The spacing graphic, in Fig. 6-7, gives some precision to the earlier remark that
the driver-to-director spacing is about twice the reflector to driver spacing. In fact,
the spacing from the driver to the reflector increases with wire diameter—between
about 0.14 and 0.17 wavelength for the span of wire sizes included in the exercise.
In contrast, the required director-to-driver spacing decreases with increases in wire
size—from about 0.32 wavelength for the smallest wire to about 0.25 wavelength
for the fattest wire.
In the antenna modeling version of the GW Basic program, note that the wire
spacing equation for the director in the left-most column of the spreadsheet adds
the driver-to-director spacing to the reflector-to-driver spacing. This move simply
indicates that on wire geometry page, the reflector has been set at zero on the boom
with the remaining elements counting in a positive direction for their positions.
The full story of what happens as we change wire sizes becomes much more
evident if we perform some frequency sweeps. So I designed quads using three
wire sizes: 0.0131" (near to #28 AWG), 0.131" (near to #8 AWG), and 1.31". The
design frequency was 28.5 MHz, and the wire sizes correspond to 0.0000316,
0.000316, and 0.00316 wavelength diameters. The frequency sweep used 0.1 MHz
intervals from 28 to 29 MHz.
In Fig. 6-8, we have the gain curves across the first MHz of 10 meters for the 3
quads. The lowest curve for the thinnest wire shows the gain peak within 0.1 MHz
of the design frequency, with a rapid drop in gain at the low end of the band. For the
middle-size wire, the gain peak is evident at 28.1 MHz, while for the fattest wire, gain
is peak but flat for the first 0.2 MHz of the passband.
Equally evident to the gain advantage of the fattest wire is the very slow rate of
gain decrease compared to the thinner wires. The thinner wires show a full half dB
variance in gain across the passband, while the range of gain is only 0.2 dB for the
fattest wire.
The front-to-back curves in Fig. 6-9 show several things. First, the slight dis-
placement of the curve for the fattest wire downward in frequency by about 15 kHz
corresponds to the flattened portion of the front-to-back curve in Fig. 6-4. Apart
from that slight offset, the three curves are remarkable congruent with each other.
The rates of decrease from the peak are similar for all three curves and are parallel
both above and below the design frequency. Finally, note the steeper rate of de-
crease below design frequency than above design frequency. These curves are
fully consistent with those for 2-element quads.
A similar narrowing of the operating bandwidth applies to the 2:1 SWR dividing
line commonly used to denote acceptable performance, as shown in Fig. 6-10.
Only the fattest wire model covers the entire passband. By judiciously lowering the
resonant frequency of the middle-size wire model, it can be set to show under 2:1
SWR just about all the way across the passband. This fact results from the more
rapid rise in SWR below design frequency than above it. Since the entire set of
resonant feedpoint impedance is between 70 and 80 Ohms for all wire sizes at the
design frequency, all SWR values in the curves are referenced to 75 Ohms.
To provide a sample of the program’s output, here are some dimensions and
performance data for a few 3-element quads.
Although the quad array modeled here has an acceptable SWR across all of 20
meters, the front-to-back ratio becomes a limiting factor. On a crowded band such
as 20 meters, front-to-back ratio is very often an important antenna design consid-
eration. For most installations, therefore, the antenna would likely be designed for
either the CW/digital end of the band or for the phone end of the band.
Let’s compare this array with another for the same frequency.
The 0.5" wire quad shows all of the dimensional characteristics in comparison to
the #12 AWG version that we have seen in the curves. As well, 0.5" performance is
slightly up, while the feedpoint impedance is slightly down relative to the #12 wire
model. Most significantly, the SWR and front-to-back operating bandwidths for the
fat-wire model are 30% or more greater than those of the thin-wire array. Of course,
it is impractical to consider construction of a quad array for 10 meters that has half-
inch diameter elements. However, we shall return to this problem before we close
the book on this exercise.
The same 4-MHz bandwidth, when moved from 6 to 2 meters, presents less of
a problem for a 3-element quad composed of 0.001 wavelength wire. The >20 dB
operating bandwidth now covers about 80% of the band. The use of 0.25" wire
would easily permit the achievement of all benchmarks across the entire 2-meter
band.
In a past 2-element quad exercise, we looked at the use of spaced #14 AWG
wires to simulate fatter single wires. In that effort, we used 2 #14 AWG copper wires
spaced 5" apart and joined at the corners. We explored two different configurations
and found no significant difference between them. The resulting 2-element quad
easily replicated the performance of a 0.5" diameter quad, with a bit to spare. The
consequences of substituting 2 thinner wires for one fatter one were a slight en-
largement of the reflector and a slight decrease in the driver circumference.
Table 6-1 provides a comparison of the dimensions (in inches) between the two
models. Note that the dimensions for the dual-wire model represent positions half-
way between the two wires, so that the actual wire positions are +/- 2.5" relative to
the coordinates that would emerge from the listed dimensions.
Although the differences are small, they are significant in arriving at the final
operating characteristics of the array. While the dual-wire reflector is slightly larger
than the single-wire elements, the dual-wire driver and director are both slightly
smaller. As well, the dual-wire director is closer to the driver, resulting in a shorter
overall boom length for the array.
Performance for the 3-element dual-wire array parallels that of its 2-element
cousin. The model shows slightly less gain at the design frequency, but whether
this minuscule gain loss is real or an artifact of the closely spaced wires in the model
remains uncertain.
Fig. 6-12 shows the gain curves for both versions of the array from 28 to 29
MHz. Immediately apparent is the fact that the dual-wire gain decreases more
slowly than the single wire gain. Shallower gain curves are generally characteristic
of fatter wires with higher overall gain—a fact which contributes to the uncertainty
over the slight gain deficit in the dual wire model at the design frequency. However,
the gain differences between versions of the antenna would make no operational
difference at all.
A second piece of evidence that the 5" spacing of the dual wire model acts
similarly to a wire somewhat fatter than the 0.5" model appears in Fig. 6-13. The
front-to-back curve of the dual-wire version is slightly wider than that of the 0.5"
single-wire model. Again, the differences make no operational difference, but their
existence is numerically significant in the process of equating dual-wire arrange-
ments with corresponding diameters of single wires.
The slightly broader operating bandwidth of the dual-wire quad over its single
fat-wire counterpart is consistent with the results obtained in Chapter 5 for the 2-
element quad comparison. The consistency strongly indicates that the technique of
replicating fatter wires with multiple strands of thinner wires is not an isolated phe-
nomenon, but is, rather, generally applicable.
Comparing the feedpoint impedances between the two version of the array does
not permit an easy chart. The dual-wire model uses a dual feed system of driver
wires fed essentially in parallel. Hence, the composite feedpoint impedance re-
quired hand calculation. (The short transmission line technique of modeling the
dual-wire 3-element quad was not used in this case. It remains applicable, should
anyone care to replicate these results.) However, a table (Table 6-2) of values may
be equally useful in exploring the feedpoint situation. Resistances (R) and reac-
tances (X) are in Ohms.
Both antennas would easily cover the first MHz of 10 meters with a VSWR under
2:1, although the 0.5" model might require a slight adjustment of the driver to bring
its resonant point lower in the band. (Such adjustments to the driver, if modest,
have no significant effects on the other operating characteristics of the array.)
The bottom line on the exercise is that a set of dual-wire loops for a quad array
can effectively improve 3-element quad performance relative to the customary single
#14 AWG quad structure. Even if one discounts the gain advantage of the dual-wire
array as operationally marginal, the improvement to both the SWR and front-to-
back operating bandwidths is undeniably significant to all except those operators
who use only small portions of the wider amateur bands.
Despite these limitations, the automated designs that emerge from the utility
program shown in this chapter provide some useful starting points for developing
realistic 3-element monoband quad arrays that live up to their theoretical potential.
This wide-band design focuses on one potential improvement in quad array perfor-
mance. In every exercise of the type used here, there is a chance that a slightly
different (or perhaps even a radically different) sequence of baseline models would
yield even better wide-bandwidth performance at equal or higher gains for a 3-ele-
ment quad. To this point, there has been insufficient exploration of the fundamental
quad element performance factors to blithely guarantee these results as the very
best obtainable. I can only note that these results are the best I have been able to
obtain so far.
Perhaps the most sound principle used in obtaining the baseline data from which
the automated program was obtained is this: for every quad design, there is a
sequence of designs that will vary as a function of the element diameter and the
original design will have a place in the sequence. How the sequence progresses is
a function of which operating characteristics one emphasizes. In this case, maxi-
mum operating bandwidth, as defined in terms of front-to-back ratio, became the
Obviously, one might well use a different set of criteria to obtain a baseline data
set for a sequence of quad array designs based on element diameter changes.
While this chapter has focused on operating bandwidth, the next chapter’s high-
gain design focuses on another potential for quad arrays.
Chapter 7
Automating the Design of 3-Element Monoband Quad Beams:
A High-Gain Model
In the first part of this small study on automating the design of 3-element quad
beams, we explored a wide-band version of the array—at least, as wide-band a
version as we could obtain while still developing reasonably good gain. The results
yielded a quite feasible set of potential designs. As with any quad design, the fatter
the element used, the better the performance.
The high-gain design has a number of other interesting properties. For ex-
ample, the gain curve is roughly centered in the passband. Fig. 7-2 shows the gain
curves for the design using three wires sizes: 0.0001 wavelength, 0.001 wave-
length, and 0.01 wavelength. For the 2 thinner wire sizes, the gain peaks on the
design frequency. The fattest wire size shows a peak just above the design fre-
quency (28.5 MHz in this case), as well as a much smaller increase in gain relative
to the next smaller wire size. There appears to be a limit in this design to the gain
increase with increasing wire size, possibly connected with a maximum degree of
inter-element coupling.
As well, the front-to-back ratio peaks are not variable from one wire size to the
next. Instead, the front-to-back ratio remains constant within +/-0.5 dB of 30 dB
throughout the series of optimized models.
Automated Design
The following GW Basic utility program encapsulates the design data. As al-
ways, LOG in GW Basic means a natural logarithm and requires a correction factor
to yield a common log. If the program is entered in another medium that treats LOG
as a common logarithm, the conversion factor can be omitted.
The program deletes one piece of data useful to the wide-band design: the rate
of change of gain per 1% of frequency change. Since the gain peaks at the design
frequency rather than outside the passband of the antenna for an amateur band, the
rate-of-change figure loses its meaningfulness.
duce the output values are completely accessible for easy transport to any desired
medium. Most spreadsheets use straightforward embellishments of Basic.
For example, the spreadsheet in the modeling program NEC-Win Plus allows
one to enter the equations and produce a model directly. In fact, the only difference
between the equations page shown in Fig. 7-3 and the one shown in the last chap-
ter are the values of the constants for the regression equations.
The program outputs can be graphically presented to show the general trends of
the high-gain design dimensions. Fig. 7-4 shows the loop circumference dimen-
sions as a function of a wavelength for wires sizes that are also functions of a
wavelength. The reflector and driver curves are familiar to those who have looked
at the wide-band design. Although the exact values differ, the growing loop size as
wire diameter increases is a familiar feature. The director loop circumference, how-
ever, is another matter. It changes very little across the span of wire diameters, with
down-turns at both ends of the scale. In fact, as the wire size decreases to the
thinnest region, the wide-band and the high-gain design begin to resemble each
other, since the gain peak comes closer to the design frequency for each version.
The element spacing required for optimized high-gain design follows rules that
differ from those applicable to the wide-band design. The reflector-to-driver spac-
ing decreases with increasing wire size—exactly the opposite of the case with the
wide-band design. In contrast, the driver-to-director spacing shows a curve similar
to the director loop circumference graph, with down-turns at both ends of the wire-
size scale. Likewise, the range of variation is very small with the high-gain design.
In contrast, the wide-band model showed a significant reduction of required spacing
between the driver and director as the wire size increased. The high-gain model
curves appear in Fig. 7-5.
Dimensions are not the only parameters that lend themselves to graphical com-
parisons. As well, we can look at some of the performance predictions for both the
wide-band and the high-gain versions of the antenna. Most significant among these
parameters are gain, the 2:1 SWR bandwidth, and the >20 dB front-to-back band-
width.
Fig. 7-6 shows the comparative gains for the two designs. Once we move
above the thinnest wires, the high-gain version of the antenna shows an average
gain advantage of about 0.5 dB over the wide-band version. The gain advantage
does not come at the expense of significant changes in pattern shape. In fact, the
free-space azimuth patterns of the two designs are quite similar, as shown in Fig. 7-
7. For any chosen height above ground, the patterns would resemble these and
remain similar to each other.
The free space azimuth patterns are predicated on 0.5" diameter copper ele-
ments with a design frequency of 28.5 MHz. As we noted in preceding chapters,
elements of this diameter on 10 meters can be simulated effectively with #14 AWG
wires spaced about 5" apart. Because of the rear side lobes of the two patterns are
so similar, the differential in front-to-back ratios will not be operationally significant.
How significant the gain differential will be must be a user measure based on an
assessment of all of the critical parameters.
Fig. 7-8 shows the comparative values of 2:1 SWR bandwidths, where the band-
widths are taken as a percentage of the operating frequency. The graph shows an
advantage of better than 1.5:1 for the wide-band version of the antenna. However,
whether that degree of increase in SWR bandwidth is necessary depends upon the
bandwidth required for the operating goals of the individual installation. With re-
spect to the SWR bandwidth, any wire size that is 0.0003 wavelength will cover any
of the wider HF ham bands (excluding 80 meters). For equivalent coverage with the
high-gain version, a wire size equal or greater than about 0.004 wavelength is nec-
essary. Consult the tables in Chapter 1 for the bandwidth of each of the amateur
bands, as well as the wires gauge vs. diameter data. I shall merely repeat the
suggestion in that chapter that the reader place paper clips on these much-used
data table pages.
More critical to many operations than the SWR bandwidth is the >20 dB operat-
ing bandwidth. Fig. 7-9 shows the differences for the two designs. Once one passes
the 0.001 wavelength wire size, the average advantage for the wide-band design is
about 20-23%. Full coverage of the first MHz of 10 meters would require a wire size
of about 2.5" for the wide-band design, while the high-gain will not quite cover this
passband with a diameter (or its equivalent in multiple wires) of over 4". Nonethe-
less, half-inch wire (or tubing) may suffice for nearly full band coverage of 2-meters
using the high-gain design. Moreover, not every 3-element quad application needs
to cover the entirety of any of the wider HF ham bands.
To get a better sense of how the two designs differ, let’s compare some models
that alternately use the wide-band and the high-gain designs.
If you have read Chapter 5, you will recognize the sample 3-element quads,
since I used the same frequencies and wire sizes for the high-gain models in the
right-hand column. In each case the dimensional differences are significant. How-
ever, for most folks, the gain and the operating bandwidth comparisons will be more
interesting. In each case, the high-gain design shows its higher gain. Likewise, the
wide-band design shows the wider SWR and front-to-back bandwidths.
Conclusion
Nonetheless, these exercises are useful in many ways. Besides the obvious
benefit of yielding relatively optimized designs for any HF/VHF frequency and wire
size, the programs provide an easy means of designing quads for comparative pur-
poses. The evaluation of antennas at a practical level is very often a matter of
comparing available designs. Yagi comparators are easy to find. Up to now, an
adequate sampling of comparable quad designs has been hard to find. These utility
programs considerably ease the process of uncovering 3-element quad designs
and their capabilities.
One final caution: the designs are monoband quads throughout. One cannot
simply plug them into a multi-band array and expect each quad to perform as speci-
fied here. In multi-band quad arrays, virtually all elements are significantly active,
adding to or subtracting from the performance of the focal set of elements for any
band. Therefore, creating an effective multi-band quad array requires considerably
more adjustment than a mere field tweak on the monoband designs.
Chapter 8
40-Meter Wide-Band 3-Element Quad Designs
Let’s begin with a pop quiz. Suppose first that you wanted a directional 3-element
quad array to cover as much of 40 meters as possible. Suppose further that space for
the array is not problem. Finally, suppose that you had a choice among the following
three designs: the “Classic #14,” “the Single W-B #12,” and the “Dual W-B #12.”
Which would you choose?
So far, the question is not fair. So let’s provide some grounds for selection with a
few performance graphs. Fig. 8-1 shows the gain curves for the three choices. The
classic #14 peaks at about 7.15 MHz, but falls off rapidly on either side of the design
frequency, especially the low frequency side. Still, this is a normal curve for a quad
beam in almost every performance category. The single W-B #12 has its gain peak at
about 7.05 MHz, although the design frequency is higher. It is not clear from the graph
whether the dual W-B #12 has a gain peak at 7.0 MHz or somewhat lower in fre-
quency. From the graph, the antenna with the most consistent gain from one end of
the band to the other is clear.
The front-to-back ratio curves in Fig. 8-2 are equally clear. The peak front-to-back
ratio in all cases occurs between 7.1 and 7.15 MHz. However, the classic #14 does
not reach 20 dB (with the possible exception of a few kHz at its peak). The single W-B
#12 does considerably better, but still manages >20 dB front-to-back ratio for only
about 85 kHz. The dual W-B #12 has the widest >20 dB front-to-back ratio bandwidth
and never drops below 14 dB across the 40-meter band.
The classic #14 array is--near the design frequency--a 50-Ohm antenna, and the
SWR curve in Fig. 8-3 is referenced to that value. Unfortunately, the <2:1 SWR
region is only about 130 kHz wide. The single W-B #12 version of the antenna covers
about 200 kHz with under 2:1 SWR--this time referenced to the array’s near-75-Ohm
feedpoint impedance. Only the dual W-B #12 model manages under 2:1 SWR (refer-
enced to 75 Ohms) across the entire band.
For someone interested in the entire 40-meter band, there is a clear winner in this
selection process: the dual W-B #12 model. But, before we look at what this model
is, let’s review the lesser arrays in the group.
For at least a quarter century, we have been given a set of formulas for construct-
ing 3-element quads:
Many hams have believed that an independently fed quad loop answers to the
driver formula, but it does not--not even close. Likewise, other hams have believed
that we can make a 2-element quad using just the driver and reflector formulas. Any
such array is less than optimal. Few folks have noticed that the spacing is left vacant,
subject to our own construction limitations.
In fact, the formulas apply to fairly short 3-element quads using quite thin wire.
The formulas are not perfect and require some optimization to bring the gain,
front-to-back ratio, and SWR “best” numbers into reasonably close alignment. The
following design in #14 AWG copper wire provides good figures at the design fre-
quency (close to 7.15 MHz):
Although this array is capable of about 8.6 dBi free-space gain at the design
frequency, it has a very narrow operating bandwidth in every performance category.
The gain drops to about 7.5 dB (1.1 dB off peak) at the low end of the band. The
front-to-back ratio is below 10 dB for much of the band. The 2:1 SWR bandwidth
covers less than half of the 40-meter band.
The selected azimuth patterns that scan 40-meters in 100 kHz intervals tell
much of the classic #14’s story in graphical detail in Fig. 8-4. Beyond the center 100
kHz of the band, the pattern of the array is more bi-directional than directional.
For those who model and might wish to adjust the design to one or the other
end of the 40-meter band, the following model description may be useful:
3-element 40-meter classic quad Frequency = 7 MHz.
How might we improve the classic design? First, we should go to a larger diam-
eter wire. However, #12 AWG copper wire is about the largest value that most hams
will use for quad construction. That fact will initially limit the improvements we can
make. #12 AWG is only about 1.26 times larger in diameter than #14 AWG. If we can
recall the data from earlier chapters, we will instantly recognize that major improve-
ments in performance—especially operating bandwidth—tend to come when we in-
crease element diameters be at least a factor of 10, if not more. Increasing a #14
element by this amount would yield a diameter of about 5/8". Hence, the move from
#14 to #12 would be marginal, indeed.
Fig. 8-5 illustrates a second design move we can make: enlarge the spacing to
something nearer to optimal. The figure shows the profiles of the classic and the
improved designs, revealing that it will take
another 20+ feet of boom to get signifi-
cant improvements in operating bandwidth
on 40 meters.
perspective, it is their differences that produces a workable design for the refigured
quad. The dimensions that resulted from the design work are these:
Even with the longer boom length, the wire diameter we have chosen continues
to limit performance. #12 wire is a very thin conductor at 40 meters--about 5E-5
wavelengths. Consequently, optimizing loop dimensions and element spacing will not
provide full 40-meter coverage.
Gain is not the problem, since it holds to a free-space value of about 8.4 dBi or
better across the band. The front-to-back ratio, although significantly improved com-
pared to the classic #14 model, remains less than stellar. It actually falls below 10 dB
at the low end of the band and exceeds 20 dB for less than 100 kHz. The 75-Ohm 2:1
SWR passband is only about 180 kHz wide--about 50 kHz wider than for the classic
#14 design. Although the single-wire #12 wide-band design provides considerable
improvements in gain and front-to-back ratio relative to the classic #14 model, it falls
short of covering the band by a wide margin. Just getting full SWR coverage would
require a wire about 3-4" in diameter.
Fig. 8-6 shows our situation graphically through spot azimuth patterns across the
band. The single-wire array will simply not do the job we specified at the beginning of
this exercise.
Those who wish to try some adjustments to the array under discussion may ben-
efit from the following model description:
It proved possible to cover all of the first MHz of 10 meters with a very good
front-to-back ratio using a single 0.5" diameter wire or 2 #14 wires spaced 5" apart. A
40-meter quad with comparable coverage would need wires spaced more than 20
inches apart. As a more modest design project, I aimed simply to achieve an SWR of
under 2:1 across the band. This goal required the use of 10" #12 wire spacing.
The optimized dimensions for the dual-wire model yields an array that is close to
4' shorter than the single-wire version. The following dimensions list the circumfer-
ences of both the inner and outer loops for each element:
For our trouble, we obtain a smooth gain curve across the band with under 0.3 dB
variation. In addition, the front-to-back ratio is above 14 dB across the band and
above 20 dB for over half the band. Finally, the 75-Ohm SWR is less than 2:1 across
the entire band. Although still less than perfect, the 10" spacing of the dual-wire
elements provides very significant improvements in performance over either of the
other models.
Fig. 8-8 provides a graphic sense of the improvements. All of the pattern ele-
ments are more tightly grouped than in the comparable azimuth pattern sweeps for
the other designs. For those who wish to examine the model structure, the following
listing may be useful.
Note in the model the use of two sources which are essentially in parallel. You
may use the standard parallel impedance equations to calculate the composite feedpoint
impedance. In practice, of course, a builder would bring the two loop wires together at
the feedpoint for a single feedline connection. For modeling purposes, creating a
short (1") transmission line using the TL facility would also provide quite accurate
impedance figures. A line of about 150 Ohms would suffice. Since the line is math-
ematical only, its physical length need not match reality, and the very short segment of
transmission line between elements minimizes errors due to impedance transforma-
tion down the line.
The dual-wire planar loop design is not the answer to all limitations of 3-element
quad bandwidth, but it goes a long way toward overcoming them. For such a large
initial array, the two-wire loop is fairly straightforward, if not simple, to implement. The
cost will be something over 400' in extra wire.
I habitually use a mixture of two NEC programs for quad design--NEC-Win Plus
and EZNEC. Although many of the graphics that appear here are from models
cross-checked on EZNEC/4, the basic design work was done with NEC-Win Plus
using the model-by-equation facility.
Fig. 8-9 provides a simple illustration of the first step in the design process--chosen
because a quad design with a single-wire structure makes a more compact and easily
read graphic. (The same illustration using the dual wire model would have required
36 lines in the “wires page” section along with additional equations to set the dimen-
sions of the second loop on each element relative to the first loop or to a center-line.)
By setting the “half-side” dimensions of the quad in terms of a fraction of a wave-
length, the task of hand-optimizing a design is considerably eased. In this case, the
director spacing is actually the spacing from the reflector to the director, and finding
the distance from the driver to the director is a simple case of subtraction. Alterna-
tively, one might set the driver at zero and use a negative value for the reflector spac-
ing and a positive value for the director. Finding the total boom length then becomes a
simple case of addition.
Fig. 8-10 shows the resultant dimensions using the equations in Fig. 8-9. Under-
lying this page is a set-up that specifies variables for each of the points in the antenna
geometry—along with + and - signs as applicable to this free-space model. See Fig.
8-11. Such models allow easy manipulation of dimensions for trial-and-error optimiz-
ing of a design. When there is only one number to change to vary a full element
(which has 8 corner points in a 2-wire arrangement), the impact of errors is less
discouraging. Nonetheless, in all hand-optimizing efforts, two subsidiary activities are
essential to making progress.
First, it is crucial to keep a side pad that records each set of values tried. Nothing
brings frustration more than the discovery that one has been covering the same ground
over and over again. Second, it is equally critical to the process to record trends that
occur with each small increment of change. These trend notes can become complex,
since there are variable “breaking” points. As one adjusts the spacing or the loop
circumference of one or more elements, the point at which a progressive gain in-
crease turns into a decrease with changes to any single dimensions will change.
Watching and recording trends will be useful in determining the best composite (or
compromise) set of dimensions.
The earlier note in this chapter on the rough equivalency of a 3-4" wire to the 10"
double #12 wire arrangement can be verified by using the 3-element wide-band auto-
mated design program. A 3" wire is the minimum diameter single copper wire that will
yield a 75-Ohm <2:1 SWR curve for a 40-meter array. The model below is one ex-
ample of such a design. Note that when run on NEC-4, the properties will show a
slight displacement (under 50 kHz) in frequency. The curve was developed in NEC-2.
To obtain a >20 dB front-to-back curve for the entirety of the US 40-meter band
requires a much greater single-wire diameter: about 20". The following model pro-
vides such a curve on NEC-2, with the usual slight frequency displacement in NEC-4.
Developing a multi-wire equivalent for a 20" diameter wire is likely not feasible
with only two wires. Even if the SWR and front-to-back curves can be replicated,
the full gain (9.2 dB at design center frequency) will not be available without several
wires per loop--perhaps 4 wires on 10" centers (as a speculative guess). The 20"
wire in the model presses the limits of wire diameter (0.01 wavelength) for which the
design program has been calibrated. Nevertheless, for one who seeks the maxi-
mum gain and operating bandwidth from a 3-element quad on 40 meters, experi-
mentation with multi-wire elements is certainly worthwhile.
Let me add one final note relating to the various automated design programs and
models that we have presented. I have been asked on occasion why I use an inde-
pendent value for the wavelength (cell D3 in most of the NEC-Win Plus equations
spreadsheets) in most of my equations. The frequency-vs.-wavelength constant
(299.7925) is surely more precise than one could possibly need in any final result.
That is, seven significant digits is more than one can possibly replicate in construc-
tion.
When letting software perform a string of calculations, I tend to prefer to use the
most precise value for physical and mathematical constants that I can find. This
includes such values as PI (3.1415927), the speed of electromagnetic radiation in
free space (299.7925 m/s), and the ratio of natural to common logarithms (2.3025851)
or its inverse (0.4342945). I then save rounding to the desired number of significant
digits for the final step of calculations performed by software programs. Others may
wish to use only the least number of significant digits in the weakest input value in all
numeric entries. However, I have found that, because various programs perform
their rounding operations at different places—usually invisibly to the user—retaining
a high calculation precision in the course of multi-step calculations produces better
correlations with the results of other programs that arrive at the same quantities by
other steps.
Chapter 9
4-Element Monoband Quad Design
Let’s begin with a comparison of three designs. The first is a standard design
wire quad for 20 meters that has the following dimensions.
The free-space gain of this array peaks just above 9.65 dBi at about 14.2 MHz.
The front-to-back ratio peaks at about 23 dB at 14.25 MHz, but falls off to about 10
dB at the low end of the band. Indeed, the front-to-back ratio is above 20 dB for only
about 150 kHz of the total width of 20 meters. The array has a 50-Ohm 2:1 SWR
bandwidth that is just about 300 kHz, not enough to cover the entirety of 20 meters.
In short, the array acts in an entirely normal way, with a front-to-back passband
(using the 20 dB standard) that is just over half the overall 2:1 SWR passband.
If we are willing to use a much longer boom length to account for the higher level
of inter-element coupling of loops compared to linear elements, we might arrive at
the following design.
This design retains the #14 AWG wire size, but extends the element spacing to
the degree feasible before losing the antenna properties. Overall, gain is up by
about 0.5 dB from the short-boom model, ranging from nearly 9.8 dBi to just under
10 dBi at mid-band. The front-to-back ratio peaks at about 40 dB, but drops to just
under 15 dBi at the band edges. The beam is designed for a source impedance
between 50 and 75 Ohms, where it misses full band coverage at under 2:1 by about
20 kHz or so. Although no single category of performance is so great as to dictate
the longer boom design over the shorter version, the composite of all of the im-
provements in both gain and operating bandwidth strongly suggest the superiority of
the longer-boom model.
As we have seen from designing quads with fewer elements, almost every cat-
egory of performance benefits from enlarging the effective diameter of the elements.
Consider the following design using 1" diameter elements. The 1" diameter ele-
ments can be synthesized from properly spaced pairs of wires in accord with prin-
ciples enumerated at length in a past chapters.
For our further efforts, we gain another half dB of gain over the #14 optimized
design, with the free-space gain ranging from 10.3 to about 10.45 dBi. The front-to-
back ratio peaks at a uselessly high 60 dB, but drops slightly below 20 dB at the
band edges. Once more, the optimal feed cable is 75 Ohms, and the array easily
holds the SWR below 2:1 across 20 meters.
formance with element diameter as well. In many cases, the effect is not very
significant, but in some instances, the effect may be worth noting.
Often, a series of strategically taken azimuth plots can show the strengths and
weakness of designs even better than graphed curves. Fig. 9-3 shows the azimuth
plots for the 20-meter band edges and middle for the short-boom design. Clearly
the design has been optimized for the upper end of the 20-meter band, although it is
not clear whether this result is intentional or accidental. The “formulas” for this
model were applied for a midband frequency of 14.175 MHz. Fatter wire with no
change in dimensions would have moved the operating peaks higher in the band,
while thinner wire would have narrowed the operating bandwidth of the array. Since
the formula-driven design provides no guidance other than the equations, the mat-
ter must rest in the “unknown” basket.
Fig. 9-4 provides the comparable set of patterns for the #14 optimized long-
boom array. In this design exercise, the design frequency of 14.15 MHz provides
well-balanced band-edge performance, although the rear lobes become significant
at both 14.0 and 14.35 MHz. A user would have to think long and hard about whether
the front-to-back performance of this array would be truly satisfactory for the types
of operation selected for a given installation.
The rear lobes of Fig. 9-5 show that the 1" diameter model belongs to the same
design sequence as the optimized #14 model. However, the band-edge rearward
performance is considerably improved over the thin-wire model. The rear lobe aligned
180-degrees from the forward lobe at 14.175 MHz shows how steep the front-to-
back curve is, since the peak occurs at 14.15 MHz.
In fact, Fig. 9-6 reveals the same information about front-to-back performance
in a different form. The peak front-to-back frequency for the #14 optimized model is
just below 14.175 MHz and exceeds 40 dB. Also of note is the fact that the short-
boom model achieved such gain as it could at the expense of front-to-back ratio.
In Fig. 9-7, we have the SWR curves, although we must remember which repre-
sent a 50-Ohm value and which represent a 75-Ohm value. In large measure, the
bandwidth quality of any array--with special attention to quads--is a function as much
of the range of reactance across a band as it is a matter of the range of the resistive
component of the feedpoint impedance. Moreover, the ratio of total change of reac-
tance to the average resistance (or to the desired feed cable, will indicate loosely
whether or not a 2:1 ratio can be maintained across the band. The short-boom
model has a median resistive component of 48.8 Ohms with a total change of reac-
tance of 85.8 Ohms--a ratio of about 1.76:1. The long-boom wire model has a
reactance range that is higher: 96.5 Ohms. However, the median resistance is 66
Ohms, for a 1.46:1 ratio. The 1" model shows a change in reactance of 67 Ohms
with median resistance of 59 Ohms, for a 1.13:1 ratio. Although the ratios are not
precise indicators of SWR performance, it is clear that a low ratio is a good indicator
of better bandwidth. More precise equations can be developed, but the complexity
of the SWR formulas make the exercise--already less than precise--somewhat su-
perfluous. A good indicator is sufficient to alert the quad designer to desired direc-
tions of improvement.
Both optimized design emerge from the same sequence of designs that will
allow for automated design after regression analysis of the baseline models that
spanned wire diameters from 3.16E-5 to 1E-2 wavelengths. As we have added
elements to the designs, it has become harder to maintain a wide operating band-
width with decreasing wire sizes. 4-element design, when applied to HF arrays that
typically use wire, significantly benefits from the development of thick-wire substi-
tutes in order to achieve maximum gain, a high front-to-back ratio over the entire
chosen band, and an SWR of less than 2:1 across the band. In general, a wire size
of 0.005 wavelengths is desirable, which is in the vicinity of 5" at 20 meters. Obvi-
ously, a 2-wire--or preferably a 3-wire--substitute is called for. (Narrow-band appli-
cations, of course, are immune from this requirement, except for achieving maxi-
mum gain and adequate front-to-back ratio.)
The selection of a design sequence also calls for some comment. The baseline
arrays in the sequence must, of course, be part of a sequence and not merely a set
of random spot designs, each of which uses the prescribed wire size. Each array
was designed so that, to the degree possible, maximum gain occurred within 1.5%
of the design frequency. The maximum front-to-back ratio was set on the design
frequency, as was array resonance. Wire size was changed in increments that
resulted in a sequence of the common logarithms of the wire size in wavelengths of
0.5. This interval ensure a regression analysis that could be carried out to the 4th
order.
Fig. 9-8 shows the reflector circumference (more correctly, the value of 1/8 of
the reflector circumference) curve developed vie regression analysis. Similar curves,
not all so precisely fitted as the example, emerged for the other parameters of the 4-
element quad design.
In practice, the number of true variables in the analysis turned out not to exceed
those required for the 3-element quads shown in preceding episodes. The spacing
required between the reflector and driver and between the driver and the first direc-
tor required very large changes before they resulted in a significant change in array
properties. In contrast, each of the other dimensions of the array were quite sensi-
tive to small changes. (These dimensions included all element circumferences and
the spacing between director 1 and director 2.) Therefore, the spacing between the
driver and its adjacent elements was allowed to stand as a pair of constants (in
terms of wavelengths).
An additional factor involved in the selection of the model sequence to form the
basis for automated design involved the rate of change of the reactance span from
the lower to the upper limits of a defined frequency span. Each model in a se-
quence will show a range of reactance change across an assigned frequency span
such that the thinner the wire, the higher the reactance range. The rate of change of
this range from one wire size to the next plays a role in the selection of the design
sequence: the lowest rate of change with wire size decrease is the most desired
sequence. This rate ensures that thinner wire designs--while not matching the per-
formance of thick-element designs in the same sequence--at least provide useful
performance.
There are, in fact, spot designs that will outperform the models in this sequence,
some of which have shorter booms. Dan Handelsman provided me with one such
design for 2 meters using a 0.5" diameter element. The design data are as follows:
The dimensions for the corresponding 0.5" diameter model from the selected
sequence are these:
The N2DT quad actually outperforms the sequenced quad by a small margin,
despite the 14" reduction in boom length. The gain is 0.15 dB higher (10.76 vs.
10.61 dBi) at the design frequency. The 20-dB bandwidth is close to 2.8% in con-
trast to the sequence design’s 2.75% value. (The bandwidth of 2 meters is about
2.78%.) Both arrays have a 50-Ohm SWR under 2:1 across the band. (For the
VHF range, with large diameter elements whose logs are between -2.5 and -2.0, the
feedpoint impedances of the sequenced designs are closer to 50 Ohms than to 75
Ohms.) The natural question is why the N2DT design was not chosen as the basis
for the design sequence.
The answer lies in the rate of change of the span of reactance with decreasing
wire sizes. for the span of wire sizes whose diameters in wavelengths result in
common logs of -2.0 to -2.5, the reactance range increased 54% for the N2DT
design, but only by 33% for the chosen sequence. Although the rate of change is
not a linear curve in all cases, it does provide an indication of the most promising
design sequence that is usable over a wide span of wire diameters.
Neither the N2DT design nor the sequence design provides the highest possible
gain for a 4-element quad. The following dimensions are for a high-gain 4-element
quad that also uses 0.5" diameter elements and a design frequency of 146 MHz.
This design will meet the <2:1 SWR standard (50 Ohms), but the front-to-back
ratio hold above the 20 dB level for only about 3/4 of the 2-meter band. As well, the
gain varies about 0.4 dB across the band. The two most prominent factors in the
design are its free-space gain, which reaches 11.0 dBi, and its length, which is about
2" short of 6'. It did not exhibit a sufficient bandwidth or a sufficiently low rate of
reactance-span change to qualify for the sequence.
Perhaps the only significant claim that can be made for the sequence of designs
that resulted in the automated design program is that they yield close to the widest
bandwidth in the listed operating categories along with the best gain potential as a
secondary criterion of any sequence that I have so far uncovered. Obviously, there
may well be other sequences awaiting discovery. Therefore, the program should be
used with due appreciation of the tentative nature of its presentation.
However, the program does give proper place to element diameter and to quad-
loop inter-element coupling in its development. As with the other programs in this
sequence, one enters the desired element diameter in a specified unit (although the
program contains no provision for entering AWG wire sizes). As well, one enters the
design frequency. For wide bands, such as the harmonically related HF amateur
bands, it may be best to select a design frequency between 0.35 to 0.4 of the way
from the lower band edge in order to achieve roughly similar front-to-back and SWR
values at both band edges.
As with past programs, the listing is for GW Basic, since that format makes all of
the mathematics visible to the user.
340 AW=.07#:BW=1.048518519#:CW=6.173055556#:DW=17.12092593#:
EW=21.34722222#
350 SW=(AW*(D1^4))+(BW*(D1^3))+(CW*(D1^2))+(DW*D1)+EW
360 AF=-.03#:BF=-.27666667#:CF=-.4475#:DF=2.348809524#:
EF=7.853214286#
370 FB=(AF*(D1^4))+(BF*(D1^3))+(CF*(D1^2))+(DF*D1)+EF
380 WL=299.7925/F:PRINT “Wavelength in Meters =”;WL;” “;
390 WF=983.5592/F:PRINT “Wavelength in Feet =”;WF
400 PRINT “Quad Dimensions in Wavelengths, Feet, and Meters:”
410 PRINT “Driver Side =”;(DE/4);” WL or”;(DE/4)*WF;”Feet or”;
(DE/4)*WL;”Meters”
420 PRINT “Driver Circumference =”;DE;” WL or”;DE*WF;”Feet
or”;DE*WL;”Meters”
430 PRINT “Reflector Side =”;(RE/4);” WL or”;(RE/4)*WF;”Feet
or”;(RE/4)*WL;”Meters”
440 PRINT “Reflector Circumference =”;RE;” WL or”;RE*WF;”Feet
or”;RE*WL;”Meters”
450 PRINT “Reflector-Driver Space =”;SP;” WL or”;SP*WF;”Feet
or”;SP*WL;”Meters”
460 PRINT “Director 1 Side =”;(IR/4);” WL or”;(IR/4)*WF;”Feet
or”;(IR/4)*WL;”Meters”
470 PRINT “Director 1 Circumference =”;IR;” WL or”;IR*WF;”Feet
or”;IR*WL;”Meters”
480 PRINT “Director 1-Reflector Space =”;IP;” WL or”;IP*WF;”Feet
or”;IP*WL;”Meters”
490 PRINT “Director 2 Side =”;(TT/4);” WL or”;(TT/4)*WF;”Feet
or”;(TT/4)*WL;”Meters”
500 PRINT “Director 2 Circumference =”;TT;” WL or”;TT*WF;”Feet
or”;TT*WL;”Meters”
510 PRINT “Director 2-Reflector Space =”;TTP;” WL or”;TTP*WF;”Feet
or”;TTP*WL;”Meters”
520 PRINT “Approx. Feedpoint Impedance =”;ZR;”Ohms “;
530 PRINT “Free-Space Gain =”;GN;”dBi”
540 PRINT “Approximate 2:1 VSWR Bandwidth =”;SW;”% of Design
Frequency”
550 PRINT “Approximate >20 dB F-B Ratio Bandwidth =”;FB;”% of
Design Frequency”
560 INPUT “Another Value = 1, Stop = 2: “;P
570 IF P=1 THEN 10 ELSE 580
580 END
Note: “LOG” in GW Basic always mean the natural logarithm. Hence, a conver-
sion factor is necessary to convert the natural log to the common log required by the
program. If the medium to which this program may be transferred already knows
the difference between “LOG” and “LN,” the conversion factor can be dropped.
As with the other GW Basic programs in this volume, a version of this program
appears in the HAMCALC suite of GW Basic utility programs available from VE3ERP.
As well, a version appears at the Nittany-Scientific web site in the form of a NEC-
Win Plus model set up in equations (http://www.nittany-scientific.com). The supple-
mental data do not appear in that program, since running a model through a fre-
quency sweep is a superior method of determining passband performance and se-
lecting the optimal design frequency.
Fig. 9-9 shows the equations page for the 4-element automated design. The
two “fixed” element spacing figures can be spotted in the left-most column, since
they do not make reference to the spreadsheet cells right of the comments column.
The forward director variables occur after the design input and element diam-
eter variables (which the user varies to change a design). This placement results
from using and expanding a previous 3-element model to arrive at the 4-element
version. The sample page uses 1" elements with a design frequency of 14.15 MHz
to equalize band edge performance to the degree possible.
Fig. 9-10 shows the set-up page for the quad. As with all of the multi-element
quad arrays, the standard material is copper. The performance data for aluminum
would not significantly differ from copper. Nor would the requisite design frequency
to equalize band edge performance.
The set-up wires page does show that as one adds only a single element to an
array, two more variables appear to make the process of optimization more com-
plex and less certain. Consequently, to treat this program and model as more ten-
tative than the preceding versions for 2- and 3-element arrays is very much in order.
Some Foreshadowings
The level of inter-element coupling among quad loops requires greater spacing,
in general, than comparable linear dipole/Yagi elements in order to optimize array
gain. To obtain high gain with a good bandwidth for other operating parameters, the
multi-element monoband quad becomes longer faster than potentially competing
quad designs. At 4 optimally spaced elements, even with thick element diameters,
the quad approaches a possible limit on its ability to provide higher gain with a
shorter boom length.
Chapter 10
Some Notes on Long-Boom Quads
As a step in the direction of such investigations, these notes offer some thoughts
on modeling studies into parasitic quad beams. More precisely, they offer three (of
many possible) approaches to the design of long boom quad beams. What the de-
signs reveal may be useful to others in further work.
The baseline for each of the approaches was to produce a quad beam model that
yielded 11 dBi free-space gain and about 20 dB or more of 180-degree front-to-back
ratio across the operating bandwidth. The operating bandwidth itself would eventually
become a consideration, but was not among initial concerns. All antennas were de-
signed and modeled for 20 meters, with a 14.175 MHz design frequency.
I selected the gain level based on the popular notion that a quad loop can achieve
about 1.8 to 1.9 dB more gain than an equivalent dipole. This notion has given rise to
the presumption that a quad beam has about the same gain advantage over a Yagi
with the same number of elements and a similar boom length. However, the presump-
tion does not work out in reality, largely because quad beams are constructed using
thin wire elements in contrast to the fatter tubing used by typical Yagis. Because the
thin wire results in lower levels of inter-element coupling, the quad’s gain advantage is
seriously reduced and is often completely negated in large arrays. Some of the gain
advantage can be restored by the use of fatter elements, which in models can be
single fat wire elements or multi-wire simulated fat elements. The restoration of gain
through the use of dual-wire elements, despite the higher material losses of the dual
thin wires, tends to demonstrate the relative dominance of inter-element coupling
over wire loss in establishing the gain of a large array. For this reason, the models we
shall discuss use a variety of wire sizes, ranging from 1/10 to 1/2 inch.
The selection of the 20-dB front-to-back criterion rests on the ability of monoband
Yagis to achieve this figure routinely across the designed operating passband. Equiva-
lent performance should be possible from a quad if it is to justify itself as a competitor
for the Yagi. (This considerations is apart from operational reports that quads have an
advantage at band openings and closings. Models cannot simulate the propagation
conditions that would either confirm of disconfirm such reports.)
The approaches taken to designing this collection of long-boom quads are varied.
The first design is a direct adaptation of parasitic element placement taken from Yagi
design. The second is a wide-band variation of Yagi design sometimes called opti-
mized wide-band arrays (OWA). The third approach uses wide-spaced principles in
which elements use approximately the same spacing throughout the design.
The first approach to quad designs uses standard Yagi spacing as a basis for
quad design, with element size and spacing optimized after initial placement. I used
the approach to design a 3-element quad with excellent performance over its pass-
band. However, several peculiarities--relative to a comparison Yagi--appeared in the
results. First, the element spacing was somewhat larger than for the Yagi, despite
elements of the same diameter. Second, the operating bandwidth for both the
front-to-back ratio and the SWR were considerably narrower than for the Yagi. Third,
instead of the traditional 50 to 100 Ohm quad feedpoint impedance, the 3-element
quad showed a feedpoint impedance close to 25 Ohms. All of these conditions and
limitations also showed up on a longer version of the antenna.
For a Yagi model, I took a 3-element 20-meter design developed by K6STI and
added 2 directors. It is equivalent to a 48' boom length Yagi that would provide about
10.1 dBi free space gain across 20 meters with good front-to-back figures. I then built
a quad based on the Yagi. After optimizing the 0.1" diameter elements and spacing
them to achieve 11 dBi free-space gain and a good front-to-back ratio, the boom
length rose to 50.36'. Fig. 10-1 presents a side view of the antenna to show the
relative element spacing.
Fig. 10-3 presents the free-space gain of the 5-element quad over its working
bandwidth. The rate of gain change--nearly 0.25 dB over 110 kHz--is quite high, and
that performance is one mark of a narrow-band antenna design. Nonetheless, the
gain figures exceed the 11 dBi target everywhere in the passband.
Fig. 10-4 shows the front-to-back ratio over the same bandwidth. The peak is
over 30 dB, but the rate of decrease from the peak is fairly rapid. With slightly more
optimizing effort, the peak front-to-back ratio might be better centered in the pass-
band. The result of centering the front-to-back peak would be better than 20 dB
across the passband.
The VSWR curve (Fig. 10-5) is the 50-Ohm SWR resulting from matching the
feedpoint of the model. The resonant 25-Ohm feed impedance was transformed via
a 1/4-wavelength section of 37.5-Ohm coax (presumptively made from parallel sec-
tions of 75-Ohm cable). The limits of the graph show values just above 1.9:1 at each
end. However, the SWR curve continues to steepen at both ends of the passband so
that only about another 10-15 kHz are available before the SWR reaches 2:1.
The 5-element quad design would be inappropriate for use on 20 meters except
for certain operators who use only small portions of the band. However, scaled to 17
or 12 meters, the antenna would provide the same performance and cover the entire
100 kHz band. Indeed, if this design has a home at all, it would be on the WARC
bands.
Of all the quad designs that we shall consider, the 5-element Yagi-based design is
the most compact. However, most quad builders shy away from antennas with odd
numbers of elements. Performance is not so much the worry as is mounting the
antenna. With an odd number of elements, the center-most element comes quite
close to the tower unless one uses a long mast above the tower-top. One can over-
come the problem to some degree by weighting the mast to one side or the other. Still,
balancing weight loads and wind loads makes this move uncertain for all but those
with considerable mechanical engineering experience.
The following table is the EZNEC model description for anyone who wishes to
experiment further with the design.
5 el quad 20 m Frequency = 14.175 MHz.
Wire Conn. --- End 1 (x,y,z : ft) Conn. --- End 2 (x,y,z : ft) Dia(in) Segs
No loads specified
-------- TRANSMISSION LINES ---------
Line Wire #/% From End 1 Wire #/% From End 1 Length Z0 Vel Rev/
Actual (Specified) Actual (Specified) Ohms Fact Norm
The second and third design approaches attempt to see what might be required
to achieve a quad beam with full 20-meter coverage. Ideally, this coverage would
include the following criteria:
In the end, I settled for some reasonable compromises. Still, the resulting de-
signs have some aspects that make them less fit for direct construction.
The second approach also uses Yagi parasitic element principles, but of a special
type. In Yagi design, the originator, NW3Z, refers to them as optimized wide-band
arrays. He has developed 20 meter Yagis for 48' booms with about 10.1 dBi average
gain and greater than 20 dB front-to-back ratio across 20 meters. These initial speci-
fications sound very much like those of other 5-element Yagis on similar booms.
However, the NW3Z design adds one more element, a director. By judicious spacing
of the reflector and the added director from the driven element, the Yagi achieves a
direct match top to a 50-Ohm feedline with low SWR (under 1.3:1) across the entire
band.
Fig. 10-6 shows the side view of the final design to provide a perspective on the
required element spacing. Note especially the reflector and first director positions.
Fig. 10-7 is a free-space azimuth pattern for the design at its design frequency, 14.175
MHz.
An added cost to achieve the design goals was the use of 0.5" diameter elements
in all of the elements. Hence, the design is unlikely to be directly implemented, al-
though alternative element construction is possible. Table 10-2 provides the array
dimensions for 20 meters. Once more, all dimensions are in feet.
If you compare the element dimensions with those for the 5-element narrow-band
quad, you will discover some interesting differences other than spacing. First, the
OWA reflector and driver have dimensions that are quite close to each other, with both
being somewhat longer than the corresponding elements in the 5-element array.
Second, the directors tend to be shorter than those in the smaller beam.
For the performance curves, we shall use the entire 20 meter band from 14.0 to
14.35 MHz. The gain of the OWA 6-element quad is quite stable, as shown in Fig. 10-
8. It ranges from 10.99 to 11.11 across the band. This is about 1/3 the variation of the
smaller array despite the 3-fold increase in operating bandwidth. The shape of the
gain curve, however, suggests that the gain stability limits of the array are not very
much wider than the 20 meter band itself. Note the increasing rates of change near
the band edges.
The front-to-back ratio across 20 meters appears in Fig. 10-9. Of all the array
approaches tested in this exercise, the OWA design is the only one to achieve greater
than 20 dB front-to-back ratio for the full passband. Part of the reason for this result
stems from the use of very large diameter elements. Single-wire elements of the
usual #12/#14 AWG material will not yield the front-to-back operating bandwidth.
However, large-diameter single-wire elements may be simulated within limits by 2-
and 3-wire elements, as described in Chapter 5.
The OWA 6-element quad provides a direct 50-Ohm match with no matching
network components, as shown in Fig. 10-10. Interestingly, the shape of the pattern
resembles that for the corresponding OWA 6-element Yagi, but with higher band-edge
values. Despite the ability of the design to cover 20 meters fully, the quad shows itself
to be inherently more narrow-banded than counterpart Yagis.
The OWA quad is quite possibly a usable design for a high performance, full band
coverage array, with one exception. The 0.5" diameter elements are not feasible
using standard quad construction techniques that employ relatively lightweight fiber-
glass or similar element support arms. Significant reductions in the effective element
diameter reduce inter-element coupling and result in gain and operating bandwidth
reductions. The solution is to redesign the array for dual-wire elements using #14 or
#12 wire. However, the substitutions will require extensive re-optimization of element
lengths to restore the performance curves. Because the closely-spaced loops would
require between 2 and 3 times the number of segments per element, with an increase
in the number of modeling wires, the slow process was not endured for this exercise.
Nonetheless, more extensive work on 2-element quads, described in past articles,
strongly suggests that the substitution is quite achievable.
For anyone who wishes to work further with this type of design, the following table
is the EZNEC model description.
Wire Conn. --- End 1 (x,y,z : ft) Conn. --- End 2 (x,y,z : ft) Dia(in) Segs
No loads specified
No transmission lines specified
Ground type is Free Space
Immediately apparent is the greater length of the array compared to the OWA
version of a wide-band quad: 70' vs. 61'. More subtle are the required variations from
uniformity in the element spacing. Although the average spacing is nearly 0.2 wave-
length, the director spacings cannot be set by simple adherence to the average. Per-
formance deteriorates rapidly using mere rules of thumb as guidance.
Fig. 10-13 shows the gain curve across 20 meters for the wide-spaced array. Like
the OWA array, the curve shows good stability, with a net variance of only 0.15 dB
across the band. Note especially that a wide-spaced design is capable of placing the
peak gain of the antenna well within the boundaries of the operating passband.
Fig. 10-14, the front-to-back curve across 20 meters, shows the effect of using
small diameter wire for the elements. The band-edge front-to-back ratio is about 17
dB, and the peak value is 21 dB. To the present, I have found no way to increase the
front-to-back performance within the constraints of the overall length and the wire
size. However, the use of large-diameter elements or dual-wire substitutes shows
promise of improving this aspect of performance considerably.
The SWR curve for the wide-spaced 6-element array appears in Fig. 10-15. Un-
like the other curves, this one is a 75-Ohm VSWR curve, the inherent feedpoint im-
pedance of the antenna. Matching the antenna to a 50-Ohm feedline requires the use
of a simple transmission-line transformer. The band-edge reactance is well under +/
-j40 Ohms for mid-band resonance of the driver.
The wide-spaced 6-element array has considerable potential for further develop-
ment through the use of larger diameter elements or substitutes. Nevertheless, the
key limiting factor in this direction is the boom length. Comparable Yagi designs with
the same boom length would likely use 7 elements and provide equal gain, but supe-
rior front-to-back performance.
Those who might wish to further optimize the design can refer to the following
EZNEC model description.
Wire Conn. --- End 1 (x,y,z : ft) Conn. --- End 2 (x,y,z : ft) Dia(in) Segs
No loads specified
No transmission lines specified
Ground type is Free Space
3. Of the designs so far surveyed, perhaps the OWA version holds the most
potential for the wider amateur bands (20, 15, and 10 meters). The 5-element
array should be adequate for 30, 17, and 12 meters, assuming that one can com-
pensate for the mechanical difficulty presented by the use of an odd number of
elements.
Chapter 11
A Place for Narrow-Band Quad Designs
One of the chief goals of this investigation has been to determine the conditions
necessary for the widest possible operating bandwidth for monoband quad arrays.
For this reason, we passed over the 5-element 20-meter quad based on Yagi prin-
ciples. Its operating bandwidth did not come close to either the OWA of uniform wide
spaced models using 6 elements.
However, there are important places in amateur radio operations for narrower
band quads. Lest you think that I have wholly neglected such applications, you might
review Volume 1 of this study. More to the point, let’s examine a couple of designs for
the non-harmonic amateur bands, otherwise known as the “WARC” bands. In par-
ticular, we shall explore adaptations of the array we neglected so that it becomes a
very fitting quad beam for 17 and for 12 meters.
The following EZNEC model description provides all of the design data. The
boom length is 39.4'. Hence, the array would fit a 40' boom, including hardware for
the support arms.
Wire Conn. --- End 1 (x,y,z : ft) Conn. --- End 2 (x,y,z : ft) Dia(in) Segs
Line Wire #/% From End 1 Wire #/% From End 1 Length Z0 Vel Rev/
Actual (Specified) Actual (Specified) Ohms Fact Norm
Fig. 11-2 presents the free-space azimuth pattern for this array. Since large graphs
of performance are unnecessary for such a narrow band as 17 meters, Table 11-1 will
summarize operating performance.
The worst-case front-to-back ratio is greater than 20 dB across the band. The
gain varies by only 0.12 dB, while the SWR values at the band edges leave a good
margin for inevitable construction variations that might displace the curve a bit.
The 12-meter array has a native feedpoint impedance similar to that of the 17-
meter beam. Therefore, as shown in the model description that follows, a quarter-
wavelength matching section is apt to this design as well.
Wire Conn. --- End 1 (x,y,z : ft) Conn. --- End 2 (x,y,z : ft) Dia(in) Segs
Line Wire #/% From End 1 Wire #/% From End 1 Length Z0 Vel Rev/
Actual (Specified) Actual (Specified) Ohms Fact Norm
As shown by Fig. 11-4, the extra element improves performance by about 0.6 dB.
Like the 17-meter array, the 12-meter quad worst-case front-to-back ratio exceeds 20
dB across the band, and the total gain change is under 0.05 dB. Table 11-2 provides
full band data on performance.
As a percentage of the center frequency, the 12-meter band is narrower than the
17-meter band. Hence, the SWR curve is shallower on the upper band, despite the
additional element.
The element lengths require only very slight adjustments to bring each array within
the 17 and 12 meter band limits. The composite array has essentially the same
performance as the independent quads that are combined. The following model de-
scription will help us locate the necessary changes.
Wire Conn. --- End 1 (x,y,z : ft) Conn. --- End 2 (x,y,z : ft) Dia(in) Segs
Line Wire #/% From End 1 Wire #/% From End 1 Length Z0 Vel Rev/
Actual (Specified) Actual (Specified) Ohms Fact Norm
The forward-most directors on each array have been altered to compensate for
interactions within the same support arm system. At the rear end of the array, adjust-
ments were made to the 17-meter reflector and driver, but the corresponding ele-
ments on the 12-meter array required no changes. Intermediate directors are un-
changed. The performance figures for the combined array (using separate feedlines)
appear in Table 11-3.
The 17-meter gain decreases by an average of 0.1 dB, while the 12-meter gain
average decrease by a mere 0.06 dB. The front-to-back curves retain a worst-case
value above 20 dB across both bands. The 50-Ohm SWR curves—with matching
sections in place—are shallower than those for independent models.
The present array is certainly usable with good results on the upper 2 WARC
bands, with appropriate construction cautions. First, finding the center of mass in this
array will require extensive analysis or experimental techniques. Second, some ele-
ments will come quite near to the mast. Hence, it will likely require a mast that ex-
tends at least 7.5' above a tower and its guy wires to ensure that there is minimal
electrical interaction and no physical interference as the array rotates.
The 2-band design specifically retained the original independent quad design val-
ues, even though this requires separate support arm structures for the intermediate
elements. The goal of the exercise was to see whether the designs—optimized as
independent arrays—could be brought together without significant change in perfor-
mance. Decreasing the number of support systems would result in considerable
weight savings, but at the cost of settling for compromise performance on at least one
of the two bands. One of the premises of this volume—and the reason for working
with monoband arrays almost exclusively—has been to look seriously at the perfor-
mance potentials of quads without contending with the compromises inherent in multi-
band designs. The present 17/12
meter design is the one concession
to multi-band quad operation and is
included because it requires no com-
promises in performance.
quencies, but instead are good “DX” bands, combining the array just discussed with a
30-meter quad seems natural.
One possibility for such an array might be the 3-element “Yagi-spaced” 20-meter
quad discussed in Volume 1, readjusted for 30 meters. For consistency, the readjust-
ment will use #12 AWG copper wire. Since the 30-meter band is only 50 kHz wide,
the thinner element size will create no bandwidth problems.
Fig. 11-6 shows the outline of the array. It is 31.5' long, about 8' shorter than the
composite array for 17 and 12 meters. A builder has several choices. He can align
one of the elements with an element in the composite array and save the weight of at
least one support system. Or he might wish to juggle the dimensions through further
modeling to arrive at element spacings that coincides with 3 of the existing support
structures. This latter exercise will require considerable effort, but the weight savings
will be significant. The model description—when compared to the composite model
for 17 and 12 meters—will indicate something of the size of the task.
Wire Conn. --- End 1 (x,y,z : ft) Conn. --- End 2 (x,y,z : ft) Dia(in) Segs
Unlike the narrow-band Yagi-spaced quads that we have so far discussed, this
design does not require a quarter-wavelength matching section. The element diam-
eter is 0.0808" (#12 AWG), which grows smaller as a fraction of a wavelength when
we decrease the operating frequency. At 30 meters, the copper wire size provides a
direct match for a 50-Ohm coaxial cable.
As shown in Fig. 11-7, the 3-element quad acquits itself well in terms of gain and
a well controlled set of rear lobes, even if it cannot match the performance of the
arrays for higher frequencies. The worst-case rear quadrant lobes are more than
19.5 dB down from the forward lobe across the band. Table 11-4 presents the perfor-
mance figures of greatest relevance.
In short, the results of this design exercise are not necessarily to be recommended
for construction. Instead, the designs are an effort to show what is possible if one
removes mechanical compromises from the design equation set. For narrow band
applications, individual high performance quads can be combined in principle, so long
as each is given the support that it needs to live up to its potential. However, it remains
one of the facts of antenna life that mass and gravity may always force upon us
compromises in performance. With this exercise as a reference—or with better exer-
cises as a more precise reference—we shall at least now have some understanding
of the degree of compromise that we accept when simplifying the mechanical design
of a multi-band quad array.
The design exercise also establishes that there is indeed a place for narrow-band
quad designs in the amateur bands—the narrow bands. They also have a place
wherever the builder decides as part of an overall operating plan (not to be confused
with mere acceptance as an act of resignation) that only part of one or more wider
ham bands is desired for primary use. For total coverage of the wider HF bands,
wide-band designs can be devised, at least for monoband arrays.
Work to Be Done
Nevertheless, new optimizer routines and programs are emerging as I write. Con-
sequently, the probability of developing a 5- or 6-element program in the future will
likely increase with time. Indeed, it may prove feasible to tweak the quads explored
here even further and to explore other design potentials as well. Not only is our
understanding of how quads work improving, but as well, new tools are arriving to
speed the process of increasing that understanding further. They likely will open up
some not-yet-dreamt-of possibilities and directions for new quad array designs.
These notes, then, are not in any way an end to the analysis and design of quad
arrays, both grand and modest. Instead, they are the barest beginning, a few steps on
a very long, challenging, and fascinating road ahead.
Chapter 12
Some Notes on Size and Height
In the very first chapter to the first volume of these notes, I stated--and showed
with some elevation patterns--that there is no significant difference in the elevation
angle of maximum radiation (take-off angle) for Yagis and quads at the same effec-
tive height. The statement had a limited context: that of eliminating one of the
coarser myths about quad performance. It remains true
However, there are some differences between Yagi and quad elevation patterns
that might make a difference to some quad builders and users. These differences
are potentially significant within a limited range of installations. They will apply to
quad designs with fewer elements and installed at lower heights--a natural amateur
operation combination. The exact degree of significance that these differences will
have will depend upon the goals and limitations of the individual user, so we must be
careful not to make too much of them. Nevertheless, the differences are interesting
and worth a brief examination.
The best way to begin is to set a baseline. Let’s examine the elevation patterns
of three 3-element Yagis having different boom lengths. We shall set each Yagi at a
height of 5/8 wavelength so that the results can be extrapolated to any of the upper
HF bands. The height is not arbitrary, and that means a short digression into the
reason why 5/8 wavelength is a good test height.
Virtually any horizontally polarized antenna shows less than a smooth curve of
gain as we increase its height from about 1/2 wavelength up to nearly 1.5 wave-
length. If we choose a simple dipole, we shall find not only variations in the rate of
gain increase as we raise height (where gain is taken at the take-off angle), but as
well, we shall find some height regions where the gain actually decreases relative to
a slightly lower height. The region around 7/8 wavelength is a prime example, rela-
tive to a height of 5/8 wavelength. Another minor decrease occurs at 1 3/8 wave-
length, but it is negligible. As the antenna rises above about 1.25 wavelength up,
the ripples in the curve of gain-increase with height largely wash out and become
invisible.
Table 12-1 provides a systematic glimpse at the phenomenon for a simple wire
dipole.
In the table, we note not only the regions in which gain is lower than surrounding
heights, but as well the heights at which the gain reaches a peak value. The peaks
recur at 1/2 wavelength intervals, as do the valleys. If you care to look back at a
study in Communication Quarterly (“The Effects of Antenna Height on Other An-
tenna Properties: A Computer Study,”) for Fall, 1992, pages 57-79, you will find a
rudimentary but detailed survey of many antenna types.
When we create directive arrays, the detection of actual peaks and valleys in the
gain curve with increasing height becomes marginal. However, we can detect
changes in the rate of gain increase with increasing antenna height. This fact has
two consequences of note here. First, when we create a baseline of data for vari-
ous horizontally polarized antennas, we must use a consistent antenna height. For
our purposes, selecting a height that results in maximum gain (or very close to it)
seems reasonable.
Second, many quad builders and users have limited space and resources. Un-
like some fully committed DX and contest operators, who place many towers over a
large parcel of ground, the average builder uses a shorter tower of lighter construc-
tion. Optimizing the height--at least for a monoband antenna--seems a reasonable
course of action.
With this background, we can return to our collection of 3-element Yagis. Let’s
place them at the 5/8-wavelength mark. We shall use good ground with a conduc-
tivity of 0.005 S/m and a dielectric constant of 13 for all of our antennas, but with
horizontally polarized antennas, changes in ground quality do not make very large
differences in performance.
Table 12-2 provides most of this data for 3 Yagis that might be classified as
having short, medium, and long boom lengths.
Perhaps the most interesting aspect of the data in the table is how little it differs
from one entry to the next. Gain, of course, increases with the increasing boom
length. The gain increase translates into a very slightly narrower vertical beamwidth
for longer boom lengths.
With respect to the formation of the secondary lobe, there is only a little over 2
dB difference in the forward-to-secondary lobe ratio. The range of actual gains for
this lobe is even smaller, once the differences of maximum forward gain are taken
into account. Moreover, as shown in the composite elevation patterns for the three
beams in Fig. 12-1, the shapes of the patterns in the region from secondary lobe to
the vertical (zenith) are very similar.
The Yagi, of course, is composed of linear elements laid out in a flat plane. That
construction determines to a large degree the resulting elevation patterns--what-
ever the height. There is very little one can do to a Yagi to change the lobe forma-
tion.
Quads are another matter. Inherently, they consist of elements laid out in 3
dimensions, with two dipoles per element spaced about 1/4 wavelength apart. The
basic geometry of the quad does not entirely overcome the fact that it remains a
horizontally polarized antenna. Hence, we can expect relatively normal lobe forma-
tion according to the height of the antenna. However, especially at lower heights,
we can expect some differences.
Table 12-3 presents modeling data on three different 3-element quad designs
having different boom lengths and design goals. The shortest quad is designed for
high-gain, narrow-bandwidth operation. The middle-length quad attempts to maxi-
mize gain over a wider operating bandwidth, while the longest quad strives solely for
the maximum operating bandwidth. For this comparison, all use copper wire. The
booms are set at 5/8 wavelength up, which will give the quad about a degree lower
TO angle than the Yagis.
The most notable entries occur in the last column that lists the relative strength
of the secondary lobe with respect to the main forward lobe. The quads as a group
show less secondary lobe development for the same height--in this lower range of
operating heights--than the Yagis. As shown in Fig. 12-2, there are other differ-
ences as well.
The key result of our first steps into looking at quad elevation patterns is this:
quads show a variability of elevation pattern development among designs that is
very much larger than the variability among Yagi designs. Second, at lower antenna
heights, the development of secondary elevation lobes at higher angles can be re-
duced by design choices. (I am indebted to Carroll Allen, AA2NN, who first called
this phenomenon to my attention.)
For the exercise, let’s place each quad at heights of 1/2, 5/8, 3/4, and 7/8 wave-
length. We shall expect the gain to rise as we elevate each antenna. However, we
shall also be interested in the emergence of the higher-angle secondary lobe, along
with the decreasing vertical beamwidth of the lowest or major lobe in the pattern.
Table 12-4 provides the data.
Table 12-4. Elevation Data for 3- and 5-Element Quads at Low Heights
3-Element Quad
Height Gain F-S Lobe V. Beam -3 dB Points
WL dBi Ratio dB degrees Lower Upper
0.5 12.58 N/A* 27.2 10.9 38.1
0.625 13.34 17.22 22.5 9.5 32.0
0.75 13.81 10.72 18.9 8.3 27.2
0.875 14.06 7.52 16.3 7.3 23.6
5-Element Quad
Height Gain F-S Lobe V. Beam -3 dB Points Gain over
WL dBi Ratio dB degrees Lower Upper 3-Element
0.5 14.02 N/A* 24.0 9.9 33.9 1.44 dB
The second notable data feature is the decreasing vertical beamwidth with in-
creasing gain and height. The differential in vertical beamwidth for any given height
between the two designs is a function of the relative gain of the two antennas. The
decreases in vertical beamwidth for each design as we increase height is a function
of the emergence and growth of the secondary lobe. At the highest level in the
tables, the 3-element quad has a secondary lobe that has a gain of 6.54 dBi at an
elevation angle of 49 degrees. Fig. 12-4 shows the elevation patterns of the an-
tenna at 5/8 and 7/8 wavelength up. At the lower height, the secondary lobe is
scarcely noteworthy. However, at the upper height, the secondary lobe is a promi-
nent feature of the elevation pattern.
The 5-element pattern for the same heights is shown in Fig. 12-5. The second-
ary lobe at 7/8 wavelength up is only 2.79 dBi at 45 degrees elevation. However,
notice that in the case of the 5-element quad, a third lobe pointing straight up is
beginning to emerge. It has about the same size and shape as secondary lobe
when the antenna is at 5/8 wavelength up. In contrast, the 3-element quad shows a
deep null near the zenith angle at both heights.
The general, but not necessarily universal, conclusion that these pattern illus-
trate is that as the gain of the quad increases as a result of increasing the boom
length and number of elements, the resulting array patterns more and more re-
semble those of comparably sized Yagis. The secondary lobes may be smaller than
those of most Yagi designs, but they are present nevertheless. The quad’s ability to
suppress high-angle lobes is most evident in smaller designs.
A third consideration arises if we compare the tabulated and pictured data with
some notes earlier in this chapter. We noted that, as a rule, hams who set up
towers with lower maximum heights generally use lighter structures. If we combine
this tendency with the data that we have been exploring, then we might suggest that
the 5/8 wavelength height using the smaller quad design may be among the best
compromise installations when viewed as a whole. The smaller design has a 1.5 dB
gain deficit, but reduces mechanical concerns relative to boom lengths over twice
as long that one needs to overcome the deficit. The nearly single forward lobe
pattern has a beamwidth that covers most of the domestic and DX angles.
I shall not recommend the summary evaluation just noted as a universal solu-
tion. Rather, let’s think of it as an example of one important aspect of antenna
planning: combining the performance of the antenna with the feasible support struc-
ture and a set of operating goals to reach the best composite judgment of how much
antenna to build and how high to place it. Of course, one would add in azimuth data
in terms of pattern shape and front-to-back ratio to fill the installation design matrix
(not to mention land and monetary resource limitations). The data suggest that in
some circumstances, there may be such a thing as mounting a given antenna too
high.
To illustrate the point, let’s return to the original 3 3-element quads whose 5/8
wavelength height data was shown in Table 12-3 and Fig. 12-2. Now, let’s raise the
antenna height to 1.25 wavelengths in order to see what changes occur in the pat-
terns and the data. First, Table 12-5.
As Fig. 12-6 shows, at the higher position, the three beams lose much of what
distinguished them at lower heights. From the perspective of elevation patterns, the
region near the zenith angle becomes almost incidental to the strengths and other
properties of the main and second major lobes. In all cases, the second lobe occurs
at 35-degrees elevation with a strength ranging from 10.94 to 11.20 dBi. It is now
only about 3.0 to 3.5 dB weaker than the lowest main lobe. Moreover, there is in all
three cases a major null between the lower and upper major lobes.
The chief advantage of the increased height lies in the lower limit of the -3dB
points. The antenna can take advantage of extremely low angle signals. However,
in doing so, it may be also limit its ability to handle domestic communications, rela-
tive to the pattern with the antennas at 5/8 wavelength up. In short, there is no
general case to be made for one height relative to the other. Each height becomes
preferable by reference to a clear set of operating goals.
Table 12-6 provides the basic data that we have been tracking throughout this
exploration of elevation patterns of quad arrays. Throughout, we must remember
that we are dealing only with design-frequency data. Operating bandwidth data
must also play a role in the final overall analysis. We are merely focusing on one
region of the overall evaluation of antenna designs.
If the table suggests that with respect to elevation concerns, the three designs
are too similar to separate, the elevation patterns in Fig. 12-8 confirm the judgment.
As quad boom lengths and numbers of elements increase, the elevation patterns at
any antenna height lose their distinctive features. There may be mechanical, band-
width, and other properties by which to differentiate the designs, but the elevation
patterns give no hint of which antenna may be better for a job.
Had we examined only the 5-element quads, we might believe that elevation
patterns have little important data to offer us in making design decisions. Fortu-
nately, the 3-element designs rebut that belief and suggest that a thorough explora-
tion of elevation behavior has much to offer. Besides, I could not conclude this
study without a few elevation patterns to offset the seeming obsession with azimuth
patterns in all of the preceding chapters.
Chapter 13
Some Notes on VHF Quad Design
In this chapter, I shall look at the question of VHF quad design, at least through
4-element arrays. The chapter will serve two purposes. First, it will provide a focus
on issues that are, while not unique to VHF quads, of perhaps more concern to
building quads above the HF range. Second, these notes will serve as a summary
of some of the major elements underlying the overall study.
In fact, I might as well begin with a summary.
In effect, compared to properly design Yagis, the quad array with the same num-
ber of elements is a narrow-band antenna. Its front-to-back performance falls off
more than 2.5 times as fast as the SWR curve deteriorates. Although some users
claim they only care about the gain, the design challenge for me is to discover what
it takes to obtain superior performance in all major categories across a given band.
2. The physical specifications for a quad array are critically dependent upon the
diameter of the elements that compose them. I have shown along the way just how
much the quad is dependent in raw performance (such as gain) upon element di-
ameter as much as it is dependent upon the loop circumference for each of the
elements and upon the element spacing. The element-diameter factor is elusive in
trial-and-error design exercises, but can become clear when optimizing basic quad
designs becomes a systematic enterprise. In fact, we have shown that monoband
2-, 3-, and 4-element quad design is amenable to computerized calculations requir-
ing only the element diameter and the design frequency as inputs.
4. The theoretical gain of a quad over a Yagi with a similar number of elements
cannot be attained if the quad elements are significantly smaller in diameter than
the Yagi elements. HF quads suffer most in the regard, since the #12 to #16 AWG
wire we generally use is often less than 0.1 the diameter of typical Yagi elements for
the same frequencies. However, VHF quads suffer similarly when they use thin
wire, while the Yagis with which they compete employ aluminum rod or tubing. In
many cases, the wire quad will lose up to half of the theoretic advantage over a
single linear element solely in terms of wire losses. If we wish a quad to achieve its
full theoretical potential relative to Yagis, we shall have to use “Yagi-size” elements.
5. A limiting factor for most quad designs with superior gain and front-to-back
ratio at their design frequency is a very rapid rate of change of reactance across a
chosen passband. Low rates of reactance change tend to accompany low perfor-
mance quad arrays.
0.1 to 0.12 wavelengths spacing, with element lengths optimized. The peak front-
to-back ratio is low--about 12 dB--and the curve is shallow as we increase the spac-
ing.
The behavior of the Yagi represents a limit for parasitic operation of two 1/2-
wavelength linear elements. It is possible to derive much higher front-to-back ratios
for any pair of such elements by any one of many methods of phase-feeding both
center-points. In fact, for any spacing and set of near-1/2-wavelength element lengths,
we can find a set of relative current magnitudes and phases for the two elements to
achieve at least 50 dB of 180-degree front-to-back ratio. However, parasitically, the
mutual coupling between elements is not sufficient to achieve more than the ap-
proximate 12 dB figure at the optimal spacing.
A quad loop may--for this exercise--be thought of as two dipoles bent so that the
ends meet each other. In this configuration, the current distribution along each of
the two dipoles is different from that of a linear dipole. The current at the square
loop corners--which approximates the distance from the center to the mid-points of
a linear element--is about 14% higher with double the phase shift of the current on
the corresponding points of a linear element. The net result is a high level of mutual
coupling between elements. Since the optimal distance for achieving maximum
front-to-back ratio is a function of coupling, we should expect that the element spac-
ing for a quad would have to be greater than for a driver-reflector Yagi. It is: in the
neighborhood of 0.17 wavelengths.
For both the Yagi and the quad, the exact spacing required for maximum front-
to-back ratio is a function of two other variables: the element lengths and the ele-
ment diameter. As the element diameter increases, the quad loop lengths (for maxi-
mum front-to-back ratio combined with driver resonance) increase and the required
spacing between elements increases. It is possible to derive a series of antenna
models using NEC--which is highly accurate in this kind of exercise--that track the
array dimensions for any element diameter (“wire size”) from 3.16E-5 up to 1E-2
wavelengths. Subjecting the results to regression analysis results in a series of
equations suitable for automated design of a 2-element quad beam having maxi-
mum front-to-back ratios of more than 50 dB. Such a program, in GW Basic format
for structural transparency, appears in Chapter 4 of this study. One need only enter
the element diameter and the design frequency to derive 2-element quad dimen-
sions and some basic performance data.
At VHF, few will be tempted to build a quad using wire in the 3.16E-5 range--
somewhere in the #80 AWG range. However, the program is calibrated for 3 to 300
MHz, and quads in the 1300 kHz range have been developed from the automated
process. There are a few cautions to observe. The program lists a design fre-
quency gain that is correct for about 30 MHz. Since skin effect and its resultant
losses do not change linearly with the change in element diameter, the actual gain
of an array will be higher than predicted for frequencies significantly lower than 30
MHz and be lower than predicted for frequencies higher than the median. As well,
changing antenna materials will result in small deviations from the predictions, es-
pecially for very thin element quads. The result of material changes will be minimal
with elements larger than 1E-3 wavelengths in diameter.
Table 13-1 lists the wires sizes that we shall sample in this study, arranged in an
overall 8:1 total ratio in 2:1 increments. The smallest size is just barely thinner than
#14 AWG (0.0641"). The largest size (0.5") represents a practical limit to modeling
accuracy for the exercise, as the diameter approaches 1E-2 wavelengths at 146
MHz. The common logs for the wire diameters (in wavelengths) are also listed,
since the properties of antennas tend to vary more directly with the common log of
wire size than with the size itself.
1. 2.
Wire Diameter: 0.0625" Wire Diameter: 0.125"
Reflector Circumference: 88.480" Reflector Circumference: 89.672"
Driver Circumference: 82.304" Driver Circumference: 82.584"
Refl-Driver Spacing: 13.140" Refl-Driver Spacing: 13.324"
Feedpoint Impedance: 141.1 Ω Feedpoint Impedance: 142.3 Ω
Free-Space Gain: 7.06 dBi Free-Space Gain: 7.11 dBi
< 2:1 SWR Bandwidth: 18.17 MHz < 2:1 SWR Bandwidth: 20.78 MHz
>20 dB F-B Bandwidth: 3.45 MHz >20 dB F-B Bandwidth: 4.19 MHz
3. 4.
Wire Diameter: 0.25" Wire Diameter: 0.5"
Reflector Circumference: 91.304" Reflector Circumference: 93.608"
Driver Circumference: 83.064" Driver Circumference: 83.936"
Refl-Driver Spacing: 13.493" Refl-Driver Spacing: 13.718"
Feedpoint Impedance: 145.0 Ω Feedpoint Impedance: 150.4 Ω
Free-Space Gain: 7.14 dBi Free-Space Gain: 7.17 dBi
< 2:1 SWR Bandwidth: 24.63 MHz < 2:1 SWR Bandwidth: 31.11 MHz
>20 dB F-B Bandwidth: 5.29 MHz >20 dB F-B Bandwidth: 6.87 MHz
The dimensional growth of the array with increasing element diameter is clear in
the table. The feedpoint impedances are resonant at the design frequency within +/
-1 Ohm reactance. We can summarize a number of the other features of the table
better in a selection of graphs that capture frequency sweeps of each design across
the 2-meter band. For example, Fig. 13-1 records not only the free-space gain of
the array at the design frequency, but as well the rate of change of gain across the
band. As one might expect, the fatter the element, the lower the rate of gain change,
thus ensuring more equal gain at both band edges. While commenting on element
diameter, we should add that, contrary to experience with linear elements, closed
loops tend to grow larger with increasing diameter elements and to have higher
feedpoint impedances.
Fig. 13-2 extracts the 180-degree front-to-back data, which reflect also the >20
dB bandwidth entry in the Table 13-2. Note that the front-to-back bandwidth--when
held to this standard--does not exceed 4 MHz until the element diameter reaches
0.125", and with a mid-band design frequency, does not exceed 20 dB at the low
end of the band until we use a 0.25" diameter element. We shall shortly explore why
it is best to design quad arrays for a position about 1/3rd up from the bottom of the
desired operating passband.
The SWR curves (relative to the resonant impedance of each array) for these 2-
element optimized arrays, as shown in Fig. 13-3, present little concern to the builder.
They are the reason so many builders classify the quad as a “low-Q” antenna, al-
though the front-to-back curves--even for these fat-element arrays--show the inac-
curacy of that claim.
Fig. 13-4 shows free-space azimuth patterns for the low end of the band for the
thinnest and thickest elements used in this study. Note that in each case, the worst-
case front-to-back ratio just approaches 18 dB--which tends to hold for all models
used across the band. What does not appear as rearward gain in one direction
tends to show up as gain in another direction within the rear quadrants. Thus, there
is an overall limit to rear quadrant performance for any 2-element single-feed quad
array.
While we are comparing the thinnest and thickest elements in our collection of
models, let’s look more closely at the wider-band performance of the array--espe-
cially at the front-to-back ratio and at the SWR curves. Fig. 13-5 presents the data
for each parameter from 140 through 175 MHz, with each array having a design
frequency of 146 MHz. Note that the front-to-back curves fall off much more rapidly
below design frequency than above it, although the two curves parallel each other
closely. With respect to SWR, we would correctly anticipate a steeper curve for the
thinner element model. However, for both the 0.625" and 0.5" model, the curves are
both more shallow above design frequency than below. The lesson of these curves
is simple: for a given desired operational passband, design the quad array for a
frequency about 1/3rd the way up from the lower end of that passband in order to
achieve roughly equal values of SWR and front-to-back ratio at both ends of the
passband. The loss of gain from using this procedure will be rather slight.
The actual resonant frequency of the driver is, for small changes in loop diam-
eters, relatively independent of changes in front-to-back ratio--which is largely con-
trolled by the spacing and loop diameter of the reflector. Hence, for special pur-
poses, one may place the resonant frequency of the driver almost anywhere within
a passband without moving the frequency of maximum front-to-back ratio signifi-
cantly. In short, the 2-element quad array can be customized within reason to the
builder’s desires.
Although 2-element quads for 2-meters are “small potatoes” by most VHF array
standards, it is necessary to understand the properties of these basic antennas in
order to better understand the properties of larger quads. (Indeed, one must under-
stand the linear dipole to understand the quad loop and the quad loop to understand
the 2-element quad.) Therefore, let’s add a director to the 2-element quad and see
what happens.
In my own efforts to optimize 3-element quad arrays, I selected two criteria ini-
tially: gain and operating bandwidth. Initially, I gave precedence to operating band-
width, using the >20 dB front-to-back standard. Later, I turned to gain as the para-
mount criterion, with reasonably-wide bandwidth as the secondary standard. The
results proved interesting. For the wire sizes we have selected as most relevant to
2-meter arrays, the arrays ranged between 32" and 36" long. This value is nearly
30% longer than most 3-element 2-meter quads, but the results turn out to be very
consistent in each category for operating properties. Table 13-3 lists the resultant
arrays for both wide-band and for high-gain applications.
Let’s begin with the wide-band arrays. Fig. 13-6 shows the frequency sweep of
gain across the 2-meter band for each array in the left column of Table 3. For wide-
band operation, the peak gain occurs below the lower limit of the operating pass-
band. Hence, gain decreases across the band. However, the decrease is signifi-
cantly lessened as we increase the element diameter, with the 0.5" element model
showing just over a 0.05 dB change. When maximum operating bandwidth is the
primary consideration in 3-element design, increasing the element diameter draws
the point of maximum gain closer to being within the operating passband so that the
gain curve is at its shallowest.
The 180-degree front-to-back curve in Fig. 13-7 reveals that the operative de-
sign technique--as used with the 2-element quad--was to place maximum front-to-
back at or very near to the design frequency. However, the addition of the 3rd
element limits the maximum value of front-to-back ratio as well as the bandwidth
over which the value exceeds 20 dB. Although any of the curves might well be
usable for many purposes, in terms of the design standards of this exercise, only
the 0.5" element version of the antenna achieves 20 dB across the entire band.
As with the 2-element quads, the SWR bandwidth of the arrays is considerably
wider than the front-to-back bandwidth, as shown in Fig. 13-8. The more rapid rise
in SWR below the design frequency is evident for all versions of the array. However,
all of the curves would be acceptable. In fact, referring to Table 13-3, any of these
antennas would be matchable directly to a 75-Ohm main feedline.
Fig. 13-9 provides--among the models given in Table 13-3--a worst-case look
at the patterns of the 0.0625" element diameter model for the entire 2-meter band.
In practical terms, the gain is reasonably stable, even if not as good as that of the
0.5" model. As well, the rear quadrants--although not up to the 20 dB standard at
the band edges--remain quite well controlled and predictable. The nearly 9 dBi free-
space gain is rarely achieved in common short-boom 4-element Yagis.
If we turn to the high-gain versions of the antenna, we should expect more gain
and a narrower operating bandwidth. In fact, the design of these arrays turned out
to be something of a surprise for the designer. As shown in Fig. 13-10, the peak
gain appears within the operating passband, although it gradually shifts toward the
high end of the band with the fattest-element version of the array. For the 0.5"
diameter model, the gain is about a half dB higher than for the wide-band version,
with excellent gain stability across the band.
The limiting factor in the high-gain arrays is the SWR curve. The second sur-
prise of the design exercise was that fact that these arrays all yielded impedance
values close to 50 Ohms. The 0.0625" element model fails to achieve a 2:1 50-
Ohm SWR curve for the entire 2-meter band. However, the larger-diameter models
fit safely within the limits, although with higher band-edge SWR values than for the
wide-band models.
Fig. 13-12 provides a set of free-space azimuth patterns across the band for the
0.0625" model--allowing a comparison with the wide-band version of the array shown
in Fig. 13-9. Whether these patterns are suitable for particular applications is a user
judgment.
The maximum design-frequency gain of the high-gain models is about 9.6 dBi
(free-space), which approaches the gain of a standard 5-element Yagi (about 10
dBi). The wide-band version achieves about 9.1 dBi for the same 0.5" element
diameter, with a proportionally wider operating bandwidth. The most interesting
aspect of this seemingly small set of differentials is that they require very different
dimensional profiles to maximize each set of design goals. An inspection of the
numbers in Table 13-3 will tell the difference in profiles, and a comparative drawing
appears in Chapter 7.
Basic listing for the wide-band array design, while Chapter 7 supplies a similar list-
ing for the high-gain version.
As with the 2-element arrays, the designs are resonated within +/-1 Ohm reac-
tance at the design frequency. However, the driven element can be adjusted to
place resonance as much as a MHz away from the design frequency with minimal
effect on the other antenna properties. Alterations to the director and reflector loop
sizes have greater consequences for performance. Thus, to move the center point
for the peak front-to-back ratio or SWR curve lower in the band, it is recommended
that one choose a slightly lower design frequency.
The foray into 4-element quad array design is fraught with further complications.
The additional director multiplies the number of variables. However, we may be
able to indicate some useful directions for design effort--and possibly some limita-
tions that may be inherent in long quads.
A useful place to begin would be with a sample or two of existing design. I have
selected two widely disseminated design, although they appear in disguise. Models
of each of them required that I frequency scale the design in order to produce any
usable results. Hence, one is redesigned 1.5 MHz below its original design fre-
quency, while the other is 2 MHz below its original design frequency. Table 13-4
summarizes the design and performance data for the two designs.
Version B:
Frequency Free-Space Front-to- Feed Impedance 50-Ohm SWR
MHz Gain dBi Ratio dB R +/- jX Ohms
144 9.55 7.89 35.6 - j 39.0 2.50
145 9.86 10.04 38.6 - j 17.3 1.60
146 10.05 12.42 42.4 + j 3.7 1.20
147 10.14 14.85 46.7 + j 23.7 1.63
148 10.17 16.86 50.8 + j 42.4 2.27
The similarity of the final designs is a function of their design premises: to use
thin wire and a short boom. There should be little wonder that the performance is
similar for each array, since there is so little difference in the design, despite their
independent sources.
The two designs have boom lengths about 6" longer than the average 3-ele-
ment quad, and their design frequency gain level is only about 0.4 dB higher than
the best of the high-gain series of 3-element quads. The point of maximum front-to-
back ratio is well above the passband of the array (that is, above 148 MHz). Finally,
the neither antenna manages to provide less than 2:1 SWR across 2-meters.
for 4 such quads with our standard range of element diameters. As shown in Fig.
13-14, even the thinnest element provides good gain across the band, although the
0.5" diameter version supplies an additional 0.4 or more dB gain.
1. 2.
Wire Diameter: 0.0625" Wire Diameter: 0.125"
Reflector Circumference: 86.658" Reflector Circumference: 87.360"
Driver Circumference: 82.448" Driver Circumference: 82.728"
Refl-Driver Spacing: 13.218" Refl-Driver Spacing: 13.218"
Dir 1 Circumference: 78.024" Dir 1 Circumference: 77.952"
Refl-Dir 1 Spacing: 38.885" Refl-Dir 1 Spacing: 38.885"
Dir 2 Circumference: 75.264" Dir 2 Circumference: 75.192"
Refl-Dir-2 Spacing: 68.338" Refl-Dir-2 Spacing: 67.947"
Feedpoint Impedance: 60.6 Ω Feedpoint Impedance: 58.5 Ω
Free-Space Gain: 10.23 dBi Free-Space Gain: 10.40 dBi
< 2:1 SWR Bandwidth: 2.82 MHz < 2:1 SWR Bandwidth: 3.08 MHz
>20 dB F-B Bandwidth: 1.74 MHz >20 dB F-B Bandwidth: 1.99 MHz
3. 4.
Wire Diameter: 0.25" Wire Diameter: 0.5"
Reflector Circumference: 88.448" Reflector Circumference: 89.976"
Driver Circumference: 83.072" Driver Circumference: 83.424"
Refl-Driver Spacing: 13.218" Refl-Driver Spacing: 13.218"
Dir 1 Circumference: 77.782" Dir 1 Circumference: 77.760"
Refl-Dir 1 Spacing: 38.885" Refl-Dir 1 Spacing: 38.885"
Dir 2 Circumference: 74.992" Dir 2 Circumference: 74.352"
Refl-Dir-2 Spacing: 67.446" Refl-Dir-2 Spacing: 66.953"
Feedpoint Impedance: 57.3 Ω Feedpoint Impedance: 55.0 Ω
Free-Space Gain: 10.52 dBi Free-Space Gain: 10.61 dBi
< 2:1 SWR Bandwidth: 3.46 MHz < 2:1 SWR Bandwidth: 4.02 MHz
>20 dB F-B Bandwidth: 2.32 MHz >20 dB F-B Bandwidth: 2.75 MHz
These wide-band 4-element quads are only relatively wide--that is, for the class
of 4-element quads. In Fig. 13-15, we can see that only the 0.5" version of this
design provides better than 20 dB 180-degree front-to-back ratio for the entirety of 2
meters, although the thinnest element version provides better than 15 dB front-to-
back ratio across the band. The SWR curves in Fig. 13-16 reveal that 3 of the 4
arrays meet the 2:1 SWR standard. Since (as shown in Table 13-5) the resonant
impedances are all close to 50 Ohms, these curves will reflect performance with a
standard 50-Ohm cable as well.
Very often gain and 180-degree front-to-back ratios come at the expense of
general pattern shape. However, the designs we are presently discussing have well
controlled patterns. The free-space azimuth patterns for the 0.5" diameter element
version of the array appear in Fig. 13-17 to verify this claim. The forward lobe is
especially consistent from one end of the band to the other, while the rear lobes
remain quite well controlled.
The arrays that we have sampled emerged from a small GW Basic program,
which is listed in Chapter 9. The optimized samples that provided the basis for the
regression analysis are extensions of the 2- and 3-element wide-band quads that
we earlier discussed.
The collection of quad designs with which we have been working reveals some
interesting trends. Fig. 13-18 graphs the design-frequency gains of all four pro-
grammed quad designs for each of the sample 2-meter element diameters. Al-
though the relative gain figures for 2-, 3-, and 4-element quads are interesting, the
relative slope of the curves is the key attention-getter. Note that the more elements
we add to the quad design, the more dependent the quad becomes on the element
diameter. Making multi-element quads of thin wire at VHF and up is one way to lose
a significant part of any design’s potential.
Fig. 13-19 reveals another fact of life about quads: the greater the number of
elements, the narrower the SWR bandwidth of the array. Above a certain number of
elements (I do not yet know the exact number), the only way to achieve a wide SWR
bandwidth may be to stagger parasitic elements. However, in every case of stag-
ger-tuned elements, the gain falls below maximum. Hence, the multi-element quad
may be self-limiting relative to competing designs wherever operating bandwidth is
a significant consideration.
Three items should be immediately apparent from Table 13-6. First, the design
uses large-diameter elements: 1/2" in this case. Second, the boom is quite long--
about 2" shy of 6'. Third, the design frequency gain is nearly a full dB higher than for
the short-boom models with which we began our foray into 4-element quads. The
question remaining is whether we have made sufficient improvements to justify the
added mechanical requirements of a quad over a Yagi.
To partially answer this question, I shall present the design data for 3 OWA Yagis
for 2 meters: a 6-element design using 3/16" elements, a 7-element design using
0.1" elements, and another 7-element design using 1/4" elements. Comparative
physical and performance data may prove instructive--at least provisionally.
As shown in Table 13-7, the boom lengths of the Yagis bracket our improved 4-
element quad array. The 70" quad boom lies roughly evenly between the 54" 6-
element Yagi boom and the 84" 7-element Yagi booms. As well, the 4-element quad
gain also lies between the Yagi numbers. However, before we make judgments
using date from just the design frequency, let’s survey the entirety of the 2-meter
band for all 4 antenna designs.
Fig. 13-20 shows the free-space gain figures for the 4 antennas--3 Yagis and a
quad. Apparent is the greater stability of the Yagi gain, as the quad changes gain by
over 0.4 dB across the band. By way of contrast, the Yagis show a maximum gain
change of about 0.2 dB. How significant this factor is to a given operation will be a
user judgment.
Of perhaps more interest is the fact that the relative average gains of the arrays
shown in the graph is almost wholly proportional to the boom lengths of the arrays.
The quad boom length is almost precisely the median between the 6- and 7-ele-
ment OWA Yagis, and its average gain also closely fits that same position. Hence,
the gain advantage of the quad relative to boom length--when optimized for a con-
siderable bandwidth--begins to disappear as we reach 4-elements. There are spot
designs that can achieve the same gain on a shorter boom--and a few such designs
may show a good operating bandwidth for certain element diameters. However, the
ones I have so far examined tend to narrow in operating bandwidth with decreasing
element diameter at a faster rate than the high-gain 4-element example used here.
All of the antennas achieve 50-Ohm SWR values of under 2:1 across 2 meters,
as shown in Fig. 13-22. All three OWAs have values of under 1.25:1 across the
band, with the 7-element 0.25" element version achieving under 1.2:1 across 2
meters. All three Yagis can be tweaked for this level of performance, although the
1.25:1 maximum of the three designs is likely to prove acceptable under most cir-
cumstances. The quad comes nowhere near the Yagi performance, although it
does stay within the <2:1 SWR standard. Whether the OWA curves are significantly
preferable to the quad curve may rest on a user judgment of the significance of line
losses within the total antenna-feedline system. For a given installation, a complete
analysis of line losses across the band would be a wise procedure to use before
reaching a conclusion.
It is interesting also to investigate pattern shape for various antennas that one
might consider using. Fig. 13-23 compares the free-space azimuth patterns for the
quad with those for the 0.25" element 7-element OWA Yagi. The Yagi displays
better control of rear quadrant radiation across the band. However, the long Yagi
also shows some “neck bulges,” that is, emergent secondary lobes. Similar lobes
are just appearing within the quad pattern, but are far less prominent. To an an-
tenna designer who is not under the press of commercial deadlines, the existence
of all such pattern bulges is a sign that further development work is in order. Such
emergent lobes are not always eliminable, but the purist tries anyway. This last note
is a way of saying that the work is still in progress and far from complete.
The conclusions to be drawn from this investigation may be obvious, but are
perhaps worth noting anyway.
2. Traditional short-boom monoband quad designs fail to realize full quad gain
and bandwidth by overlooking the naturally higher mutual coupling of quad elements-
-a factor that dictates longer boom lengths for optimal performance (relative to ar-
rays with linear elements).
5. When it comes down to simple and cheap utility antennas with some gain
and directivity, the Yagi-quad decision becomes more a matter of what materials are
available and depends less upon the purist considerations that have gone into these
notes. Coat hangers, scrap wire, wood, and PVC have yielded generations of utility
antennas of either the Yagi or the quad type.
Chapter 14
Quad Horizons
What I think this volume might have accomplished has been summed up in the
preceding chapter. What remains to be done I shall list here, at least as I see it
today.
First, there are a number of spot designs that seem to violate the progressions
uncovered in the work so far. They manage to achieve high gain and reasonably
broad-banded operation on shorter than expected boom lengths, for a given num-
ber of elements. Finding a progression of such designs with consistent rates of
reactance change has so far been elusive. However, elusiveness does not mean
impossible-to-find, but only that I have not found the desired progressions. Whether
such progressions exist may rest on the translation of these findings into more theo-
retically sound fundamental relationships governing quad loop interaction. Indeed,
there may be some “periodic” functions associated with quad loop interaction.
The nature of this investigation has been systematic, but not theoretically based-
-other than the degree to which the modeling programs themselves are theoretically
based. At most, I have uncovered some progressions of quad operation that sug-
gest fundamental properties, but the discoveries do not establish the suggested
principles to the exclusion of all others. Much remains to be done with respect to
quad theory.
Second, there are a number of directions that I have not taken in this study. The
potentials of parasitic arrays composed of asymmetrical and symmetrical rectangles
of two or more loops each is work that others are exploring. Such loops show great
promise of providing wide operating bandwidths on shorter booms with fewer ele-
ments--but, of course, at the expense of a far more complex structure within each
element of the array.
showed great promise, generally outperforming Yagis of the period. However, those
Yagis were generally of the era before computer optimization. How much a quagi
design might add to the present generation of Yagi designs remains to be deter-
mined. Of course, we might also let the quad loop elements be multi-loop elements,
either asymmetrical or symmetrical. Whether there is a design that combines the
best of both worlds remains on the horizon.
Finally, but not exhaustively, much remains to be done in the region of multi-
band quads. Volume 1 of this effort established that the concentric rings of quad
elements not only interact, but do so in good designs so as to enhance operation on
at least some bands. The enhancement may show up as gain or as operating
bandwidth. However, to have shown that these effects exist is a long way from
understanding how they occur and an even longer way from being able to control
them so as to optimize multi-band quads to desired levels of performance.
To the degree, then, that this study has solved any problems in quad design,
only the easiest questions have met with some success. Many more theoretical
and functional challenges still remain. Hopefully, this study, besides providing some
practical analysis and design guidance, will inspire others to look into the fascinating
world of quad loops and arrays. The more investigators and investigations, the
greater the chance that we shall eventually understand how to put this class of
antennas to its optimal set of uses.
A Publication by
antenneX Online Magazine
http://www.antennex.com/
POB 72022
Corpus Christi, Texas 78472 USA
Copyright © 2001 by L. B. Cebik jointly with antenneX Online Magazine. All rights
reserved. No part of this book may be reproduced or transmitted in any form, by any
means (electronic, photocopying, recording, or otherwise) without the prior written
permission of the author and publisher jointly.
ISBN: 1-877992-10-0