A

2
“ayrete

2S
pee SESS
1 Se ES
a a~
ae

Aecoeni <n ne cbt

os
o-4 4 Bat 5 FRY gel BR CRN ene oi RP etek i
Ai iJ 5 m] ae Sy |S es Se e iw i.
: $ wor NP) a: TRAN 4.5

WR ory ata)
Sure bias Ama No!lc RL
Ay *, = ¥% Mar ¥ rr
SOC nt ars nee
Aen ; y
ult wy

as
BLY ae Af A 4 ee" § in”.
> iy
=f } aed My
| On ie ae
“aa &hy 7 1 4 hey
? t x A Lh | 7% yaa rea is o ye
ie ees tL MOE Ty Wile Ces
Rah
ae
77)
Pera

:
"
als
a omg
Me
alt a
Airy a
Mi
rr

:
te
ves
ue
dah
ef

oe
Yo

‘
i a

a reo 7 Wine i? i i
_ S f :'? ’ 4) j k \. ‘

ey a, a: ae ae a
‘ aN 4 - Wi ar rlJ wa \ + xy r . .

‘i ba‘ I, a3 MT : >
72% i ah ; j
Zi 2@ r a * ; ally
cet ‘ | Ste
¥ } t =
i, uh ity 4 Ne
; , i fe Un
“v a pw <a

al S41 ed

ti! v
\
iy ‘Ae

' { , y
v. a”

i ( j
Ly >
1 , ‘ f ‘
+ 5
)
nt
; Sy ’

~J

2 ry, ‘

{ +4
: ge by
, j i

a ~ , t
tb
= '

Vey,

i 1s

oe!

‘ ome SS
. th ly et
pin } Rn: ‘=
 st,
aa

aA Me Yo é
ig Pa> ¢
ar
J Ly a uo
alee ve } rie
Pv; i Pe ta a
Le Ciel
ia'p)
rs Ks of

f 7 ~

i i U . Yaw!
“a in’ a

. ‘

‘chy \ eat

yi ‘

y ~
* P| {

ine ’ - tia Tit w . a BSa?

; , a ; Pe a ee
F | - Lie

ERS
a
ALY
4
tian
Nery
PS
a
a tb,
ith
oa elde DALES
rie rid
7. our ar ORF
Fe a ih Yi Oe is
Lie : 2
r ie TAREat |
eM
i et
; > at e. nt
i a y 4
i Wy ets - nh, , 4 net ¢ ? os . ny
| - HAG PO e ine, | ; t
a PS tok an ee Pie he , Che SST or Pi ' k PATS ; i= 5 ' |

Seite
7 ‘6 , Y : Ae

ken MURCR Ret Wah st eae

whims ¢ ae { Vv :

oy waihe tae el
tin Ny" 4 styVECO
hy re :ul7 Mar
anes
nied ahs
aed Br
WALI
tesavi a
ie wT
Way OMEN eR
|
yA ame
P

or Ie Wha wig a eee hy A \ al iy. Ve jorhae Ws, Bat hi Pt i

ret : ~ ey a ee yo ’ a ‘ :
. Se eh ie. a Vv eh re j
i?
. nl Lae)
T
i ‘ ; ; | ‘ aM+ ‘ef
\ heehee P
|
bh ig Ved
PU tom 9 At Ae etna ty |
Ay 4 i 5 Pees \
\ e ‘ \ rte ‘
iy rf A ih oust: f
P| 24 ‘ J ive Pent e ver ‘ | 1
Aas . ial? /

i ant
ri , ul 4
* r e) i
i
 , r xh Nene
li t eo. i wi \ [ _* ne ‘ ry
ete Sen14BATea og0 te ee :
id ‘ 5 ’ ae i"; ms es Frits Pee oan
fe ‘

1 y
vhf
' t
APs) Y
met
UF
hay Cahier
:
« Ay ¥ D4
ye
;
oer
ce | TS
i
Aral
ry tht
i
\:

s ’ . 1 .

- 4 . j i,
x nt ’
{ - f ‘ he

| i
4 a
ti NI 1 a h yiy
hf antennas
for all locations

L. A. Moxon, BSc, CEng, MIEE, G6XN

RADIO SOCIETY OF GREAT BRITAIN

Published by the Radio Society of Great Britain,
35 Doughty Street, London WCIN 2AE

First published 1982

© Radio Society of Great Britain, 1982. All rights reserved. No part of this
publication may be reproduced, stored in a retrieval system, or transmitted, in any
form or by any means, electronic, mechanical, photocopying, recording or
otherwise, without the prior written permission of the Radio Society of Great
Britain.

ISBN 0 900612 57 6

Text set in 9/10 pt Linotron 202 Times, printed and bound

in Great Britain at The Pitman Press, Bath
Contents

PART [ — HOW ANTENNAS WORK

Mlamnoa new Ook at hhantennas: fii ..o-4. ae ts ees 1

wes ANC HeMIS . ee, dee a ee ee eee 4
RaCoHIN An GHlOSGESS. wwe. Seek. cs. Seat evecare
Ee, ewcall!s 20
Mere Cei ote ANCHiia™, Jett cc. Meet eee, ee 30
MIO SESHACC CRO CANIS cy ag fi ont Wane ou)eoen nd hiaadanda ide 67
. Arrays, long wires, and ground reflections ............ 93
PEMD Ae Gilids | pies bese el eee ee 106
ES ATIONTIGPee want adeees. .. , Bieta tilede tit cgacce’ Se etae 121
PeyAniteniia GESiIgn LOL LeCEPUON eis
.ag eee cote st ce 126
PaLNe Aten
—NHN
HBRWN
SOwuoMAanNANA ara its CUVILOD INCI aes ctees ce ak kc: iteW

PART II — THEORY INTO PRACTICE

molnglecclementantennasyr » ys sees eit, Soeeaeee. 147
PREIOTIZON Cal DCAITIS fee een os 15 ern tee ete 166
SASH TEEN ISBY=71
11 Ee oe Oi arn 183
PIC AGOCEAIT AVS wxas RIEU. sk Aces Gan eaten eet amicSs ois ls 198
Ma ate lies tats a Or ey aca 205
-aviebile. and portableantennas. awstats.
cee. ee. 208
Myviine king Olanteniay 2. eee
eee es 216
Sees Obert LOT ABVMK. a2 waar ise Oi cuisine ines pita 225
x Antenna construction and erection Me. es 245
Acknowledgements
The author is indebted to numerous other amateurs for advising him of their
queries, difficulties and experiences which have helped considerably in the choice
and presentation of subject matter. He is grateful also to Pat Hawker, G3VA, John
Piggot, G2PT, and others for helpful criticisms leading to some revision of earlier
chapters. Thanks also go to Robert Meadows for the cartoons appearing on pp145,
179 and 205.
Much use has been made of material from other sources as indicated by
appropriate references.
CHAPTER 1

Taking a new look at hf antennas

Amateurs have been using hf antennas for 60 years or There is also an urgent and increasing need for reduc-
more so that some slowing down in the rate of progress tion in interfering signal levels, and the designer of a
might be expected. As evidence of this, some of the beam antenna is often faced with the dilemma of
best antennas in use today were designed 20 or 30 years whether to maximize gain or minimize interference.
ago yet, impelled by strong incentives, the search for There is, however, no mandate insisting that antenna
“better” ones continues unabated. At times of slow characteristics have to remain unchanged when one
progress it is not a bad idea to look at problems from switches from transmission to reception, so in principle
new angles, and a possible first step is to erase from this should be a non-issue; in practice, however, there
one’s mind the picture of “things as they are’’ and is still a long way to go. In the case of a rotary beam the
concentrate, for a few moments at least, on the way one time required to change beam direction is a serious
would like them to be. Listing the reasons why they are nuisance yet, with two or three elements, instantaneous
not and challenging each in turn can be good fun; beam reversal can be readily provided and combined
moreover, before accepting once again the status quo, with other advantages.
one feels entitled to convincing evidence that there are The contents of this book reflect the author’s view
no alternatives. that more often than not these objectives can best be
In the following pages the reader will find few such achieved by the use of two dipole elements only. The
proofs but instead several challenges to existing beliefs argument that one can do a lot better with loops or, in
and practices, ‘proof of impossibility’ being reserved the case of dipoles, three elements is disposed of by
mainly for the antenna gain figures frequently claimed challenging the usual implicit assumption that dipoles
by authors and advertisers. have to be “‘straight”’, thereby providing a good exam-
ple of the critical approach suggested above.
On the basis of what used to be the only type of
The need for a new approach dipole array in normal use the superiority of three
Most amateurs are subject to more or less severe elements can be demonstrated mathematically but this
restrictions in the erection of hf antennas. Those able to conclusion has been invalidated by new forms of con-
erect beams at a height of SOft (15m) can count struction such as the VK2ABQ antenna and, pursuing
themselves lucky, and any sense of frustration they may the matter further, it turns out that by introducing
feel at not being able to double this will not be relieved different but equally realistic sets of practical con-
by evidence reviewed in the next chapter pointing to straints a case can be established either way.
the desirability of heights of 1,000ft (300m) or so! The
constraints by which amateurs are afflicted can be
reckoned as ranging all the way from “severe” to What is possible?
“impossible” with each installation tending to be a Apart from the items already noted perhaps the most
special case. One has therefore the problem of selecting insistent demand, given the possibility, would be for
the antenna systems which provide the best ‘‘match”’ to more gain and although basically there is very little
any given set of circumstances and, apart from possible scope in this direction much can be done to avoid
improvements to antennas as such, the author believes wastage; for example, by avoiding unsound designs and
there is an urgent need for better guidelines. This in the case of multiband operation eliminating traps,
conviction arises from a study of the available litera- the whole of each of the lower-frequency elements (or
ture, from his own personal experiences, and from an equivalent length of element) being used at the
numerous enquiries which in most cases cannot be dealt higher frequencies. Better performance may also be
with adequately short of writing a book! achievable in those cases where a reduction of size and
What then are the improvements needed, and how weight as mentioned above allows a substantial in-
are they to be achieved? Perhaps the most urgent need, crease in height.
and as will be shown one of the easiest to meet, is a Possibilities appear to exist also for improvements in
substantial reduction in the size, weight and cost of indoor antennas and other methods of achieving opera-
beam antennas without compromising performance. tion under difficult circumstances. Many amateurs are
2 HF ANTENNAS FOR ALL LOCATIONS

severely handicapped by height restrictions and, in this the additive case elements have to be widely spaced
context, the author’s investigations have pointed to- otherwise the addition tends to be equally effective (or
wards a much more important role for vertical beams ineffective) in all directions, but with this proviso the ©
which at the time of writing appear to be almost gain obtainable is equal to n, the number of elements.
non-existent for the higher frequencies. In this way it For large gain therefore one needs a lot of elements and
has been found possible to achieve reasonably competi- a lot of space so the first step is to invest in a lot of real
tive dx performance even when height is severely estate—which puts it beyond the reach of the average |
restricted, and to this end improved designs with amateur who must either settle for a small gain or hope
simpler and more efficient methods of feed not requir- for better luck using the subtractive method. In this
ing extensive ground systems have been worked out. case close spacing of elements is not merely possible
but essential since wide spacing produces large phase
differences, in which case, in some direction or other,
Reasons for the present situation there will be addition.
The fact that much scope still exists for improvements Most amateur beams are in fact of this latter type,
in amateur hf antennas can be attributed to widespread and so operate on the subtractive (or “‘supergain’’)
ignorance of a number of basic principles, leading to principle; unfortunately this appears to be little under-
proliferation of unsound or inefficient designs. Text- stood, thus accounting for numerous mistakes as well as
books have been geared mainly to the needs of failure to realize the full potential of close-spaced
students and professional engineers, and amateur arrays.
handbooks have tended to be concerned more with the
“how” than the “‘why”’ so that experimenters have for
the most part lacked the insight which would enable Subtractive-gain arrays
them to challenge an observation by posing the all- Most of the difficulty stems from the fact that when
important question: does it make sense? This is vital in elements are closely spaced they couple tightly
any experimental work and especially in the case of hf together; it is the nature of this coupling and the way in
antennas in view of the difficulty of making reliable gain which it is exploited that determines the behaviour of
measurements, as discussed in Chapter 18. the antenna. Let us for the moment assume that in
some way or other this has been optimized with a view
to achieving as much gain as possible.
Towards a better understanding
The first point to note is that, compared with additive
In view of serious errors by some of the experts and the arrays, the theoretical gain for a given number of
apparent complexity of the hf antenna scene it is felt elements is much greater, being equal to n? and inde-
that the amateur can hardly be blamed for a situation pendent of size provided only that this is small enough!
which is somewhat less than satisfactory, or perhaps for Unfortunately, for reasons explained in Chapter 3 in
thinking there is not much he can do about it. Never- more detail, it is not possible to get very far along this
theless most of the main essentials can be grasped particular road but by way of a practical, or at least
intuitively; thus most of what one needs to know about “not impossible” example, consider the case of four
radiation from the simpler types of antenna can be elements. If these are arranged in an additive formation
deduced merely by observing the way in which the they would provide a gain of four times or 6dB,
apparent length varies as one moves around, towards, requiring for this purpose at 14MHz a space of about
or away from them as explained in Chapter 2. This 80ft by 35ft (24m by 11m) or, in the case of a rotary
point is elaborated in Chapter 3 to show that all such beam, a turning-circle diameter of about 90ft (27m).
antennas behave in the same way regardless of shape The gain is roughly equivalent to that provided by the
or, up to a certain limit, size so that a bewildering average linear amplifier; in contrast to this the n2
assortment of apparently different antennas can be formula predicts a gain of 16 times, equivalent to two
seen as a single, quite simple device in various dis- linear amplifiers each giving 6dB gain.
guises. This constitutes a basic ‘“‘building brick’’, and Unfortunately, as the reader may suspect, matters
beam antennas consist of two or more such sources of are not quite as simple in practice, as will be evident
signal assembled in such a way that the radiated energy from the ‘small print’ in Chapter 5, but one design
tends to be focussed in a particular direction rather (p88) aimed at this objective succeeding in achieving a
than spread relatively thinly throughout the surround- gain of 8-7dB with a turning circle diameter (when
ing space. scaled for 14MHz) of 55ft (17m) only. The snag is that
Current misunderstandings are attributable largely to large “‘subtractive’’ gains involve very large currents
the failure to realize that there are two completely and a point is quickly reached where any increase in
different methods for achieving this result, signals being theoretical gain is offset in practice by increased loss
arranged in the one case to add in the wanted direction in the form of heating of the antenna. There is also a
and, in the other, to subtract in unwanted directions. In small theoretical deduction to be made from the n2
 TAKING A NEW LOOK AT HF ANTENNAS 3

formula, ie 0-8dB for two elements, rising to 2-0dB for closely-spaced elements can be made to work, have
five and over. These effects lead to an upper gain limit been disposed not to bother unduly with explanations.
in most practical cases of about 7-8dB for subtractive Unfortunately, although knowing “how” has resulted
(or “supergain”’) arrays, with the distinction between in lots of antennas capable of producing big signals,
the two mechanisms tending to become somewhat lack of the why has led to absurdities such as the quad
blurred as the number of elements increases. versus Yagi controversy which has been a prominent
These complications need not worry the reader un- feature of the amateur hf antenna scene for more than
duly since reasonably accurate guidelines based on data two decades.
from a number of sources have been provided in There has also been a consequent waste of much
Chapter 5. Furthermore, application of the n? formula, time, money and effort in the pursuit of better results
even without corrections, is often sufficient for asses- by methods which either have no chance of success or
sing the credibility of claimed gain figures, many of lead to undesirable compromises of one sort or
which are well in excess of n?. Due to ignorance of the another; in addition it has ensured that virtually all such
subtractive principle and its consequences, gains in antenna designs have been in some degree (or in some
excessof n? are liable to be attributed to some unlikely respect) sub-optimum. For this reason particular im-
property of the element shape or general arrangement portance attaches to Chapter 5 which shows that for
and there is an evident need to set the record straight, two elements the problem can be reduced to simple
this being one of the tasks of Chapter 3. dimensions merely by taking success for granted, ie one
For a given form of construction there is a practical assumes the elements to have equal-amplitude currents
lower size limit determined by losses as discussed in which are correctly phased. In this case the answers
Chapter 3, and there is in all cases an upper spacing emerge from a few lines of algebra, affording con-
limit in the region of 0-2 above which “‘supergain” siderable insight into the basic processes and with the
decreases significantly as explained in Chapter 5, conditions for success spelt out in detail.
though wider spacings tend to provide a better com- It may be objected at this point that existing beams
promise as the number of elements is increased. do not have equal currents, that the obvious way to
obtain equal currents is to drive both elements, but that
such designs have enjoyed only brief periods of popu-
Mutual coupling and its significance larity. Reasons for this will be clear from the analysis in
At this point let us return to the question of the mutual Chapter 5 which also indicates that except in the case of
coupling between elements which differs from that some common types of driven arrays there is no need
between tuned circuits only to the extent that one has for the currents to be unequal. The currents are
to consider mutual resistance as well as mutual react- unequal because the coupling is insufficient but this
ance. This means that if two closely-spaced elements only means that we need to increase it! In practice, in
are connected to a transmitter, but the opposite way the case of a two-element Yagi with straight elements
round, each subtracts resistance from the other, thus one need only bend them so that the ends are closer
making possible the large currents which are needed to together. Alternatively one could merely reduce the
offset all the subtraction which is going on! It is spacing but this does have some practical disadvantages
important to note that but for the mutual resistance (p79). A further alternative is to use loop elements
there could be no gain. The snag is that in the case of which couple more tightly than dipole elements, and
large gains or very small sizes of antenna the signal is this is believed to be the real reason for the reputation
the difference between two or more much larger quan- of the quad. On the other hand, constructions such as
tities and the practical consequences add up to severe the modified VK2ABQ antennas described in Chapter
losses, infinitesimal bandwidth and almost non-existent 12 allow a free hand for adjustment of coupling besides
tolerances. providing smaller, lighter systems of equivalent per-
The role of mutual reactance on the other hand is formance. Adjustable coupling is an inherent feature of
quite different and of particular importance in the case each of the smaller beams described in Chapter 12, with
of two-element beams. In this case one requires equal the exception of the small quad to which it can be
currents suitably phased so that cancellation can take added if necessary.
place in unwanted directions but when (as in the most It is common practice to disregard the reactive
usual case) one element is connected to the transmitter component of mutual coupling, the resistance compo-
and the other energized from it via the mutual cou- nent being more important insofar as it determines the
pling, current equality is impossible without mutual gain. However, Chapter 5 shows that the reactive
reactance. component is also very important since without it in
Attempts to resolve the problems of mutual coupling one form or another there can be no equalization of
by calculation have tended to end up as algebraic currents; gain is reduced and directivity seriously
equations of unmanageable proportions and most ex- affected. On the other hand, the provision of additional
perimenters, observing that almost any assortment of coupling reactance means that little or no detuning is
4 HF ANTENNAS FOR ALL LOCATIONS

required in the case of parasitic elements and the other mode. Thus in the case of the driven arrays
construction of efficient reversible beams is made much mentioned above, the absence of power in the line
easier; radiation resistance is also increased with advan- going to the reflector should leave this free to operate
tages in respect of bandwidth. as a parasitic element with the mutual reactance provid-
In these and a number of other respects Chapter 5 ing the necessary phase shift. However it is equally
breaks new ground. Extension of the analysis to three likely that poor results will be obtained, in which case
or four elements is much more difficult but references the user (unless equipped with better-than-average test
appended to Chapter 5 between them cover the ground facilities) may well be unaware of what he is losing.
fairly fully, and attention is also drawn to a particularly This pinpoints the desirability of avoiding the use of
important paper by Bloch, Medhurst and Pool de- unpredictable systems and methods on the grounds that
scribing in detail the four-element “superdirective”’ they are usually non-optimum.
array mentioned earlier in this chapter. The vector The “why” of correct design is the main concern of
diagram for this, as plotted by the author from the Part I and the “how” that of Part II, though the author
theoretical design data given in the reference and has tried throughout to maintain close links between
compared with that for a typical two-element amateur theory and practice: in the case of feeding power to the
beam (p88), demonstrates convincingly that amateur antenna (Chapter 4) separation was felt to be neither
beams are members of the ‘‘supergain’”’ family. This feasible nor desirable.
fact, though pointed out on several occasions by the
author and not challenged, has failed to attract general
notice and might well be dismissed as ‘‘academic’”’ but Feeding the antenna
for the importance of the n? rule and, for example, the Chapter 4 aims at presenting the simplest possible
prevalence of mistakes such as attempting to apply explanation of transmission lines and draws attention to
additive-gain principles to amateur beams. Further to the value, for the solution of tuning and matching
this, inspection of the diagrams may well suggest to the problems, of regarding the antenna itself as a “‘trans-
reader an intuitive approach to the task of obtaining mission line”. Particular importance has been attached
maximum gain from more than two elements and the to ensuring that the feeder acts only as a feeder and not
improbability of comparable gains being achieved in competition with the radiating system; thus an
along conventional lines. inherent loss of 3dB has been deduced for the Windom
antenna, considerable doubt cast on most ground-plane
systems, and stress placed on the need for symmetry
ideas to be avoided including where appropriate the use of baluns. The
The need for the ‘“‘why”’ extends not oniy to the best discussion of baluns concentrates on the essential sim-
ways of doing things but also those which are best plicity and efficiency of these devices, particularly the
avoided. Hence Chapter 5 includes a detailed discus- ferrite-rod type with bifilar or trifilar windings. It how-
sion of X-beams and Chapter 4 draws attention to the ever would appear that here, as in so many other
highly non-linear variation of phase along mismatched aspects of antenna design, failure to appreciate the
lines. This effectively rules out the use of non-resonant main essentials has led to the production of a number of
phasing lines for close-spaced driven arrays, although inefficient devices, thus tending to bring baluns as a
many such arrays have been described in the literature whole into some degree of disrepute.
and can be expected to work after a fashion. In view of the important role which the author
These examples highlight one of the main difficulties foresees for vertical beams, particular attention has
in seeking to simplify the hf antenna scene. Virtually all been given to problems of feeding, with the object of
existing designs are open to criticism in one or more avoiding ground losses or losses by radiation from the
respects, not excluding some of the author’s own earlier feeder or radial systems. These are suspected reasons
efforts such as phased arrays based on resonant lines. for the relatively poor performance of vertical antennas
These are predictable and afford a foolproof method of in general.
achieving optimum performance at a given frequency It is concluded that high efficiency can be achieved
but deteriorate when the frequency is changed, tending with very short radials or counterpoises or by the use of
to become comparable with conventional parasitic dipole elements, and that it is best to avoid the use of
arrays at the band edges unless re-adjusted. Any depar- 4 radials. Lower-end feeding of vertical dipoles by
ture from optimum usually results in simplifying means of the Zepp feed is recommended, subject to use
assumptions having to be abandoned so that theoretical of a balancing stub as proposed by G6CJ. The Zepp
analysis becomes more difficult and in some cases feed in its traditional form “does not work”’ and the
virtually impossible. reasons for this have been analysed in some detail, the
Often one can predict, for example, the impossibility only puzzle being the persistence with which this device
of an antenna working in some intended manner but has been featured in reference books, professional as
this does not necessarily preclude operation in some well as amateur, although many years have elapsed
 TAKING A NEW LOOK AT HF ANTENNAS 5

since G6CJ drew attention to its defects and his case in conjunction with antenna height and polarization
has been fully supported by the experiences of the determines the angles of radiation which can be suc-
author and a number of correspondents. cessfully achieved. Whereas high angles present few
problems and are optimum for the shorter ranges much
confusion exists in regard to very long ranges. It is
Multiband antennas almost universally supposed that these involve multiple
Methods for achieving multiband operation are the reflections between the ionosphere and the earth’s
subject of Chapter 7. In this area also, inefficient surface, and that there are optimum angles of radiation
methods which sacrifice bandwidth and potential gain, which are greater at lower frequencies.
particularly at the higher frequencies, have hitherto These beliefs have persisted and continued to form
been the order of the day. These seem unlikely to the basis of ionospheric propagation predictions despite
survive the impact of the new frequency allocations; convincing evidence from a number of sources to the
methods using centrally-located tuning units as pro- effect that (a) “multihop” modes are abnormal, at least
posed by a number of writers including the present under those conditions of relatively low path loss which
author overcome the disadvantages. The interleaving permit of amateur or other low-power communication,
of two- or three-band beams of these types aided by the and (b) for optimum dx performance it is best to aim
use of relays for switching appears to offer the best always for the lowest possible angles of radiation at all
prospect for ‘“‘no compromise” performance covering frequencies. This evidence, largely from professional
all bands, the rival attractions of log-periodic anten- sources, is reviewed in Chapter 2.
nas being offset by increased weight and windage. This It is ironic that whereas chordal hop (tilt-mode)
type of antenna is discussed in Chapter 7 and explained theories of propagation, which were first put forward
in greater detail with practical examples in Chapter 14. by Albrecht in the early ’fifties, owed their origin to the
experience of amateurs, the amateur literature has
persistently followed the ‘official line’? whereby com-
Other aspects of performance mercial services have for more than a quarter of a
Chapter 8 discusses the important topic of bandwidth, century denied themselves the advantages which accrue
drawing attention to the several different meanings from exploitation of prevailing layer tilts. According to
which this can have. Chapter 9 discusses the difference an investigation by the German broadcasting authority,
between transmitting and receiving requirements and official predictions have now been found liable to
suggests the separate optimization of antennas in these errors of up to 25dB, this being accounted for by
respects as an important line for future development. ionospheric focussing and the elimination of losses at
ground reflections. The amateur has meanwhile been
exploiting these modes very successfully, being denied
Effects of the environment only the true explanation of his consistent S9+ signal
Differences in performance between amateur stations reports over distances in excess of 12,000 miles
can be attributed very largely to differences in the (19,000km).
environment and Chapter 10 attempts to put together
as many as possible of the known facts. In addition it
draws attention to the unknowns and indulges in con- Practical antenna design
siderable speculation. Particular attention is given to The first chapter of Part II presents a comprehensive
differences between horizontal and vertical polariza- summary of types of single-element antenna, their
tion, including the variation of signal strength with respective merits or demerits and a detailed discussion
height and angle of radiation for various types of of practical aspects. Chapter 12 recommends appropri-
ground as calculated from handbook data. In conjunc- ate types of horizontal beam antenna (including minia-
tion with experimental evidence this suggests the de- ture beams) for a wide variety of circumstances and
sirability in many cases of experiments with both suggests methods of construction aimed at reduction of
polarizations before making final decisions as to a weight and windage by bracing with polythene cord.
permanent antenna system, an important factor being The same cords can also be used to support thin-wire
the ability of vertical antennas in many cases to work elements suitable for the higher frequencies, thereby
equally well or nearly as well at very low heights. hopefully solving the problems of multibanding posed
Indoor and other unfavourable environments together by the new frequency allocations. Although the author
with the effects of obstructions are also discussed. and others have had some success with this technique it
has yet to be fully exploited and could thus be an
interesting new field for the experimenter.
Chordal-hop modes of propagation Chapter 13 consists of a detailed discussion of vertic-
Consideration of the environment is linked to the al beams with numerous examples. It has been neces-
requirements of the ionosphere, since the environment sary to evolve completely new designs based on bent or
6 HF ANTENNAS FOR ALL LOCATIONS

end-loaded dipole elements, or monopoles working is believed to be fairly reliable but the reader looking
against very short inductively-loaded radials or for a precise blueprint of some well-established design,
counterpoises; these can alternatively be regarded as an guaranteed to work at the first attempt, must necessari-
asymmetrical form of dipole which, though fed “‘off- ly be disappointed. There is some doubt whether such
centre’’, appears none the worse for this. In regard to guarantees are ever fully justified, a conclusion borne
optimum groupings of vertical elements, considerable out by reports of manufactured antennas tested at the
reliance has been placed on a computer study reported factory which have been found to be quite badly out of
in The ARRL Antenna Anthology. However, like most adjustment after erection, due probably to environ-
of the literature, this assumes the use of direct low- mental differences.
resistance earth connections which the present author On the other hand, though not yet tested on a
considers to be in general neither feasible nor desirable sufficiently large scale, the exploitation of critical reac-
for amateur systems. tive coupling between a pair of close-spaced elements
Chapter 14 contains brief descriptions of various (as described in Chapter 5) implies a possible break-
large arrays and also includes an explanation of the through in this respect. This is because it allows both
principles of log-periodic arrays, together with some elements to be adjusted to mid-band resonance before
performance details. For the average reader this is erection with reasonable certainty that the beam will
probably border country with which he is not directly function correctly over at any rate most of the relevant
concerned, but it is necessary for completion of the frequency bands. The use of two feeders provides not
overall picture and hopefully of some general interest. only the bonus of instantaneous beam reversal but an
It includes some space-saving ideas culled from various insurance against any likely degree of maladjustment,
sources. and such beams have the further advantage of relatively
Chapter 15 comes down to earth again with the topic small size. Ease of adjustment is a further valuable
of “invisible antennas’. This includes ideas from a feature of the vertical arrays described in Chapter 13
number of sources as well as speculations by the author since, being at ground level, the elements are readily
which, even if not directly applicable, may encourage accessible.
those with a “difficult problem” to treat it as a chal- In seeking to clarify the various aspects of hf antenna
lenge rather than an obstacle. design the author feels that more questions have
Chapter 16 analyses the technical problems of mobile perhaps been raised than answered. Be this as it may,
operation and provides suitable guidelines as well as there remains still a wide-open field for the experimen-
some rather more speculative suggestions for the ex- ter who, it is hoped, will be able to set out from a firmer
perimenter. It also puts at the disposal of the reader the base. Anyone who feels that the subject is beyond their
author’s fairly considerable experience of portable op- competence should take due note of the bent-element
eration using carefully selected locations to achieve a example mentioned earlier in which the ‘“‘obvious’’, or
low angle of radiation for dx communication with very to be more precise the “‘very obvious’’, so successfully
low power and simple antennas. eluded the experts (all of us). He should then have no
Chapter 17 endeavours to provide practical guide- further qualms about his own abilities.
lines for the choice of an optimum antenna or system of The importance of subjecting all findings to the test of
antennas to meet any given set of user requirements. It “‘does it make sense?’’ must however be stressed.
does this largely in terms of a random, but hopefully Failures in this respect can be discerned as one of the
representative, set of scenarios and includes informa- main causes of the prevailing confusion and it has been
tion in regard to UK planning requirements, a topic the author’s aim to convey to the amateur reader the
featured also in Chapters 15 and 19. kind of understanding needed for the application of this
Chapter 18 is concerned with providing simple and criterion, even at the risk of some sacrifice of scientific
inexpensive ways of ensuring that the antenna is work- precision which in amateur work is usually not needed.
ing correctly, and Chapter 19 tackles the problems of Assumptions are not always stated if judged to be
antenna construction and erection with a bias in the sufficiently obvious or implied, and students should
direction of making the most of limited resources. perhaps beware of the absence of ‘‘fine print” which
has been included only where necessary in the amateur
context.
Opportunities for the experimenter
It is regretted that in a few places it has been found
Part of the author’s objective throughout the book has necessary to dig into some of the deeper theoretical
been, by the provision of suggestions and guidelines, to recesses and even engage in some elementary algebra.
encourage the experimenter to tackle problems which If difficulty is experienced the reader may be content
have not yet been fully resolved or may even involve for the moment to note the conclusions since, when
some considerable degree of projection into the future. rightly nourished, understanding grows with experi-
The design data for three-element arrays given on p79 ence. |
 CHAPTER 2

Waves and fields

The reader is assumed to have already some acquaint- the current in an antenna, travel at the speed of light,
ance with electricity including a knowledge of Ohm’s and are comprised (more often than not) of at least two
law, the difference between direct and alternating components which need to be added or subtracted with
currents, the meaning of inductance and capacitance, due regard for phase. Nevertheless, somewhat deeper
_ and the idea of magnetic and electric fields. He will insight into the process of radio-wave propagation
already be aware that when an alternating current flows helps in the avoidance of pitfalls and may enable
in a wire it generates a similarly varying magnetic field opportunities to be more fully grasped.
which induces voltages in other wires within its sphere Just as the alternating current flowing in a loop of
of influence and suitably oriented. From this it is only a wire, Fig 2.1(a), can induce voltages and therefore
short step to grasping the idea of communication at a currents in other loops, so an electric field between,
distance, except that if one is aware of the very limited say, two metal rods, Fig 2.1(b), can induce voltages
range of electromagnetic induction effects it becomes between other similar objects in the vicinity and, as
necessary to suppose the existence of some other before, a current flows if the circuit is completed. From
mechanism whereby the field surrounding a transmit- this point of view the end portions of an antenna wire
ting antenna is able to influence receiving antennas at can be thought of as the plates of a capacitor, the circuit
great distances. being completed via the inductance of the wire between
Most amateur antenna problems can in fact be them. This may alternatively be regarded as a magnetic
tackled with no more knowledge of the structure of loop, the circuit of which is completed by capacitance
radio waves than the fact that they can be generated by between the ends.
An antenna can therefore be energized by either type
Lines of magnetic force of field, and conversely when it is energized both forms
of energy can be identified by walking round it with
Low current or voltage probes as is often done in the course
resistance
R
of adjustments. The fields detected in this way, how-
Generator ever, are known as induction fields and decrease rapidly
(7) Meter with distance, whereas an electromagnetic wave can be
visualized as a special relationship between an electric
(a) and a magnetic field as a result of which energy travels
in the form of waves to great distances at the speed of
light, as depicted in Fig 2.2.
The wave consists of an electric field and a magnetic
field at right-angles to each other and to the direction of
Paes High
propagation. The two fields are in phase, that is to say
resistance the maxima occur at the same point of time and space
R
Generator

Meter
—-—g-—--—
---—qg-
(b) psn
en a
Of ta
-—<—g-— —
10

Fig 2.1. Induced currents. At (a) part of the magnetic field

surrounding the conductor a—b also encircles c-d so that a
current is induced as shown, and power is transferred from the
generator to the resistance R. At (b), the voltage between a and b
sets up an electric field which causes a voltage to appear UE (a) (b)
between c and d via the capacitive coupling between the rods,
and power is transferred as before. If the generator is powerful Fig 2.2. Instantaneous representation of an electromagnetic wave
and the measuring device sufficiently sensitive the circuits may (a) along the path of travel and (b) as seen by an observer looking
be moved a long way apart towards its source
8 HF ANTENNAS FOR ALL LOCATIONS

in both cases, the radiated energy being divided equally Seasiakeresl

between the electric and magnetic components. Max-
ima and minima are separated by a quarter of a ae

wavelength in space and a quarter of a cycle in time. Wave

travelling
The wave in effect provides the coupling between at the
speed of
transmitting and receiving antennas, the processes of light
transmitting a wave and extracting signal energy from it Observer notes the number of crests
passing (forexample) this point
~

being closely related. Usually in amateur radio com- in one second

munication the same antenna serves for both purposes
Fig 2.3. Magnetic or electric component of electromagnetic wave
although, as explained in Chapter 9, certain departures as seen passing the observer. The frequency, defined as the
from this rule may provide important advantages. number of cycles per second (hertz) passing the observation
point, is equal to c/A where c is the velocity of light

Wavelength and frequency

The plane of polarization of an antenna may be
In Fig 2.3 the wavelength, that is to say the distance
inclined at any angle. Inclined polarization is often
between the crests of the waves, is denoted by the
wrongly identified with elliptical polarization, which is
symbol A in accordance with usual practice. Since the
completely different since it involves rotation of the
velocity is 300 x 10° metres per second the number of
plane of polarization corkscrew fashion as the wave
crests passing the observer in one secondis 300 x 10°/X
progresses and is produced by radiating two waves,
if A is measured in metres. This is the frequency of the
polarized at right-angles to each other but with a phase
oscillations used to generate the wave and also of the
difference of 90°. This means that a maximum of one
current induced in a receiving antenna; for application
wave coincides with a zero of the other, polarization
to the hf band it is convenient to express fin megacycles
being momentarily horizontal at the instants when the
per second* in which case
vertical component is zero and vice versa.
f = 300/ If both waves are of equal amplitude the polarization
is described as circular and rotates at a constant angular
Since the relationships between the electric and magne-
speed but in each case there is one complete revolution
tic fields is fixed, specifying either component auto-
per cycle of rf oscillation. Rotation can be left or right
matically determines the other. However, the usual
handed and must be the same for reception as transmis-
procedure is to specify field strength at the receiver in
sion. Reception with an ordinary (ie plane polarized)
terms of the voltage ¢ which would be induced in one
antenna halves the signal power since, in effect, only
metre of wire lying parallel with the direction of the
one of the two waves is received. At hf, however, one is
electric field, ¢ being dependent on the radiated power
concerned mainly with signals reflected by the iono-
and the attenuating properties of the path between
sphere and these contain on average equal horizontal
transmitting and receiving antennas.
and vertical components regardless of the polarization
of the transmitted signal. The phase relation between
Polarization these is random, which means there is no advantage in
trying to combine them in any simple manner, though
An important property of the wave is its polarization
use is sometimes made of polarization diversity techni-
which means in practical terms that the transmitting
ques which exploit lack of correlation between the two
and receiving antennas should be “the same way up’’,
components in order to reduce. fading in commercial
though this statement will later be qualified. One can
systems. |
perhaps liken the waves surrounding an antenna to the
ripples on the surface of a pond, inducing currents in
other antennas which they encircle momentarily as they Radiation from a conductor
travel outwards, much as the ripples might cause a cork
to bob up and down. The wave is described as vertically The next step is to build up a picture of the radiation
polarized if the transmitting antenna is vertical, in surrounding an antenna and the factors by which it is
which case the electric field is also vertical; similarly a influenced. As a convenient starting point consider a
horizontal antenna radiates horizontally polarized short length of wire, or “‘doublet’’, carrying a radio-
waves. frequency current J. A receiver is located at some
distance d in the broadside direction and experiences
* By international ruling the descriptive term ‘“‘cycle per second” has from the doublet a voltage V which is proportional to J
been replaced by “‘hertz’’, thereby commemorating the pioneer work of and /. If now a second conductor is added as in Fig 2.4,
one of the founders of modern communication by making it more
difficult to explain. One can but hope that readers have passed the having the same length / and carrying an identical
stage where translation from “cycles/sec’’ to “‘hertz’’ requires a con- current, the voltage at the receiver is increased to 2V.
scious mental effort. This must obviously be true whether the conductors are
 WAVES AND FIELDS 9

cad § Receiving Length of an equivalent dipole carrying a uniform current

antenna
I !L hee Ver

Current = I at this point |

|
f$ md —___—__e Voltmeter \
on
=
=
Average as
an
current an
—------- an
on
abl
an
fale
on
an
Fig 2.4. Method used for estimation of total field from wires gfe
Ae Sei
having a non-uniform current distribution. A current /through the song
ann
length / produces a signal voltage V. A second doublet carrying
an identical current increases the signal to 2V but usually the
doublets are portions of a single wire and carry different currents, Square counts for each segment
their lengths being assigned so that they are just short enough to
be regarded as having the same current throughout | 270 | 270 | 110
Square counts for central and end portions
Total square count = 760
Separate, as illustrated, or joined to form a single
A/>
conductor of length 21.
Typically the antenna current varies along the length
of the conductor which can be pictured as divided up ~~ hr=0-225A —_——»|
into lengths /,, 1,, 1, etc, just short enough to be
regarded as having the same current throughout,
though they can be shorter and it may be more
convenient to make them all the same. If the individual (b)
currents are J,, J,, J, etc, values J, 1,, I, 1, etc can be
assigned to each contribution; these can then be added
and the sum divided for example by J, to obtain the Square count = Square count =
110
length of antenna which would produce the same field
if fed with a uniform current of this magnitude. Alter-
natively one could divide by the total length of conduc-
tor to obtain the average current, but the first proce-
dure tends to be more useful since most calculations
need to be referred to the feedpoint of the antenna
where the current is most likely to be known. Width of equivalent
fr—rectangle—=|
The process just described is known as integration
and the mathematician has various shortcuts at his
disposal. For the rest of us it is simpler, though
admittedly more laborious, to draw out the current
distribution on a piece of graph paper, count the
squares under the curve, and divide by the length as SRSC SES Reeeo
Pi olet) ep aletoeyeiar
ee.
py
et fohete ta: oreo epPetree yy
illustrated in Fig 2.5. This diagram shows the sine-wave
current distribution typical of 4/2 dipoles which form a
basic building block for the majority of hf antenna —Loading stub
systems. At this stage there is no need to be unduly
concerned with how this current distribution comes ’

about or the special significance of the half-wavelength

which are topics for the next two chapters. However, A/a"
since the circuit is completed via the capacitance be- Fig 2.5. Estimation of radiation field from an antenna having a
tween the two halves of the dipole and this capacitance known current distribution by dividing it into elementary doub-
is distributed along the wires, it follows that the current lets. The example shows a }/2 dipole and also the effects of
reducing its length by bending over the ends (b) or folding in the
gradually decreases to zero as one moves outward from centre as at (c). The figures indicate the relative areas of each
the centre. segment based on a count of 0-1in squares with the current
It will be seen that for a sine wave the average distribution drawn on graph paper to a scale of 6in by 2in
current is 64 per cent of the maximum or alternatively
we can regard the maximum current as flowing through shorter antenna is required it is better to sacrifice the
64 per cent of the length. It is interesting also to note ends rather than the middle. For the moment, how-
that 71 per cent of the field is produced by the middle ever, the main essential is the direct relationship be-
half and only 29 per cent by the end portions so that if a tween field strength at the receiver and the product of
10 HF ANTENNAS FOR ALL LOCATIONS

Y a
“atid
Ry ia
as
straight conductor, field strength at a given point P is
proportional to average current times the apparent
length as seen from that direction’’, ie the length /’ in
Fig 2.6(a). The ratio /'// therefore describes the way in
Observer walks round at which signal strength varies with the direction relative
es a constant distance, d to the line of the antenna; this is the polar diagram of
s the antenna and is shown in Fig 2.6(b). |
\ To obtain the radiation pattern for a more compli-
C Ay \ (a) cated shape of antenna it is often useful to resolve the
radiation from a wire into two components at right-
! pabershb chsddd eg A angles to each other. For this purpose each “‘doublet”’
Mes
Doublet
| can be replaced by an equivalent pair at right-angles,
thereby forming a “‘staircase”’ as shown dotted in Fig
2.7 which shows two observers looking at the same
antenna from directions at right-angles to one another
and seeing lengths /,, /, respectively. Provided the wire
still carries the same current the zigzag shape makes
not the slightest difference to either observer, though
the portions of antenna seen by one observer are now
completely different from those seen by the other. The
zigzag wire may now be further divided into straight
portions, all these running in the same direction and
being joined end to end, so that each observer now has
his own personal antenna.
Throughout the proceedings, provided he stays in the
same place, neither observer is aware of any change or
the fact that he has not been looking at a dipole of
length J, or /, oriented for his direction. Similarly an
observer at A, the direction of maximum radiation, is
also unaware of change; it might be argued that an
Fig 2.6. (a) Derivation of directional pattern. As observer walks observer at C, Fig 2.6(a), will now see something, but
round from A to B at constant distance the apparent length of the this is merely a matter of reducing the size of the steps!
antenna shrinks from / to /'. At C, it disappears completely; d is
assumed large compared with /. (b) Shows the directivity in polar It follows that the two separate dipoles or doublets at
coordinates of a A/2 (or shorter) dipole: e, and e,, represent the right-angles are strictly equivalent so that the radiation
relative field strengths in the directions indicated from, for example, an inclined wire may be resolved
into a horizontal and vertical component. Often two or
more such wires may be combined so that one or other
length and current. For example, if it is found that of these components is cancelled, leaving the user with,
reducing the length has no effect on field strength this
can only mean that the current has increased. This
conclusion is subject to certain conditions since it ~@J Observer (1)
requires elaboration for the case of large antennas or if
there are phase reversals.

Directional pattern
Next we have to consider the directional pattern of the
antenna, ie the effect of walking around it, or turning it
round, as in Fig 2.6(a). In the end-on position it
“shrinks to a point”’ and becomes invisible. This applies
equally to the radio signal, the electric field being no
longer correctly aligned with the direction of propaga-
A
tion, and it is not surprising to find that for intermedi- Direction of maximum
radiation or reception
ate positions also the radio signal varies in exactly the Observer (2)
same way as the apparent length of the antenna. The
previous conclusion can now be expanded to read ‘‘for Fig 2.7. Resolution of a single dipole or doublet into an equivalent
pair mutually at right-angles. The length / may be replaced by
a transmitting antenna consisting of a single short two sets of elementary doublets corresponding to the steps
 WAVES AND FIELDS 11

resolved each portion of the antenna into two compo-

nents, it is a simple matter to add those seen by the
observer while ignoring the others; the addition how-
ever must be carried out with due regard for any phase
differences, as will now be explained.

Phase
In most practical cases there is not just one source of
signal but at least two, since even if the antenna itself
can be regarded as a single “‘point source”’ of radiation
there is a wave reflected from the ground as well as the
direct wave from the antenna. It is convenient to regard
the reflected wave as coming from the mirror image of
the antenna in the ground though in many cases this is
an oversimplification and the topic is discussed at some
length in later chapters. For the moment all we need to
Current lags 63° ° V,
relative to the Re YR know is that there are two or more sources of signal
voltage which are going to have to be added, or maybe
subtracted, in order to discover the effective signal in
(e) V_L=2VR
any given direction. This brings in the important ques-
Fig 2.8. Explanation of phase. (a) Typical ac circuit. (b) Voltage tion of phase and an understanding of this is essential to
across L drops to zero as current reaches its maximum value V/R further progress.
(ie current lags behind voltage). (c) Current is initially V/R but falls
to zero as soon as C is fully charged (ie current leads). (d) There are two kinds of phase relationships with
Resonant circuit: current phase lags if wL>1/wC, leads if 1/wC which we need to be familiar, that between a voltage
>wL, otherwise | =V/R. (e) Phase relations for circuit shown at and the current producing it, Fig 2.8(a), and the phase
(a), assuming wl = 2R
relations between waves arriving from a number of
different sources such as A and B in Fig 2.9.
for example, some constructional advantage as in the To understand the difference in phase between cur-
case of the quad antenna which can be fitted in to a rent and voltage in Fig 2.8(a) consider the situation at
much narrower space than a A/2 dipole. (b) where the current flowing in the coil of wire (or
The trick of replacing any short piece of antenna wire inductance) generates a magnetic field which repre-
by two equivalent wires at right-angles can be carried sents stored energy. After closing the switch it takes a
out for any direction of observation but is then valid for little time for this store to be filled so that there is some
all observers, provided those who are able to see both delay before the current has risen to its steady value
parts of a step add or subtract the contributions accord- V/R in accordance with Ohm’s law. Similarly the
ing to the apparent lengths and whether the currents capacitance in Fig 2.8(c) takes time to charge up to the
appear to be flowing in the same or opposite directions. full voltage of the circuit, the current at the moment the
The argument thus far has assumed, implicitly, that switch is closed being equal to V/R. This falls to zero by
all the wires are lying in the same plane but the the time the capacitor has been charged to the voltage
principle can be applied to three dimensions by resolu- V so that there is no longer any “‘potential difference”
tion into three orthogonal (ie mutually at right-angles) across the resistor. In case (b) the current Jags behind
components. The dimensions of each wire must be the voltage across L and in case (c) it leads the voltage
small enough for the radiation from all parts of it to be across C.
regarded as arriving simultaneously at the observer, In the case of ac or rf circuits the switching takes the
longer wires being subdivided, but they do not need to form of reversals of polarity at the appropriate frequen-
be very short. For example, two concentric A/2 dipoles cy with the current reversals leading, lagging, in phase,
at right-angles carrying equal currents have the same or in antiphase with the voltage reversals, except that
radiation pattern as a single dipole lying in a direction usually one is not concerned with a switching process
half-way between them. but the smooth transitions of a sine wave. Inductances
This result is of direct practical significance since it and capacitances in ac or rf circuits impede the passage
allows the directional pattern to be switched in 45° of current by acting as stores for energy which is
steps, thus obtaining better all-round coverage. The subsequently released, but they do not consume power.
main point of the method, however, is to allow the The impedance is equal in magnitude to wL and 1/(wC)
directional pattern of any antenna to be estimated, respectively where w is the “angular frequency” in
however complicated its shape, provided the current radians per second, being equal to 2x times the actual
distribution is known or can be measured. Having frequency.
12 HF ANTENNAS FOR ALL LOCATIONS

combine at a remote point such as P. Thus if both

To P signals start their travels in the same phase that from B
A will have further to travel by an amount BC. Fig 2.9(b)
(a)
has been drawn for the case where BC = 1/3 so that the
90°
phase difference is 3 cycle or 120°. The field due to B is
assumed to be only 70 per cent of that from A, and the
resultant field is 93 per cent of that from Q alone. On
Cc Q the other hand, by using L and C as shown in Fig 2.9(c)
B
it would be possible to give B a phase lead of 120° so
Field at P Field at P
from A+B from B that the fields at P add directly in phase to give a
(b) resultant OT which is 1-7 times that from A alone.
5
Field at P
fromA 7
Although this example has been chosen solely to
P demonstrate a principle, the design or evaluation of
antenna systems consists very largely of operations of
this kind.

Antenna A Antenna B (c)

Free space
It often simplifies antenna problems to divide the
Fig 2.9. Combination of voltages subject to a phase difference: antenna into two or more parts, and as a rule the first
the device shown at (c) uses the principle of Fig 2.8(a) to apply a step is to separate it from its environment. Thus we
phase lag at A such that the fields at P from the two antennas A
and B add in phase. Any number of voltages may be added in this may first decide how an antenna would behave if it
way were in free space, that is to say far enough out in space
for the presence of the earth to be ignored, and then
These quantities are normally referred to as the consider as a separate problem how this behaviour will
reactance (symbol X), the more general term impe- be modified by the presence of the earth or other
dance (symbol Z) being applied to combinations of surroundings. It then transpires as explained in Chapter
resistance and reactance so that Ohms’ law for ac 6 that for a given height the effect of the earth is the same
circuits becomes V/J = Z = R + jX where j is used to for all antennas provided that they are ‘“‘the same way
indicate “‘addition at right-angles” as explained below. up”’.
The negative term denotes capacitive reactance and, Failure to simplify the problem in this way has led to
if Z includes both inductance and capacitance as in many claims for this or that type of antenna based on a
Fig 2.8(e), we have Z = R + jwL — j/wC so that the “lowering of the angle of radiation” relative to its
normal Ohm’s law expression V/J = R applies if competitors. For the time being therefore, unless
wl = 1/wC. otherwise indicated, all antennas under discussion are «
This is the condition known as resonance and is an assumed to be in free space, the effect of the ground
important condition for the transfer of power from a being explained in Chapter 6 after which the reader
transmitter to an antenna. The voltage across L as we should be well-placed for relating the basic characteris-
have seen leads the current through it whereas the tics of an antenna system to any given practical situa-
resistance provides a simple example of Ohm’s law, the tion.
voltage and current being in phase. In adding these two It will be recalled that, in general, signal strength is a
voltages it is necessary to allow for the fact that they are maximum when both antennas are the same way up, ie
displaced in time by a quarter of a cycle and this may be have the same polarization, and this is always true for
done graphically as shown in Fig 2.8(e). This also shows short antennas in free space. To appreciate this, im-
the current which, because the circuit is inductive, lags agine trying to inject a signal from one wire into
behind the voltage. It is important to realize that, another one arranged symmetrically at right-angles.
although Z determines the current, it is R alone that There is clearly nothing to tell the induced current
consumes power which is equal to /?R just as in the case which direction to flow so therefore there can be no
of a de circuit. induction, and similar reasoning can be applied to
In antenna work one is concerned largely with re- electrostatic couplings. If the polarization is not identic-
sonant systems but phase differences arise due to al at both ends, signal strength decreases with the angle
detuning and to the fact that the time of travel of a of non-alignment in exactly the same way as it did with
wave between different parts of an antenna system may direction in the case of Fig 2.6(a), the analogy being
represent a large fraction of a cycle, or even several obvious if one thinks, say, of a vertical receiving anten-
cycles. This situation is illustrated in Fig 2.9 which na as “‘seeing”’ only the vertical extent of the transmit-
demonstrates the way in which signals from A and B ting antenna. This is also to be inferred from Fig 2.7.
 WAVES AND FIELDS 13

Field strength versus distance given in watts by (h,€)?/4R or, for a A/2 dipole, by 3-5
Referring again to Fig 2.6(a), the reader may be ready €2,2 x 10-4.
to accept that the apparent length /' can be used as a The alternative calculation is known as the Poynting
general indicator of relative field strength. If the dis- vector method and is based on the concept of the
tance is doubled /’ subtends only half the angle at one’s transmitter being located at the centre of a very large
eye and thus looks only half as long, so one can infer sphere representing the whole of space, the power flux
that in free space signal voltage is inversely proportion- at any point being given by €2/377 watts/sq metre. From
al to distance. Alternatively one could perhaps imagine Chapter 9 (p126) the aperture of a A/2 dipole is 0-132
oneself as a receiving antenna which is grabbing all the and multiplication of these two figures leads to the
energy within its reach; at twice the range the energy is same answer as before. The figure of 377Q is often
spread over four times the area so the fraction of it referred to as the “‘impedance of free space’.
within reach is only a quarter of what it was before. It will be noticed that the longer the wavelength the
This is known as the inverse square law, and “‘the area greater the received power since the receiving dipole
within reach” is a loose description of the aperture occupies more space and therefore captures more of
_which is explained in more detail on p126. the radiated energy. The transmitting antenna, assum-
These concepts provide two methods of obtaining the ing this also to be a A/2 dipole, is similarly increased in
power available at the terminals of a receiving antenna. length but the directional pattern is unchanged and this
The first depends on the knowledge that the field means that the way in which the transmitted energy is
strength produced at a distance d by a uniform current J distributed throughout the surrounding space is also
amps flowing through a short wire of length / as in Fig unchanged; in other words for a given power radiated
2.2(a) is given for the direction at right-angles to the the power flux or field strength at any given point in
wire by space is independent of frequency. In contrast the
received signal power, by virtue of the greater capture
_ 607A Il
ipa volts/metre area, is proportional to the square of the wavelength.
To illustrate the calculation of signal strength, consid-
er the case of 100W radiated from a 2/2 dipole at
The ratio of /' to / is a property of the angle 8, namely 14MHz, and let us assume a similar antenna for recep-
its cosine, so for other directions the expression can be tion. As explained in the next chapter the radiated
written power is accounted for by a resistance of 73Q referred
to the centre of the dipole so that, since power = /*R,
= Il cos 8 volts/metre I = V 100/73 = 1-17A. To allow for the sine-wave dis-
tribution the actual dipole length must be multiplied by
In these expressions /, d and i are all expressed in 0-64 as explained above, so that the expression for €
metres. Since in practice J is usually not uniform the becomes:
product J] must be determined by the method of Fig
2.5. To find ¢ for any given direction, distance and
wavelength all that is then needed is the above formula
plus a pocket calculator having a button labelled ‘‘cos”’. = 60-3 J/d volts/metre
However, as a matter of practical convenience / is
This is the free space field and is only the first step in a
usually taken to be the current in the centre of the
calculation which must also take account of propaga-
dipole and to make the sums ‘‘come right” it is / instead
tion factors such as those discussed in the next part of
of J that is multiplied by the appropriate factor.
this chapter. Anticipating this discussion, it will make
The new value of / is usually known as the effective
the example more interesting to assume a long anti-
height and designated by the symbol fA for reasons
podal path of 20,000km, ie d = 20 X 10°, and it will be
which go back to the early days of radio before the
more convenient to express € in microvolts/metre so
advent of “‘short waves’, heights then being small
that
compared with the wavelength and closely linked with BOA oy lid eel O”
efficiency. In deriving h from / it is usually accurate
enough merely to multiply by 1, 0-64 or 0-5 depending
ree S30 x 106
on whether the current distribution most closely resem- = 3-53uV/m
bles the middle, the whole or the end portions of a sine
For the receiving antenna the value of h, is 0-64)/2 or
wave. approximately 6-8m and the open-circuit voltage at the
The voltage induced in a receiving antenna having an
terminals of a matched feeder connected to the centre
effective height 4, metres by the field € is by definition
of the antenna will be 6:8 X 3:53 = 24uV. The power
h,e volts so that, knowing the radiation resistance R as
available at the receiver is given by:
defined in the next section, the signal power available
at the receiver in the case of free-space propagation is 3°5 x. 244.x 21-22) 1074 x 107%
14 HF ANTENNAS FOR ALL LOCATIONS

= 9-06 x 10-!W to specify ‘‘optimum”’ radiation angles and antenna

It is not possible to allow accurately for propagation heights, despite convincing evidence [1, 2] that the
losses but these may well be offset by ionospheric lower the angle that can be radiated the better the
focussing as discussed on p16, so that the above figure average dx performance. Apart from this, evidence
stands as a possible value for the received signal power. against multihop propagation as an important element
In case the reader is wondering what purpose is in amateur radio communication over very long dis-
served by this type of calculation it might well be asked tances has been mounting steadily over the last 25 years
for example whether the 100W is really necessary. For or more [3, 4, 5, 6] and the reader may like the chance
example, supposing there is no interference how much to judge this for himself.
signal power would be needed to produce a usable At amore mundane level, most amateurs in planning
signal? The answer depends on the required signal/ their requirements are faced with the need for com-
noise ratio, which in turn depends on whether the promise and this may well become more acute as more
signal is cw or ssb and in the latter case, whether rf hf amateur bands become available. A decision as to
speech processing is used. There is a shortage of what band or bands are needed, for what purposes and
reliable figures but a signal equal to the noise level in what order of priority is likely to be the first step in
would probably be of interest, though somewhat mar- planning a new antenna or system of antennas. To this
ginal. The background noise in a quiet location will be end it is important to know what waves can or cannot
mainly of cosmic origin and from handbook data is do after they leave the antenna and the ways in which
some 23dB up on the noise level of a resistance at room this affects the selection of optimum frequencies of
temperature. This is 10-!’W of available power for a operation. It may then be necessary to decide whether
bandwidth of 2kHz so that for the signal to be equal to to use several antennas or make a single antenna do
the noise at 14MHz the received power must be at least several jobs, and what performance compromises, if
5 x 10-#*W. any, can be accepted in order to simplify the installa-
In this example we have about 2,000 times more power tion. This in turn may react back on the initial require-
than this so that, if the noise level is reckoned as S1 and ments and rather than try to achieve something which
the author’s unofficial estimate of 3-5dB to a typical “does everything all the time”’ it may be found best to
average S-point (p230) is used, a signal report of S9 can use an adaptive approach whereby changes are made
be expected. Alternatively, marginal communication from time to time to suit the prevailing phase of the
may be possible with a radiated power of 0-05W only. In solar cycle.
practice allowance must also be made for gains or Radiation from an hf antenna includes a ‘“‘surface
losses (a) in antenna systems (b) due to ground reflec- wave” which is attenuated rather rapidly so that max-
tions at both ends (c) focussing and attenuation in the imum ranges are typically of the order of no more than
ionosphere. 50 miles (80km) or so, depending on antenna height,
polarization and the intervening terrain. It is often
pointed out that vertical polarization is “‘necessary”’ for
efficient communication by this means though in prac-
lonospheric propagation tice, at least on the higher frequency bands, the author
In designing an hf antenna or system of antennas there has found just the reverse, probably because most
are sO many interesting and challenging problems to be other stations use horizontal antennas and polarization
faced that it is easy to lose sight of the main objective, needs to be the same at both ends. Moreover, ground-
namely to make the best possible use of the ionosphere. wave ranges of 50 miles (80km) or more are often
One can perhaps take refuge in the impossibility of achieved even with horizontal antennas at modest
defining “best”? since it depends on the particular heights whereas vertical antennas are too often ineffi-
interest of the individual user, but there are some cient.
aspects of ionospheric propagation which are relevant The majority of amateur contacts in the hf bands,
in the antenna context and need to be more widely however, involve reflections from the ionosphere, and
known. To this end it is necessary to descend once as a first approximation this can be regarded as a
again briefly from ‘‘free” space into that which is conducting “‘shell’’ surrounding the earth and acting as
bounded by two concentric spheres, the earth and the a reflecting mirror. Upper layers of the earth’s atmos-
ionosphere. phere become ionized due to the bombardment of
The characteristics of the ionosphere have been atoms by ultraviolet radiation from the sun which sets
studied in great detail over the 60-odd years since it was free electrons, thereby leaving the parent nuclei posi-
first discovered but mainly in the context of relatively tively charged. At night the bombardment ceases and
short ranges requiring only a single reflection from the eventually the electrons (being negatively charged) get
ionosphere, communication over longer ranges being attracted to other nuclei so that the amount of ioniza-
assumed to take place by multiple reflections between tion decreases, but in the very rarified upper levels of
the earth and the ionosphere. This has led to attempts the atmosphere this is a slow process.
 WAVES AND FIELDS 15

exists only in daytime, is relatively weak and not of

much practical importance.
Polarization is “lost” on reflection and this entails a
3dB loss in signal strength as mentioned earlier, but
there are also losses by absorption; these vary greatly
but are least just below the muf so the optimum
frequency band to use is generally the one nearest to
R the muf but on the low side of it. Below this there is a
Vi}7 gradual increase in absorption resulting in a lowest
(b) usable frequency (luf), but this depends to a large
extent on signal power and, unlike the muf, is not
Fig 2.10. (a) Conventional picture of the ionosphere showing
multiple reflection. (b) Actual path of wave corresponding to sharply defined.
the virtual height h Apart from regular diurnal and seasonal variations
dependent on the zenith angles of the sun, both the muf
and the absorption indices (and hence the luf) vary with
A wave entering the ionosphere is bent (refracted) the number and activity of sunspots. During the solar
by encounter with the electrons and if the bending is cycle of about 11 years the “‘sunspot count’’, suitably
sufficient it is returned to earth again. At the higher averaged, varies from a low of three or four to a peak
frequencies and vertical incidence the bending is very which over the last 150 years has varied between 60 and
slight and the wave goes straight through into outer 200. There are however considerable day-to-day varia-
space; however, when the wave meets the ionosphere tions with a tendency for repetition at 27-28 day
at an oblique angle it has further to go before escaping intervals, this being the period of rotation of the sun.
and, as a result, experiences a lot more bending so that To get this in perspective a count of at least 50 is
it is returned back to earth as illustrated in Fig 2.10. needed to provide consistent dx openings on 28MHz,
Due to optical analogies it is usual in this context to although occasional contacts are possible over north-
talk about rays rather than waves, and the bending just south paths even during sunspot minimum years.
described is equivalent, as shown enlarged at (b), to The above picture of the ionosphere is much simpli-
reflection from a slightly greater height than that fied and there are many anomalies, not all of them fully
actually reached by the ray. understood. Commercial and broadcasting services
The height thus defined is the ‘‘virtual height” of the have an urgent need for up-to-date information on the
layer and there are several layers, the one of most best frequencies to use, including advance warning of
importance for hf communication being known as the when to make changes of frequency, and to this end
F2 layer or, at night, merely as the F layer. This is at a ionospheric prediction services have been operating for
height of about 130 to 260 miles (210 to 420km) many years. Predictions have hitherto been based on
depending on the season and time of day, and the single-hop paths, making use of the latest sunspot
degree of ionization. Therefore the reflecting prop- numbers in conjunction with solar zenith angles and
erties are strongly influenced by the 11-year solar cycle, time of day etc. As mentioned earlier these have been
ie by the number of ‘‘spots” on the sun since these universally applied to longer distances on the assump-
largely determine the amount of ultraviolet radiation. tion of propagation by multiple reflections between the
The frequency at which a ray entering the ionosphere earth and the ionosphere as illustrated by Fig 2.10(a).
at vertical incidence just fails to escape and is reflected This method of prediction has led to serious difficul-
(or refracted) back to earth is known as the critical ties in the case of long-distance paths, eg between
frequency, f,. At oblique angles much higher frequen- Europe and Australia, and its persistence has been
cies can be reflected up to and including the maximum difficult to reconcile with the discovery of chordal-hop
usable frequency (muf) which is typically in the region propagation (Fig 2.11) by H. J. Albrecht in 1953 [3],
of two to four times f,. It is given by f,/cos i where i is the work of many other investigators, particularly Stein
the angle of the incident ray to the vertical as shown in [4], and the experience of radio amateurs over more
Fig 2.10. than half a century [5].
The F2 layer is the highest layer, the lowest being the As far back as 1948 it seemed obvious to the author,
D layer which in daytime has an absorbing effect on as it must have been to anyone with experience of the
rays passing through it. The D layer disappears at night long path to Australia over a sufficient period of time,
because at this level the atmosphere is much more that its behaviour was not consistent with the existence
dense and ionization does not persist after removal of of intermediate ground reflections. However, Albrecht
the cause. The E layer is at a more or less constant (who also based his conclusions on the observation of
height of 72 miles (116km) and, though useful for amateur signals between Europe and Australia) was
medium-range daytime communication, also _ dis- the first to offer an explanation. The experiences of
appears at night. Above this there is the Fl layer which radio amateurs also provided encouragement for the
16 HF ANTENNAS FOR ALL LOCATIONS

age:
(dB)
GF

s)O

Fig 2.11. Showing how the change of effective layer height

around dusk and dawn produces the layer tilts required for
chordal-hop propagation between near-antipodal points, thus O 10 20 30 40
providing a night-time path with little absorption and the avoi-
D/103 (km)
dance of ground-reflection losses
Fig 2.12. Antipodal focussing gain, showing variation of gain with
distance for one complete circuit of the earth. Maximum gain is
more recent work on chordal hop [6] in which it was theoretically infinite but for practical chordal-hop paths the curve
found that field strengths predicted by the International should be rounded off as shown (Telecommunication Journal)
Radio Consultative Committee (CCIR) method were
too low by no less than 25dB (equivalent to 300 times in the change in layer height between the light and dark
power), this being explained in terms of a focussing sides of the earth, there are regions of transition where
gain of 16dB plus a 9dB advantage due to the absence the dotted line representing the ionosphere ceases to be
of ground-reflection losses! parallel with the earth’s surface. These tilts result, as
An explanation of focussing gain follows from the shown, in a chordal-hop mode of propagation via the
fact that great circles passing through any point on a dark side of the earth. This is believed to be responsible
sphere also pass through the antipodal point. Waves for virtually all amateur communication over, for ex-
transmitted from one point in whatever direction meet ample, the long path between Europe and Australasia
up again in phase at the antipode: if the ionosphere was but any transequatorial path tends to experience similar
a perfectly conducting shell 100W of transmitted power tilts over a large part of the day.
could be used to light a 100W bulb 12,000 miles away at The various types of reflection at the ionosphere can
full brilliance! The reason this does not happen is that be better understood by referring to Fig 2.13 which
the ionosphere is neither perfectly conducting nor a
perfect sphere, moreover the explanation has been
oversimplified since it presupposes that the transmitter DAY NIGHT
and receiver are in contact with the spherical surface
lonization gradually reducing
which in this case is the underside of the ionosphere.
Nevertheless the gain of 16dB is quite impressive, more
than compensating for an estimated absorption loss of
15dB and in line with the author’s own earlier conclu-
sion [5], arising from low-power portable operation on
steep ground slopes, that focussing gain was at least
comparable with the absorption whatever this might
be.
It should be noted that focussing gain (a) applies only
to chordal hop since otherwise it is neutralized by
defocussing at the ground reflections in a manner very
similar to the situation depicted in Fig 10.13, and (b)
requires the angle of incidence at the ionosphere to be
as small as possible. The dependence of focussing
gain on range is shown in Fig 2.12 taken from reference
Fig 2.13. Chordal-hop conditions. High-angle ray passes through
[6], where it is stated that as the theoretical gain tends the ionosphere with little deviation, lower-angle ray is reflected in
to infinity at the antipodal points it must be limited to the usual way. At a still lower angle the wave finds itself in a
an appropriate value such as 30dB for practical applica- region of gradually reducing ionization so that the rate of bend-
ing becomes less and less until it emerges at a shallow angle;
tions. there is no return to earth and the next reflection also takes place
Referring to Fig 2.9 it will be seen that, because of at the ionosphere (note: not to scale)
 WAVES AND FIELDS 17

Table 2.1

Highest Lowest
amateur band amateur band
Period and time Height of F2 layer f, Average required for required for
(miles) (km) (MHz) muf (MHz) dx (MHz) dx (MHz)
Solar min day 260 420 4-5 18
(summer) night 190 305 2-7 9:3 18 7
Solar min day 140 225 6-0 18
(winter) night 190 305 2:3 8
Solar max day 240 390 8:8 34
(summer) night 190 305 7:0 24 28 14
Solar max day 140 225 13-6 40
(winter) night 190 305 4-9 17

illustrates a typical situation in greater detail. The much accepted dogma, does not derive solely from
high-angle wave, after being bent slightly by its encoun- theoretical requirements for maximum focussing gain
ter with electrons, passes straight through into outer but rests also on direct observations by Utlaut [1] over
space. Meanwhile a wave at a lower angle is bent the North Atlantic path and by Epstein et al [2] over
sufficiently to cause its return to earth as if reflected considerably longer paths, low angles being achieved
from the height indicated. Another wave at an even by the use of very high sites (eg 1,000ft or 300m)
lower angle encounters a “‘tilt’’; not a tilt in the literal looking out to sea. In reference [2] much greater signal
sense but its equivalent in the form of a varying levels (typically 10dB) were consistently obtained from
refractive index. Thus at the angle shown, after the sites at 1,000ft (300m) compared with sites at 150ft
initial bending and before leaving the ionosphere it (45m) or so, though even the latter would qualify as
finds itself in a region where the electrons are becoming exceptional by normal amateur standards, the advan-
fewer and the rate of bending is less so that eventually it tage being greatest at the lower frequencies. This
leaves the ionosphere as shown, only to travel round by implies radiation angles less than 1° and, although it
successive chordal hops until it meets another tilt. was sometimes found that high angles resulted in
As well as large signal levels another marked charac- signals which were as good or better, average signal
teristic of chordal-hop paths is an absence of slow levels were increased and operating periods greatly
fading, such as occurs due to interaction between extended by the inclusion of very low angles in the
different modes (eg two- and three-hop paths) when radiation pattern. On the available evidence, however,
multihop propagation exists, as is usual for example it would be unwise (even if it were possible) to exclude
across the North Atlantic. On this basis, as well as the higher angles up to at least 10° or so.
usual absence of strong signals from intermediate For ranges less than 2,000 miles (3,200km) higher
ranges [5], the author is convinced that multihop angles are required, as will be evident from Fig 2.14.
propagation is unusual in amateur communication over Typical values of critical frequencies for summer and
very long distances, greater than say 5,000 miles winter periods at the two extremes of a typical sunspot
(8,000km), and this finding would be consistent with cycle [8] are shown in Fig 2.15. The values of muf are
those of Stein [4] based on back-scatter soundings of obtainable by dividing this with the cosine of the angle
the ionosphere. of incidence i (Fig 2.10) of the wave entering the
It may however be necessary to differentiate in this ionosphere. For very low angles of radiation we have
respect between amateur and commercial services, sin i ~ i (in radians) ~ Vh/2,000 where h is the layer
since the latter tend to be concerned with maintaining height in miles as given in Table 2.1, figures for the F2
communication between specific points at all times, layer (daytime) being very approximate.
whereas amateurs are more interested in making the At typical angles of radiation i is slightly greater,
best use they can of whatever opportunities happen to being given approximately by V//1,600 for a 6° angle.
exist at a given time. This obviously generates a bias Based on a radiation angle of 6° and temperate lati-
towards the use of low-loss modes irrespective of the tudes, Table 2.1 shows the highest and lowest of the
destinations which they serve. As an instance, a British amateur frequency bands needed for continuous dx
station wishing to communicate with Australia is communication at different periods based on the above
usually content with the 2 or 3h time slot during which data. In general at least one band higher would be
this is easy and, except perhaps in a contest, is not usable for north-south paths. It should also be noted
very interested in periods when hundreds of times that day-to-day variations are quite large and higher
more power is needed due to the absence of layer tilts. frequencies when usable wiil provide bigger signals.
The argument that there is no optimum radiation The frequency pairs on the right of the table should
angle other than the lowest possible, though opposed to result in a dx capability at nearly all times for some
18 HF ANTENNAS FOR ALL LOCATIONS

nH ay RCO OREAbisdenins
re an er as ey ae ade
ENNONILTAN
NY Hc EUUBB ARIE
fi N h aa aaa Giada eaueg
T

i RR CN TNS era [ead

:
\ NJ
EN

Ea NaN PENTEL
N- ka4 hth eo a
aN i Zz
Fig 2.14. Radiation angle involved in

Nek
Fee
SN one-hop and two-hop paths via E and Fo
\\
a
layers. (From NBS publication lonos-
hallANedieid Pheric Radio Propagation, p191)

habsLLNS Mea Bs
IST NET Pee
BRRERS
BE fa} bel CAN NECNCEEL
(degrees)
Aangle
Radiation
— Me Adeirae SOR RNUNGEEINGE
BE | Soha IN TETN NUTT NEE
id HE TTT TANTTONG
Pg
pee
oe
Tell
ye
ae
ee
ee
ee
oe
eae
ee
te MyEatear:LTTE ENE ANAT TD
iSian
eee
aaa
caer
Gearieee
eee
tied
UanRineetanid
eannsiaie
TE EEE TENET SET
SE
es)
a)RECS
ae
|eee
SS
Bd
PY
el
Ea: a TET
AF
Paro
Ee
a
ee PII,
EC TTT ETT PNT NS
200 300 400 600 8001000 2000 3000 4000
L18 Path length (km)

ZN

ee:

Fig 2.15. Typical diurnal variations of

layer critical frequencies for summer
and winter at the extremes of the solar
cycle (Radio Communication Handbook)

LLL
VAL
0 ~ Sea ce

20 an
SOLAR 2
MAXIMUM

directions and apply also to single-hop ranges near the frequencies may result in failure to communicate dur-
right-hand edge of Fig 2.14. Absorption is a maximum ing most of the night and especially for an hour or so
in daytime and the higher frequencies are needed for before dawn.
best propagation around midday, and in summer the At intermediate times and intermediate portions of
greater part of the afternoon. Exceeding the lower the solar cycle there may be considerable benefit from
 WAVES AND FIELDS 19

intermediate values of frequency, but an antenna sys- as explained in later chapters. Unfortunately many
tem covering a range of one octave should meet most antenna systems described in the amateur literature
dx and medium-range requirements, excluding contest have been arrived at by trial and error without
work at the solar maximum. Solar minimum conditions regard for the underlying principles. The result is
create a more difficult antenna problem due to the sometimes successful but usually suffers from at least
larger ratio of frequencies as well as the lower values, some minor deficiencies, and in the absence of the
though if operation at night is dispensed with the necessary logic the mode of operation may well prove
10MHz band can be substituted for 7MHz. incomprehensible.
From the f, column it will be seen that for very
short-range (zero-skip) skywave working the 1-8 or
3-5MHz bands are required at solar minimum but in References
addition the 7MHz or even the 10MHz bands may be [1] ‘Effect of antenna radiation angles upon hf radio
usable at solar maximum. For ranges of the order of signals propagated over long distances”, W. F.
600-1,200 miles (950-1,900km) the lower of the dx Utlaut, Journal of Research of the National Bureau
frequencies will be suitable for a large part of the time of Standards, Vol 65D, No 2, March/April 1967,
but one band lower may be needed, each of the bands p167.
being usable at times. [2 “A comparison of long-distance hf radio signal
wees!

It would not be appropriate here to discuss the reception at high and low receiving sites’, M. R.
numerous anomalies and freak modes of propagation Epstein et al, Radio Science, Vol 1, 1966, pp751-
which can occur but it is useful to be aware of Dellinger 762.
fade-outs (sids) which occur very suddenly during [3 “Investigations on great-circle propagation be-
——

periods of high solar activity, may last for several tween Eastern Australia and Western Europe”’,
hours, can be total, and are accompanied by the H.J. Albrecht, Geophysica Purae Applicata, Vol 38,
disappearance of the cosmic noise levels which normal- 1957, pp169-180; see also Journal of the Wireless
ly set a limit to the usable sensitivity of receivers. At Institute of Australia, Vol 21, No 5, 1953, and Vol
such times increased sensitivity may allow communica- 24, No 10, 1956.
tion to continue in some cases, or to be resumed at an [4 “The role of layer tilts in ionospheric radio prop-
—

earlier point in time, and some of the measures used for agation”, S. Stein, Journal of Geophysical Re-
improving antenna and receiver performance under search, Vol 68, p217.
normal conditions in respect of interfering signals be- [5 ——_—
“High-frequency antennae and propagation modes
come counter-productive. These occurrences are in relation to the amateur service’, L. A. Moxon,
however fairly infrequent and are virtually unknown IEE Conference on Antennas and Propagation,
during the minimum years. 28-30 November 1978.
[6
—
“On the propagation of short waves over very long
distances: predictions and observations’, K. J.
Hortenbach and F. Rogler of Deutsche Welle (the
Summary and conclusions German external broadcasting service), Telecom-
The aim in this chapter has been to provide the reader munications Journal, Vol 46 vI/1979, (ITU, Gene-
with simple ways of visualizing the radiation from an va), pp320-7.
antenna and its progress through space to the receiver. [7] Ionospheric Radio Propagation, NBS Monograph
However complicated an antenna may appear, it can 80, Washington, 1965.
usually be understood by thinking of it as a number of [8] Radio Communication Handbook, Vol 2, 5th edn,
simple components assembled in some logical manner RSGB, 1977, p11.17.
CHAPTER 3

Gains and losses

In judging the suitability of a transmitting antenna for value of resistance, by 20 log (V,/V2). This is often
its task there are many questions to be asked, of which more convenient to measure but it is essential to realize
the following (apart from their obvious importance) are that these are the same decibels, the expressions being
the most basic: mathematically identical subject to the stated condi-
(a) Does it radiate most of the power available, or is tions. There is no such thing as a “‘voltage decibel”
a substantial part wasted in the heating of wires, though this point seems to have caused a lot of confu-
components, or surroundings? sion.
(b) Does it, as far as possible, concentrate the Useful figures to remember include 1dB for a 25 per
radiated energy in wanted directions or is there wastage cent increase in power and 10dB for a tenfold increase,
due to avoidable radiation in other directions? with 1dB representing in most cases a just-appreciable
Put briefly, the aim is to achieve negligible /osses and difference in signal strength. Signal reports are based
as much gain as possible; this involves not only the on the S-unit which has been frequently defined as a
antenna as such but the relationship between the 6dB step, but since its inception and up to the time of
antenna and its environment which is the subject of a writing the average S-unit has been closer to 3-4dB
later chapter. We need however to proceed in two which the author believes to be a much more realistic
stages, first considering the antenna in isolation and figure. Unfortunately reliance by some experimenters
ignoring for the moment some important aspects such on the official definition appears to have been the
as whether it continues to work when the frequency is origin of some wildly inflated published figures for
shifted to another part of the band, how easy it is to antenna gain.
make sure of correct operation, and whether it is likely If more energy is radiated in one direction there is
to survive the first gale. less available for others so that gain and directivity are
The aim of this chapter is to provide a clear outline closely linked. To have no gain in any direction an
without too much involvement in detail of the princi- antenna would have to radiate the same signal in all
ples involved in maximizing gain and minimizing losses directions, in which case it is said to be isotropic. The
in the antenna itself, the problem of ensuring that all gain as defined above is therefore the gain relative to an
the available power finds its way to the antenna being isotropic source and if two antennas have the same
the concern of the next chapter. Later chapters seek to directivity they must necessarily have the same gain.
interpret these principles in a practical context and One disadvantage of isotropic antennas is that nobody
Chapter 9 discusses a number of important differences has yet succeeded in making one, but this in no way
between transmission and reception. detracts from their usefulness to mathematicians, an
author trying to explain antenna gain, or an advertiser
trying to make his gain figures look as good as possible
Gain and directivity without actually cheating!
Referring back to Fig 2.6 it will be noticed that the On the other hand, the real interest for the amateur
radiated signal is stronger in some directions than lies in the relative gain of various practical antenna
others. If in a given direction the signal is stronger than systems and it is common practice to express gain in
the average the antenna can be said to have gain in that decibels relative to that of a A/2 dipole (Fig 2.5). The
direction, this being normally expressed in decibels. latter’s gain exceeds that of an isotropic antenna by
The decibel (dB) is the natural measure to use for this 2-15dB so that use of an isotropic reference adds this
purpose since it denotes a power ratio which is the kind amount to the gain figure.
of difference that the ear hears. Thus “3dB” means Antennas may consist of single wires, long or short,
“equivalent to a 2:1 difference in power” and each 3dB or numbers of separate wires or other conductors
step indicates multiplication by 2, or division if there is co-ordinated in such a way that there is a considerable
a minus sign in front of it. concentration of energy in one direction. There is no
The decibel is therefore a logarithmic unit, a power precise definition of a beam antenna but any arrange-
ratio of P,/P, being expressed in decibels as 10 log ment using two or more elements is normally classed as
(P,/P,) or, in terms of the voltage ratio across any given a beam. Further discussion of antenna gain is divided
20
 GAINS AND LOSSES 21

or
dipole, and is one of the most useful of all amateur hf
antennas in view of the ease with which it can be
supported, leading to numerous applications as de-
scribed in later chapters. We have thus made no
progress in the search for worse performance, and
although by making the apex angle sufficiently acute
(a) (b) Ly)
losses could be increased it would probably be easier to
Fig 3.1. Bending of elements. In case (a) the effective field fit an attenuator! .
strength is halved for a given current if the receiving and Another approach might be to try using not one but
transmitting antennas initially have the same polarization, but several bends, but it will be recalled that the worst that
current is increased by \’2 so the loss is only 3dB. (b) Re-orienting
(a) as shown restores field strength to its original value. (c) is can be achieved even theoretically is an isotropic
equivalent to (b) antenna which has a loss of 2:15dB compared with a d/2
dipole as discussed on p20, and it is difficult to get
between single wires and beam arrays, leading to the within even 1dB of this in practice. This tends to the
conclusion that within each of these categories, assum- conclusion that, subject to remaining in free space and
ing a given height and polarization, there is a tendency avoiding extremes, all single-element antennas have
for all antennas of interest in the amateur context to be roughly equal gain, thus leaving the designer free to
equal. Thus it turns out to be remarkably difficult to concentrate on other aspects of performance or con-
influence appreciably the gain or directivity of a single struction.
piece of wire shorter than about 3A/4, unless losses
occur or unless at some point there is a reversal in the
direction of current flow. Unwanted modes of radiation
This is obvious in the case of a straight wire since the The close equivalence of different types of simple
method of calculating the directional pattern described antenna discussed above is not generally realized and
on p10 breaks down only if the various contributions to the author has heard, for example, strong criticisms of
the field cannot be directly added. Such a breakdown the inverted-V on the score of the power wasted due to
occurs if, and only if, important contributions to the “radiation off the ends”. Such unwanted modes are a
field are of opposite sign or come from points so far common feature of antenna patterns especially when
apart that the distances travelled to the receiver are elements are “‘bent into odd shapes’’, and it may be
widely different, thus resulting in a phase difference. necessary, if only for reassurance, to work out how
This in its turn can only happen if the antenna is quite much power is being wasted in them. They need also to
large but, to get this in perspective, it is useful to be considered as potential sources of interference to
consider radiation in a nearly-endwise direction from reception.
the dipole of Fig 2.5. Contributions from near the ends The inverted-V provides a particularly important and
will indeed be separated by nearly )/2 so that they are instructive example. It is obvious from inspection that
out of phase but, after allowing for this, it turns out that the endwise mode exists, and it can be readily analysed
the 4/2 dipole has a gain of only 0-4dB (5 per cent in with the help of Fig 2.5, noting first that the U
signal voltage) compared with the short doublet of Fig arrangement, Fig 3.1(c), though of less practical in-
2.4 and even a 50 per cent increase in length yields no terest, produces the same field strength for a given
more than an extra 0-7dB. current as the inverted-V, Fig 3.1(b). This follows from
It might be thought that a simple way to degrade the the 45° slope of the wires of the V which reduces the
performance of an antenna would be to bend it, eg as in field from a given current by V2, whereas the loss of
Fig 3.1(a) where one half of a dipole has been bent over the ends in case (c) reduces the field by 30 per cent (ie
at right-angles. Assuming optimum conditions to start divides it by V2) as discussed in the last chapter.
with and maintaining the current constant, this will The inverted-U, with its similar tendency to produce
halve the field strength but, out of the total difference endwise radiation, has the advantage, important for the
of 6dB, half is the result of reducing the effective length present purpose, of being much easier to analyse. From
and half is due to tilting the plane of polarization as Fig 2.5 the ratio of field produced by each of the end
explained in Chapter 2 (p12). Since as already pieces to the field in the wanted direction is 1-4/0-3, and
explained gain is virtually independent of length, the although the “‘ends” are out-of-phase there is enough
current does not stay constant but increases to compen- phase shift in the endwise direction to prevent cancella-
sate for the reduction in length, whereas the polariza- tion. Using the method of addition described on p12,
tion is immaterial subject to conformity between the the field strength ratio for the unwanted directions is
two ends of the link. To restore the field strength to its given by (0-3 x V2)/1-4=0-30. To find the power
original value, more or less, it is therefore only neces- ratios (assuming the same directional pattern for both
sary to rotate the bent dipole through 45° as in Fig modes) this has to be squared, ie the efficiency on this
3.1(b). This is usually referred to as an inverted-V basis would be 1/1-09, a loss of 0-37dB only. However,
22 HF ANTENNAS FOR ALL LOCATIONS

the directional patterns are not the same since the ends to the radiated field are in the same phase-.and can be
(considered by themselves) resemble the W8JK an- lumped together. An antenna ceases to be a point
tenna described on p24. This has a gain of 4bB so that source only if there are major current concentrations
less power has to be radiated to account for the observed separated by distances of the order of at least \/4, or if it
field strength, thus bringing the loss down to only includes antiphase sources in which case small antennas
0-15dB. The advantage of the U lies in the ease of can usually be represented by two such sources.
calculation. To obtain significant gain relative to a single source it
With such low losses there is no need to worry much is necessary to assemble two or more sources to form
about precision but a more accurate figure for the V beam antennas in ways to be described later. Formulae
obtained by graphical integration along the lines of Fig describing the directivity then become more compli-
2.5 comes to 0-23dB. Even this is much reduced when cated but with closely spaced elements, which is the
two elements are assembled as a beam or if the apex case of most practical importance to amateurs, still
angle, which should be as large as possible and prefer- contain only cos 6 so that again there are no dimensions
ably not less than 90°, is increased to 120°. The effective involved.
mean height of the V is somewhat less than the mean The explanation, as we shall see, is simple provided
height and increased interference might be expected the number of elements is restricted to two or three,
from a vertically-polarized local station if there hap- and thereafter ceases to be of much practical interest,
pens to be one within 20° or so of the end-on directions, though a number of mathematical studies have yielded
though for high-angle (short-skip) signals there is little conclusions bordering on science fiction. It has been
to choose between straight wires and bent ones. As will shown that there is no limit to the gain theoretically
be evident from later chapters, the inverted-V has obtainable from an antenna provided only that there are
important practical advantages, particularly for “‘invisi- enough elements packed together in a small enough
ble” antennas, and the above example should allay space! In particular Uzkov [1] has shown that the
doubts which have probably caused many operators to maximum gain theoretically possible, expressed as a
opt for more expensive, larger and more visible power ratio relative to an isotropic source, is equal to n?
antennas. where n is the number of elements; unfortunately with
In the case of other types of antenna it is usually increasing gain and reducing size, the efficiency, band-
possible to deduce from the shape what unwanted width and allowable constructional tolerances decrease
modes of radiation are likely and arrive at a rough at an astronomical rate due to the requirement for
estimate of their magnitude along the lines of the above much larger currents as explained below.
examples. Antennas such as these are known as supergain
antennas and it has been shown [1] that the special
characteristics just outlined are typical also of the
Small antennas operation of amateur close-spaced beams. This identity
In Fig 2.6(b), following usual practice, the direction of sets clearly defined limits on possible performance and
the receiver relative to the best direction of transmis- explains rapid widening of the gap between theory and
sion has been denoted by the Greek letter 6. Since field practice as the number of beam elements is increased.
strength is proportional to /’//, which is by definition the Its failure to attract much attention can perhaps be
cosine of the angle 0, the radiation pattern shown in Fig attributed to the fact that in practice it is possible to
2.6(b) is commonly referred to as a “‘cos 8” pattern. Its travel only a very short distance down the supergain
derivation here has involved no more than the use of a road, so that typical amateur rotary monoband beams
ruler and references to cos @ here and subsequently with three closely-spaced elements represent, more or
should be regarded as a form of shorthand, not as an less, the upper limit of practical achievement for trans-
attempt to inflict advanced mathematics on the reader. mitting antennas based on supergain principles.
The important point is that cos 6 provides a full Large beams, typical of commercial installations and
description of the antenna directivity and therefore beyond the means of most amateurs, are based on a
gain without any reference to the dimensions which are different set of principles, and in the next section the
therefore of no direct significance. Later when consider- two modes of operation are examined and compared.
ing losses it will be necessary to qualify this but the Further discussion of the Uzkov limit and its implica-
primary fact is that most amateur antennas for the hf tions for the design of amateur beams will be found at
band other than beams and long wires can be regarded the end of Chapter 5.
virtually as point sources of radiation and a point by
definition has no size.
In practical terms “point source”’ means that from all Beam antennas
angles of view the distance to any part of the antenna is It is useful, particularly in the case of beam antennas, to
the same within some small fraction of a wavelength as think of the transmitter as being in the centre of a
the distance to all other parts, so that all contributions sphere, Fig 3.2, whose radius r represents the distance
 GAINS AND LOSSES 23

A..... Illuminated area a gain of 16/(1-37 x 0-82) times or 11-5dB. Subtracting

Z..... Vertical beamwidth 2:15dB, one obtains a gain of 9-35dB relative to a
(in plane YZ)
6...... Horizontal beamwidth dipole which is absurd, though not alas untypical of
(in plane XZ) published figures. Suspecting the use of a square-law
meter so that the quantity plotted was relative power,
not field strength, a half-beamwidth of 45° was
obtained for the vertical plane and, multiplying the
vertical pattern by cos 8 (as explained in Chapter 6) to
obtain the horizontal pattern, the new values of a,, a,
came to 1-58 and 1-22 approximately. This brought the
gain (relative to a dipole) down to 7-2dB, in line with
the text of the article and within the possible theoretical
limits, though still slightly improbable as a practical
Direction
maximum
of figure. This example is important insofar as it illustrates
radiation aw the only method, albeit an indirect one, open to the
Power gain =
Surface area of sphere ordinary amateur for obtaining a meaningful figure for
Surface area of "Illuminated" region the measured gain of an hf antenna.
Fig 3.2. Spherical representation of space surrounding an Appreciable errors arise (especially with low-gain
antenna. An isotropic radiator at the central point O provides arrays) from the fact that beam edges are not sharply
uniform illumination of the sphere surrounding it. A directional
antenna illuminates only a portion of the sphere, as shown by the defined, but the width between half-power points pro-
shaded area. The terms “horizontal’’ and “vertical” are not vides a useful approximation. This may be measured as
applicable in free space but ¢ and 6 can be identified with a, and described on p235 or obtained by plotting the directional
Gz in the text (Radio Communication Handbook)
pattern as explained below. This provides a simple
OA to the receiver. The surface area of a sphere is method for obtaining a rough idea of gain whereas
given by 4r? and an isotropic transmitting antenna attempts at direct measurement in the hf bands with
would, by definition, illuminate the whole of the sur- normal amateur resources can be highly misleading
face evenly whereas the object of a beam is to concen- (p236). In all cases beams are formed by making use of
trate all the available energy into a small area which phase differences so that signals transmitted from (or
includes the receiver. received by) different parts of the antenna add together
If the beam is circular with a radius 7, at the surface in different ways depending on the direction.
of the sphere, the area which it illuminates is approx- Figures 3.3 and 3.4 illustrate the two methods using
imately equal to the cross-sectional area mr provided n, phase differences to favour preferred directions at the
is not too large. The gain, being the ratio of the two expense of other directions, and the inset is a reminder
areas, is given by 4r*/ré and, since 2r,/r is equal to the of the graphical method for adding two signals when
width a of the beam in radians,* the gain is given by there is a phase difference between them. In Fig 3.3 the
16/a2. If the beam is not circular but has widths of a,, a, elements A and B are fed in the same phase, causing the
in planes at right-angles to each other, a,a, may be
substituted for a. Mm ToQ
To illustrate this a circular beam with a width of 28-5° 4

(a =%) would have a gain of 16/(1/4), ie 64 times or 7

¢
4
7
7
a

18dB. This formula is not strictly accurate for large 7 7

beam angles since (a) the cross-sectional area of the Vertical dipoles fed
in Same phase
beam is not quite the same as the corresponding area of
spherical surface, (b) the angle has been used instead of
its sine. These errors are however more or less self
compensating. In practice beams always have side-
lobes, ie they radiate some energy in unwanted direc-
tions and this reduces the gain slightly. The gain as
defined above is relative to an isotropic source so that Phase-shift due to BC, givenin
2-15dB must be subtracted for comparison with a radians by AB sin@. oe
dipole.
As a more typical example, some published directiv-
EA
ity patterns for a three-element Yagi array suggested
vertical and horizontal free-space beamwidths (a, a) Fig 3.3. Additive gain. Two widely-spaced point-source elements
not exceeding 1-37 and 0-82 radians respectively, giving fed in same phase produce maximum field in direction P. To
obtain relative field in any other direction Q, it is expressed in
terms of the equivalent phase shift (AB sin 6 x 27/) radians) and
* To convert degrees to radians it is necessary to divide them by 57-3. the fields produced by A and B added as shown at inset
24 HF ANTENNAS FOR ALL LOCATIONS

Each dipole can be regarded as a

point source at A,B, andd is To Distant point, P
where field-strengths 8JK, Vertical plane,
small compared with A
w from A,B=—F and F (cos 8)
Uc a respectively, i.e.
8JK, Horizontal! plane,
3 odi FR (cos? @)
nig 4 ay,
Pat we F

tak ee re ati aks Dp

Fe=2F sin
Maximum B Maximum R 2
radiation radiation
-1| fh: Fr=field-strength left over \
after combining F and —F Dipole, Vertical
To produce same field Fr j plane (circle)
with a dipole of same |
Signal from A travels a distance length requires a |
AC further than signal from B. current Ip = oh x I Dipole, Horizontal
Therefore phase difference Dp / Plane (cos ©)
=d cos @ x 390° /

Fig 3.4. Mechanism of W8JK antenna

radiation to be additive in the direction P, whereas in

the direction Q the contribution from B has further to
travel and therefore lags in phase so that the addition is
less effective. In this way the width of the radiation
pattern compared with that of a single element is
reduced, thereby causing more energy to be radiated in Fig 3.6. Comparison of dipole and W8JK radiation patterns (both
the direction P. It is however important to note that for antennas horizontal)
this effect to be appreciable A and B must be widely
separated, since otherwise the phase difference is in-
sufficient. The directivity of a simple array of this type wanted than the unwanted directions. This narrows the
is shown in Fig 3.5. radiation pattern so that gain ensues, provided all the
In the case of Fig 3.4, which illustrates the W8JK energy is radiated. Unfortunately gain implies large
array, the mechanism is subtractive, the elements currents in the elements since the signal arises from the
working against each other but less effectively for the difference between two relatively large quantities, and
losses may occur as described in the next section.
Referring to Fig 3.4 it will be seen that along the line
AB radiation from one element lags slightly behind that
from the other because it has further to travel, the
Antiphased pair
eg asfig 3.4
phase shift being ‘‘equal to the spacing’’, eg one eighth
of a cycle or 45° if the spacing is A/8. In the direction P
—Pair as fig 3.3
the field is reduced in proportion to the distance AC.
The resulting directivity is shown by the solid lines in
Fig 3.6.
It should be noted that in contrast to additive gain,
which requires wide spacings, the subtractive principle

NS comes into operation only if A and B are closely spaced,

otherwise there will be some directions in which the
fields are additive and the operative principle reverts to
that of case (a). Due to their wide spacing, elements of
an additive array can be regarded as more or less
independent of each other, so that if two elements each
receive half the total power the current in each is I/V2
where / is the current which would flow in a single
element. However, due to in-phase addition this is
Fig 3.5. Polar diagram corresponding to Fig 3.3, assuming \/2
spacing of co-phased elements fed with equal currents. This is equivalent to a single-element current 2//V/2 = 1-414,
the diagram for the horizontal plane assuming vertical elements, ie a gain of 3dB.
and requires multiplication by cos 6 if these are replaced by Extending the argument to n elements, the power
horizontal elements aligned for the direction P. The pattern for
similarly aligned but close-spaced elements fed in antiphase is gain becomes n times but to satisfy spacing require-
shown for comparison; this is the cos? 6 pattern from Fig 3.6 ments the volume of space occupied by the array
 GAINS AND LOSSES 25

(a)
"BJK'
(b) 30° Phase shift
F Fa (Direction of 5dB Gain
maximum
radiation) Double Null
pattern

(a) 45° Phase shift

4.2dB Gain

Cardioid pattern
Forward field
F Fp=Fx/2 if
a mate (b)
[SOK
+

F
Phase shift Fig 3.8. Radiation patterns (a) corresponding to Fig 3.7(b), and (b)
showing the result of a 15° reduction in phase shift

; Backward field
(pp28, 29 etc) and (b) major improvements require
only a small modification, in the course of which the
WS8JK antenna gets deprived (perhaps rather unfairly)
Fig 3.7. Effect of phase shift. Vector diagrams are drawn to scale
of its label. On the other hand, its virtues include
for the case of \/8 spacing. Note that Fp is almost unchanged in extreme simplicity and the ease with which an under-
going from @ =0 (the W8JK condition) to ~ = go but the radia- standing of its mode of operation can be extended to
tion is now unidirectional. Also F, and therefore the element
current /, is much smaller. @o is the phase shift ((d/A) x 360°)
include all other small beams.
corresponding to the spacing d The directional pattern of the W8JK antenna is very
easily calculated since, referring again to Fig 3.4, the
increases directly as the number of elements. High gain distance BC varies with direction in exactly the same
by this method therefore requires a large amount of way as the variation in the apparent length of the
real estate which puts it beyond the reach of most individual elements, so that there are now two cos 8
amateurs. The principle may however be applied to the patterns superimposed. The new cos 8 pattern applies
combining of pairs of antennas; if these are equally equally to horizontal and vertical axes, so that to obtain
good, adequately spaced and have the same polariza- the overall directional pattern of the antenna the
tion, an extra 3dB gain is obtainable in this way. The pattern of the single element (which is cos @ in one
effects of unequal performance or power sharing are plane and omnidirectional in the other) is multiplied by
described on p100. cos 8 to give cos* 6 and cos 6 respectively. This is
Spacings between the elements of additive arrays illustrated in Fig 3.6 where the patterns for this antenna
normally need to be at least A/2, although a row of are compared with those for a short dipole, reproduced
elements may be backed by a row of reflectors spaced from Fig 2.6(b). The corresponding half-power beam-
4 and phase-shifted by- 90°. In this case in-phase widths are 66° and 90°, so that converting to radians and
addition takes place in the forward direction and substituting for a,a, gives a gain of 6-36dB (4-2dB
cancellation in the back direction. relative to a dipole) in excellent (though perhaps to
Additive methods were the only ones generally rec- some extent fortuitous) agreement with more accurate
ognized until the appearance of a classic paper by G. H. calculations described in Chapter 5 and featured in Fig
Brown [5] in 1937 which demonstrated the practical 5.4.
possibility of gains in excess of 5dB from pairs of The modification required to overcome the limita-
closely-spaced elements. One of the sequels was the tions of the W8JK beam is the introduction of a phase
development of the W8JK array, to which reference shift between the elements and if this is “equal to the
has already been made, but, despite its deep roots in spacing’’, the net phase shift becomes zero for one
the history of amateur radio and the sanctity conferred direction and doubled for the other as shown in Fig 3.7.
by long and extensive use, it has to be said that the Zero phase difference implies cancellation of the signal
W8JK antenna has few practical merits. so that the pattern becomes unidirectional as shown in
This is because (a) achievement of gain despite close Fig 3.8 and a further advantage is that the field strength
spacing and antiphase excitation implies large currents in the beam is almost doubled for a given current. Since
with consequent reduction in efficiency and bandwidth the gain is about the same the current is halved which
26 HF ANTENNAS FOR ALL LOCATIONS

wasted. For comparison Fig 3.9 also shows the patterns

for additive gain arrays comprised of two elements
phased vertical
array, $/Q =1
spaced )/2.
0-8
ria
»~
It will be seen from results presented in Chapter 5
Dw
that the assumptions start to break down at spacings
=e 0-6-+ ie
ie in-phase
in-p \ approaching A/4, at which point the additive principle
Cy pair spaced \] a
a Al2 \ becomes fully operative, so that spacings of 4/5 to 4/4 |
for AS nS cos 8 (1+ cos 8) represent a region of transition from additive gain
ap XY NNN eg horizontal close- proportional to size to subtractive gain independent of
s collinear array NiRS spaced phased array,
or (broadside curve Rye 1 size.
x cos @) The modification to the W8JK antenna links it
closely to the paper by G. H. Brown [5] which put an
end to the belief, current until 1937, that spacings of /4
were mandatory, and showed how gains and directional
O 10.4520 we SO pee O DOF 6075/0
Angle relative to beam heading
.8Oi2590
patterns vary with spacing and phase shift when ele-
ments are closely spaced. Practical implementation can
Fig 3.9. Cartesian plot comparing the radiation patterns of va- take various forms, of which the two-element driven
rious types of two-element beam. In view of symmetry the plot is
restricted to 90°. The unidirectional patterns are shown for the array bears the closest resemblance to the original.
forward direction only. Patterns for the back direction (various There are however other and perhaps easier ways
values of ¢/@g) can be found in Chapter 5, p70. The pattern for a
horizontal dipole is also included
(p73) of avoiding the current inequalities which consti-
tute a serious defect of many designs of close-spaced
beam.
reduces losses and leads to further practical advan- Beams with more than two close-spaced elements,
tages, including the possibility of making the antenna though clearly members of the same family, are not
much smaller (p83). In addition, the swr bandwidth amenable to simple theory. Their design is usually best
(p121) tends to be increased in the same ratio. Instead tackled along experimental lines, bearing in mind the
of being given by cos*@ the directional pattern in this general principles which have emerged from study of
case is cos 8 (1 + cos 8) as shown in Fig 3.9. two-element beams and making use of design charts
The energy no longer being radiated in the back such as those on p80. As stated earlier the Uzkov
direction is now being used to broaden the beam, thus formula predicts an upper limit of n* for the gain from n
providing increased angular coverage without altera- closely spaced isotropic elements. After applying small
tion of beam direction. These patterns are for the corrections for dipoles [6] the possible gain becomes
horizontal plane, assuming the antenna to be horizon- 5-3dB for two and 8-4dB for three elements. In a
tal, the vertical patterns being given by cos 0 and separate study [7] a gain of 7-6dB was obtained for a
(1 + cos 8) respectively. Since none of these patterns three-element Yagi array using A/2 dipole elements
contains any dimensions the gain and directivity are not spaced 0-15).
directly dependent on size, which is really stating the Unfortunately the extra gain from the third element,
obvious since the way in which apparent length and except for about 1dB, is not realizable in practice. This
apparent spacing vary with direction has nothing to do is because the signal, however large it may be, is the
with the actual values of these quantities. It is assumed difference between much larger quantities and the
for this purpose that the ratio of phase shift to spacing is balance becomes rapidly more critical with increasing
maintained constant, though the actual choice of phase gain or reduced spacings. This is reflected in very tight
shift is optional between fairly wide limits, 0-8dB of constructional tolerances, narrow bandwidth (Chapter
extra gain being obtainable with two thirds of the above 8), and a very low value of radiation resistance which is
phase shift. Fig 3.9 compares the cos 0, cos’ 0, (1 + cos less than 2Q for the above-mentioned Yagi array. The
6) and cos (1 + cos 9) patterns in Cartesian form. The significance of this will be appreciated from the discus-
last two are for the forward direction only, back- sion which now follows.
direction curves for various values of phase shift being
given in Fig 5.6.
- This form of plot, though perhaps less familiar than a
Efficiency, radiation resistance and loss
polar diagram, is more informative since it allows resistance
beamwidth and null angles (if any) to be read at a The proportion of power radiated depends on the ratio
glance and it can also be used for estimating (by the of radiation resistance and loss resistance, and there are
method demonstrated in Fig 2.5) the fraction of the also close links between radiation resistance and band-
total power which is radiated in the rear direction. It width as explained in Chapter 8. It has been shown that
requires only a modest front/back ratio to reduce this to directivity and therefore gain does not depend directly
a small value, typically 6dB for less than 1dB of power on antenna size, but if the length is halved the field
 GAINS AND LOSSES 27

normal temperatures. This means that a fraction

R,/(R + R,) of the total power from the transmitter is
wasted in heating the antenna wire. Alternatively the
efficiency, defined as the ratio of power radiated/total
power supplied by the transmitter, is given by
R/(R + R,). Since rf current flows only on the surface
of a conductor of normal dimensions, R for a given
length of wire is inversely proportional to surface area.
Values of R can be obtained directly from Fig 3.10
which shows the variation of loss resistance with fre-
quency for typical tubing and wire sizes, or Fig 3.11
which shows the dependence on wire gauge for several
frequencies in the hf band. For comparison with the
radiation resistance, the loss resistance of an antenna
wire must be divided by two since the current along it
Ohms
wavelength
half-
per is more or less sinusoidal, and the mean square current
is therefore only half the value which it has at the usual
feedpoint, ie the centre of a current loop. Approx-
imately, for non-magnetic materials, the rf resistance
varies as the square root of the resistivity and inversely
as the surface area. Values obtained from the curves
should be increased by about 25 per cent for aluminium
and rather more for aluminium alloy. For magnetic
1 2 5 10 20 50 100
Frequency (MHz) materials the losses are much higher.
The radiation resistance R is usually taken as 73Q for
Fig 3.10. RF resistance of copper conductors. Resistance per unit
length is proportional to the circumference of the conductor and a \/2 dipole in free space, though it varies slightly with
the square root of the frequency. Uniform current distribution length/diameter ratio, mainly because of the slight
assumed effect this has on resonant length (p122). Values for
other lengths and shapes can be readily deduced from
strength can only remain unchanged if the average this using the above principles, though allowance must
current is doubled. In accordance with basic principles, 14
it must also meet the condition P = J?R where P
represents the power radiated, and the resistance R
needed to complete the relationship is defined as the
radiation resistance. This is not a resistance in the usual
physical sense but a notion conjured up to account for
energy travelling outwards from the antenna, much of
it vanishing for ever into outer space. However, unlike 3
some “fictitious” resistances it is real enough to
account for the measured values of antenna impedance
less any losses which may arise as discussed below.
The current J may be observed at any point in an
antenna system and a value of R can be assigned to that
point in such a way that the power relationship is
satisfied, though unless otherwise specified the point of
reference is usually taken to be the point of maximum
current. It is often useful to remember that, barring
losses, P must be the same at all points along an
antenna element plus any associated feeder system, so half-wavelength
ohms
in
resistance
Loss
per
that given for example relative values of J for two
points in the system and the value of R for one of them,
R for the other (which may be the radiation resistance
at the feedpoint) can be readily determined.
It will be seen that as length is reduced R drops
rapidly because of the increase in J, and power will be
Standard wire gauge (swg)
wasted if this becomes comparable with the loss resist-
ance (R;) which is possessed by all conductors at Fig 3.11. RF resistance of thin copper wires as a function of swg
28 HF ANTENNAS FOR ALL LOCATIONS

Table 3.1. Current distributions and multiplying himself (a) dividing by a further factor of two and (b)
factors for estimation of radiation resistance adding 8 per cent to the loss resistance because the total
wire length has to be increased by about this amount!
Type of distribution Factor
This brings R down to about 1-4 only, compared with
Uniform (ie rectangular) 2-46 36Q for end-loading.
Middle half of a sine wave 2:00 The large currents referred to in the context of the
Sine wave 1:00 W8JK antenna, and the far larger currents experienced
End half of a sine wave 0-69
Triangular 0-50 with three or more elements operating close to the
Uzkov (gain =n?) limit, imply correspondingly low
Example: lf a dipole is halved in length by folding over the ends values of R and, beyond certain modest gain levels, a
the second line of the table applies and R = 73 x 4 X 2 = 36-52. reduction in efficiency. It is this fact, and this fact
alone, which sets a limit to the possible reduction in size
be made for any differences in the way the current is of antennas. Admittedly, long before this limit is
distributed along the wire. This can be done by compar- reached the user may be complaining of narrow band-
ing areas under the curve as in Fig 2.5 but for rough width (Chapter 8) which forces him always to operate
estimates it is sufficient to assume one of the current near the same frequency, but this restriction was at one
distributions listed in Table 3.1, whichever seems most time commonplace, the best transmitters then being
appropriate, and use the appropriate multiplying fac- “rock bound”’ (ie crystal-controlled) and some oper-
tor. This is in addition to multiplying R by the square of ators still have strong frequency preferences.
the length ratio.
The radiation resistance of dipoles shortened in
various ways is shown in Fig 3.12 which is based on Fig
2.5. This shows clearly the enormous advantage of end
loading, but does not tell the full story which is even Signal
generator
less favourable to centre loading since the “missing”
section of the antenna has to be replaced either by a
loading coil (which is a relatively inefficient device) or
by a stub as in Fig 2.5(c). For comparison with the loss
resistance of the wire the radiation resistance must be Fig 3.13. RF signal source supplies power to a load resistance R.
For maximum power in load A, = Rg and any reactance X must
referred to the lower end of the stub, and for an be removed by suitable tuning such as the connection of a
antenna length of 4/4, assuming a sinusoidal current reactance — X in series with the generator
distribution throughout, it is necessary to divide by a
further factor of two. Even this is not the end of it, A consequence of reducing the length of the antenna
since after reading Chapter 4 the reader will be rightly is the need for loading devices such as coils, stubs or
suspicious of the discontinuity at the feedpoint and, “capacitive hats” to maintain resonance, without which
having mastered the use of the Smith chart, will find it is not possible to feed the antenna efficiently. The
need for resonance can be demonstrated by considering
the situation illustrated in Fig 3.13. This shows a
“generator’’, or rf source, of internal resistance Rg and
open-circuit voltage Vg, feeding power into a load

Al
resistance R,. From Ohm’s law the current J = V,/(Rg
+ R,) and the power in the load is V€R)/(R;, + Ro)?,
Capacitive provided the reactance is zero. This has a maximum
end-loading
value when R, = Rag, and VelaRg |is therefore defined
40 as the available power of the generator. Any reactance
X in series with the load reduces the load current in
Inductive
proportion to the increase in impedance, and if the
Nn{o)
centre (stub) reactance is not otherwise removable it must be tuned
loading
out by an equal reactance of opposite sign. Similarly if
measured
at
system
of
centre

aes.
Radiation
ohms
in
resistance
— the generator is reactive this requires a reactive load,
the reactive components being equal and opposite in
40 60 80 100
all cases. This condition is known as a conjugate
ieathe of element in percentage of 4 match and requires the generator and load to have
equal resistances, any reactances being not only equal
Fig 3.12. Radiation resistance of short dipoles with alternative but of opposite sign, ie the load must be inductive if the
methods of loading. For close-spaced beams the radiation resist-
ance varies in the same way but values are reduced according to generator is capacitive and vice versa.
the rules given in Chapter 5, p70 A wire is resonant subject to a small correction for
 GAINS AND LOSSES 29

end effects when the length is a multiple of 4/2 or, if one both are reduced in the same ratio as discussed in
end is grounded, any odd multiple of 4/4 but apart from Chapter 9. For reception therefore it is the directivity
convenience there is no special merit in such lengths. If gain which is important and losses can be largely
a shorter antenna is required one way to achieve ignored.
resonance is to take A/2 of wire and merely fold it to fit
the available space, in which case, as we have just seen,
R varies greatly depending on the method of folding Summary
whereas R, (being determined almost entirely by the In terms of signal strength all single-element antennas
actual length and diameter of wire) tends to be more or have nearly the same effective performance for a given
less independent of how the folding is done. Some height and polarization unless there are losses or
adjustment of length is usually needed if the shape is complications arising from large size or current rever-
changed but this is not important for the rough estima- sals.
tion of efficiency. Gain and loss mechanisms have been explained and
The simplicity of this approach to the design of small directional patterns provided for two-element arrays of
antenna elements makes it easy to calculate the per- various types. It has been shown that most amateur hf
formance and arrive at optimum designs as described in beams consist in effect of two or more closely-spaced
a later chapter (p84). Unfortunately most existing point sources of radiation; the apparent magnitude (or
designs of small antennas are based on the use of “brightness’’) and relative spacings vary with direction
loading coils which add considerably to the effective in a way which is independent of the size of the
value of R;. Others use centre loading but due to the antenna, and this in turn means that the radiation
rapid escalation of losses this allows only a very limited pattern and gain are also fixed, subject to maintaining
degree of size reduction. correct phase and amplitude ratios. For a given number
Losses may occur also in the surroundings of a of elements gain is inseparably linked with the direc-
antenna since the electrostatic field associated with the tional pattern, subject to a limited range of options
voltages at the ends penetrates into adjacent dielectric within which gain may be traded for better directivity as
material such as trees and brickwork etc which may explained in Chapter 5.
have a high loss-factor. These losses are not amenable Gain is theoretically unlimited provided an antenna
to calculation since they involve dielectric constants, is small enough, tolerances close enough, losses zero
power factors and voltage gradients, none of which is and bandwidth considerations ignored; unfortunately
likely to be known. However, being proportional to the the gain obtainable is generally limited in practice to
square of the voltage, they vary with radiation resist- about 6dB. Larger directive gains are associated with
ance in the same way as the losses in the antenna wire, very large currents and high losses unless antenna size
and it will be seen that the smaller the antenna or the is increased, in which case although the gain theoreti-
lower its radiation resistance the more important it is to cally possible is reduced more of it can be realized in
keep it well away from its surroundings. If for example practice.
an antenna has to be fitted into an attic or other
confined space, it is difficult to imagine anything less
suitable than the W8JK antenna since the antiphase
References
connection ensures the lowest radiation resistance and
highest end voltages that it is theoretically possible to [1] “An approach to the problem of optimum directive
achieve for a given design of element. antennae design’, A. I. Uzkov, Comptes Rendus de
Earlier it was shown that gain and directivity are l’Academie des Sciences de l'URSS, Vol 53, p35.
inseparable, but there is nevertheless an important [2] ‘Evaluating aerial performance’, L. A. Moxon,
distinction to be made between gain in this sense and Wireless World February and March 1959.
the actual increase in level of transmitted signal, which [3] “‘Supergain aerials”, L. A. Moxon, Radio Com-
in the case of close-spaced elements is obtainable only munication September 1972.
if there is a rise in the actual current flowing in the [4] ‘“‘“Gains and losses in hf aerials”, L. A. Moxon,
elements as discussed above. On the other hand, to Radio Communication December 1973 and January
establish a given directive pattern and obtain the 1974.
corresponding directivity gain it is only necessary to set [5] “Directional antennas”, G. H. Brown, Proceedings
up the correct current phases and relative amplitudes. of the IRE, Vol 25, January 1937, p78.
Losses, therefore, although they reduce the current [6] “‘A new approach to the design of superdirective
and thus restrict the transmitting gain, need have no aerial arrays”, Bloch, Medhurst and Pool, Proceed-
effect on directivity gain. In the hf band, noise levels ings of the IEE, Part III, September 1953.
are high so that in the case of reception, up to a certain [7] ‘“‘The gain of an idealized Yagi array”, D. G. Reid,
point, losses have no effect on signal/noise ratios since Journal of the IEE, Vol 93, Part IIIA, 1946, p546.
CHAPTER 4

Feeding the antenna

In the early days of radio, antennas often consisted of screen and is likewise prevented from radiating.
odd lengths of wire connected directly to the transmit- Although a balanced line is assumed for the following
ter but this results in radiation of harmonics and other explanation the principles apply equally to all types of
defects, making it unacceptable under modern condi- line.
tions. These difficulties can be overcome and ‘‘odd bits The wires have inductance and there is capacitance
of wire”’ still have their uses but usually the antenna is between them so that the line can be represented by the
located in some ‘“‘best possible” position which may be equivalent circuit, Fig 4.2. If a dc voltage is applied to
a long way from the transmitter, 50-100ft (15-30m) the line the current starts to build up in an inductance
being typical and 200yd (180m) by no means exception- and as it does so the associated capacitor is gradually
al. Transmission lines, otherwise known as feeders, are charged up to the applied voltage, this process being
used to carry power from the transmitter to the antenna repeated all the way along the line. In terms of rf
and constitute an important part of the antenna system. signals this translates into a wave propagating along the
They can take various forms but there are two main line with a velocity given by 1/VLC, which for air-
types, balanced lines and coaxial lines as illustrated in spaced lines is virtually equal to the velocity of light.
Fig 4.1. If the line is filled with insulating material this
In the case of the balanced line the two wires carry increases the capacitance per unit length, thus making
currents in opposite directions, radiation being pre- it appear longer than it really is to a wave travelling
vented because the fields generated are of opposite sign along it, both the velocity and the wavelength in the
and the spacing is small enough to ensure that phase line being decreased by VK where K is the dielectric
differences (such as those which form the basis of the constant of the filling. Imagine now a transmitter
W8JK antenna) are negligible. In the case of coaxial connected to an infinitely long line in which there are
lines the outer conductor is at ground potential and acts no losses; energy leaving the transmitter will travel on
as a screen to prevent radiation from the inner conduc- for ever but the transmitter is aware only that ‘‘some-
tor. The return current flows down the inside of the thing” is absorbing power from it and interprets this as
a resistance. The value of this resistance is known as the
characteristic impedance of the line, and is usually
denoted by the symbol Zp.
The greater the value of the reactances which consti-
Open-wire line tute the line as illustrated in Fig 4.2, the larger the value
of Z,) which is simply the geometric mean of these
reactances, ie Z) = V L/C and is therefore also inverse-
Generator ( a) Load
ly proportional to VK. Apart from the effect of K on
C, both L and C are determined solely by the

Coaxial line \

Generator >= (b) Load

Fig 4.1. Generator connected to load via transmission line. Vg, Rg

and R, are as in Fig 3.13. In case (a) radiation is cancelled since the Fig 4.2. Equivalent circuit of balanced line
two wires, viewed from a distance, produce equal and opposite Velocity = 1/\/(LC) metres/second
fields. In case (b) the outer conductor acts as an earthed screen to 2) = V(L/C) ohms
prevent radiation from the inner conductor. Transfer of power
from generator to load is unaffected by the presence of the line, where L is in microhenrys per metre and C is in picofarads per
provided this is matched as explained later metre
 FEEDING THE ANTENNA 31

dimensions of the line and are given for two-wire

open lines by the formulae
L = 0-92 log), (2S/d) microhenrys per metre
A A

—= log.)
1205(25/d)
Ks picofarads
cofarad z zs ris £ j
per metre {

(a)
from which Z, (ohms) = 276 log;, (2S/d) for air-spaced
lines (K = 1), S/d being the ratio of spacing to diameter
of the conductors.
Similarly for concentric lines
L =0-46 log.) (D/d) microhenrys per metre
C = 24-1K/log,, (D/d) picofarads per metre
Zo = 138/(log;9 (D/d)) ohms
where (D/d) is the ratio of the diameter of the outer
conductor to that of the inner conductor and the line is
air spaced, but since concentric lines are usually filled
with insulating material it is necessary to divide the
value of Z) obtained from the formula by VK as
explained above. Design charts based on these formu-
lae will be found on p37, methods of obtaining awk- Current
ward values of impedance by the pairing of lines being
Fig 4.3. Lines with total reflection from the far end (a) open-circuit
also featured. (b) short-circuit. /; represents the amplitude and phase of the
The next step is to imagine that at some distance from forward current wave and /p that of the reflected current wave.
the transmitter the line is cut. What has been removed Zero current at the end of the line means that the waves are in
opposite phase at that point. Phases are indicated at }/4 intervals
is still an infinitely long line which looks like a resist- moving back down the line towards the transmitter, using the
ance Z, so that if it is replaced by a resistance of this convention of clockwise rotation to indicate advancing phase.
value the transmitter is unaware of any change; all the Moving closer to the transmitter the forward wave arrives sooner
and is phase advanced whereas the reflected wave has further to
power going into the line is absorbed in this terminating travel and is retarded. (c) shows the sinusoidal voltage and
resistance and the line is said to be matched. If now the current distribution along one wavelength of open-circuited line
resistance is replaced by an efficient antenna designed
to look like Z, all the power previously going into Zp two reversals in all so the two waves are always in phase
will be radiated from the antenna. at this point and there is a current maximum. Another
At any point in the matched line the voltage and \/4 back down the line the waves are again out of phase
current are in phase, the ratio V/I of the rms values is and the whole pattern repeats itself at \/2 intervals.
equal to Z, and the power flow is therefore given by VJ, In the short-circuited case, Fig 4.3(b), the voltage is
I?Z, or V7/Z,. These expressions are identical, though zero at the point of reflection, the short-circuit being
sometimes one or the other may be more convenient. (as one might expect) a point of maximum current, so
If, on the other hand, the line is not accurately that the direct and reflected waves are in phase at this
matched, some of the power reaching the termination is point. This is perhaps best appreciated by replacing the
reflected back down the line. To understand this it is short-circuit with a low value of resistance which, in the
helpful to start by considering a total mismatch, the line absence of losses, absorbs all the available power (P).
being either open-circuited as in Fig 4.3(a) or short- The usual formula P = /?R = V’/R is applicable at all
circuited as in Fig 4.3(b). On reaching the discontinui- points in the line and this means that the smaller the
ty, the wave finds itself with nowhere to go and, since it value of R the closer J gets to infinity. It follows also
cannot just disappear, has to return back down the line that a low value of J implies a large value of R and thus
in which case it is said to be reflected. a large voltage so that voltage maxima coincide with
Considering first the open-circuit case; there is no current minima and vice versa. Although in the short-
current at the end of the line so that at this point the circuit case there is no phase reversal on reflection, it is
forward and reflected waves must be equal and oppo- only necessary to move backwards down the line for a
site. Next let us stand back A/4 from the end of the line distance /4 to find the two waves out of phase so that
and observe (a) the forward current wave, and (b) the with 100 per cent reflection 7/=0 and the situation
reflected current wave, which left the transmitter two precisely duplicates that existing at the end of the line
quarter-cycles earlier and is now on its way back. in case (a).
Because of phase reversal at the reflection there are It may be noticed that consecutive maxima are
32 HF ANTENNAS FOR ALL LOCATIONS

opposite in phase, a matter of no great importance greater than Z, there is a voltage maximum at the end
when working with lines but useful to remember later of the line and, if less, a current maximum.
when applying the same ideas to antennas. It will now The foregoing theory predicts infinitely large cur-
be evident that there is a pattern of current and voltage rents and voltages if R = 0 and there are no line losses.
along the line with the maximum and minimum values In practice there is always some loss as explained below
of each not moving along the line but always occurring but very large currents and voltages can be experienced
in the same positions. This is therefore described as a under fault conditions, and a short- or open-circuit can
standing-wave pattern, the voltage and current waves result in damage to feeders, meters, transmitters and
being separated by A/4 as in Fig 4.3(c). Although this other components. There can even be a fire hazard as
implies a phase difference of 90° between adjacent the author found on one occasion when a short-circuit
current and voltage maxima, it must not be confused developed in an open-wire feeder, causing high voltage
with the much more important relationship between to appear at the lead-in which consisted of plastic-
the current and voltage at any given fixed point on the insulated mains cable pushed through holes in a win-
line, which obeys the normal rules of ac circuits and can dow frame! The insulation broke down at this point,
be worked out as explained below. resulting in fairly impressive fireworks and thus empha-
In the most usual cases the line is neither perfectly sizing the need for adequate insulation of lead-in or
matched nor completely mismatched, the load resist- indoor wiring if there is a chance of high voltages
ance being different from Z, and also having some developing under fault conditions.
reactance associated with it. On the face of it this may Similar considerations apply to resonant lines, this
seem a bit complicated but the difficulties can be being a description commonly applied to any line
resolved by proceeding in two stages, considering first a operating with a large mismatch. Such lines are often
mismatch involving resistance only. It is then found used as explained later to allow the use of a non-
that the effect of reactance can be translated quite resonant antenna, in which case tuning of the antenna
easily into an alteration in the value of the resistance plus its feeder is carried out at the transmitter and
plus a shift in the positions of the maxima and minima. relatively large currents and voltages exist in the line.
Still looking at Fig 4.3, let us suppose a resistive The current in a resistance must by definition be in
mismatch such that /, is less than J;, say 0-5J;. It will phase with the voltage, and this condition exists at all
perhaps be obvious that maxima and minima occur as points along a matched line, at a resistive load in the
before and in the same positions, but now the max- case of a mismatched line and, moving back towards
imum value is 1:5/; and the minimum is 0-5/;. the transmitter, at intervals of 4/4 from a resistive load.
The ratio of maximum to minimum, in this case 3-0, At any given intermediate point there will be some
is defined as the standing wave ratio (k) and I,/I; as the other fixed phase relationship between forward and
reflection coefficient (r). Putting this into more general reflected waves depending on the relative distances
terms travelled. This causes both amplitude and phase to vary
along the line as shown in Figs 4.4 and 4.5.
k
Ce: =
It will be noticed that phase shift along the line is
~l-r
highly non-linear, and at the intermediate points there
This of course is always greater than 1 unless the are substantial phase differences between the voltage
matching is perfect, in which case r = 0 and k = 1. and current waves. These differences, which increase
Sometimes, mainly in microwave work, the inverse with the reflection coefficient, make nonsense of the
expression is used so that k is always less than 1 but in widely-used assumption that phase shift is directly
such cases the need for inversion to comply with normal proportional to line length, although this has often
amateur usage will be obvious. Since the power flowing been employed in the design of two-element driven
along the line must be the same whether it is observed arrays. In these systems both elements are connected to
at points of maximum or minimum voltage and the the transmitter and it is usually assumed that a differ-
power is always equal to V7/R, it follows that the ence in feeder length of A/8 produces a phase shift of
resistance seen looking outwards at a voltage maximum 360/8 = 45° or pro rata. This is despite the fact that it is
must be k? times its value at the minimum. difficult and often impossible to guarantee accurate
This condition is met if matching of both elements, one of which (as explained
in Chapter 5) can have a value of R which is zero or
= 7H0 =p!E ie Z)= VRuR even negative!
The standing wave pattern with a moderate mis-
which is the geometric mean of the two resistance match is shown in Fig 4.5(a), the ripple being nearly
values, denoted here by Ry and R,. This relationship is sinusoidal in shape, whereas with a greater degree of
basic and may also be expressed in the form Ry = mismatch it takes on the appearance of Fig 4.5(b)
Z°/R;, which allows lines to be used as transformers which closely resembles a plot of the amplitude of a
as described on p35. Note that if the load resistance is sine wave.
 FEEDING THE ANTENNA 33

A
4 a A4 ~ A
4
A

Time

| ‘ Distance

=== Phase of current |

= | : wave, swr=3 +
rH 150 "| V ; : Ci 4
® maul Sy Phase of current wave ‘
c | we for total reflection
A ay iA (swr=OO) -
2 ‘ :
v : tas ~ Phase of voltage (b) 0
a 04 <a wave, swr=3 ~
Ps ~
. Perfectly : Hf | Fig 4.5. Standing wave pattern corresponding to Fig 4.4(a)
a .—Matched line Nese: & 2 shown over a full wavelength for two cases: (a) a nearly matched
a (phase of both = line, (b) a short-circuited line. Note that there are two maxima
current and voltage) and two minima in each wavelength
50-- ———
and minima occur twice per wavelength along the line.
It follows that as the incident wave moves along the line
at the velocity of propagation, the reflected wave vector
rotates relative to it at twice the signal frequency.
Fortunately this harmonic frequency does not exist as
Fig 4.4. (a) Development of the standing wave pattern due to such, the situation being analogous to that of two trains
interaction of the incident and reflected current vectors. Variation
of line current with distance is derived directly from the variation approaching each other at SOmph (80km/h) and passing
with time of the two vectors. At any given point on the line the therefore at a relative velocity of 100mph (160km/h). In
relative phase of the vectors is fixed, but as the incident wave the same way the incident wave travelling outwards
travels along the line the relative phase changes with time at the
rate of 2wt. (b) Variation of phase along the line corresponding to along the line ‘‘sees”’ the reflected wave rushing back
the current wave in (a). The dotted line shows the phase variation past it at what appears to be twice its own velocity. In
for the voltage wave. Phase variation is also shown for a perfectly most cases there is no need to think in terms of
matched line and a completely mismatched line
reflection coefficient since swr meters are readily avail-
It is important to note that the voltage pattern, able or easily constructed, and swr is related to the load
though similar in other respects to the current distribu- impedance R by the simple formula k = Z)/R or R/Z,
tion, is displaced from it by A/4 so that voltage maxima whichever is larger.
coincide with current minima and vice versa. This It is important to appreciate that power reflected is
follows from the facts that voltage must be zero at a not lost. The process of ‘‘matching”’ the transmitter to
short-circuit where the current is a maximum, and the antenna system ensures that in the absence of
current must be zero at an open-circuit. The voltage feeder losses all the available power goes into the
standing waves are shown in Fig 4.5 by dotted lines. antenna if only because it has nowhere else to go!
From Fig 4.3 or Fig 4.5 it will be seen that the same Matching of the transmitter to a mismatched line is
relative phases exist at any two points along the line achieved as described on p28 and can be regarded as a
separated by half a cycle, which means that maxima kind of deliberate and carefully organized mismatch
34 HF ANTENNAS FOR ALL LOCATIONS

component of the antenna impedance resonates with

the reactance of the line which therefore behaves as if it
were terminated by some value of resistance R, at the
point P. This will be a position of maximum voltage or
current, and in the case of high-impedance lines can
easily be located using any convenient rf indicator such
== as the current probe described on p237.
If matching cannot be achieved at the antenna itself,
the connection of a transformer at P or appropriate
matching stubs can be used as explained on p41 to
achieve a low value of swr in the feeder between P and
dia
Sap)
Ss)
PIR
eee the transmitter. If P is not accessible a point A/2 closer
to the transmitter can be substituted. The actual value
(dB)
LOSS
REFLECTED of R, may be obtained from the swr since k = R,/Zo
or Z)/R,. Conversely, having determined P and k or
R,, the value Z, of the antenna impedance may be read
soy
BA
3]
|e
a
3 8 ee
ae
Lee
from a Smith chart and this knowledge used to achieve
better matching at the antenna as explained on p60.
In the case of low-impedance lines the determination
of P or Z, is more difficult because there is no access to
Fig 4.6. Curve showing loss due to mismatch when this is not the centre conductor but, given a measurement of
compensated at the input end of the line
impedance at the transmitting end of the line, Z, can be
which ensures that the reflected power is re-reflected read from the Smith chart in this case also. A bridge for
and thus sent back to the antenna, the process being impedance measurement is described on p239.
repeated until all the power has been radiated. This is Despite a somewhat fearsome appearance the Smith
fortunately not a procedure which has to be carried out chart is basically simple and has so many applications
consciously by the operator but a description of what that a little time devoted to its mastery with the help of
actually happens when “knobs are twiddled for best pp56-65 is well spent. In the meantime it will be
results’. sufficient to note the characteristic behaviour of lines
On the other hand, if a transmitter is adjusted to with different reactive terminations as illustrated in
work into a matched line and a subsequent mismatch is Fig 4.7.
not corrected by re-adjustments at the transmitting end
there will be a mismatch loss given as a power ratio by
4r/(1 +r) as plotted in Fig 4.6. Misinterpretation of Line losses
this formula may be responsible for the prevailing There is always some loss in a transmission line due to
tendency to overemphasize the importance of low swr, the ohmic loss of the conductors and losses in insulating
though changes in swr due to variation with frequency materials, these being proportional to the square of the
of antenna impedance can make it impossible to tune current and voltage respectively. Since increased swr
over the band without the complication of having to means that parts of the line experience higher currents
rematch. In general, values of swr up to 2 or 3 have little and voltages, any losses are correspondingly increased,
ill effect, although very high values may produce the though if small enough to start with they may still be
side-effect just described or an unacceptable amount of negligible.
additional loss in accordance with Fig 4.9. It is true that The effect of line losses is illustrated in Fig 4.8 which
some transmitters are not designed to cope with impe- shows how losses cause the reflected wave to be weaker
dance variations but these can be corrected by the use at the transmitter where the forward wave is strongest.
of an external tuning unit (p54). Thus the swr measurement at the transmitter may be
much less than the swr at the antenna. Measurement of
a low swr at the transmitter therefore does not neces-
Reactive load
sarily indicate a good match since it could also be due to
As mentioned earlier the line termination is likely to a large line loss. The author has found that such a loss
include some reactance. For example, if an antenna is can occur due to radiation from the line in the absence
tuned and matched at ground level and then raised into of a balun, even though the measured loss in the line
its operating position there will almost certainly be when this is disconnected from the antenna may be
some change. If this proves unacceptable it is important negligible.
to know how to correct it. This presents no great Such effects can be highly elusive and an arrange-
difficulty since there is always some point P along the ment which works well in one set of circumstances may
line, within A/4 of the antenna, where the reactive fail badly in another. To insure against this, careful note
 FEEDING THE ANTENNA 35

R! (>Zo)

Fig 4.7. Behaviour of lines with ee 4A 4 Bie ie 4 |

a Ae
4
ae
typical resistance and reactive |

terminations. The figure shows |

the sign of the reactance seen From |
|
looking outwards along the line a |
!
from positions indicated. When transmitter ()
1
the termination is partly reactive I
there is always some line length 1
I
PQ, less than 4/4, which results . 7 |
|
in the equivalent of a resistive \
!
termination A’. If R' (or R) ex- + | R>Zo
ceeds Zp the sign of the react- !

ance varies along the line as |

| Equivalent
shown by the top figure; if low- | R series
er, by the lower figure. With ! .
terminations
.

large swr the line presents a

|
reactive impedance equal to Zp |
in the centre of each A/4 interval; 7 |
|
this can be assumed as an |
|
approximation (error less than |
10 per cent) if the swr is less 1
!
than 3. Series/parallel equiva- |
1
lents and length of the line PQ |

can be obtained from the Smith |

|
chart. This figure may also be |
applied to centre-fed antennas
by regarding them as opened- '
i
out lines (p60) + |
!

should be taken of the precautions listed on pp49-51. R! (<Zo)

Additional line loss caused by a high swr is directly
related to matched-line loss and can be obtained from
Fig 4.9. Fig 4.10 shows the reduction due to line losses
of the swr observed at the transmitting end of the line. of 37:-5Q could be matched to 150Q by using a A/4
Short lengths of line have many uses as transformers, section of 75Q impedance, and this in turn to 600Q by
resonators (eg in lieu of tuned circuits), capacitances, using a i/4 section of 300Q line. These methods unfor-
inductances and matching stubs, some of which are tunately are not easy to adapt for operation on more
described below. than one frequency band.
The explanation has been simplified by considering

Lines as transformers Ij Zo
It has been shown that if a line is terminated by a
resistance R this appears as a resistance Zj/R when LOSS=3db

viewed from a distance of 4/4 down the line. It follows

that to transform a resistance R, to some new value R3,
all that is required is 4/4 of line having Z) equal to
V(R,R>). For example, to match a A/2 dipole having a INCIDENT
radiation resistance of 73Q into a 300Q feeder CURRENT
(MAGNITUDE
Zy = V300 X 73 = 1482 ONLY)

Line of this impedance is not readily available but can REFLECTED

be contrived by using two lengths of 75Q line as CURRENT

explained on p37. ONLY)

Approximations to other likely impedance values can
be similarly achieved using pairs of lines available on REFL COEFF 0-5 0-7 1-0
the market or readily constructed. If necessary several -—--—— tb ——————oH4
SWR——53:1 5°73] co
sections of line having characteristic impedances Zp),
Zo, Zo3 etc can be used to obtain a required trans- Fig 4.8. Effect of line attenuation on incident and reflected
formation in several stages. Thus an antenna resistance currents shown for the extreme case of a short-circuit
 36 HF ANTENNAS FOR ALL LOCATIONS

-)
i
i)

1-0
Nate
PEN
et
are
[esa
ZS.

: =
7 tea

0-5 Gel
NAIR
BRANNAN
MERGER
aes
SWR=2
0-4
So ° ie)

BAZAN
SNe
RUEN
OS
SE
Bs
Pd
iT
SS
MSL
ToTy
Io
cleat
oa
INGEN
Nol
oS nN
TNIN
KENNA
ANCE
IN
[as
ROLT
ERO
ey
eee
GLUINIGAN
Naame ae
i
Bak
Gey antenna
at
SWR

“|
Sa
x 0-2
INTENTS
NINU
AAAS
AC
AMA
aN 4 5 678 10
ONE Loss iN db WHEN MATCHED

Fig 4.9. Additional line loss due to the indicated values of swr, for
any given matched-line loss. The swr is the value measured at the
load, swr at the transmitter being reduced in line with Fig 4.10.
These losses are due to extra heating of the line with increased
voltages and currents, and will be incurred in addition to the
matched line loss regardless of whether or not the generator
matches the line

1 2 3 4 5) ao 8 10
SWR at transmitter

Fig 4.10. SWR at input end of transmission line versus swr at load
for various values of matched line loss (ARRL Antenna Book)
h|> aa “Tt b|> aX

Line impedance mo resistive impedances only, but the transformation is

R sag =
equally valid for reactances and can be expressed in the
A B A 'B more general form
Z> = ZWZ,
a a wees!
t 8
x
| 8 Xi
This involves quite complicated algebra except that
R = Zo? once again the Smith chart comes to the rescue by
providing answers at a glance.
(c) There is a very close analogy between the action of
Fig 4.11. Comparison between lines and tuned circuits. Left-hand lines and that of tuned circuits. Thus, referring to Fig
figures contain series resistances (R <<Zp or wL), right-hand side
shows equivalent parallel resistances. At (c), half the line has
4.11, the resistance R in series with the coil is equiva-
been replaced by a capacitor having the value X = Zo, but any line lent to a resistance X’/R in parallel, and the Q of the
length less than )/4 can be used subject to appropriate choice of coil (which is equal to the voltage step-up ratio) is X/R
capacitance. As the line length is reduced below 1/8 it resembles compared with Z,/R for the line. A short-circuited line,
a lumped inductance more and more closely so that the distinc-
tion between (a) and (c) rapidly disappears. At (b), AB illustrates often referred to as a stub, can be used in lieu of a tuned
the effective contraction in line length due to a reduction Af in circuit or, if this is not convenient, a shorter line may be
frequency, the ratio of AB to 1/4 being equal to Afff. The lowest
possible value of R is the self-resistance Ro of the coil or line but
used with additional capacitance to bring it to reso-
any of the arrangements shown may be used for transforming nance. The advantage takes the form of higher Q
larger values of R up to the values indicated on the right (ie lower losses) since usually a larger diameter of
 FEEDING THE ANTENNA 37

500

400 s
5 ia
N 300 bead
He
vt Ee
—

“te
100 a

im th 2B 10 20 S07 A050 100 200 300 500 1000

( a) Ratio = or s

200

~ LEER Zo

Ries prior ie
Characteristic impedance 2 balanced

©
oa at fof eet ce
ng i pase:

LATTE Te LEP oT
a Se oncerrT||
PECTIC a
A EEE
4-wire open-wire line viewed from end-on
Zo x 3000hms

:a eer TTI CATT Fey)

3 4 2
(b) Ratio 2 or ~s

Fig 4.12. Chart giving characteristic impedance of concentric and two-wire lines in terms of their dimensional ratios. Air insulation is
assumed so that the values of Zp) must be divided by the square root of the dielectric constant, ie by the velocity factor which is usually
about 0-67 for coaxial lines. Even for open-wire lines it is slightly less than 1-0 due to spacers and end effects, and can be as low as 0:9.
Methods of obtaining awkward values of Zp by using lines in pairs are shown at (c)

conductor can be used and the “proximity effect” five per cent would move the short-circuit from A to B
between adjacent coil turns is avoided. and the portion of line AB would present at A the
inductive reactance appropriate to a line whose length
is five per cent of /. Similarly, a decrease in frequency
Lines as reactances of five per cent would result in a capacitive (ie negative)
Reactances making use of short lengths of transmission reactance of the same amount which is given for both
line have many uses, one of which has just been cases approximately by
described, and the physical resemblance of very short
lines to single-turn coils (if the line is short-circuited) or +X = 2nZy.
small capacitances (if the line is open-circuited) is fairly
obvious. Somewhat less obvious but equally useful is ‘“‘A” indicates that the quantity of interest (AJ) is not
the fact that in the case of a resonant line as in Fig the line length / itself but a small change in /, so that
4.11(b) any change in frequency is equivalent to insert- starting for example with a resonant line of length 4/4
ing a short length of line where previously there was a and subjecting it to one per cent change of frequency,
short-circuit. Al/X becomes 1/400.
In the case of Fig 4.11(b) a decrease in frequency of For lines, Z) is obtainable from Fig 4.12, and for
38 HF ANTENNAS FOR ALL LOCATIONS

where C is in picofarads, / in centimetres, and v is the

velocity factor of the feeder, ie 1-0 for air-spaced lines
and approximately 0-66 for coaxial cable.

a Hee
Bs (12 swg)
reali Antennas as transmission lines
The idéa of taking a resonant 4/4 line and opening it out
- 20m dipol
to form a i/2 dipole can be extended to any type of
peal (tubing oon dia) antenna which can then be regarded as a balanced line
Characteristic
(Zo)
Impedance if fed symmetrically or, if fed at one end, as an
ee 2m dipole
unbalanced line with the ground providing the return
Z— (tubing 2cm aia) | path. Although Z, is increased and radiation resistance
100 500 1000 5000 10000 50 000 100 000 is acquired, all transmission-line rules are still applic-
Length/radius of wire
able with the radiation resistance providing the ter-
Fig 4.13. Characteristic impedance of straight antenna wires. mination. It will be recalled however that R is usually
Points marked on the curve are typical examples. (Note that the defined with respect to a current maximum and there-
addition of 1202 gives the Zp) of parallel wires spaced by the
length of the antenna; this is useful as a rough approximation in fore appears as a large resistance Z,2/R at the end of the
calculations involving antennas with bent ends) line, assuming this to be open-circuited as in the case of
a dipole. The larger the value of Z) or the smaller the
antennas from Fig 4.13. A particularly interesting
value of R, the larger the end voltages and the greater
aspect of this example is its application to A/2 dipoles,
any losses due to the close proximity of trees and
which can be regarded as 1/4 lines opened out, thus brickwork. This effect may however be partly offset
decreasing the capacitance and increasing Zp) in typical
due to the reduction in capacitance resulting from
cases to about 1,000@. For this example X comes to
increased Zp). Unfortunately, in this respect few guide-
15-7Q and for equality with the radiation resistance of
lines are available.
73Q it has to be multiplied by 4-65. This corresponds to
It is possible to terminate antenna wifes in their
4-65 per cent of detuning which is the half bandwidth of
characteristic impedance so that no standing-wave pat-
the antenna as explained in Chapter 7. It is also the
tern appears and the bandwidth is increased, but in the
condition for an swr of 2-0, assuming the line to be
case of very short wires most of the energy is then
matched at resonance. The formula can also be used for
absorbed in the termination instead of being radiated.
finding the reactance of a short length of line which is
A more interesting situation arises however with long
short-circuited at one end, in which case the actual
wires, since the energy absorbed is that which would
length / is substituted for Al.
otherwise have been radiated in the back direction
The formula is accurate within 10 per cent provided
(p95), and an amateur fortunate enough to have a few
Al/h is less than 4/12 but for greater accuracy of larger
acres of ground at his disposal can use a number of such
values of A/ the exact formula should be used, ie
wires to form V and rhombic arrays providing high
Al gain in a small number of fixed directions.
X = Z, tan 2n
d By regarding antennas as transmission lines many
mistakes can be avoided and anomalies explained.
Assuming access to a pocket calculator this is just as
Thus many of the antennas described in later chapters
convenient but the simpler formula can be taken as 162
have been folded to reduce their overall length, a
for each per cent of detuning per 1,0002 of Z, and has
procedure which nearly always results in a bad mis-
proved invaluable while doing odd jobs on the antenna
match unless it is realized that the usual length formu-
farm or working out some new idea in the bath! If it is
lae are no longer applicable. For example, if the
supplemented by remembering that X = Z)if Al = 4/8
antenna is shortened by folding in the centre to form a
the need for the “‘tan” rarely arises.
stub, reactance is removed at high Z) and replaced at
For an open-ended line the following exact formula is
low Z, so that the length of the stub must be increased
applicable
in proportion to the difference in the values of Zp.
> —Ly Let us suppose for example a reduction in length of
~ tan(2Al/n) 2m in the case of a A/2 dipole for 14MHz (A = 21m).
Remembering to picture the antenna as a line which
Again the “‘tan’”’ can be omitted if A//d is less than A/12 has been opened out, this amounts to a reduction in
but the main use is for calculations of capacitance and line length of 1m so that, assuming a typical value of
for short lines it can be rewritten in the form 1,000@2 for Zp, a reactance 27(1,000/21)Q has been
removed. Since Z, for the stub is only about 600Q this
33.1 must be increased in length by 1,000/600 or 1-67m to
compensate for the reduction in antenna length. This
 FEEDING THE ANTENNA 39

ro|» Y exist at the lower end of the feeder into the value
required for correct loading of the transmitter.
ea 722 twin-lead For frequencies up to 30MHz any length of antenna
(or coaxial cable plus 1:1 balun) from about A/2 upwards may be centre fed with up to
100ft (30m) or so of open-wire resonant line as in Fig
4.14(e) without incurring significant losses, all adjust-
ments being carried out in the shack. Such an arrange-
ment has the advantage of being ‘“‘multiband”’ since it
allows a 3-SMHz dipole to be used at any of the higher
60082 line frequencies, although with a long feedline the adjust-
ments in the shack may be extremely frequency-

|
sensitive. The swr is 8-0 in the case of a /2 dipole,
dropping to 4-0 or less when the same dipole is used at
(c) the higher frequencies.
1
! 50 or 72 $82 coaxial cable
Losses if any are predictable in accordance with Figs
'
3.10 or 3.11 in conjunction with Fig. 4.9. By way of
8 example, 140ft (43m) of 600@ line made from l6swg
(1-6mm) copper has an rf resistance at 283MHz of 3022,
(d) te corresponding to a loss ratio of 630/600 or 0-22dB
‘ 30082 twin- lead under matched conditions, the additional loss for an
(or coaxial cable plus 4:1 balun) swr Of 8 being 0-5dB. The total loss, 0-72dB, is less than
the typical loss of 1-5dB for the same antenna fed with a
Almost any length
matched high-grade coaxial line as in Fig 4.14(a).
a ay. Though expensive and less efficient the coaxial line is
usually considered to be more convenient, and in the
(e) Open-wire
600 or 30082
line
case of a dipole works also on the third harmonic, eg a
but not 300 twin 7JMHz dipole operates also on 21MHz although (as
explained in Chapter 6) the radiation pattern will be
Fig 4.14. Methods for centre-feeding antennas: at (a) and (e) a different. Some types of multiband dipole (pp108, 110)
series connection is used and at (b) to (d) the antenna is shunt (or may also be fed in this way but it is important in the
parallel) fed
case of an unbalanced feeder to use a balance-to-
unbalance transformer (or balun for short, pp51—54)
means a nett increase of 1-34m in the total length of at the antenna. These devices, as described later in this
wire in the antenna system. chapter, are extremely efficient as well as simple to
construct.
In the case of parallel-feed systems such as Fig
Matching the antenna to the line 4.14(d) it has been found convenient to match the
Usually the antenna is arranged to be resonant with the antenna to a 200 or 300 impedance, in which case it
help if necessary of tuning devices such as the stub in may be fed with 50 or 72Q coaxial line plus a 4:1 balun.
the previous example, which is strictly speaking not Alternatively it may be fed with 300Q line or with an
part of the antenna since it does not contribute to the even number of half-wavelengths of 600Q line, open-
radiation. In most cases the feeder is matched to the wire line being more efficient even with an swr of 2 or
antenna at a point of maximum current, mainly because 3. In these cases the balun is still required for present-
of the low impedance of available transmission lines, ing the transmitter with a 50 or 72Q impedance but can
although it also helps to reduce losses by keeping points in many cases be located in the shack so that it does not
of high voltage away from supporting structures. need weatherproofing, the author’s personal prefer-
Many antennas have a radiation resistance which ence being for a combination of methods.
matches the impedance of available types of transmis- Thus if 600Q line is used in lengths which are
sion line so that a simple series connection can be made multiples of 4/2 at 7MHz (about 62-65ft (19-20m)
as in Fig 4.14(a). In other cases a parallel connection as physical length) it will be similarly resonant at 14, 21
in (b), (c) or (d) may be necessary. Often, however, and 28MHz. Any length deficit is made up by the use of
there is a case for the use of so-called ‘‘resonant”’ lines, coaxial line (<62ft) plus balun. With a 4:1 impedance
ie lines operating with a high value of swr as a result of ratio balance-to-balance transformer at the top end this
transferring the tuning and matching process from the arrangement can also be substituted for the low-
antenna to the transmitter or, more usually, to an impedance twin lead of Fig 4.14(a), the losses in the
external antenna tuning unit (atu or “‘transmatch’’). pair of transformers usually being less than the extra
The latter converts whatever impedance happens to losses in the twin lead or coaxial line which it replaces.
40 HF ANTENNAS FOR ALL LOCATIONS

Cc A p B D an extension of the 600Q line, the slight change in Zp

over quite a small length being immaterial.
It is unfortunate that the delta match, though ideal
for its purpose, has had a “‘bad press’ owing to a
Provided by the number of misconceptions. The case against it rests
capacitance between mainly on the belief that radiation takes place from the
VAG, BD
delta; moreover the shape of the delta is said to be
critical, though values have been published (and are in
most of the handbooks) for a single dipole.
/
/ \
\ There is no foundation for these beliefs. Consider
ne \

4
/ \
\
ws
first radiation from the delta, of which there are two
obi Se
components, the only significant one being aligned with
Lm cn]fenLm the antenna radiation and part of it! The other is the
‘“‘wide-spaced line” effect for which the formula given
Fig 4.15. Equivalent circuit for shunt-fed antenna. This is drawn later in this chapter (p51) predicts a loss of only
for the case of a T-match or balanced version of Fig 4.14(c), but
the principles apply to all shunt-fed systems. L represents the 0-25dB for a line with a spacing of 4/10 (ie 7ft (2-1m) at
inductance of AB, and the radiation resistance (though in fact 14MHz) and a length of at least 5A [1]. A delta match
distributed along the whole antenna) can be represented by an with typical dimensions of 3 by 4ft (0-9 by 1-2m) can be
equivalent resistance R in series with L which is tuned to
resonance by AC, BD so that a resistive impedance w“L“/R expected to produce a few hundred times less than this
appears across AB; in the case of a T-match this result is modified amount of radiation loss though there will be some
to take account of L,,, C,,, and the coupling between L,, and L increase if the line is mismatched.
As for the shape of the delta, one is assured that this
still works after experiencing such severe distortion that
Alternatively if an atu (or ‘“‘transmatch’’) is used in the it turns into a T-match so obviously there can be no
shack the 600Q line may be of any length, and this need to worry about minor shape differences! For a
would seem to be the simplest course if the antenna is fixed antenna the author prefers to drop the two wires
required to operate on the 10, 18 and 24MHz bands as straight down so that they only come together at
well as those just listed. ground level, thus obviating the need for spacers. This
In the case of beam antennas or short dipoles R may has been found to work perfectly, possibly helped by
be much less than 50Q and other methods of matching the fact that, calculating the dimensions in accordance
may be required. The gamma match, Fig 4.14(c), with the caption to Fig 4.15, the delta spacing is very
though popular, is not a balanced system and does not much smaller than the figures usually quoted: typically
necessarily overcome the problem of radiation from the about 44in (112cm) for a two-element beam with 1/8
feeder. Use of the delta match, Fig 4.14(b), or its close spacing at 14MHz. In the case of a rotary beam it has
relative the T-match, Fig 4.15, is preferable and with been found convenient to bring the sides of the delta in
one type of antenna to be described (p112) this proved to a point on the mast just above the beam rotator
essential. Any feeder impedance can be used with the which can be bypassed mechanically by splicing in a
delta match but with 600Q line the capacitors can be short length of 3002 twin line.
omitted. With the help of the Smith chart, matching Having thus, it is hoped, finally vindicated the delta
networks can be devised using inductance and capaci- match for high-impedance feedlines, it only remains to
tance to match any impedance to any other impedance point out that its advantages apply also to use in
but are usually restricted to single-band operation. conjunction with a 4:1 balun for feeding an antenna
Fig 4.15 has been included to provide a further from low-impedance lines. As for other alternatives,
insight into the operation of shunt-fed systems. This the gamma match has already been discussed to its
can function in one or more of several ways, for disadvantage, though this does not necessarily mean
example as indicated in the caption which provides a that it will produce large amounts of line radiation in all
good method of matching to a high-impedance feeder. (or even most) circumstances. If existing results are
Alternatively if C,,, is short-circuited, M = 0 and2L,, = entirely satisfactory and there are no obvious signs of
L then half the current flows through the matching feeder radiation (such as a skewed radiation pattern
circuit and the feeder sees an impedance 4R, but it has when observed using a field-strength meter sensitive to
been found that this can lead to reduced bandwidth. horizontal and vertical fields) it might be best to leave
It is evident that due to the large number of variables well alone* (p228).
there can be more than one solution, and that solutions Another method of obtaining an impedance
are not all of equal merit. Difficulties arise mainly in transformation is the folded dipole (p147). This is
matching to low-impedance lines but there is no prob- * Reference [17] reports on a severe case of tvi cured by changing from
lem whatsoever in the direct delta match to 600Q line, coaxial line to 75Q or 600Q balanced line and reference [18] analyses in
Fig 4.14(b). The delta can in fact be regarded merely as greater detail the deficiencies of the gamma match.
 FEEDING THE ANTENNA 41

If the half-length of the antenna plus the distance AB

is an odd number of quarter-wavelengths the radiation
resistance R appears at B in the absence of the stub.
Moving towards the antenna by a short distance / to the
point C a capacitive reactance equal to Z)2m//d
appears in series with R, and this is shunted by the
inductive reactance of BC which, since the system is
resonant, must also have the value Z,2ml/\. This leads
to the equivalent circuit in Fig 4.16 and, since the
voltage across L is X/R times that across R, the
impedance R, seen looking across the line at C is given
by X*/R where X = Z,)2al/i. For matching R to Zp, in
X=(Zo.2n
&/) the >caseq Of? dasN2es.dipole.iweh require 4X.
V(RZo) = ViGa0% 600)'=.2099 2, so.«).that, </-=
209/(6:28 x 600) = 0-055 which is a stub length of
Fig 4.16. Stub matching of open-wire line. The antenna plus the 3-8ft (1-1m) at 14MHz.
line AB is resonant and matching is correct when
Provided the swr is not too low (in which case a stub
is hardly necessary) it is easy with BC disconnected to
determine approximately the points of minimum vol-
where R is the radiation resistance; instead of the shorted stub tage or current with suitable probes such as that shown
BC the open-circuited stub B’C’ may be used though this is much in Fig 18.12 (or in the case of voltage, a neon lamp), the
more likely to disturb the line balance
voltage nodes being separated by exactly 4/4 from the
mechanically awkward in the case of tubing elements current nodes. Having located point D in this way and
but can be a useful alternative to use of a transformer in calculated the length BC the stub is attached to the line
the case of wire elements. By use of multiple wires or ACD so that DC = BC.
unequal diameters a wide range of transformation For greater precision /; and J, should now be
ratios becomes possible [2], though this restricts opera- compared; if /,; is less than J, it means that the
tion to a single frequency. The delta match has been reactance of the stub is greater than that of the antenna
found more generally useful in these cases also. and therefore AB is too long. Moving the shorting strap
Methods which require the cutting of tubing elements at B until the currents are equal results in the correct
in the centre without some good reason should be length for AB, and if necessary C may now be moved to
avoided because it weakens the elements and compli- restore BC to the correct length. If the correct stub
cates the structure. It is better to use ‘“‘plumber’s length is one of the unknowns then, havitig located B, it
delight’’ construction in which beam elements are is necessary to observe the current at a point along the
electrically bonded to the supporting boom. Apart line exactly 4/4 from C. If greater than the current at C
from the mechanical advantage, this construction may it means that the line is terminated in too high an
reduce the risk of lightning strikes or static discharges impedance and so must be tapped further down onto
but requires careful attention to symmetry to avoid the stub, and vice versa.
currents in the boom which waste power and ‘“‘skew”’ It is possible in theory, and with sufficient care even
the radiation pattern. in practice, to use an open-circuited stub as shown
dotted, the distance DC’ being equal to DC and DB’ to
4. The problem with this is line unbalance which
occurs unless the stub wires are exactly equal in length
Use of matching stubs and symmetrical also with respect to any surrounding
Stubs may be connected at suitable points on a trans- objects, so that in the absence of some overriding
mission line to eliminate standing waves between these necessity it is much better to use the shorted stub CD.
points and the transmitter, though the swr on the It should be noted that having gone through this
antenna side of the stub remains unaffected. Some- procedure at ground level where everything is accessi-
times this can be done at or close to the antenna, but it ble the stub may be moved nearer to the antenna by A/2
often happens that a feeder is brought down to a point or, in appropriate cases, any multiple of this. Charts are
near the base of a mast where it is accessible, though available in many handbooks [2, 3] for determining the
there may still be a long run of feeder from there to position and length of stubs when the swr and the
the shack. A stub at this point can be used to improve positions of voltage minima are known. However, the
the average swr even though the antenna itself may not author has found the above procedure to be particular-
be accessible, and Fig 4.16 shows how this can be used ly simple and relatively foolproof, the only rider to this
to improve the “‘resonant line’’ feeder system described being the desirability of making sure that J, is less than
above. For multiband operation a different stub may be I,, thus guarding against the remote possibility of
used for each band. getting one’s lengths so badly out that BC finds itself
42 HF ANTENNAS FOR ALL LOCATIONS

----_ —-<—<--O-
connected to C’ instead of C, which has been known to ~ ~ .eO- - ~~ el

happen. Instead of an inductance resonating with a

capacitance one then has two equal inductances in ——>
parallel! The most obvious way to determine the I “Voltage
resonant length would be by means of a gdo and this is ee
~
Se eo
oO

a convenient method with a simple dipole. However, in

the case of a beam the resonant length is affected by the
presence of the other element or elements, there can be
more than one “‘peak’’ and the method is unreliable.
In the case of coaxial lines the inner conductor is not
accessible and a somewhat different procedure is
needed based on measurement of impedance, swr and
use of the Smith chart as explained in Chapter 18,
though the principles involved are similar.
(b)
Unsymmetrical feed
A generator of rf may in principle be inserted at any
point in a resonant wire, eg as shown in Fig 4.17(a). It -—.
~~

then ‘‘sees” the capacitive reactance of the left-hand

wire in series with the inductance of the right-hand wire
and some appropriate value of radiation resistance to
which it can be matched so that all is well. What is not
admissible is the hanging on of odd bits of wire at points
=~
_ OF es cs

(c)
where there is appreciable rf voltage as shown in Fig
4.17(b).
Capacitance between end sections
Although the absurdity of this is obvious in the case
of the left-hand sketch the equivalent arrangement on
the right is sometimes advocated. There are now in Equivalent circuit of (c)
L Cc
effect two antennas AC and BD, plus a tuning error,
which can be resolved by means of an atu or transmatch Inductance to the right of the centre
(p54) at the transmitter so that power will still be
radiated, though the system bears little resemblance to A B ——2e Cc
Oe eos oa
Fig 4.17(a). The pattern may be difficult to predict and mY

efficiency is likely to be poor, particularly if there is a /

long feeder run at low height. Vhs

tee
7

There is a way out of this problem as shown in Fig ve

4.17(c). Here the left-hand side of the dipole is tuned to na | eee Capacitive
resonance with a series inductance and the right-hand “55 oer A coupling

side with capacitance, thereby presenting the feeder

with a resistance load equal to the radiation resistance | “ (d)
referred to that point in the antenna system. This is
simply the usual value (which is referred to the centre
of a /2 dipole) multiplied by the square of the current Fig 4.17. Off-centre feed. (a) Small self-contained transmitter, off
ratio, eg 73 X 2 = 146Q for a dipole fed at 4/8 from one centre, produces normal current and voltage distribution. (b)
end. Due to the series resonance the voltage distribu- Hanging things onto the antenna wire produces chaos; this
includes the off-centre feed DB shown on the right. (c) Correct
tion is now as shown dotted, the voltage to ground at method of off-centre feed, the antenna being resonant as viewed
the end of the feeder being approximately zero as in the in each direction from the feedpoint. Voltage distribution is
centre of the element. virtually the same as at (a) but, turned on end, this can be
regarded either as a dipole or as a vertical monopole with a short
There is however still one pitfall, since the length BD counterpoise earth. (d) Difficulty arising if BD is an odd multiple
could be A/4 or some odd multiple of this. In other of }/4; voltage induced in BD from BC can result in unsymmetrical
words it could present an alternative low-impedance current distribution and feeder radiation. The effect is probably
negligible unless ABD is resonant
path at B so that the current would again be shared
between the branches. Similarly any voltage induced in
the feeder, say from BC, could result in the situation BD is equal to A/2 it looks like a dipole with one end
shown at (d); these effects are both avoided by exercis- short-circuited so that it “doesn’t work’’, and the
ing care in the choice of feeder lengths. For example, if unwanted response is thereby eliminated. This problem
 FEEDING THE ANTENNA 43

Counterpoise can be very short

is not confined to off-centre feed but can arise with any
system that is not inherently balanced, and will be / cs a Z at this point
encountered again in the general discussion of feeder
radiation on p49. ( a)
For the extreme case of “‘off-centre feed”’ we should
perhaps come down from free space to the earth’s 1 Link coupling to transmitter

surface and consider the Marconi antenna, Fig 6.19. Counterpoise (fairly long if possible)
Here the earth has replaced one half of the dipole and a
large proportion of the available power is usually cee eRe A
Low Z at this point

wasted in the resistance of the earth connection. To Lee (b)

overcome this, an extensive system of ground radials is
usually recommended, advice which if followed to the
letter would usually involve digging up not only the
whole of one’s own garden but those of several neigh-
bours as well. Link coupling to transmitter

This topic recurs in Chapter 10, but it is interesting to Fig 4.18. Methods of feeding ‘‘odd lengths of wire”. Provided the
observe in passing that if the arrangement of Fig 4.17(c) impedance is fairly high method (a) can be used with only a very
is rotated anticlockwise into a vertical position the short counterpoise. Sometimes the counterpoise has been omit-
ted, relying solely on the stray capacitance of the coil but this is
left-hand half can be regarded as the “‘earth connec- not recommended. (b) is recommended for low impedances; X
tion”. As such it can be further improved by using two tunes out the reactance (if any) of the antenna. The shorter the
or more short radial wires in place of AB, the effective antenna the more important it is to have a relatively long or
ideally a multi-wire counterpoise, though lengths of 4/4 should
“earth resistance’’ amounting to no more than the small not be exceeded and }/8 may be better in some cases. Depending
ohmic loss in the loading coil. on antenna length positive or negative X may be required as
Such an arrangement is often described as a counter- shown, this arrangement being very similar to Fig 4.17(c)
poise, a term which applies in general to any grouping
of wires insulated from earth and acting in lieu of an
earth connection; such groupings are normally In Fig 4.18(a) the impedance Z, to earth at the end of
arranged so that the wires generate opposing fields, as a the antenna wire is in series with the impedance Z, of
result of which little or no radiation takes place from the counterpoise, and it is desired to match the trans-
the counterpoise. mitter into the impedance (Z, + Z,). If the length of
This discussion thus far has failed to take account of the antenna is anywhere near a multiple of 4/2, Z, will
losses due to currents induced in the ground adjacent to be high and it is easy to make Z, small compared with
the antenna; these exist even in the case of dipoles and Z, So that it has little effect on the matching process.
may well be increased if advantage is taken of the Typically, at resonance Z, may be about 8,000 and a
shortening of AB in order to erect the antenna at the capacitance of a few picofarads to ground will suffice
lowest possible height. Dipoles can however also be for Z.. At 14MHz this can be provided by about 3ft
reduced in height by the use of end-loading (p155) and (1m) of wire clipped onto the picture rail but at lower
it is concluded that for most purposes the distinction frequencies or as one departs from resonance it be-
between ‘‘dipole’’ and ‘““monopole”’ has little relevance comes necessary to increase the size of the counter-
apart from such incidental differences in height. A poise. However, no real difficulty arises until the
height of only 1 or 2m should be sufficient for reducing antenna approaches the A/4 resonance or an odd multi-
losses due to earth currents to an acceptable level [6]. ple of this, in which case Fig 4.18(b) should be used.
A number of practical situations involving off-centre Matching can usually be achieved with one or other
feed will next be considered.* of the arrangements shown, with the matching circuit
providing extra selectivity for the rejection of harmo-
Feeding the “odd bit of wire” nics etc. It is advisable that there should be no direct
The random length of wire, end-fed, was one of the connection to the transmitter since this can result in
commonest antennas in the early days of amateur radio currents “wandering vaguely” to earth via mains wiring
and can still be very useful as discussed in Chapter 15. etc.
Fig 4.18 shows the method of feeding such antennas
and this can also be applied to vertical antennas which Ground-plane antennas
are often much easier to feed at their lower ends than in
their centres. These usually consist of A/4 vertical antennas grounded
by means of a set of radial wires insulated from earth as
* Long-wire resonant antennas containing two or more _half- in Fig 4.19. A single A/4 radial can be used but in this
wavelengths can be fed in the centre of any current loops. Such feeds
are asymmetrica! only in the physical sense and not appropriate to the case it forms a considerable part of the radiating system
present discussion. and seriously distorts the pattern. It is better to use at
44 HF ANTENNAS FOR ALL LOCATIONS

because the counterpoise as well as the antenna itself

becomes reactive when the frequency is changed,
thereby tending to halve the bandwidth as will be
evident from Chapter 8.
A loading inductance of 3uH can consist of a six-turn
coil of 3in (7-6cm) diameter and lin (2:5cm) long, in
which case a good match at 14MHz can be obtained by
overwinding with a single-turn coupling coil spaced
about 4in (6mm) from the main winding. The feeder
may be connected to the coupling coil so that no direct
connection is made to the antenna and, despite some
(a) capacitive coupling, radiation from the feeder should
be negligible provided resonant lengths of feeder are
avoided. On the other hand, loading stubs have lower
losses than coils, should need no weather protection,
and are better suited to the requirements of multiband
Fig 4.19. Ground-plane antenna. (a) Basic system; any number of operation, though less suitable for indirect coupling.
radials can be used but preferably at least two. (b) Typical The use of direct connection is subject to two rules:
installation with guy wires acting as radials. Radials are usually
/4 but see recommendations in text
first, the point of connection must be the true electrical
centre of the antenna system and, second, any current
flowing, say, in the outer of the coaxial cable due to
least two radials, three or four being more usual, and unbalanced coupling from the antenna must not be able
the antenna is usually fed with low-impedance coaxial to find its way to the inner conductor. The electrical
line despite the considerable risk of radiation from the centre can be checked by first exciting the driven
feeder as discussed on p49. The radials can be much element with radiation from another antenna which
shorter than the A/4 usually specified. The author should be spaced at least \/4. Using a simple current
obtained identical antenna currents and signal reports indicator (p237) the position of minimum voltage can
at 14MHz when comparing four A/4 radials with a set of be located by touch, ie one should be able to touch it
four 40in (100cm) radials or two 60in (150cm) radials, without affecting the current. This should be the lower
using in each of the latter cases a common loading end of the radiator. It may be found necessary to alter
inductance of approximately 3uH. the length of the radiator slightly, retuning the counter-
The size of counterpoise can be further reduced if poise. The feeder should be run straight down to the
necessary, the measured loss being only 1dB in the case ground and if possible buried.
of a single helix wound on a 30in (75cm) length of plastic If connecting the feeder upsets the electrical centre,
pipe. It should be noted however that, although the or if any current can be detected on the outer of the
height of the vertical antenna can be considerably coaxial cable (using for this check as much power as
reduced by top-loading as explained on p155, the lower possible), further measures are needed. It may be
value of radiation resistance means that any loss resist- sufficient to wind a few turns of feeder round a ferrite
ance in the counterpoise system becomes correspon- core or alter the feeder lengths, otherwise the outer of
dingly more significant. the feeder may have to be detuned by using a linear
For example, at 14MHz a 6ft (1-8m) vertical element resonator as described on p112. One method that has
with top loading has a radiation resistance when close proved satisfactory is the use of a series tuned coupling
to the ground of about 9Q, and if three are arranged as coil which is symmetrical with respect to the voltage
a close-spaced beam this may drop to, say, 4Q2. A Sft node, the coaxial feeder being connected to it via a
(1-5m) open-wire inductive stub will provide suitable balun as suggested on p163.
loading for a centre-connected counterpoise wire of These precautions are likely to be even more impor-
about 10-12ft (3-0-3-7m) in length and, if 12swg tant when full-sized (A/4) radials are used because these
(2-6mm) wire is used, it should be possible (from Fig couple more tightly into the feeder system. Though by
3.11) to get the loss resistance down to about 0-5Q. no means always needed, failure to adopt such mea-
Allowing for the smaller but still substantial currents in sures can lead to reduced efficiency, rf getting into the
the parasitic elements, the total effective loss resistance mains wiring and causing tvi, or electrical noise being
will be at least 1Q and the drop in overall performance picked up from the mains wiring.
roughly 1dB. A recommended form of counterpoise, designed to
Apart from any effect due to the possible reduction minimize coupling into the feeder and make the best
in loss resistance, the use of a ground-plane or counter- use of space, is shown in Fig 4.20. An existing system
poise results in some reduction of bandwidth compared using 4/4 radials need not be scrapped but should be
with a system using a direct ground connection. This is checked using a current probe (p237) for symmetry at
 FEEDING THE ANTENNA 45

Ca) bs

Fig 4.20. Compact two-radial ground plane (plan view); the

["] (b) :
loading inductance L is not essential but further reduces the size
and, by reducing the required lengths below resonance, ensures Fig 4.22. (a) Single-wire feed to }/2 dipole. (b) Long wire antenna
that both radials are capacitive and avoids any risk of imbalance identical with Fig 4.21 except that a )/2 dipole is used as the
due to phase differences. Current distribution in the radials is terminating resistance, and the earth is conventional. The feed is
marked by arrows displaced from the centres of the dipoles by 0-067)

Typical values range from about 500Q for a continuous

two frequencies. This is because current equality at one thin wire to about 150Q for a capacitively stretched
frequency does not rule out the possibility of a phase wire (p199). A suitable counterpoise for 14MHz might
reversal but, in such an event, the situation will be have a capacitance of 6pF (2,000) to ground tuned by
changing rapidly as the frequency is varied. a coil having a loss resistance of 1002. The loss would be
A further point which seems to have been generally less than 0-3dB and with three per cent detuning the
missed is the need for radials (if resonant) to be of reactance of the counterpoise would be only 120Q ie
identical length. If two 1/4 radials are respectively 6in less than Z,, so that swr should remain acceptable over
(15cm) long and short at 14MHz this will generate a the width of any band likely to be required. Because of
phase difference of about 60° between them and radia- the inherent asymmetry, balanced feeders present
tion (upwards) from the radial system will be compara- problems but use of coaxial line with its outer conduc-
ble with that from the vertical element. tor earthed by the counterpoise is feasible, provided the
More practical information on systems of this type line length is such that it does not also provide a low
will be found on pp164 and 195. impedance to ground.

Feeding long-wire antennas Feeder systems to be avoided

V- and rhombic antennas are _ high-impedance At this point the reader should perhaps be alerted to
balanced systems which can be fed with open-wire two very attractive-looking feeder systems which, de-
lines, but single-wire systems present a more difficult spite serious deficiencies, have survived since the early
problem unless one end can be brought into the shack days of hf radio, thereby casting some doubt on the
as described earlier. Such wires may be terminated in “test of time?:
their characteristic impedance as shown in Fig 4.21 and
this is also the impedance presented at the feedpoint. (a) Single-wire feeder (‘‘Windom’’ antenna)
A single wire of any length may be used as a transmis-
CaS escA.,
205410 B sion line with the ground providing the return path.
+ ae At least 2A a ae Such a line may be matched by connecting it to the
correct point on an antenna system, eg at a distance of
about 0-067A from the centre of a A/2 dipole which is
Loaded- L Artificial
thereby energized as in Fig 4.22(a). .
counterpoise
1 earth Before proceeding it is interesting to digress for a
earth
11502 moment and consider further the case of the long
|
! antenna wire terminated in its characteristic impe-
dance. As with a transmission line this removes stand-
Fig 4.21. Feeding of terminated long-wire antennas. The principle
is similar to that of Fig 4.17(c) but the antenna can be any length, ing waves but, as explained in Chapter 6, the resistance
though preferably at least 2\, and no tuning is needed. The absorbs the power which would otherwise be radiated
counterpoise can be replaced by a 1/4 “‘single radial”. The opti- in the back direction. This leaves the gain in the
mum value of R is about 500Q; this absorbs the back radiation
resulting in a unidirectional pattern as explained in Chapter 6. For forward direction of the wave unaffected, and if the
multiband operation “fans” of “artificial earths” or counterpoises wire is long enough a large gain is possible. The
may be used, having wires cut to length or inductively loaded as terminating resistance can take the form of the right
in Fig 11.14(c) for each frequency. The tuned circuit LC provides a
10:1 step-up in impedance, suitable values for 14MHz being point on a d/2 dipole, Fig 4.22(b), in which case the
L=1-86uH, C = 68pF power in the termination is radiated instead of being
46 HF ANTENNAS FOR ALL LOCATIONS

converted into heat. However, as the radiation from

the dipole is distributed over a wide angle, the high- j<— X —> =< xX >| + a
gain undirectional pattern of a long wire is not greatly
affected.
By this time the reader may be puzzling over the
similarity between (a) and (b). There is in fact no
difference, and the argument clearly implies a feeder JK X—>| xX oh + >
loss of 3dB in the case of the Windom, a figure in good
agreement with the author’s experience, though some (a)
i
of
allowance should also be made for losses in the earth +
connection. The loss will be reduced if the feeder is
very short but it is difficult to imagine a problem for
which the single-wire feed provides the best answer.
Some readers may feel this point has been laboured
unduly but, although the Windom is more or less
obsolescent, attempts are made to revive it from time
- aes
A

to time and the author is not aware of any previous

derivation of the basic 3dB loss figure.
(b) i 4 ae
(b) Zepp feed or ‘‘J-match’’—standard version
There is an urgent need for the end-feeding of some
types of antenna, particularly vertical dipoles or in-
verted ground-planes which are often difficult for jx
[>>
mechanical reasons to feed anywhere other than at
Antenna
their base. End-feeding of resonant horizontal anten- tuning unit
nas is also desirable in many cases, and although the
method shown in Fig 4.18 offers one solution there may
be problems in housing the components.
The ‘“‘obvious”’ answer in these cases is the Zepp feed
which has deep roots in the history of radio and an ent)
Cx. iC
honoured place in most of the standard textbooks. It is
a pity therefore that in its recognized form it “‘does not
work” [5].
To appreciate this, consider Fig 4.23(a) which shows
replacement of the tuned circuit of Fig 4.18(a) by a 4/4 waa
,—r
=

resonant stub, the antenna being made slightly longer

Fig 4.23. Voltage-fed antenna. (a) is very similar to 4.18(a) and
than )/2 so that it looks like a capacitive reactance in lengths are not important if a tuned circuit is used. If this is
shunt with a considerably larger value of resistance. replaced by a stub the lengths x, x must be equal or the line will
The counterpoise is made equal in length to the extra be unbalanced; x must not be too short, ie the reactance of C, in
the equivalent circuit shown at (c) must be small compared with
length of the antenna so that it has the same capaci- R, or again there will be unbalance. The arrangement at (b) ought .
tance to ground, and the system is balanced as illus- not to work but may do so after a fashion as a result of stray
trated by the equivalent circuit, Fig 4.23(b). It will be capacitances; this is the basis of the Zepp feed or J-match. At (c)
note that R, is equivalent to a resistance 4A, across C. R is the
seen that if the length of the antenna is reduced to 0/2, current-loop radiation resistance and Z, the characteristic impe-
balance requires a counterpoise length of zero as in Fig dance of the antenna
4.23(b), which means that the circuit is not completed
and the antenna is no longer energized. was in the process of clocking up yet another addition
With an LC type tuner, balance is not required so to a long list of antenna failures, most of which have
there is no need to get into this predicament, but since been attributed to use of the Zepp feed. To
suppose that as shown in the lower figures the tuner overcome this problem G6CJ proposed the addition of
circuit is replaced by 4/4 of resonant line. This has a \/4 stub as shown in Fig 4.24. Starting from the end of
basically the same properties as the tuned circuit except the antenna note that there are two current paths but,
that to prevent radiation from this line it is necessary to because current flowing into the new stub has i/2
ensure balance and this is a further reason why the further to travel, there is a phase reversal so that the
arrangement shown at (b) ought not to work. This is the voltage across the stub is balanced relative to earth.
traditional ‘“‘Zepp feed’’. It can be deduced from Fig 4.23(c) that if at the end
This problem was first pointed out by G6CJ in the of the antenna there is a voltage V across the impe-
RSGB Bulletin for December 1955 just as the author dance R, the Zepp feed will see a voltage 2V across
 FEEDING THE ANTENNA 47

addition to the capacitance between the wires, there is

also capacitance to earth from each wire separately,
thus re-creating the situation shown in Fig 4.23(a)
despite the absence of the counterpoise.
Just to remind the reader, we are still in ‘‘free
(a) space’’, having been left there in Chapter 3 in order to
simplify matters, and by definition the earth does not
exist. If on the other hand we come down to earth, one
might intuitively expect the Zepp feed to work after
all, provided the wires are far enough apart and close
enough to the ground to ensure that currents induced in
the ground by one wire are not all cancelled by equal
cee
Vv
and opposite currents induced from the other.
It has in fact been found by the author in two cases
Fig 4.24. Unbalanced antenna—balanced line that the balancing stub, Fig 4.24, was not essential but
in each of these cases the wires were spaced about 7in
4R.. Hence the swr tends to be rather large with (18cm) and only some 2-3ft (0-6-0:9m) from the
consequent reduction in bandwidth and possible in- ground. Even so the system did seem to be less critical,
crease in losses, so that it is unwise to exceed a resonant particularly in respect of balance in the main feeder,
feeder length of A/4 or to use for this any type of line when the stub was in use.
other than 6000 open-wire. At the low-impedance end The tuned-line version of Fig 4.23(a) suffers from
of this line any of the usual methods may be employed narrow bandwidth because due to the greater length
for matching into the remainder of the feeder system. the impedance on the right-hand side of the line
The reader may well be wondering how it is that if changes with frequency more rapidly than that on the
the Zepp feed does not work it managed to avoid left, thus upsetting the balance.
suspicion of its bona-fides for so many years, being
ousted from popularity mainly by the swing of fashion
which currently favours low-impedance coaxial lines. What type of feeder?
The reason is believed to be in part that antenna Coaxial feeder is well ahead of the field in popularity
current used to be regarded as a good indication of due in part to the mistaken idea that it does not radiate,
antenna efficiency and plenty of current flows in the but it is also the most convenient since it can be “run
tuned circuit of Fig 4.23(c), likewise in the tuned anywhere’’, even underground, with no special precau-
feeders of Fig 4.23(b). tions except to make sure that water cannot get into any
This of course is not a sufficient reason in itself since joins or terminations. The disadvantages are high cost
no amount of amperage would be convincing without and high losses, ranging between about 0-8dB and
some radiation but, though not normally a good 2:4dB per 100ft (30m) at 30MHz and roughly pro-
method of obtaining radiation, Fig 4.23(b) is not a good portional to the square root of frequency. Many of the
method of preventing it either and under certain condi- older cables also have a limited life.
tions it can even work efficiently. This happens if, in The best low-impedance flat twin (eg type 214-023)

Table 4.1. UK UR series coaxial cables

Maximum
Nominal Overall Inner Capaci- operating Approximate attenuation (dB per 100ft) Approx
UR No impedance diameter conductor tance voltage —_—_— __ RG
Zo (9) (in) (in) (pF/ft) (rms) 10MHz 100MHz 300MHz 1,000MHz_ equivalent
43 52 0-195 0-032 29 2,750 1:3 4:3 8-7 18-1 58/U
57 75 0-405 0-044 20-6 5,000 0-6 1:9 5 7-1 11A/U
63* 75 0-853 0-175 14 4,400 0-15 0-5 0-9 1:7
67 50 0-405 7/0-:029 30 4,800 0-6 2:0 3:7 75 213/U
74 51 0-870 0-188 30:7 15,000 0:3 1:0 1:9 4:2 218/U
76 51 0-195 19/0-0066 29 1,800 1:6 5:3 9-6 22-0 58C/U
77 75 0-870 0-104 20:5 12,500 0:3 1:0 1-9 4-2 164/U
79* 50 0-855 0-265 21 6,000 0:16 0-5 0-9 1:8
83* 50 0-555 0-168 21 2,600 0:25 0-8 1:5 2:8
85* 75 0-555 0-109 14 2,600 0-2 0:7 1-3 2:5
90 75 0-242 0-022 0 2,500 1-1 3:5 6:3 12-3 59B/U

All the above cables have solid dielectric with a velocity factor of 0-66 with the exception of those marked with an asterisk which are helical
membrane and have a velocity factor of 0-96. This table is compiled from information kindly supplied by Aerialite Ltd and BICC Ltd, and
includes data extracted from Defence Specification DEF-14-A (HMSO).
48 HF ANTENNAS FOR ALL LOCATIONS

Table 4.2. USA RG series coaxial cables

Cable | Maximum
Cable Nominal outside Velocity Approximate attenuation (dB per 100ft) Capacitance operating
no. impedance diameter factor (pitt) voltage
Z,.(ohms) (in) 1MHz 10MHz 100MHz 1,000MHz 3,000MHz (rms)
RG-5/U 52-5 0-332 0-659 0:21 0:77 2:9 11-5 22-0 28:5 3,000

RG-5B/U 50-0 0-332 0-659 0-16 0-66 2:4 8-8 16-7 29-5 3,000

RG-6A/U © 75:0 0-332 0-659 0:21 0-78 2:9 11:2 21:0 20:0 2,700

RG-8A/U 50-0 0-405 0-659 0-16 0:55 2:0 8-0 16:5 30:5 4,000

RG-9/U 51-0 0-420 0-659 0-16 0:57 2:0 7:3 15-5 30:0 4,000

RG-9B/U 50-0 0-425 0-659 0-175 0-61 2:1 9-0 18-0 30:5 4,000

RG-10A/U 50-0 0-475 0-659 0-16 0-55 2:0 8-0 16-5 30:5 4,000

RG-11A/U 75-0 0-405 0-66 0-18 0-7 2:3 7:8 16-5 20:5 5,000

RG-12A/U 75-0 0-475 0-659 0-18 0-66 2:3 8-0 16:5 20-5 4,000

RG-13A/U 75:0 0-425 0-659 0-18 0-66 2:3 8-0 16-5 20:5 4,000

RG-14A/U 50-0 0-545 0-659 0-12 0-41 1:4 5-5 12:0 30-0 5,500

RG-16A/U 52:0 0-630 0-670 0-1 0-4 1:2 6:7 16-0 29:5 6,000

RG-17A/U 50-0 0-870 0-659 0-066 0-225 0-80 3-4 8-5 30-0 11,000

RG-18A/U 50-0 0-945 0-659 0-066 0-225 0-80 3-4 8-5 30:5 11,000

RG-19A/U 50-0 1-120 0-659 0-04 0-17 0-68 3-5 77 30:5 14,000

RG-20A/U 50-0 1:195 0-659 0-04 0-17 0-68 3-5 7:7 30:5 14,000

RG-21A/U 50-0 0-332 0-659 1:4 4-4 13-0 43-0 85-0 30:0 2,700

RG-29/U 53-5 0-184 0-659 0-33 1:2 4-4 16-0 30-0 28-5 1,900

RG-34A/U 75-0 0-630 0-659 0-065 0:29 1-3 6-0 pled 20-5 5,200

RG-34B/U a5 0-630 0-66 0:3 1:4 5:8 21:5 | 6,500

RG-35A/U 75:0 0-945 0-659 0-07 0-235 0-85 3-5 8-60 20:5 10,000

RG-54A/U 58-0 0-250 0-659 0-18 0-74 3:1 11-5 21-5 26:5 3,000

RG-55B/U 53:5 0-206 0-659 0-36 1-3 4:8 17-0 32:0 28:5 1,900

RG-55A/U 50-0 0-216 0-659 0-36 1:3 4:8 17-0 32:0 29:5 1,900

RG-58A/U 53-5 0-195 0-659 0-33 1:25 4-65 17-5 37:5 28:5 1,900

RG-58C/U 50-0 0-195 0-659 0-42 1:4 4:9 24-0 45-0 30:0 > 1,900

RG-59A/U 75-0 0-242 0-659 0:34 1-10 3-40 12-0 26:0 20:5 2,300

RG-59B/U 75 0-242 0-66 1-1 3-4 12 21 2,300

RG-62A/U 93-0 0-242 0-84 0-25 0-85 2:70 8-6 18-5 13-5 700
RG-74A/U 50-0 0-615 0-659 0-10 0-38 1:5 6-0 11-5 30-0 5,500
RG-83/U 35-0 0-405 0-66 0-23 0-80 2:8 9-6 24-0 44-0 2,000
RG-213/U* 50 0-405 0-66 0-16 0-6 9 8-0 29-5 5,000
RG-218/Ut 50 0-870 0-66 0-066 0-2 1:0 4:4 29:5 11,000
RG-220/Ut 50 1-120 0-66 0-04 0-2 0-7 3-6 29-5 14,000

* Formerly RG8A/U t Formerly RG17A/U + Formerly RG19A/U

 FEEDING THE ANTENNA 49

is comparable with the best coaxial cable but it is baluns which provide 50Q terminations. The line oper-
advisable to space it out from any supporting struc- ates at a nominal swr of 3-0, and with the further
tures. The velocity factor, at least of some types of flat addition of an atu the overall loss is only 1:6dB at
twin, is greatly affected by moisture, making it unsuit- 14mHz. This includes estimated losses as follows:
able for outdoor use where phase stability is required.
Matched line loss 0-6dB
The type of 300Q twin ribbon available at the time of
Extra loss due to mismatch 0-3dB
writing in the UK, though often recommended as an
Two baluns 0-2dB
alternative to open-wire line, is extremely sensitive to
moisture. The velocity factor can change by up to 20 This makes no allowance for bends which are an
per cent and in the case of the semi-transparent plastic additional source of losses and probably account for the
types severe deterioration has been found after a short residual 0-5dB loss. After much experiment with high-
period of use, resulting in a weather-dependent loss of ratio transformers, additional atus and the like, the
several decibels per 100ft (30m). Better types exist and above arrangement was found to be best for conveni-
hopefully may become available since this is a very ence and efficiency, and can be recommended to
useful type of line, the rated losses being in general anyone with a similar ‘“‘long-line” requirement.
about half those of coaxial cable. Like 75Q twin lead it Obviously the losses could be much reduced by the use
needs to be kept more or less clear of supporting of, say, 14swg (2mm) wire and the avoidance of sharp
structures, though the odd point of contact with, say, a bends. This would bring the estimated loss down to less
wooden mast seems to have little effect. than 1dB.
Although open-wire line has to be rated lowest in
terms of convenience, there are usually no major
problems except for long indoor runs and the mecha- Radiation from feeders
nical bypassing of beam rotators, a problem which can Feeder radiation, though it may result in the occasional
be overcome without detriment to performance by the “unlikely” contact, is more commonly associated with
use of short 300Q line inserts, and which hardly arises tvi, additional feeder loss (which can be large), and
anyway in the case of the reversible beams described in increased noise or interference in reception. Radiation
later chapters. from feeders can arise in several different ways [8].
It is much the cheapest and most efficient type of Fig 4.25 illustrates a typical antenna fed at the centre
feeder and there is no foundation for the belief that it is with a feeder which may be of any type, and it is
any more liable to produce radiation than other types of assumed for the moment that the system is energized by
feeder. This can be easily checked by means of the inductive coupling into an atu with no earthing at the
formula on p51 but in effect the radiation from a long transmitter. If the feeder has an effective electrical
line is as if from a dipole having a length equal to the length / as defined by Fig 4.25(a) which (allowing for
spacing and carrying the line current which, in view of top loading by the dipole) is an even number of
the high Zp, is in any case low if the line is matched. quarter-wavelengths, it will resonate as an antenna and
Radiation can of course result from imbalance but, can be used as such. In fact a 14MHz dipole with about
assuming reasonable care in this respect, the more 44ft (13-4m) of vertical feeder, voltage fed at the lower
likely causes of radiation are those which tend to affect end, performs well as a low-angle radiator for 7MHz
all types of feeder equally as discussed in the next dx. On the other hand a length of 17 or 51ft (5-2 or
section. Attenuation using 14swg (2mm) conductors [1] 15-5m) would enable it to radiate efficiently on 14MHz,
is of the order of 5dB per mile (3dB/km) or 0-1dB per though probably in an unwanted direction or at an
100ft (30m) at 30MHz and this can be reduced by using unsuitable angle of elevation.
thicker conductors, the attenuation being inversely With the arrangement shown the unwanted mode
proportional to the surface area of the copper so that could be excited by capacitive coupling at the transmit-
each time the attenuation is halved the weight and ter unless the coupling coil is balanced to earth and
approximate cost is multiplied by four. For losses shielded. The two halves of the dipole can also induce
comparable with coaxial feeder the wire size could be currents in the vertical feeder (the outer in the case of a
reduced to 36swg (0-2mm), and for runs of up to 100m coaxial line), but given perfect symmetry these are
there seems little point in going to anything thicker equal and opposite and therefore cancel. On the other
than 18 or 19swg (1:2mm). hand any asymmetry due to failure to bring the feeder
It is evident that if the shack is a long way from the away from the dipole at right-angles, or not connecting
antenna there is no alternative to the use of open-wire it to the exact centre of the dipole, or proximity of one
line. By way of illustration the author has a shack on half of the dipole to a tree or building, will lead to a net
the “antenna farm” and a more convenient one in the current in the down lead and hence to radiation in an
house linked by 150yd (137m) of line made from unwanted and probably lossy mode with considerable
odd bits of wire with several changes of gauge, averag- potential for tvi.
ing about 1mm diameter and incorporating two 4:1 Earthing the feeder at the transmitter alters the
50 HF ANTENNAS FOR ALL LOCATIONS

connecting an earth or short counterpoise to the centre

point of the coupling coil as shown dotted. Avoiding
resonance minimizes but does not necessarily eliminate
the unwanted radiation at the fundamental frequency,
and the problem (though basically similar) may prove
much more intractable at harmonic frequencies re-
sponsible for tvi.
In the case of balanced lines the unwanted current is
induced in the same direction in both conductors,
subtracting from the transmission line current in one
conductor, adding in the other, and displacing the swr
pattern in one relative to the other. This makes it
impossible to obtain reliable measurements of swr and
a similar problem arises when current flows on the
outer conductor of a coaxial line.
If a dipole fed with coaxial line is symmetrical with
respect to earth, half the terminal voltage of the
Current
distribution
antenna can be expected to appear between the outer
on outer of of the coaxial line at its top end and earth, Fig 4.25(c).
co-axial
cable The feeder must then appear like an “inverted ground
plane” antenna for which the radiation resistance at
resonance is about 1802 and the power radiated in the
unwanted mode will be
2
ik watts

ie as much power is radiated from the feeder as from

the antenna. More likely, symmetry will not be
Fig 4.25. (a) If ! is an odd number of wavelengths the feeder is achieved and feeder radiation will take place as pre-
energized as a vertical radiator via the stray capacitance C,. (b) If viously described. As before, the situation can be more
not brought away at right-angles the feeder is energized by
unequal inductive or capacitive coupling as indicated by arrows.
or less rectified for the fundamental frequency by
(c) Excitation of standing waves in coaxial feeder if current avoiding resonance but this does not necessarily elimin-
distribution is symmetrical. Even if feeder is at earth potential ate harmonic radiation.
excitation can take place as at (a) or (b)
A long line will have many resonances which may be
fairly broad, and it is impossible to avoid all of them if
length required for resonance and leads to a rather less the antenna is required to operate on all bands includ-
predictable situation, as the earth resistance and effec- ing 10, 18 and 24MHz. However, if only a few
tive length of the earth connection are introduced into frequencies are involved / may be chosen with the help
the problem. The safest course is probably to avoid of charts such as Fig 4.26. Careful attention to symme-
earthing, keeping the electrical length of the system as try will go a long way towards avoiding problems, the
far from resonance as possible. If necessary the reso- use of a balun being mandatory in the case of coaxial —
nance can be shifted by altering the feeder length or by lines. It remains important, however, to suppress at the

| | | |
L F
28— n ' ae) (oe =

:
11 ' ie 1 1
aMod
=x
oe i!it
tl
it
n
i!
il
11 |
ae
a
1
1a
mn
See
14
of
4
H
1
!|
i
!
J
2 i! 1! ts vase I H
v eee i! i ca i Hee =f \
i! i! it rele ber \ H
as “ HF ne \! tot : I
5 ee 1! jh We Ht bits Ha: jaLamsef ‘
ti iI \ | ar i : 1
i! \| ie Cae Hy \ | |
i 1| tee ee
35—. i!if Gf
1;
1
Let
ote
igs eat
Hate}
\ ! |
i! 1 tJ it hoa! 1 1 '
i] i! ot tt fp 1 ! '
it i! reat i! ' t 1 1 1
C
10 20 30 40 50 60 70 90 100 110 120 ro o xN3
Length (feet)

Fig. 4.26. Lengths shown by solid lines along the horizontal axis avoid exact resonance in the 3-5, 7, 14, 21 or 28MHz amateur bands in
systems where the coupling apparatus is isolated from ground. Best operating lengths corresponding to /in Fig 4.25(a) are at the centres
of the wider ranges as shown by the arrows. To include 10, 18 and 24MHz while avoiding undue proximity to band edges it will be
necessary to make some provision for alteration of /when the waveband is changed (ARRL Antenna Book)
 FEEDING THE ANTENNA 51

source any harmonics, which by radiation from the 50 0r 752

balanced output
feeder could cause tvi, since baluns have a finite
bandwidth and are likely to be ineffective at the higher
harmonic frequencies.
By the provision of one or more counterpoise wires it 4R (balanced
| load)
may be possible to alter the effective value of /, Fig
4.25(a), as required. The precise length usually has to
be found by experiment since it depends on stray
capacitance, impedance discontinuities and, if the
antenna is joined to the case of the transmitter, a
number of indeterminate paths to earth which may
50 or 750
even include the mains wiring. COaxial cable
Particular difficulty attaches to vertical and end-fed
antennas since in these cases perfect symmetry is
Fig 4.27. (a) Trifilar balun, 1:1 impedance ratio but can also
unobtainable, but use of non-resonant lengths in con- provide 1:9 ratio (balanced or unbalanced) or 4:9 unbalanced on
junction with loose couplings as described on p44 both sides. (b) Bifilar balun, 4:1 ratio. (c) Trifilar balun redrawn as
or the “balanced” Zepp feed described _ earlier auto-transformer to illustrate the principle of operation. The
three (or two) windings are wound as one, with the least possible
(p46) provide a satisfactory solution in most cases. spacing between wires, though individual turns may be spaced
Unscreened lines are in themselves potential out along the core as shown in Fig 4.28
radiators, but subject to accurate balancing this should
be negligible for frequencies in the hf band. The power
radiated for an rms line current of J amps and a spacing should be avoided since line balance is particularly
D is given by: sensitive to the proximity of the open ends of such stubs to
surrounding objects.
P, = 160 (22)'r watts

D and jd being measured in the same units so that D/A is Baluns

the line spacing measured in wavelengths. This is the Most transmitters have an unbalanced output connec-
same power as would be radiated from the same tion intended for use with a coaxial cable whereas the
current in a short dipole of length equal to the line majority of antennas are balanced, so that conversion
spacing|1]._ from an unbalanced to a balanced system has to take
For 60082 line with 6in (15cm) spacing and a place at one end or the other of the feedline. There are
current of 1A (ie 600W) the line radiation at 14MHz is many ways of carrying out this process but, ignoring
therefore such commonly-used devices as the gamma match
nm \2 which is inherently unbalanced and a lot of other
160 (33) =0-09W arrangements which are unattractive for various
reasons, one is left with the trifilar and bifilar wound
The proportion of any harmonic power radiated by the auto-transformers shown in Fig 4.27.
line increases as the square of the harmonic number, but These devices, contrary to much that has been
seems likely to remain insignificant. The figures given by written about them, are highly efficient as well as
the above formula must be multiplied by four to account extremely simple both in theory and practice. All one
for radiation from the terminal connections, but even so needs to do is take two or three lengths of enamelled
for a 75Q line with 0-08in (2mm) spacing only about one copper wire about 10in (25cm) long, twist or bind them
millionth of the incident power is radiated, as compared very tightly together (this is the vital part of the
with the possibility of up to one-half as previously process), wind them onto any odd bit of ferrite rod (Fig
discussed for an unbalanced line. 4.28) that happens to be lying around, and connect
This is still not the full story. A major advantage of them as shown. For a 3—30MHz balun the theory states
coaxial line is that it can be laid anywhere, even buried merely that the inductance must be large enough not to
underground. In contrast to this, low-impedance twin shunt the line significantly at 3MHz and the /eakage
line generates a field which can be detected externally reactance which appears in series with the output must
and, though it spreads only a short distance, makes it be negligible at 30MHz. This is a very small amount
advisable to keep the line clear of trees and buildings and failure to realize the crucial nature of this require-
etc, at least to the extent of mounting it on stand-off ment may be responsible for the difficulties which have
insulators. It is a minor disadvantage of open-wire lines frequently been reported.
that the balance is easily disturbed and the presence of Another possible explanation is the prevalence of
objects near the line can cause asymmetry and hence balun designs which are claimed to involve “‘transmis-
radiation. The use of open-circuited matching stubs sion line’ principles. The leakage requirement also
52 HF ANTENNAS FOR ALL LOCATIONS

Ferrite rod
literature on this topic may find it hard to believe that it
can be simplified to this extent. Nevertheless, using
these rules the author has constructed large numbers of
baluns of many shapes and sizes. In one test no less
Start of winding1 Start of winding 2 End of
winding 2
than six ill-assorted small and large baluns in pairs,
back-to-back, were inserted between a 400W p.e.p.
winding 1 transmitter and the antenna with an overall loss of
only about 0-5dB and no untoward effects except that
one very small one got too hot to touch!
Many other baluns have been constructed and the
procedure outlined seems to be virtually foolproof
Fig 4.28. Construction of bifilar (4:1) balun. Although the wires provided impedances are restricted to the range of
must be bound tightly together spacing between turns is not about 25—300@. Moreover similar devices of compara-
critical unless impedances are very low, in which case leads must ble performance are available from a number of manu-
be as short as possible. Terminals AB provide a balanced output
of 200-3002 facturers. Baluns of this type do not require any special
care in matching, adjustment of the antenna being
apparently neither more nor less critical as a result of
rules out the tapping of baluns which has frequently their use.
been recommended for obtaining impedance ratios There are, nevertheless, some limitations. At impe-
other than the usual 1:1 or 4:1, although this is feasible dance levels around 12, for example, leakage react-
for spot frequencies subject to some re-tuning of the ance can be a problem, requiring particular care with
antenna. connecting leads, the lowest possible value of induct-
For further insight into the principles reference may ance, and possibly the use of capacitive compensation
be made to Fig 4.29 which shows at (a) two wires of as described above. At the other extreme the author’s
identical length laid alongside each other, so that if the attempts by various means (including quadrifilar wind-
spacing is small enough all the lines of magnetic force ings) to achieve direct matching from 50 to 600@ met
resulting from a current flowing in one wire also with little success, probably due to the self-capacitance
surround the other wire, and there is no leakage. of windings. The solution adopted, as described else-
Separate the wires, allow them to be unequal in length, where (p49) was to use 4:1 baluns into and out of the
or attach long leads to their ends and the above 600Q line which was allowed to operate with an swr of 2
condition is no longer satisfied. The auto-transformer or 3, an arrangement which proved highly efficient and
connection further reduces the leakage since a large much more convenient than alternatives based on the
part of the winding is common to input and output sides use of pi-couplers or atus as described in the next
of the transformer. section.
Fig 4.29(b) shows the principle of operation; assum- Readers will recall the stress placed on the use of
ing a 50Q load and a shunt reactance of 2002 at 3MHz ferrite rods. Ferrite and powder-type toroid rings have
a glance at the Smith chart (p59) gives an swr of 1-25.
At 30MHz the shunt reactance is 2,000Q and a leakage Ae ec A op ae ai!
reactance of 12-5Q causes the swr to rise to 1-25. This is
a leakage coefficient of only 0-625 per cent but can \ LS vies

nevertheless be realized easily enough if the above

instructions are carefully followed. (a) NA |
It is important to realize that the core has no role NAS) SSS

other than to enable the required specification to be

Mutual inductance M=kL
met at 3MHz while using the smallest possible length of
oi eatage reactance L'=L (I—k)
wire so that the leakage is minimized. Any additional
coupling provided by the core is trivial as may be
proved by removing it, in which event the balun
becomes useless on the lower-frequency bands but is (b)
barely affected at 28MHz. Frequency-independent Inductance L (each winding)
compensation for the leakage reactance (so long as it
does not exceed about 25Q) is possible by means of a Fig 4.29. (a) Unity-coupled transformer formed by laying two
straight wires close together. The alternating current / produces a
parallel capacitor connected on the output side. It will voltage V, as shown. The voltage V, is virtually equal to V,
be evident from this explanation that the tighter the because almost all the magnetic lines of force surrounding the
coupling the greater the frequency range that can be top wire also encircle the lower wire. (b) Equivalent circuit of
transformer; leakage reactance = L(1 —k) where kis the coupling
covered. coefficient. k is very close to unity in mains transformers and in rf
Readers familiar with the large amount of recent transformers based on the principle of (a) above
 FEEDING THE ANTENNA 53

also been used successfully but appear to have fewer properties, six or seven turns (which need not be
advantages. The powder type were supplied as a kit and closely spaced) provide a good starting point, the
from the instructions it seemed that the core was being completed balun being tested at low power on all bands
relied on for ensuring adequate coupling, but in this with an swr meter and a dummy load consisting of, say,
respect it was found to contribute very little and swr at two 5W 100Q carbon resistors in parallel. If the induct-
28MHz was in excess of 2-0. After evolving the proce- ance is just right there will be a just-perceptible in-
dure outlined above this was reduced to 1-4 for a 1:1 crease in swr at the lowest frequency. If the balun is to
balun and around 1-2 for a 4:1 balun, the reason for be used, say, for a triband beam covering 14-28MHz
rather worse leakage in the case of the powder core the swr may be allowed to rise from a typical value of
being linked to the need for twice as many turns which 1-1-15 at 14MHz to 1-3-1-4 at 7MHz.
followed from the lower permeability of the material. It A good test procedure is to construct two identical
should perhaps be mentioned that the unsuccessful baluns, connect them back to back, and insert them in
attempts to achieve a 600Q impedance were based on the feeder, first on one and then on the other side of the
powder cores, and self-capacitance effects may have swr meter. On the antenna side they should have little
been accentuated due to the requirement for a larger effect on the swr reading, whereas placing them on the
number of turns. transmitting side of the meter provides an indication of
Ferrite rings are less convenient than rods and there the total insertion loss (if any) for the two baluns.
is a possibility of cross-modulation at high power levels For weather protection many types of plastic contain-
due to saturation of the core [9]. The risk of this should er may be pressed into service, using for example
be much less with rods, due to the greater reluctance of ordinary screw-type terminals with appropriate washers
the magnetic circuit which is mostly “‘air gap’. Balun to ensure an adequate seal. Immersion of a completely
transformers are so efficient that rods of less than 0-5in unprotected transformer with twisted windings in a
(13mm) diameter and a length of 2in (S1mm) or less can bucket of water produced an immediate loss of several
be used for powers up to the legal limit (400W p.e.p.) decibels, some of which persisted after removal, but a
in the UK. A 1:1 balun using 17 turns on a powder core “temporary” arrangement with taped windings and no
0-69in (18mm) diameter (Amidon toroid type T-68-6) other protection was used for several months in all
had measured losses averaging less than 0-1dB, an swr weathers with little or no variation in relative signal
better than 1-35 over the range 3-5—28-5MHz without reports. Nevertheless adequate protection is essential if
compensation, and handled a power of 20W cw with the transformer is to be used in an inaccessible position.
only a modest temperature rise.
It is important to note with all types however that
under some conditions of severe mismatch high Applications
temperatures can develop, and even for normal opera- A 1:1 balun is needed for the centre feeding of dipoles
tion some means of escape for heat should be provided. with low-impedance coaxial line; this includes two-
Failure was experienced in the case of a balun encapsu- element arrays based on multiband dipoles as described
lated in Araldite for weather protection, and attributed on p167. It can also be used for conventional designs of
to a combination of several factors: the Araldite, a bad quad antennas.
mismatch and operation for too long under key-down The 4:1 balun is required in conjunction with T and
conditions. Other failures have been attributed to the delta matches as favoured on p40 and is particularly
use of enamelled wire from an old mains transformer, suitable for feeding into and out of “‘mismatched” 6002
the enamel having a tendency to chip off. Cotton lines as recommended above. Similar devices can be
covering, though perhaps obviously unsuitable, was used and are often required as impedance transformers
once used in an emergency thereby merely creating a in balance-balance (p39) or unbalance-unbalance
further emergency! situations (p49). Despite some increase in leakage
reactance, the trifilar construction can be used as a
Construction double-wound transformer (balanced on both sides),
Twisting of the wires is satisfactory, avoiding the giving a 4:1 impedance ratio as required for example
extremes of too tight a spiral or enough slack for when feeding dipole arrays with 6002 line as described
daylight to be visible between the wires. It is probably on p39.
better however to use straight wires side by side and An alternative to this is to combine a 1:1 balun with a
bind them together as tightly as possible with insulating 4:1 balun to obtain a 4:1 balance-to-balance trans-
tape. The wire diameter should ideally be as large as formation. Two such transformer pairs have been
will conveniently fit on to the core, but it is not critical. tested back-to-back as described above with satisfac-
Some commercial baluns use a single-layer trifilar tory results, though the trifilar arrangement is simpler
winding but, though reasonably satisfactory, this results and appeared to be equally good. A rather complicated
in a slightly worse specification. split-winding version of this was evolved to provide
Starting from scratch with a ferrite core of unknown symmetry in respect of stray capacitances, but these do
54 HF ANTENNAS FOR ALL LOCATIONS

to be too short for such effects to be present. Maybe

there is still some point here which needs to be cleared
up but attempts to complicate the issue should be
ignored since, whatever may be said to the contrary,
the auto-transformer type of balun is a necessary,
highly efficient and inherently simple device which,
within the specified limits, presents no difficulties of
Fig 4.30. Transmission-line balun. The winding acts as a transmis-
sion line in respect of the current J, but the two wires act construction or use.
together as a choke in respect of an in-phase current /, such as
may be induced due to lack of symmetry in the antenna system. Antenna tuning unit (atu)
The load impedance Ra, representing the antenna system, is
normally balanced with respect to earth. This type of balun is not The use of an atu allows any antenna system to be
recommended: see text connected to any transmitter. Some designs are in-
tended to cope with any situation, however extreme,
and others are suitable only for correcting a typical
not appear to be a problem when working at the 200 mismatch in a coaxial line. Most transmitters can be
impedance level. adjusted to compensate for an swr up to at least 2-0 but
A trifilar winding can also be used to provide 9:1 or others require an swr of 1-0, and in this case the use of
9:4 unbalance-unbalance or 9:1 balance-balance trans- some form of atu is essential because most antennas
formation and on occasion these ratios have been found show some variation in swr over a frequency band.
useful. Typically a three-band trapped beam peaked for the
Nothing has yet been said about transmission-line centre of a phone band may have an swr rising to three
transformers, to which there have been many refer- or more in the cw band, although it is still capable of
ences in the literature, and the reader may well be radiating a good signal and can be used if an atu is
expecting enlightenment. The two wires of Fig 4.29(a) available.
may be regarded as a transmission line so that if a A “universal’’ unit must be able to cope not only with
voltage is applied to one end an equal voltage appears this type of situation but also such diverse requirements
at the other if the line is properly matched. as the end-fed antenna of Fig 4.18 and short dipoles fed
This in itself does not get us far but if the line is coiled with resonant open-wire lines. This adds up to a
onto a ferrite core it then acts as a choke for in-phase complex set of design problems requiring some degree
currents so that only equal and opposite currents can of compromise and it is difficult to guarantee loss-free
flow in the line. If one side of the input and the centre performance under all circumstances. Designs are
of the output load is earthed, the necessary balance available in various handbooks [12, 13] and commercial
may be established as shown in Fig 4.30. By using two units are available at prices which are perhaps an
such lines with their inputs in parallel and outputs in indication of the size of the problem. The author
series a 4:1 ratio is obtainable as before. prefers to recommend a more “piecemeal” approach
There are difficult problems in the matching of the on the following lines:
line and reference [10] points out that “design is (1) For end-fed antennas including odd bits of wire
complicated because there is mutual coupling between: use tuning units which are an integral part of the
turns, which modifies the characteristic impedance. antenna system, eg as in Fig 4.18. Coupling may be
However, suitable units are available commercially”’. adjusted to provide a good match to low-impedance
The reference also points out that matching is critical so line.
that it is “essential that the antenna match the 300Q (2) With 60082 feeder connect a balun plus low-
twin line at all frequencies at which the balun assembly impedance line at a point of low impedance, in or as
will be used”’. near as possible to the shack. This should be possible
Problems with “‘transmission-line baluns” are discus- except in some cases of multiband operation with
sed in reference [11], leading by an entirely different resonant lines. |
chain of reasoning to a design very similar to Fig 4.27. (3) In these latter cases use a tuner on the lines of
This article also stresses the connection between low Fig 4.31, with the help of Fig 4.32 for determining
leakage reactance and wide frequency range which, whether the series or parallel configuration should be
though it appears to be ignored elsewhere in the used [15].
literature, was pointed out also in reference [12] and (4) With matched 300Q line use a 4:1 balun.
has been well known for a long time to at least some (5) Having thus obtained a low impedance, use an
designers of commercial baluns. elementary form of atu based on a “simple” pi- or
As a final comment it should be noted that a L-network such as Fig 4.33(a) or (b) to correct any
transmission line differs from a “lumped circuit” only residual mismatch. 3
by virtue of time or phase delays along the line, and Despite the simple appearance a full analysis of these
balun transformer windings of any type would appear circuits is very complicated and it is frequently possible
 FEEDING THE ANTENNA 55

+— 66feet 7 MHz——+| +—
66 feet 14MHz—+

eg.
-——-

(aoe? (bdecP
< Fig 4.33. L and pi matching networks for 50Q lines are shown at
(a) and (b) respectively. The antenna impedance Z, may be
oo
(a) (b) expressed in the form A, + jX,, and X,, if capacitive, tuned out by
part of the reactance X. If R, is less than 50Q it can be made to
appear as 502 across AB by increasing the value of X, the
inductive reactance resulting from this oe tuned out by C. If R,
exceeds 50 the input and output terminals may be reversed. If X,
is inductive it can be tuned to resonance by the left-hand
capacitor of the pi-circuit, (b), and over a limited range the
additional control in the case of the pi-circuit can obviate the
need for a variable inductance

Fig 4.31. Centre-fed antennas with tuned feeders. The antenna to

the left is 1/2 long and has a low input impedance; the feeder is
to obtain a match with two completely different compo-
also )/2 long and thus repeats the low impedance so that series nent settings, at least one of which involves a substan-
tuning should be used. On the right is a similar antenna operating tial loss. The theory is simple enough assuming a small
at twice the frequency so that high rf voltage exists at both ends
of the feeder, and parallel tuning is best. The effect of feeder
value of L, in which case after allowing for the tuning
length is illustrated in Fig 4.32. The tuning capacitor sections out of any antenna reactance there is an impedance
should each be 50pF maximum for 14MHz and higher frequen- transformation given by the inverse square of the
cies, proportionately larger values being needed for lower
frequencies. The voltage ratings of the capacitors should be
capacitance ratio. On paper this method of design is
comparable with those required for the transmitter tank circuit demonstrably foolproof, but it is disconcerting when
(as happened to the author) the unit falls apart in the
middle of a contact due to the solder melting. The
connections, incidentally, were not the author’s usual
“haywire” but could fairly have been described as
“heavy duty” and the power was well under the legal
limit.
A recommended design of pi-network atu is shown in
Fig 4.34. There are two cases to be considered, depend-
ing on whether the antenna impedance is less or more
than Z,, but since the network can be turned round it is
only necessary to consider the low case as in the figure,
where the tuned circuit L’C’ is used to remove the
8 antenna reactance X, which may be positive or nega-
tive. Reference to the Smith chart shows that for an swr
of 2-0, X, will be not less than 63Q, and a value slightly
(Feet)
L
Length lower than this should be chosen for the reactance of
80 L’. The lowest value of capacitive reactance provided
by C’ should be half this since it may have not merely to
tune out the reactance of L’ but provide an equal
reactance of opposite sign. For any value of swr less
100
than 2 the antenna reactance can now be removed,
leaving a resistance R,.

. 120
~angy

Fig 4.32. Standing-wave chart for tuned feeders. A line through

the length L of feeder plus half-antenna and the appropriate
frequency shows, at its intersection with the wavelength scale,
the nature of the input impedance. Rectangles to the left of the
line indicate capacitive impedance, those to the right being
inductive. The shaded areas indicate high impedance, the blank
areas low impedance. This diagram is suitable for use with dipole ._ Fig 4.34. Practical pi-network for use with 50Q line. For 14MHz
antennas fed with open-wire lines. It may be applied to a L1 = 0-4uH, L’ = 0-6uH, C1 = C2 = 450pF maximum. Note that in
bi-square loop by measuring L from the open-circuited top, and this case X, is the equivalent parallel reactance component of the
to quad or delta loops by measuring L from points )/4 out from antenna impedance and C’, which is required for resonating X, if
the top centre this happens to be inductive, can be provided by C1
56 HF ANTENNAS FOR ALL LOCATIONS

There remains now the more difficult task of choos- usually all that is needed is knowledge of a few such
ing the components on the right of the diagram so that a basic principles, common sense and access to the
pure resistance of 50Q appears at the output terminals. appropriate data.
In principle Cl is not needed, the L-network formed by Transmission lines are another story, and for a long
L1 and C2 being sufficient. The combination of L1 and time the author—having an impressive record of
R, results in a resistance failures in mathematics examinations—was content to
leave them to the experts. They usually seemed to
id
Ratt work, after a fashion anyway, and the antenna was the
Ri interesting part! In fact, although feeder matching can
which can be arranged to have any value at the output usually be achieved by trial and error, use of the Smith
terminals, and this is shunted by an inductive reactance chart is straightforward, saves time and reduces the risk
of mistakes. The delightful simplicity of the chart is,
wl, 7s
(i alas, belied by a somewhat fearsome appearance, and
w*L,? the aim is to penetrate the disguise and provide suffi-
cient insight into the principles and methods of use for
which is tuned out by C2. This means that L1 must be the chart to take its rightful place in the armoury of
variable from zero upwards and, although suitable readers faced with problems of antenna design and
components exist, they are relatively inconvenient to feeder matching.
use since they employ a slider which travels the entire Nevertheless the potential of the chart for some
length of the coil, making some 40 turns in all. All of rather complex manoeuvres poses a problem since it is
this range of adjustment may be required when chang- felt that a description covering all the important uses
ing frequency within the 3-5MHz band! would not be the best way of explaining it to everyone
Design charts are available for pi-networks [16] and who might benefit, bearing in mind that many amateur
although the procedure is still quite involved the values transmission-line calculations are of a simple type such
shown have been obtained for 14MHz. Cl incorporates as discovering the impedance of an antenna from
C’ and these values can be scaled as necessary for measurements at the lower end of the feeder. In an
coverage of other bands, the appropriate network effort to get round this problem, the chart is presented
being selected by a two-pole switch. Variable capaci- in Chapter 18 as a “piece of test equipment” complete
tors similar to those used in old valve-type broadcast with a set of instructions. It is hoped that readers who
receivers are suitable for powers up to 400W p.e.p., have persevered thus far will now read on but if they
though unless three- or four-gang specimens can be find it sends them to sleep they should turn at once to
obtained it may be necessary to switch in additional p242 where the subject is presented in lighter vein.
fixed capacitors on 3-5MHz. The range may be ex- Among the topics not covered in Chapter 18, the
tended by switching in a trifilar-balun-type transfor- application to antennas, as distinct from feeder systems,
mer, the connections being rearranged to give a 2:3 has been largely overlooked and is of particular interest
voltage ratio, ie an impedance step-up or step-down of as it provides an insight into many peculiar aspects of
4:9. This circuit will correct most mismatches up to at antenna behaviour and often saves a lot of arithmetic.
least 3 to 1, or at worst bring them within the range that Moreover, it has particular relevance for amateur ap-
can be corrected by adjustment of the transmitter. plications since it reveals at a glance how to tune any
With Cl or C2 and L’ tuned to resonance and LI odd bit of wire to resonance even when, for example, -
variable Fig 4.34 takes on a close resemblance to the the wire has to be bent to fit it into a confined space.
transmatch described in reference [14] which is de- For this application the antenna has to be regarded as
signed to match any random-length antenna. Perhaps a transmission line with the appropriate value, or
one should however add a note of warning from [16], values, of characteristic impedance assigned to it. Un-
which points out that losses in an L-network increase fortunately this information, although given in profes-
with increasing transformation ratio and can be large. sional textbooks, is not readily available to the average
amateur. It would in any case be impossible to present
The Smith chart accurate data covering all possible ways of bending an
antenna wire, but with the aid of intelligent guesswork
As the author has tried to show, most antenna problems the data presented here should be adequate for most
with which the amateur is faced can be tackled without amateur purposes.
much knowledge of mathematics. Most of us can It is necessary at this point to say a little about ‘‘j
visualize a wave travelling through space and are ready notation” if only to assure readers who have survived
to accept, for example, that field strength at a distance their first glance at the chart that they are not going to
is proportional to current multiplied by the length of be asked to master yet another new concept. Neverthe-
antenna wire through which it flows, without needing less, }does appear on the charts produced commercial-
to master the intricacies of Maxwell’s equations; ly and it is felt that most readers, after working through
 FEEDING THE ANTENNA 57

readers, and, as the examples prove, this approach is

adequate for a wide variety of applications. Neverthe-
less, as we shall see, an understanding of admittance
enables the chart to be used as an admittance diagram
and as a simple calculator for making impedance-to-
admittance conversions, which greatly extends its use-
fulness.
In many cases the chart requires an input which can
only be derived by direct measurement of impedance or
admittance at whatever points happen to be accessible
in an antenna or feeder system. However, it will be
shown that good use can be made of the charts even if
the reader is equipped with nothing more than an
understanding of the difference between resistance and
reactance, some simple form of swr indicator or rf
voltage probe, and a few odd lengths of cable. If
open-wire feeders are used the only instrument needed
consists of a small pick-up loop, rectifier and meter.
Holding the pick-up loop at a roughly-constant distance
from the feeder, it can be moved along the line to
determine the swr and the positions of the current
minima, with sufficient accuracy for most purposes.
oO O-1 0-2 O05
Wavelengths 2 aerial
Properties of lines
Fig 4.35. Variations of resistance and reactance along a 502 As explained earlier, if, for example, a 50Q resistance
feeder when connected to a resonant antenna having a feedpoint is connected to one end of a 50Q line, the impedance
resistance of 25Q (swr = 2:0)
observed at the other end will also be 50, and this
hoids true for any length of line. For any other termina-
a few examples using the charts, will find it useful as tion, the observed impedance varies with line length;
shorthand since it is much easier to write “50 + 35)” for example, Fig 4.35 shows what happens when the
than ‘‘50Q of resistance and 35Q of inductive react- 50Q resistance is replaced by one of 25, correspond-
ance’. This is the general form in which the chart ing to a standing-wave ratio of 2-0. These are the curves
produces answers to questions, and at first glance it which would te traced out by walking backwards down
may look like an unfinished sum requiring an under- the line towards the transmitter, observing the appa-
standing of j for its completion. In fact, the 50 and the rent value of the ‘‘load impedance” as seen from each
35 are “‘different kinds of ohms”’ and tend to be of more point on the line. In the absence of line losses, the
interest when they are separated than after they have pattern repeats itself every half-wavelength, and a
been added. given resistance always has the same value of associated
Typically the 50Q represents the radiation resistance reactance although the sign of this changes from posi-
of an antenna, or some other load into which one tive to negative at intervals of )/4.
wishes to deliver power, and the 3592 is a reactance Reactance can therefore be plotted against resist-
which needs to be removed by tuning, ie by connecting ance, as shown in Fig 4.36. The fact that this particular
an equal and opposite reactance in series. Conven- plot happens to be a circle need not unduly concern the
tionally a negative sign is used to denote capacitance, reader, who is asked to note only that a pattern has
and in this case the chart indicates the use of 35Q of been traced out which repeats itself at A/2 intervals. A
capacitance so that the complete answer to the problem distance (measured in wavelengths) scale can be
becomes 50 + 35j — 35j = 50Q which provides a good marked off (as illustrated) round it so that, given any
match to standard low-impedance cable. For the pre- point on this pattern, the impedance at any other point
sent purpose this is all the reader needs to know about on the line can be obtained merely by moving the
j, although a further acquaintance with it gives added correct distance round the pattern. All points on the
flexibility in the use of the chart and is essential to a line, including the antenna itself, must fit on to this
proper understanding of many aspects of electrical and pattern so that if the antenna impedance and line length
radio technique. An attempt to provide a simple ex- are known, the impedance at the transmitting end can
planation of j has been included as an appendix to this be read directly from Fig 4.36.
chapter. It is then a simple matter to devise, for example, a
Here the chart is presented initially as an impedance matching network such that the antenna is presented to
diagram since impedance is a familiar concept to most the transmitter as a matched load. Alternatively, from
58 HF ANTENNAS FOR ALL LOCATIONS

To make Fig 4.37 into a useful tool one further step is

necessary—the addition of suitable scale lines as in Fig
4.38 which shows a simplified version of the chart or Fig
4.39 which is typical of charts available from commer-
cial sources. In the case of Figs 4.35 and 4.36 the reader
Wavelengths could easily fill these lines in for himself since the
from aerial
originals were plotted on ordinary squared paper, ruled
with sets of equally spaced parallel lines at right angles.
Unfortunately when the graph paper is “‘bent’’ these
lines get bent with it and, since by definition parallel
lines meet at infinity, the scale markings appropriate to
Fig 4.36 all converge on the point marked © when
transferred to the Smith chart. To avoid undue conges-
—75 —50 —25 O 25 50 TAs. tion at this point, therefore, most of the lines have to
Reactance (ohms)
stop short, and the curious pattern results from this
Fig 4.36. Similar to Fig 4.35 except that reactance is plotted convergence plus the obvious practical need to main-
against resistance
tain a reasonably uniform density of scale lifes
measurements at the transmitter the antenna impe- throughout the diagram.
dance can be read from the chart and used to determine Despite omission of most of the lines shown in Fig
whether, in order to improve the swr, the antenna need 4.39, Fig 4.38 would be adequate for amateur use since
be lengthened or shortened, or some other adjustment usually no great accuracy is either needed or possible
made. and it has important advantages. Unlike Fig 4.39 it
cannot be accused of the ‘‘fearsome appearance”’ which
The circle diagram seems to have had a considerable deterrent effect on
So far only one value of swr has been considered, would-be users, and it includes swr circles enabling it to
though obviously a family of circles could be drawn be used without the addition of a cursor or having to
covering a fairly wide range of values. Useful though keep a pair of compasses handy. Nevertheless, after
this might be, the diagram in this form leaves much to working through a few examples the user will appreci-
be desired: the scales are too cramped at their lower ate the convenience of having more scale divisions as in
end and cover too small a range of impedances; diffe- Fig 4.39 as well as the swr circles of Fig 4.38. Unfortu-
rent sets of curves are needed for each value of feeder nately their addition in monochrome to Fig 4.39 merely
impedance, and the non-linear nature of the distance increases its resemblance to a spider’s web but this
scale makes it rather inconvenient to use. problem is easily resolved by the use of red ink for the
To overcome these problems the obvious first step is circles. .
to make the resistance and reactance scales non-linear, The bending of the scale lines should cause no
expanding the lower ends and compressing the upper difficulty if it is appreciated that lines of constant
ends. In this way any desired range of values, even
extending all the way from zero to infinity, can be
accommodated. The next step, rather less obvious, has
been aptly described [6] as ‘bending the graph paper’’.
In effect, the reactance scales are bent round so that
they form a complete circle with their ends, now
scale-marked infinity, meeting at infinity on the resist-
ance axis as shown in Fig 4.37, which is the basis of the
Smith chart. Negative Positive
reactance 50
Fig 4.37 is identical to Fig 4.36 except for the (ohms)
bee ai 50 reactance
(ohms)
distortion and bending of the scales, which has been 40
done in such a way that the impedance plot remains
circular although the wavelength scale marked round it
is now linear. In other words, movement round the
circle of any given number of degrees always corres-
ponds to the same fraction of a wavelength whatever
the starting point. A further advantage of the new
diagram is that the circles of constant swr now share a Fig 4.37. The circle diagram or Smith chart. This is similar to Fig
common centre and make use of the same wavelength 4.36 except for “bending the graph paper”, the centre line being
scale, which is usually marked around the outer cir- the resistance scale. Two lines of constant resistance and two of |
constant reactance have been shown dotted. The inner circle is
cumference of the chart. drawn for an swr of 2 and the markings on it are steps of 0-05)
 FEEDING THE ANTENNA 59

Resistance and
SWR or |/SWR
Towards Towards
antenna —0-2 transmitter

—O0-5

Negative Positive
Reactance Reactance

Fig 4.38. Simplified Smith chart. -1°O

Each circle represents a con-
stant value of swr and meets the
resistance scale at points corres-
ponding to the value of the ter-
minating resistance and its re-
ciprocal, the swr being equal to
the higher of the two figures.
Note that the reciprocal of any
point on any circle is the oppo-
site point on the same circle.
The markings on the circles are
steps of 0-05). The _ points
marked A, B and C correspond
to the example in the text

WAVELENGTHS
°
DECIBELS ATTENUATION

TQ Q SHS
ae ERR PLES eo
e, e
LT TERROR D> oS

Bao
SL RRS
S555 SOS | RRR SOO
RR SREY SSSI ay, ERRORS SSS

[ors
PRESS
re
:
QL RESO
ey, 2 SES
S E HERERO
TEER
OK
SS SIE
OS SSG c)
peruens
Orn
L]

LT
Fig 4.39. Typical Smith chart. The additional
It
BER
DO xa re,
lines make for greater accuracy and con-
weene, Ye venience despite the adverse effect on
PEEL
appearances. SWR circles need to be added
for most calculations. Points D, E illustrate
the application of the chart to finding the
new value of resonant length which results
from folding part of an antenna wire, as
discussed in the text. Subtraction of a
pee length DF at a Zp of 1,00022 can be compen-
ae sated by adding a length EF at a Zp of 60002
60 HF ANTENNAS FOR ALL LOCATIONS

resistance are circles or parts of circles which touch at wavelengths we arrive at point B where the antenna
infinity but also pass through the appropriate point on impedance is read off as (1-0 + 0-7j) Zo, ie (50 + 35j)Q.
the resistance scale (ie the vertical line in the centre of This was the value cited earlier as ‘“‘the sort of answer
the diagram), and lines of constant reactance are arcs of the chart might give” so we already know what to do
circles centred off the diagram but also meeting at about it, ie connect a capacitive reactance of 35Q in
infinity as explained earlier. series with the antenna.
Another important respect in which Fig 4.38 differs There are, however, various alternatives; note that
from Fig 4.37 is the scaling, which suggests that it has had we stopped short of point B by 0-15\, ie at point C,
been drawn for a rather improbable line impedance of the answer would have been 100 with no reactance so
1Q. In fact, this is another bit of clever trickery and that a feeder system consisting of 0-154 of 50Q line
overcomes the remaining defect of Fig 4.36 which, as adjacent to the antenna plus any length of 100@ line
readers may recall, was the need to prepare different would be perfectly matched except for the short length
charts for different values of line impedance. The of 50Q line. Feeder which is close to 10082 impedance is
method is known as “‘normalization” and is a simple unobtainable but the nearest popular value would be
standard conversion process of the kind with which 75Q, giving an swr of 1-33 which is quite good enough.
anyone who travels abroad is familiar—the difference Note that had we used 7502 feeder “‘all the way” the
being that instead of converting pounds into francs or normalized antenna impedance would have been
miles into kilometres, when entering on to a Smith (0-47 + 0-33j) which sits on an swr = 2-4 circle, an even
chart ohms have to be converted into ‘“‘zednoughts’’, Z) worse result than in the case of the 50@ feeder cable.
being the symbol used to represent the line impedance.
For example, if the line impedance is 50, then 50Q is Application to antennas
one Zp, and if the line is used to connect a transmitter [t is common practice to make a clear distinction
operating at 14MHz to a load consisting of 1000 of between ‘‘antennas” and “transmission lines”. This —
resistance plus a reactance of 85 (ie an inductance of attitude appears to be reflected in the prevailing usage
1uH), the chart is entered at the point corresponding to of the Smith chart which is of course quite correctly
100/50 (ie 2 zednoughts) on the resistance scale and regarded as a “‘transmission line calculator’. Neverthe-
85/50 = +1-7 on the reactance scale. Putting this into j less, as mentioned earlier, the antenna can also be
notation the impedance is (100 + 85j)§$2 and becomes regarded as a transmission line, and from the point of
(2:0 + 1-7j) after normalizing to 50Q. In the latter case view of tuning and matching there are important advan-
there is no need to fill in the unit of impedance since a tages in so doing. The alleged “‘bandwidth” of an
Zy is the only one recognized by the chart. antenna is usually not that of the antenna but some
The charts illustrated in Fig 4.39 are available in combination of antenna and feeder system, and in
graph-paper form. They can be used as graph paper for dealing with such a combination the Smith chart can be
the permanent recording of such experimental results particularly useful.
as the variation of antenna impedance and swr across a In treating the antenna as a transmission line, it is
band of frequencies, but for the present purpose the essential to think of it as a line which has been unfolded |
chart is of more interest as a calculating device. For this so that, for example, a A/4 line turns into a 4/2 dipole.
purpose one method is to glue it onto a hard surface This unfolding must of course lead to an increase in the
and employ a rotating cursor of transparent material characteristic impedance (Z,)) since it reduces the
with marks corresponding to the points on the resist- capacitance, and the load resistance (being identifiable
ance scale (greater than one) over which they pass. As with the radiation resistance) is usually small compared —
the cursor is rotated, the points marked 1-5, 2, etc with Z, so that the swr is high. As well as being an
describe circles representing these values of swr. antenna, the wire remains basically a transmission line
Nevertheless the red ink method is simple, labour whose length is half that of the antenna, and if Z, and
saving, and adequate unless hard use is anticipated. If the radiation resistance are known, the Smith chart can
any other marks required are made in pencil and be used to discover the impedance at the centre or ends
rubbed out when no longer required, a single chart of the wire. The wire may be bent into any shape, in
marked up with red swr circles can be re-used many which case different parts of it may have different
times. values of Zp), and in dealing with these cases the
The simpler types of calculation are those confined to performance of the Smith chart is particularly impres-
a single circle on the chart, such as points A, B and Cin sive since each change of impedance along the antenna
Fig 4.38. For this example it is assumed that one wishes and feeder system can be regarded merely as a frontier
to know the impedance of the antenna, given that a at which we have to change currency, ie normalize to
measured value of (35 — 25j)Q has been obtained at the the new impedance.
“shack” end of a 50Q line. Dividing by 50 this is plotted The main difficulty so far has been lack of impedance
as the point A, ie (0-7 — 0-5j) on the chart. Given that data; this is readily available for transmission lines, but
the feeder length is an odd number of quarter access to professional textbooks is usually needed in the
 FEEDING THE ANTENNA 61

very poor since the reference followed the usual prac-

End-loaded elements tice of omitting to state that the required total length of
(eg, ends bent overx_
at right-angles) 4 wire is no longer A/2. The questions are, what should be
the length of the stub and what will be the radiation
Straight single-
resistance ‘“‘seen” by a feeder connected to its lower

aap
wire elements
end?
iy Lin
‘7
Use right-
hand scale
The antenna is treated as a 0-2) line having a typical
Zo Of 1,000 and, since the swr in this “‘line” is rather

rere
large, one is restricted to operations on or very close to

<2
ti Ke
0-3 0-4 O-
the outer edge of the chart. Starting from the far end of
the line (the bottom of Fig 4.39) we may proceed
Length of element in wavelengths clockwise round the edge for a distance 0-2), arriving at
Fig 4.40. Radiation resistance expressed as an equivalent series point D; this is a reactance —0-33j, ie it is required to
resistance at the centre of the element. These are the figures add 330Q of inductance in series. Since Z, for the stub
required when working with the chart in its impedance form as is 600Q2, the normalized value of X is 330/600, ie 0-55,
described here. Unlike the equivalent parallel resistances they are
independent of wire diameter. For two-element beams the resis- which comes opposite the 0-08A mark (point E) on the
tances and the multiplying factors given on pp70-71 are applic- wavelength scale marked round the edge of the chart.
able However, had we taken A/2 of wire and merely put a
fold in the middle to take up the slack this would have
case of antenna wires. Fig 4.13 rectifies this omission to resulted in a stub length of 0-05( only. The stub must
the extent that it should cover most cases of importance therefore be lengthened by 0-03A or 4-2ft (128m), ie a
for amateur applications. Data on radiation resistance total wire length of 8-4ft (2:56m)!
is widely available but usually relates to positions of To obtain the feedpoint resistance, a value of 40Q is.
current maxima; given this information, however, the given by Fig 4.40 but this is the value at the antenna end
radiation resistance referred to the antenna terminals, of the stub, not that seen by the feeder. In principle we
when the antenna wire is not a resonant length, may be could have entered the chart at the point (0-04 ¥ 0-55j)
roughly determined with the help of the chart. How- and, proceeding clockwise by 0-05A, arrived at the
ever, from this point of view values of R tend to be point 0-036 or 3682. However this part of the scale is so
rather on the low side, and if the distribution is known cramped that, although one can just about see that the
they may be derived more accurately by using the usual reduction in R is not enough to matter, it is much easier
PR formula, J being the current and R the radiation to work out R from the basic power relationship as
resistance at any point in the system, and /?R the same suggested above. Thus we note that the antenna is only
for all points. This has been done for single wires with 18 electrical degrees from the end of the stub which is a
lengths up to 3/4 in Fig 4.40, which also includes the point of maximum current, and cos 18° is 0-95 so the
case of elements which have been physically shortened increase in current at the feedpoint compared with that
by bending back the ends of the wires. at the antenna is five per cent only.
In the case of full-wave dipoles the terminal resist- Since /*R must be the same at all points, R at the
ance is best obtained from the current loop radiation feeder terminals is 10 per cent less than 40Q, ie 360
resistance (180Q) by regarding the antenna as a A/4 which, though a bit low, is still a reasonable match to
transformer, obtaining Z, from Fig 4.13 and evaluating 50Q feeder. Had the dipole length been a full 1/2 rt
Z9/180. In principle the chart can be used for this would have been better of course to use 75 feeder.
purpose by entering it at the point 180/Z) and proceed-
ing round it for a distance A/4. This procedure, inciden- Multiple changes of impedance
tally, demonstrates the inversion of impedances where- In an antenna system there may be several changes of
by they turn into admittances, a particularly important impedance. The chart makes light work of this situation
feature of the chart as explained later. as illustrated by the following example which is plotted
Note that if Z) = 900, Z,/180 = 5 and use of the as points K, M, M’, N and R in Fig 4.41.
chart produces the answer in the form 5Zp, ie A 14MHz antenna is assumed to have an impedance
5 X 900 = 4,500 which, as some readers may be of 40Q providing an acceptable match to 50Q cable.
quick to point out, could have been obtained with much The required feeder length is 106ft (32-3m) but unfor-
less trouble by simple arithmetic. The situation is very tunately there is only 95ft (29m) available and at
different, however, if even a small amount of reactance current prices we cannot afford any more. In the junk
is introduced, ie if the antenna is no longer exactly box however there is plenty of 75@ cable, and a
resonant as in the following example. velocity factor of 0-67 is assumed for both cables. The
Suppose it is desired to erect a 4/2 wire dipole for required overall length could be obtained by splicing in
7MHz but there is only room for a length of 0-4). 4 of 75Q cable, and the question is where should we
Having read elsewhere that the middle 0-14 can be put it? Next to the antenna, next to the transmitter or
folded into an open-wire stub this is tried, but the swr is somewhere in between?
62 HF ANTENNAS FOR ALL LOCATIONS

Transmission loss
(1dB steps)

Fig 4.41. Application of Smith chart

to typical problems. Points K, M, M’,
N, R illustrate the effect of a discon-
tinuity in the line impedance. The
effect of a typical line loss is illus-
trated by the spiral transition from ~
the inner circle at P to the outer
circle at Q’

The electrical length of the 50Q feeder is 2-125), and re-normalizing to 50Q this becomes 0-95, providing
which takes us from the antenna, point K to a point M an swr of 1-05 in the 50Q cable which may be used for
on the chart, an impedance of 0-97 + 0-22). Adding on the remainder of the feeder system.
the 4/4 of 75Q cable, this value must be normalized for In the examples above it has been assumed that the
the new impedance, multiplying by 50 and dividing by swr is constant along each section of line though it
75, ie multiplying by two-thirds to given an impedance should be noted that losses in a transmission line cause .
of 0-65 + 0-15j. This is point M’ and fits on to a swr the swr to vary along the line. In effect, looking into the
circle of 1-6 which is the value applicable to the 75Q lower end of a lossy line, it appears to a large extent to
line. If it is desired to know the impedance at the be matched by its own losses whereas at the antenna
transmitting end of the line, this can be found by end the swr is determined only by the relationship
proceeding 4/4 round the 1-6 swr circle to the point N. between Z, and the load impedance. The relation
Next suppose the extra length is placed at the between line loss and the swr at its ends is illustrated in
antenna end of the line. The starting point is now 40/75 Fig 4.10, and this information is also incorporated in
but is inverted by the 4/4 line to become 75/40 at the the transmission loss scales on the left of the chart.
start of the 50 cable. Normalizing to 50Q involves a The method of allowing for line loss is illustrated in
multiplying factor of 1-5, bringing the resistance and Fig 4.41. The point P represents an impedance
therefore the swr up to 2-74 which is much higher than measurement at the transmitter with its swr circle
before and applies to most of the feeder run, whereas drawn in, and the outer circle corresponds to the swr at
the previous highest value 1-6 applied only to the last the antenna. The distance between the two circles
11ft. depends on the line loss, and if the loss is known (eg —
Much time can be spent in trying to improve on the from data tables) the swr at the antenna can be
first result, since the chart allows experimentation on obtained from Fig 4.10 or more conveniently by using
paper with a wide range of cable lengths and positions, the scale at the left of the chart which is marked in steps
but the following is a good solution. Proceeding from of 1dB, a loss of 1dB being assumed for the present
the antenna at 50Q for A/8 and then changing to 75Q example. Horizontal lines may be drawn from this scale
again gives point M’. After a distance of 0-476) the so as to transfer the decibel step to the centre line of the
point R is reached. This is a resistance of 0-63, ie 47-5Q chart, the outer circle then being drawn so that the
 FEEDING THE ANTENNA 63

ie F
!
ee vv Y=G+jB= ithe, AGUae
15g a 7, Neen OP aaPS

QO ~s [}x
Xs H
From this equation it is possible to derive the following
Rs identities:
| — L--.(?)---
(a) A (b)
Fig 4.42. Equivalent impedance circuits. For every series arrange-
ment of resistance and reactance as shown at (a) there is an
equivalent parallel circuit (b) such that measurement of impe-
dance across the terminals AB is identical in both cases
Referring to Fig 4.43, the point Z represents a
normalized impedance 0-31 + 0-45j, and to find its
distance between the circles corresponds to the reciprocal one need only proceed round the circle to
appropriate portion of the transmission loss scale. It is the opposite point Y where the admittance can be read
convenient to set a protractor to the radius of the inner off as 1-03 — 1-5j. These figures are respectively the
circle and then place it against the loss scale, enlarging conductance and susceptance in admittance units, ie
the radius by the appropriate amount. This is then used “Y-noughts” which are reciprocals of zednoughts so
_ for drawing the outer circle. that if Z) = 50@ then Y) = 1/SOmho, a susceptance of
The distance from P to Q corresponds to the line 1-5 “Y-noughts” being therefore equal to 1-5/50mho.
length but, in moving from the transmitter to the The change in sign which accompanies the inversion is
antenna, the track is now a spiral as shown dotted, so discussed later, but for the moment is best ignored.
that instead of ending up at Q we find ourselves at Q’ For better or worse, most of us are used to thinking
which gives the impedance of the antenna. Conversely in terms of impedance rather than admittance and
if Q’ is known we can find P, and in this case the 1dB ohms rather than ‘“‘zednoughts”, and to obtain the
line loss has changed the impedance “‘in the shack”’ required component values in ohms, ie R, and_X,, the
from (0-88 — 1-14j) to (1-0 — 0-9)). admittance terms have to be inverted to give 50/1:03 =
This brings us to another important aspect of the 48-5 and 50/1-5j = —j33Q respectively. The step of
Smith chart, namely its use for the inversion of impe- converting the figures read from the chart into admit-
dances so that they turn into admittances. This allows tances and then back into ohms can be skipped as soon
parallel combinations of resistance and reactance to be as the user is familiar with the principles, the 1-03 and
substituted for the equivalent series networks and vice 1-5 being divided into Z, to obtain R, and X, whereas
versa. In many cases it is much more convenient to use the initial figures, 0-31 and 0-45 were multipliers so that
the parallel approach and, once familiar with the chart, going through the reverse process we have R, + jX, =
the reader will find [6] that it is a highly flexible device (0-31 + 0-45j)Z, = (16:5 + 22-5j)Q.
allowing him to wander at will through impedance, Thus we find ourselves back at the starting point
admittance and transmission line territory uninhibited, from which indeed we have never moved, the half-
not merely by changes of Z, as explained above, but circuit of the chart being no more than a paper exercise
also such obstacles as the cumbersome formulae pre- for carrying out the series-parallel conversion without
viously needed for making series-parallel substitutions. resort to the rather cumbersome algebraic formulae
It will be recalled (p35) that a A/4 line can be used quoted above.
for impedance inversion whereby an impedance Z is The importance of this procedure may be seen by
changed to Z,*/Z and, since the Smith chart is based on supposing that the (16-5 + 22-5j)Q is a measured value
a “zednought”’ of one, it follows that moving half-way of the impedance of an antenna which it is required to
round it (ie a distance of A/4) is equivalent to converting match to a 50@ feeder. If the chart is treated solely as
Z into 1/Z so that the impedance R, + jX, illustrated in an impedance chart the best we can do is tune out the
Fig 4.42(a) becomes inductance with a series capacitor, reducing the swr to
3-0 from 4-0, but the parallel conversion indicates an
1
almost perfect match subject only to getting rid of the
R, + jX," equivalent parallel reactance of j33Q2. This requires the
The equivalent parallel circuit is shown at Fig. t.42(b) simple addition of a parallel capacitor having a react-
and, following the normal rules for impeaances in ance of —j33Q, ie 345pF for a frequency of 14MHz.
parallel, we have Had the series reactive term been missing at the start
of the exercise we should have been less fortunate, and
=seabird |
he 1
for a more general solution of the matching problem
ai « eensVe
reference may be made to Fig 4.44 taken from refer-
These quantities are defined respectively as the admit- ence [7] where the solution is obtained from the
tance Y, conductance G and susceptance jB, so that formulae. The top diagram illustrates the problem and
64 HF ANTENNAS FOR ALL LOCATIONS

Fig 4.43. Illustration of impedance inversion

and conversion of impedance into admitt-
ance or vice versa by use of the Smith
chart. Z, Y are equidistant from C which
means that each is the reciprocal of the
LY other
igPER
os ®, ERB
ay O SS
% NySOA gua eS > S

ESSSO []
re
wee,

as a first step the reactance is tuned out at (b), leaving a values of G + jB directly, and hence the required value
5:1 resistive mismatch. By manipulation of formulae it of X, (ie 1/B), without reference to the chart. How-
is found that a series reactance +30j results in a value ever, in another instance the value of G might be just as
of 75Q for R,. The addition of a capacitor to tune out unsuitable as the value of R, in the first example, so
X, results in an impedance match and the final solution that the above procedure would need to be applied in
is shown at (e). reverse.
Being thus conversant with the essential logic, the An alternative method of solving the matching prob-
procedure can be greatly simplified by using the Smith lem is to proceed down the line from the antenna to the
chart along the lines of the previous example. A point at which R, is equal to Z, or, re-phrasing this in
straight-edge is placed across the centre of the chart “admittance language”’, G = Yy. After determining the
and rotated so that the intercepts with the R = R, and corresponding value of X,, connection of a parallel
R= Z, scale lines are equidistant from the centre. The reactance — X, will establish correct matching, though
intercept points (which are similar to Z, Y in Fig 4.43) the portion of line between this point and the antenna
correspond respectively to X, and X, but, whereas X, is continues to operate at a high value of swr. |
the actual value of the component required, X, includes In general we can use either procedure and have a
the reactance of the antenna which must be added or choice of two quite different solutions to the problem,
subtracted depending on the respective signs. In the one of which is often much more convenient than the
case of Fig 4.44 we have Zy = 75Q and R, = 15Q. There other. On a point of mathematical detail, equating
is a series reactance of —j200Q which as a first step is 1/jX, to jB means that 1/X, = j°B = — B, so that if X, is
tuned out, being later replaced by the correct value, positive B is negative and vice versa. In other words a
+30Q in conjunction with the required value of + capacitor, though its reactance X is negative, is regarded
37-5Q for X,, as shown in Fig 4.44(d). The plotting of as having a positive susceptance (B). Similarly an
this example on the chart is an exercise the reader may inductor has a negative susceptance. This should not
like to try for himself, particularly if he has experience confuse the reader since the sign is a matter of conven-
already of the algebraic method. tion, the change being a formality needed for “keeping
In the case of the previous example use of an the record straight’’. In the same vein it may be noticed
admittance bridge (Fig 18.16) would have given the that j is shown as negative on one side of the chart and
 FEEDING THE ANTENNA 65

insertion in a 600Q line to bypass the beam rotator, or

for the paralleling of dipoles in a multiband system as
described in Chapter 7. The transparent type of 300Q
line deteriorates very rapidly with exposure to the
Cay weather and should only be used indoors.
Baluns, preferably of the auto-transformer type
g 15.2 wound on ferrite rods, should always be used when
going from a balanced to an unbalanced system or vice
versa. These devices are broadband, inexpensive to
+2002 construct, and highly efficient but there are some
restrictions on their use and p51 should be consulted.
a _s008 An atu can also operate as a balun but experience has
75 +jO 15+ j0O 2a (b ) led to a preference for separate baluns in conjunction
with an unbalanced atu for feedline impedances less
than 300Q, or a balanced atu if the impedance is high,
eg a high-voltage point on an open-wire line. An atu
can usually be designed for a specific task using inex-
75 +j0 Co) pensive components such as can be found in any
well-stocked junk-box, but universal units are more
difficult to design, expensive to buy, and not recom-
mended.
For the feeding of vertical antennas there is no need
for the extensive buried-earth systems usually recom-
75+j0 (d) mended, and advice apparently requiring one to buy up
several neighbouring properties in which to bury sever-
al miles of expensive copper wire should be ignored.
Moreover, in the case of above-ground radial systems
10H the length usually recommended (A/4) is the one above
758 we (e) all others to be avoided for reasons explained on p45,
ene and existing systems should be checked for lack of
symmetry as well as feeder radiation by the methods
described.
Fig 4.44. Example to illustrate the matching of two arbitrary
impedances. Tuning out the reactance of the antenna results in a
““Monopole”’ vertical elements should be “‘earthed”’
5:1 resistance mismatch. Detuning slightly, to the extent of to short inductively-loaded counterpoises and fed with
+j30Q, matches the resistances but results in a reactive mis- due regard for the advice given on p44. Symmetrical
match which is removed by a reactance +j37:5Q. The practical
solution of the problem is shown at (e) for 3-5MHz
(dipole) vertical elements should be centre-fed if the
feeder can be brought away at right-angles and without
positive on the other, so that inverting jX one obtains a coupling into other elements. Usually it is however more
susceptance —jB, meaning that X and B are both convenient to use a “‘Zepp” feed at the lower end.
inductive. However, the Zepp feed does not work in its convention-
al form; it must be terminated at the antenna end by aA/4
open-wire stub closed at its far end, as first pointed out by
Recommendations GO6CJ.
The correct choice of feeder system depends on the The Windom (or single-wire) feed is not recom-
type of antenna and the layout of the station. Coaxial mended.
feeder is usually the most convenient but is inefficient
and very expensive. In contrast, open-wire line can be
quickly constructed from inexpensive materials and
Appendix—use of j
more often than not can be used either on its own or in Most readers will be familiar with the fact that when
conjunction with relatively short lengths of coaxial 10Q of resistance is added to 10Q of reactance, the
cable, in which case the length of the open-wire line impedance offered to the passage of an electric current
should preferably be a multiple of 2/2. is not 20Q but something less. To avoid doing the sum,
Due to a large decrease in velocity factor when wet, the impedance in this case would usually be written
plastic twin-line (75 or 300982) should preferably be down in the form 10 + 10j, the j being merely a form of
avoided. If phase stability is required, eg for phased shorthand to indicate that the answer (if needed) may
arrays or the remote tuning of a reflector, it must only be obtained by adding the two quantities “at right-
be used in short lengths such as might be required for angles”. For example, if we represent them by drawing
66 HF ANTENNAS FOR ALL LOCATIONS

two lines at right-angles using a scale of lcm to an ohm Moreover, since j can be handled like any other
and then make this into a complete right-angled tri- algebraic symbol, it enables those whose knowledge of
angle, the length of the line which completes the mathematics extends no further than simple algebra to
triangle is 14cm, and 14Q is the answer to the problem. tackle quite elaborate calculations involving complex
Suppose, however, that before adding the two lengths, numbers, such as coupled circuits, antennas (including
the second one is turned through a right-angle not close-spaced beam elements), phase-shift networks,
once, but twice. It is now going in the opposite and the series/parallel conversions mentioned earlier;
direction and the sum of the two quantities is zero, ie it expressions for the latter are to be found on p63, but
is the answer to the sum “10 minus 10”. the manipulations can be very cumbersome and those
Turning the second quantity through two right-angles with experience of using them will be the first to
has, in effect, multiplied it by (—1) and with this picture appreciate the delightful simplicity of the Smith chart
in mind the reader will probably have little difficulty in for this purpose.
identifying —1 with j Xj or j*, so that j can be consi- Since expressions containing j carry implications of
dered as shorthand for V—1. This is not a “real” direction or phase, the associated calculations are
number since, however useful it may be to the mathe- known as vector algebra but in its simpler forms this
matician, there is no such thing in nature, hence the differs from ordinary algebra only to the extent of
usual description of the 10j as the “‘unreal’’ part of the substituting a ‘‘minus’’ for j*, as explained earlier.
impedance. It is also unreal in another sense—when a
current flows through the circuit, power is consumed
only in the resistance. If the j term is positive, it
represents an inductance, in which case energy is stored References
in a magnetic field and then released. [1] Radio Engineers Handbook, F. E. Terman,
When a voltage is applied to an inductance the power McGraw-Hill, 1943, p193.
going into the magnetic field constitutes a load on the [2] Radio Communication Handbook, Vol 2, 5th edn,
source and the flow of current is impeded until the field RSGB, 1977, p12.34.
has been built up to a value corresponding to whatever [3] The ARRL Antenna Book, 12th edn, ARRL,
maximum current the circuit is capable of supporting. 1970, p117. See also reference [1], p189.
In other words, the current is said to lag behind the [4] ‘“‘Aerial reflection’, F. Charman, BEM, G6CJ,
voltage, and in ac circuits this lag is a quarter of a cycle, RSGB Bulletin December 1955.
or one right-angle, hence the j. Conversely the current [5] ‘““The Smith chart”, L. A. Moxon, Radio Com-
going into a capacitor is initially large but drops to zero munication January 1977.
when it is fully charged, so the current is ahead of the [6] ““More on the Smith chart”, G. Garside, Radio
voltage and the j is of opposite sign, ie negative, though Communication December 1977. .
the + and — are dictated only by convention. [7] Radio Communication Handbook, Vol 2, 5th edn,
Although the reactive part of an impedance does not RSGB, 1977, p12.40.
consume power, it limits the flow of current and [8] Radio Communication Handbook, Vol 2, 5th edn,
prevents a generator from delivering all the power of RSGB, 1977, p12.31.
which it is capable. Fortunately it can usually be [9] ‘“‘New class of coaxial line transformers’’, G. Bad-
neutralized by a reactance of opposite sign—the pro- ger, W6YC, Ham Radio March 1980.
cess known as tuning. Reactance is therefore the basis [10] The ARRL Antenna Book, 12th edn, ARRL,
of selectivity, and performs other useful functions, so it 1970, p98.
should not be regarded merely as something to be got [11] ‘““High performance broadband balun”, J. J.
rid of—even though this is what one is usually trying to Nagle, K4KJ, Ham Radio February 1980.
do when using the Smith chart for the solution of an [12] Radio Communication Handbook, Vol 2, Sth edn,
antenna or feeder problem. It is important to realize p12.51.
that when working with complex impedances, ie those [13] The ARRL Antenna Book, 13th edn, 1974, p181.
containing both resistance R and reactance X, the two [14] The ARRL Antenna Anthology, 1978, p92.
parts play quite different roles. In general it is more [15] Radio Communication Handbook, Vol 2, 5th edn,
useful to know the separate constituents of an impe- RSGB, 1977, p12.45.
dance than its absolute magnitude which can of course [16] Radio Engineers Handbook, F. E. Terman,
easily be worked out from them, being given by McGraw-Hill, 1943, p212.
V(R?2 +X?) or by graphical methods as described [17] ‘““A 14Mc/s co-ax fed dipole and tvi’’, F. G. Rayer,
earlier. When, therefore, an impedance is given in the G30GR, RSGB Bulletin March 1965.
form R + jX this is not an unfinished sum, but a [18] ‘““Why coax’, E. M. Wagner, G3BID, 73 Novem-
particularly informative bit of shorthand. ber 1971.
CHAPTER 5

Close-spaced beams

More often than not a frequency band is either “‘dead”’ phase shift or built-in losses, whereas higher figures
or capable of supporting communication even with low may be due to failure to realize the difficulties of
powers and simple antennas, a point underlined by measurement (pp229, 236) plus the enthusiasm normal-
experiences recorded in Chapter 10, so why, it might be ly engendered by optimistic results.
argued, go to the extra trouble and expense of a beam? To avoid errors of this nature the reader need only
It is thought that the main benefit accrues from the select the most appropriate curves from Fig 3.9 and
reduction of interference during reception, though the insert the 3dB beamwidths in the (16/a,a,) formula,
4 to 6dB gain provided by typical amateur beams is an p23, not forgetting that 2:15dB must be deducted to
important bonus and probably the reason which carries obtain the gain relative to a dipole. Even this method,
most weight with the majority of amateurs. For these however, can lead to slightly optimistic results, the
reasons, and because of the need for beams which can object in Chapter 2 having been to establish basic
be erected in a limited space and pointed in any principles without involving the reader in too many of
direction, the close-spaced beam with two or three the finer details. Nevertheless, attention was drawn to
dipole or loop elements provides the mainstay of sources of error and these can add up to about 1dB.
amateur long-distance communication on the higher- Though this is not a lot in itself, there are other
frequency bands. Such beams are widely available from aspects involved and a fuller understanding of the
commercial sources but many have been constructed mutual coupling between closely-spaced pairs of
from published data, and new designs are a common elements leads, not merely to greater precision, but to
feature of amateur journals. conclusions which are somewhat unorthodox in so far
The need for a multiplicity of antenna designs might as they demonstrate the necessity of bending the ends
be questioned in view of the close equivalence between of dipole elements inwards to increase capacitive cou-
different antenna systems which was demonstrated in pling. This greatly improves the front/back ratio of
Chapter 3, but this was primarily in respect of gain; two-element arrays and overcomes certain practical
important differences arise when account is taken of disadvantages of two-feeder systems. This in turn paves
other aspects in conjunction with local circumstances, the way for highly efficient reversible beams with some
individual interests and abilities, and the cost factor. In degree of remote tuning, and it will be shown that only
consequence there is a need for many different types of one feeder at a time needs to be matched, matching of
beam affording plenty of scope for ingenuity. Many the other being impossible but also irrelevant.
important innovations owe more to patience and com-
mon sense than technical expertise but questions often
arise for which there is ‘‘no answer in the books” and
experimenters frequently complain of lack of expert
Mutual impedance
guidance. This is reflected in a tendency to attribute When two circuits tuned to the same frequency, Fig
good results to impossible causes such as “‘lowering the 5.1, are placed near each other there is usually some
angle of radiation” (p101), proliferation of designs open degree of coupling between them, that is to say a
to more or less serious criticisms, improbable claims for current /, flowing in one circuit induces a voltage V, in
some systems and waste of time and money in the the other. By close analogy with Ohm’s law the ratio
search for better results by methods which have no V,/I, is defined as the mutual impedance Z,,. It is the
chance of success. same with antennas except that by meekly accepting
Extensive browsing through the literature has unear- such conventional constraints as the use of straight
thed gain figures for two-element beams ranging all the elements the antenna designer usually denies himself
way from 3dB in some professional journals to 14dB for any control over the coupling. Furthermore, whereas
the Swiss quad [7], yet theory is quite specific as to what normal coupling between tuned circuits can be repre-
the gain ought to be, and for the cases in question sented by a ‘‘mutual reactance’’, that between antennas
admits of nothing outside the limits of 4 to 5-7dB can be reactive, resistive or complex which means that
relative to a dipole. Lower figures are difficult to it includes both resistance and reactance. The tendency
explain, except in some cases on the basis of excessive to ignore the reactive component has unfortunately led
67
68 HF ANTENNAS FOR ALL LOCATIONS

between the two types of operation, amounting to

virtual identity when both are designed correctly.
Mutual coupling and gain are closely linked and
without mutual resistance there can be no gain. This
applies to all close-spaced beams and can be readily
(a)
understood by taking yet another close look at the
W8JK system; the way in which this achieves directivity
and therefore gain was explained on p28 and it was
shown to depend on values of R which are much lower
than those for the elements in isolation. It follows that
ohms are being subtracted from each element as a
result of the presence of the other, and these subtracted
ohms are the mutual resistance (R,,), the subtraction
sign being due to the antiphase connection which is a
(b) feature of the W8JK system and its derivatives. On the
other hand, with in-phase connection and close-enough
spacing, two elements become indistinguishable from a
single element except that each takes half the current.
In this case the ohms are added, giving a total of
R,, + R = 2R for each element, this being equivalent
to R for the two in parallel which therefore behave as
one.
Equivalent dipole
current 1-54-
Two-element close-spaced beams
(c)
It can now be seen that the properties of close-spaced
beams as described in Chapter 3 are inseparable from
their behaviour as ‘“‘coupled circuits” so that, given
I, = 1-29 accurate values for mutual impedance, it should be
possible to calculate the gain with equal precision.
Fig 5.1. (a) Coupled circuits. The current /, flowing in the primary
induces a voltage /,Z,, in the secondary, resulting in a current Studying the subject from this angle sheds light also on
b =1,2,/(R +jX) where X =0 at resonance (ie wl = 1/wC). For important practical aspects, bringing out for example
tuned circuits as shown Z,, =jwM where M is the mutual induct- the need to increase coupling by bringing the end of
ance. For closely spaced “antenna elementsZ,, is in general a
complex impedance, ie it has the form (RA, +jXm,4 though X,, can elements closer together as already mentioned.
be zero. (b) shows phase shift ~ due to X for the case of Xm =0 First it will be useful to pursue further the example of
but is equally applicable to the case of X = 0 if R, X are replaced the W8JK since this provides a simple illustration of the
by the appropriate values of R,,,, X,,. (c) shows calculation of gain
for a parasitic array with )/8 spacing and @ = 45°; obtaining /,, b use of mutual impedance for calculation of gain. Refer-
and 9 (see text) the triangle is drawn and the gain, expressed as a ring back to Fig 3.4 the elements are assumed to be
voltage ratio, is measured off to scale “driven” by voltages of opposite phase E and —E
respectively, the crossover connection being extremely
important since otherwise, as already indicated in the
(as we shall see) to unsolved problems and missed discussion of mutual impedance, one ends up with a
opportunities. very close approximation to a dipole and the gain can
Critical coupling (kK = 1) exists between identical be specified without further calculation as almost exact-
resonant circuits when /, = J,, the ratio J,/I, being equal ly OdB! The currents in the elements are J and —I
to the coupling factor k. A similar situation exists in the respectively, the voltage /Z,, induced from the other
case of antennas, though it is necessary to distinguish element being subtracted from E in both cases. It then
between driven operation in which each element is follows from Ohm’s law that E — (—J)Z,, or in other
connected to the transmitter through some network or words E + IZ,,, must be equal to JR. Tuning out the
arrangement of feeders intended to enforce correct reactive part (if any) of Z,, we have E/J = R— R,, and,
phasing and power sharing, and parasitic operation substituting known values (p69) for parallel 4/2 dipoles
which is closely analogous to the action of coupled spaced 4/8, R — R,, = 73 — 64:5 = 8-5Q. This is the
circuits as just described. In this latter case one element effective radiation resistance of each element. Adding
only is connected to the transmitter, the other element the two resistances, Rg. = 17 whereas for a dipole
or elements being energized via the mutual coupling. It we have Rp= 73Q.
will be seen that this in fact plays an equally important The current in each 8JK element is therefore V(73/
role in both cases and there is a close relationship 17) = 2-06 times that in a dipole. Adding these with a
 CLOSE-SPACED BEAMS 69

Collinear dipoles To calculate mutual resistance between

Parallel dipoles staggered dipoles
Mutual=Rr Cos?@ + Rp Sin?
Mutual resistance of where Rr is mutual resistance between
Collinear dipoles 4 two parallel dipoles spaced x apart.
Two dipoles decr Vn es Rr is mutual resistance between two
end-to-end, ri aoe 2? (distances? collinear dipoles spaced x apart
ie Full-wave after 2A spacing between centres
Maxima and minima at
half wavelength points
and zeros at quarter
dane st points

(ohms)
resistance
Mutual

nae |e ahh alpet ae paar as

Fig 5.2. Variation of mutual resistance between parallel and collinear }/2 dipoles as the spacing is altered

phase difference of 45° (Fig 3.7), it is found that the To keep the algebra as simple as possible we need
forward field from the W8JK antenna is 1-58 times that X,, = 0 which happens very conveniently for S = 8.
from a dipole, equivalent to a gain of 4-0dB. (Usually This leaves Z,, = R,, which can be obtained from Fig
the elements are not connected in series but in parallel 5.2. Now E/I, is the impedance “‘seen”’ by the feeder
so that the actual feedpoint impedance is 17/4 and the j terms are eliminated by the normal tuning and
= 4-25Q, thougn this is not material to the calcula- matching process, leaving the radiation resistance
tion of gain). which is given by
It is instructive at this point to work out the radiation 2
resistance, power gain, polar diagram and reflector Rey Tes R(I " ase) (4)
tuning for a two-element parasitic array with straight
elements, treating it like an ordinary pair of mutually- The phase angle @ is determined by R and_X as shown
coupled tuned circuits (Fig 5.1) with which many in Fig 5.1(b) but can be found more quickly by obtain-
readers are already familiar. Those allergic to algebra ing the angle tan-! (X/R) with the help of a pocket
should merely note the conclusions and the remarkable calculator. This is the phase difference between the two
simplification achieved, as shown later, by more logical currents J, and —J,.
design procedures based on elements with equal cur- Recalling Chapter 3, p25, we may decide to make @
rents. “equal to the spacing” Qo, ie 45° for a spacing of 1/8, so
Referring to Fig 5.1, the voltage induced in the that for this example we need X = R and from equa-
secondary is J, Z,, and the resulting current J, develops tion (2) the magnitude of J, is given by
a voltage 1,(R + jX) across the impedance of the
secondary. This is assumed to be a parasitic element IL,| = LR! V(R? + X?) = 0:7071,R,/R

detuned in the direction causing it to act as a reflector;

however, since this impedance is short-circuited the From Fig 5.2, R = 64:5Q; putting R = 73Q and
sum of the two voltages is zero and we have assuming a power of 73W so that the current in a dipole
would be 1A, we have Rpg = 44Q, J; = 1:29A and
LZ, + L(R + jX) = 0 (1)
I, = 0-805A. Since (@ + @p) = 90° the triangle may be
Hence completed as in Fig 5.1(c) to obtain the forward field
lb = —L\Zpl(R + jX)= Zn pre ©) which is represented by AB. This gives an equivalent
dipole current of 1-54A so that the gain in decibels is 20
log 1-54 = 3-75dB. This is accurate to better than
But J, induces a voltage /,Z,, back into the primary so
0-1dB and represents a loss of only 0-45dB compared
that if this is resonant E = 1,R + J,Z,, and substituting
with the equal-current condition assumed in Chapter 3.
for I, we have
On the other hand, the field in the back direction,
E/I, = R — Zy2h(R + jX) (3) which in this case may be obtained by subtracting /,
70 HF ANTENNAS FOR ALL LOCATIONS

0-20 0-25) 0-3

Fig 5.3. Variation of mutual
Spacing (d) in wavelengths reactance between parallel )/2
dipoles as the spacing is altered
(ohms)
Reactance

directly from /,, is 0-485/1-54 times the forward field, a

front/back ratio of only 10dB.

Lt is
This performance is rather poor compared with what
can be achieved with two-element arrays since about a Nominal F/B Ratio
Multiply LH 100

Ea
third of the possible gain has been sacrificed. In
practice more than half the missing decibels can be
recovered merely by retuning the reflector, but to
realize the full potential of two-element beams it is
necessary, as will be seen, to equalize the currents and
shift the null away from the back direction. This makes
it possible to obtain gains of the order of 5dB combined
with better than 20dB rejection over all or most of the
180° of back direction.
It used to be assumed that for equalization of cur- (dB)
Gain
rents both elements would have to be driven. The
calculations for this case are featured in Figs 5.45.6
which tell most of the story [1, 23]. These show how
Resista
Radiati
Relativ
gain, directivity and radiation resistance vary with the
ratio of phase shift to spacing (@/q,), and it should be
/
noted that except for corrections explained below the /

S
ef
Pf
curves are applicable to all two-element close-spaced 4 4s

beams with equal currents (including quads). There are

however some differences in the case of parasitic beams
due to current inequalities which vary between wide
limits, depending on the type of beam and whether the
parasitic element is tuned as a director or reflector. Fig 5.4. Variation of gain, rediation resistance and front/back ratio
Though these curves were calculated prior to evolving with qg/qp, where Gp is the phase shift corresponding to the
the simpler method described below, the task was distances between the elements and q is the electrical phase shift
relative to the antiphase condition. Resistance scale is correct in
eased by assuming a spacing of 1/8. ohms for each of a pair of dipoles spaced 1/8 and carrying equal
It will be recalled from Chapter 3 that subject to the currents
 CLOSE-SPACED BEAMS 71

the individual elements become unequal, and although

\
Parasitic beam,=N the total is fixed, one of them can be zero or negative.
Depth of nulls This means that in the case of a driven array one
element is actually returning power to the transmitter
Parasitic for re-issue, as it were, to the other element, the
Equal dipole array radiation resistance of this being more than doubled
current
which can be of considerable help in the matching
process. This will cause little surprise to those familiar
with the operation of coupled circuits in receivers;
moreover, since the second feeder is handling very little
power it does not matter in the least that the swr may
well be close to infinity as explained later. A further
advantage of this situation is that each feeder can be
separately matched for its power-handling role and
beam reversal can then be effected from the shack by a
F/B
(dB)
ratio
simple interchange of feeders without affecting swr, a
different situation as we shall see from that which exists
(dB)
Gain Nominal F/B ratio when X,, = 0.
for Parasitic beam
An attempt to repeat the calculation for the case of
4/ X,, not being equal to zero defeated the author for a
44_Nominal F/B ratio
for equal current arrays long time due to the increase in the complexity of the
algebra, but eventually yielded the curves of Fig 5.7
and 5.8 which show the extent to which supergain
assumptions break down as the A/4 limit is approached.
At this spacing there is a drop of 0-5dB in maximum
10 20 30 40 45 gain with driven operation and 1-0dB for conventional
Phase shift
parasitic operation. For the m = @, condition the gain
(45°= Do)
drops even further, from 4-2dB with close spacing to
Fig 5.5. Gain and front/back ratio for two-element arrays with only 3dB with i/4 spacing, and front/back ratios are
unequal currents. Top curve from Fig 5.4 is reproduced for substantially worse.
comparison. Spacing = \/8
Another effect of reactive coupling is to improve
either reflector or director operation while rendering
spacing not being excessive there is no direct connec- the other more or less useless, with negative reactance
tion between size and gain or directivity. On the other favouring reflectors and positive reactance directors.
hand, the current required to maintain a given signal This makes driven operation unnecessary since all that
level varies inversely as the length and width of the is required to achieve optimum results is to increase
array. This means that the radiation resistances given in
Fig 5.4 have to be multiplied by factors proportional to
D=0-5 Dy (22-5°)
the inverse square of each of the dimensions and if, for
example, the spacing is halved the radiation resistance
is divided by four.
It may also be necessary to make a further correction
to allow for any change in shape of the current distribu- B= 0-67 Bo (30°)
tion. This can be combined with the length correction
by use of Fig 3.12 which gives radiation resistances for
elements which have been shortened by bending over PD = 0:78 Gq (35°)
the ends and discourages the use of other methods.
When two such elements are used as a beam the
radiation resistance is reduced in substantially the same
ratio as that of \/2 dipole elements, Fig 5.4.
strength
field-
forward
of
Percent
X/| P=Dy (45°)
The total radiation resistance for the two elements is 90 110 130 150 170 180
(End Bearing (degrees) (Reciprocal of
twice the value given in Fig 5.4; it is this total value on) beam heading)
which must be used for calculating the current and it is Fig 5.6. Variation of backward field strength with bearing and
given correctly by use of the scaling factors. These phase angle (q@). Angles in brackets apply for spacing = )/8. Curve
(a) corresponds to maximum gain and curve (d) to maximum
Operations conceal a hidden factor of considerable (nominal) front/back ratio. Curve (b) provides near-maximum
practical importance, namely that if Z,, includes react- gain, and better interference rejection than (d) for all bearings
ance as well as resistance, the radiation resistances of less than 147°
72 HF ANTENNAS FOR ALL LOCATIONS

many pages and wasted hours of algebra, besides

causing others to resort to computers or experimental
modelling techniques which have in some cases pro-
duced the wrong answers. In fact all one needs to know
is Ohm’s law and the definition of mutual impedance
from which, for the driven element
E a IR 3 LZn

Little more remains except to express J, in terms of J;

and then write this expression out in full.
To explain the process in more detail Fig 5.9 shows
the current J, shifted in phase relative to —J, by a
(dB)
Gain typical angle as required by Fig 5.4, and it will be seen
that in this position it can also be described as the sum
of two currents at right-angles, ie al, + jbl,. Pythagoras
tells us that for equal amplitudes a? + b* = 1? = 1
and as in previous examples the phase shift is given
by tang = bD/a. Putting 1, = —I,(a+ jb), Z,=
(R,, + jXm), and, assuming any j terms to be
removed by the normal tuning process, we have for the
radiation resistance
E
=7 = R-aRy + bXp
1
©)
Initially one is not concerned with how to meet the
Ratio /d, stated condition, but is merely looking at the situation
Fig 5.7. Variation of gain with phase shift at large spacings. (Top
curve is reproduced from Fig 5.4 for comparison)

capacitive coupling between the elements by bringing Maximum

the ends towards each other. It is a characteristic of the ain
quad antenna (as well as some of the more recently
introduced compact dipole arrays) that this desirable
condition is achieved, more or less, without conscious
effort. This seems to be the only possible explanation
for the particular reputation of the quad since, as will i
later appear (p98), there is no way that the gain
difference can exceed about 0-5dB.
Driven elements have attracted numerous ex- > iT)

perimenters but, despite some outstanding successes in

individual cases, the results have been somewhat vari-
able for reasons which will become obvious. The use of (dB)
Gain
controlled coupling on the other hand is simple, com- (dB)
ratio
FB
pletely effective, and in retrospect it seems amazing
that it should have escaped attention for so long,
particularly in view of the close analogy with coupled
circuits. Moreover, on the basis of parasitic operation
with controlled coupling a simple explanation emerges
for the curious but extremely useful effects of reactance
described above, and indeed all the more important
aspects of the operation of close-spaced beams in
general.
It,turns out in fact that merely by assuming that
Spacing between elements (A)
somehow or other we are going to succeed in making J,
equal in magnitude to J,, all the answers can be written Fig 5.8. Variation of gain with spacing for the conditions of
maximum gain, maximum nominal front/back ratio (~ =qo) and
down straight away. This avoids going through the @ =0. Dotted curve shows nominal front/back ratio correspond-
normal processes which have involved the author in ing to the maximum gain curve
 CLOSE-SPACED BEAMS 73

bI; (or I;sing) maximum gain and, though slightly on the low side for
best interference rejection, this is of little consequence,
particularly if separate tuners are used for transmission
and reception.
There is one more useful result of this design proce-
dure to be noted—the condition bR,, = aX,, also gets
rid of the j terms so that there is no detuning of either
element by the other. In other words, it follows from
Figs 5.6 and 5.10 that the process of making both
-(a+jb) I, ,ieI2 elements individually self-resonant and bending the
ends together so that the currents are equal should
automatically ensure infinite rejection of signals on
bearings of about —130° relative to the beam heading.
,
A

Fig 5.9. Derivation of [> from /,. A fraction a of I, has added to it at The radiation resistance is exactly double the values
right-angles another fraction bso that OA = (a = jb)/,. Marking off read from Fig 5.4 apart from any corrections which may
an equal length OA’ in the opposite direction, we obtain —(a +
jb)l,. Suitable choice of a and b provides the desired phase angle
be necessary in accordance with the rules stated earlier.
gq. Since we are dealing with a right-angled triangle OA = The use of two feeders allows remote beam reversal
|, V(a? + b?) and it is only necessary to put a* + b? = 1 to make /, and separate tuning for reception and transmission,
and /, equal in amplitude though the nulls will not be complete for bearings other
than —130° unless both phase and amplitude are ad-
from the point of view of an element having another
justed, eg by the use of a phasing unit as described on
one alongside it carrying some particular value of
pil.
current, regardless of how this may have come about.
Referring back to (5) and (6) it will be seen that if X,,
To obtain the conditions for equal amplitudes one need
is negative (ie capacitive) and b is positive (phase
only put
advanced) the radiation resistance of the driven ele-
—f,
ment is equal to 2bX,,. If on the other hand we want to
es a+ jb Hila (a 18) know the radiation resistance of the retarded element
substituting this in equation (1) we obtain (reflector), b must be replaced by —b in which case (6)
and (7) are no longer satisfied and the effect on (5) is to
R + jX = (a — jb)(Rm + jXm) make E/I, = 0, ie the radiation resistance is zero. In a
aR,, + bX,, — j(bR — aX) way this is a glimpse of the obvious since if the
adjustment is already correct the last thing one wants to
and since the “real’”’ and the ‘‘j’’ terms have to be equal
do is either to supply power to or take it from the
to each other separately (because there is no other way

HESS
of constructing the triangle)
R = aR,, + Xn (6)
X ='aX,, — dR,, (7)
Notice that (5), (6) and (7) are essential conditions
but it does not matter how they are satisfied. In
principle one can use parasitic operation with either
element driven, or drive both elements, but in practice
there are restrictions. In particular, since aR,, is always
less than R, condition (6) can only be met if bX, is
positive; thus if X,, is negative b/a must also be negative,
which means that in this case parasitic elements must be
tuned as reflectors. It is of interest to note from (7) that if
aX,, = bR,,we have X = Oso there is no detuning of the
reflector and the beam can be reversed merely by
switching from one element to the other. From (6) we find
this implies a further condition a = R,,/R.
Fig 5.10 shows the variation of @/q@, with a for 0-6 0:8
spacings between 0-1A and 0-19), the point correspond- a(=cos¢)
ing to a = R,,/R being marked on each curve, from
which it will be seen that the X = 0 requirement is met Fig 5.10. Variation of ~/q@g with a (or cos ¢) for various spacings.
The dotted line links the points for which b/a= F,,/R, showing
for ~/@) ~ 0:63 over the whole range of spacings. This that no detuning is required for the parasitic element provided
is sufficiently low to provide a close approach to @/Pg is made equal to 0-63
74 HF ANTENNAS FOR ALL LOCATIONS

reflector. On the other hand, if equations (6) and (7) to simplify the calculations and created a situation
are not satisfied the driving of both elements is one way which is extremely difficult to analyse.
of putting matters right. In this case (with X,, negative) Let us suppose then that, realizing its importance,
we have from (5) for the radiation resistances of the the necessary effort has been made to acquire,a Zp of
two elements: 10192. Calculation can now proceed, the total radiation
resistance being equal to 146 so that the current is
Rai ie ier aRn aL bX in
(8) that of a dipole divided by V2, and the two fields add in
Rret = R—- aR, wai bX in phase to give a gain of only 3-0dB over a dipole.
Suppose that, nevertheless, this very poor performance
When these are added X,, disappears, ie is accepted in return for the apparent ease with which
the beam direction can be reversed. Now we could be
Rrotai = Rair cts Reeg = 2(R ae aR») a 2(R "2 in COS %
9
in trouble again since the calculations have assumed
resonance, ie any reactive components of impedance
Note that X,, can cause one resistance to be zero or have been tuned out. Note that (5) was derived from
negative but has no effect on the total. the ““Ohm’s law” relationship EF = 1,R + 1,Z,, which
The usefulness of these equations is illustrated by the for this case is equal to /,(R + X,,) so that in fact there
following example. In one well-known type of driven is no reactance and no problem but this again is a stroke
array, elements are spaced /4 and a phasing-line of of luck, deriving from the condition a = 0. In general
electrical length \/4 is used between them in the belief (+) reactance is transferred, the sign being reversed
that this produces 90° phase shift. Such arrays are when the direction is reversed, so that if the beam is
commonly assumed to be reversible. For the moment required to be reversible any attempt to tune out the
let us assume that equal currents with a phase differ- reactance only makes matters worse as shown below.
ence of 90° (ie b =+ 1-0, a = 0) have been achieved; For the next example, consider a spacing of \/8 and a
in this case aR,, disappears and for a dipole X,,, from phase shift of 45°, ie the antenna is tuned as before for
Fig 5.3, is equal to —28Q. Putting b =—1-0 so that the infinite rejection in the direction opposite the beam
element furthest from the transmitter lags in phase and heading. We have a = b = 0-707 and, from Figs 5.2
therefore acts as a reflector, we have for dipole ele- and 5.3 respectively, X,, = 0 and R,, =64-5. The radia-
ments R = (73 + 28)Q@ for the front element and tion resistance is 73 — (0-707 X 64:5) = 27-4Q but
(73 — 28) for the rear element, ie 10192 and 45Q putting
respectively. EF=]R+ LR, = LR -— (a+ jB)Ry
Reference to Fig 4.4(b) shows that provided the = 1,(R — aR,, + jbR,)
phasing line is terminated in a resistive impedance a
length of A/4 will indeed produce a phase difference of we find there is a reactance term of 45-69 to be tuned
90°. (Note however that this is a special case and does out. For the reverse direction the reactance is reversed
not imply for example that a line of length 4/8 will in sign, and unless these reactances are tuned out there
produce 45° phase shift). What we now have is two will obviously be a very large swr which can be found by
pairs of terminals between which the impedances are entering the Smith chart (p59) at a point
respectively 10192 and (Z)7/45)€2 where Z, is the char-
acteristic impedance of the phasing line. Ns pe
27-4
geet ae :
It must next be noted that the equal-current condi- which corresponds to an swr of 4-5. If the j term is
tion requires the reflector to receive less power in the tuned out for one direction (it cannot be done for both)
ratio 45/101. This will be achieved by connecting the beam reversal will result in twice the amount of detun-
two impedances in parallel provided Z)2/45 = ing, and the j term is therefore doubled, resulting in an
101 x (101/45), ie Z) = 101Q and the reflector impe- swr of 12!
dance then comes to 227. Unfortunately for this Another interesting aspect which may be noticed
example, 1012 is not a standard value of impedance and with an swr meter connected in the reflector lead is
so far as is known to the author, the required conditions reversal of the direction of power flow as the reflector is
have never been correctly implemented. Indeed a Z, of tuned, ie R,.¢ goes negative. Lack of reference to this in
300 or 600 is most usual and clearly quite useless since the amateur literature suggests that those who have
the impedance which would be presented by the reflec- observed it may have judged it best to keep quiet!
tor line (if it were possible by some extraneous means Nevertheless if it causes surprise and concern this can
to satisfy the required amplitude and phase conditions) only be due to the failure to identify amateur beams
would be 2,000 or 8,000Q respectively. Nevertheless, with “‘supergain”’ antennas, for which negative values
the fact that very little power flows along the feeder of R are a typical feature [1]. Note that if X,, is positive
going to the reflector may leave this free to operate as a (inductive) parasitic elements should be tuned as
parasitic element. In practice, failure to establish the directors not reflectors.
correct conditions has invalidated the assumptions used Suppose that despite X,, being negative we want to
 CLOSE-SPACED BEAMS 75

tune a parasitic element as a director. From equation and the use of straight or inverted-V elements with
(7) we require X = aX,, + DR, = —2aX,, if nothing open-wire resonant feeders joined in “‘parallel anti-
else is changed; this implies considerable shortening of phase”’ and fed slightly off-centre has been the basis
the parasitic element but from (2) there is then, as of a number of systems described by the author [2, 3, 4]
might be expected, a reduction in the value of J, and reviewed briefly below. A limitation of this
relative to /,. This is exactly what happens in the case of approach arises from the desirability of avoiding ex-
_ the quad where the negative reactance is due to the treme values of swr even in open-wire lines and the
wide spacing between the top of one loop and the difficulty of reconciling this with requirements for
bottom of the other, negative reactance being typical of compact multiband systems.
spacings between about 0-15A and 0-65) as shown in Fig Some readers may be worried about extra losses
5.2. In the case of the quad it is a matter of common when two feeders are used, and there is some justifica-
observation that despite the impressive performance of tion for this. Suppose that having met the X = 0
reflectors attempts to retune these as directors are condition there is a current J flowing in each element,
singularly unsuccessful. and let the feeder length be a whole number of
Similarly with 4/4 spacing of dipoles it is well known half-wavelengths so that for correct operation a short-
that reflector operation is more efficient; meanwhile circuit is required at the lower end of the reflector line,
director operation is impossible with the VK2ABQ the wave travelling down it being totally reflected. This
(which has element ends very close together) and the establishes a sinusoidal current distribution in the
miniature beams described later. The opposite effect feeder, J being the maximum value and [7/2 the mean-
has been observed in the case of a single parasitic square value. Since R is the same for both feeders and the
element used with a collinear pair, indicating a positive losses in the main feeder are given by /?R this represents
value of X,, though it has to be confessed that the a 50 per cent increase in the total feeder loss. Subject
reason for this is still obscure. to the practical compromise suggested in another
It will be clear by now that there is no difference in chapter (p39) the length of coaxial line will not exceed
principle between a driven array and a parasitic array 40ft (12:2m), so that assuming a typical loss of 1-0dB/
but the practical differences are very important. To 100ft (30m) at 30MHz the extra loss comes to 0-2dB in
realize the above example in practice, both elements the worst possible case, which is not unduly serious. It
must be driven, otherwise the condition X,, = 0 means may be interesting to note in passing that with X,, = 0
that J, cannot equal /,. Returning to the above exam- the feeders share the work equally and there is no extra
ple, the feedpoint impedance of element 1 is given by loss, except that due to the high swr in the event of
beam reversal being required; this also comes to 0-2dB.
s&h
F= R- aRy + jbRn (10)
1

Two-element driven arrays

= 27-4 + 45-6] ohms Many experimenters have used driven operation of
close-spaced elements (including elements spaced A/4)
to overcome the limitations of parasitic operation with
For element 2, b is negative and we have
conventional designs using straight elements.
The development of controlled coupling as just de-
© = 27-4 ~ 45-6) ohms scribed casts some doubt on the continued need for
2. driven systems but does not rule them out as alterna-
tives, and a combination of the two methods might be
It follows from this that if each feeder is made equal to expected to reduce the need for working to close
an integral number of half-wavelengths and the system tolerances, a matter of particular importance in the
fed off-centre to obtain the required phase shift, the design of very small beams as will be appreciated in due
impedance Z; seen by the transmitter in the absence of course. The usual method is to employ a phasing line
retuning is (27-4 + 45-6j)Q in parallel with (27-4 — with its length ‘‘equal to the spacing” so that a null is
45-6j)Q. obtained in the —180° direction, but this usually makes
The system as a whole is therefore resonant and with matters a lot worse for reasons which by now may be
the help of the chart it is found that Z; comes to 35Q, obvious to the reader. This is anyway the wrong
so that for 50Q line the swr for the whole system is 1-35 direction for the null even for best interference rejec-
for both directions of fire. This unfortunately applies tion, but more serious consequences arise from the
only to the short length of line between the transmitter misconception that a line length of A/8 produces a 45°
and the phasing device, and effectively conceals the phase shift and pro rata. As explained in Chapter 4 this
undesirable situation in each of the individual feeders. happens only if the line is matched, and most of the
The resonant condition is nevertheless of some interest, arrangements thus far described have an inherent
76 HF ANTENNAS FOR ALL LOCATIONS

I +X case of close-spaced beams, however, large phase shifts

are not required, and when mutual impedance comes
into the picture the “+X” method takes the situation in
its stride. Ignoring for the moment mutual coupling
between the loads, the generator ‘‘sees’’ an impedance
(R + jX) looking one way and (R — jX) the other so
that the magnitude of the two impedances (ie
V(R2 + X2)) is the same in both cases, the total phase
shift @ being given by 2 tan-!(X/R).
This principle was invoked by the author in the
design of his very first close-spaced beam, the consequ-
ences being even more startling though a great deal less
useful than the findings described in the last section,
creating the impression of some extremely powerful
demon in charge of the whole proceedings. To put it in
yoretion of beam
a nutshell the system:
(a) worked perfectly
Fig 5.11. (a) The resonance or “+X” method of phasing; total (b) ceased to work at all when the frequency was
phase difference between the currents in the two loads (RA) is shifted a mere 10kHz or thereabouts!
given by tan (q~/2) = X/R or approximately by p = 2X/R radians if X
is less than about A/2. For parallel \/2 dipoles spaced )/8 and fed At that time phasing-line systems were much in
out of phase R = 137:-5Q, the mutual resistance being added to vogue, the desired total phase shift of 135° being
the radiation resistance. Note that with equal values of AR the obtained sometimes with an in-phase connection and a
currents are equal only if the reactances are of equal magnitude.
(b) Application to the W8JK antenna. The total electrical length 3/8 phasing line, sometimes with out-of-phase connec-
from the ends of one element through to the ends of the other is tion and a i/8 phasing line. The author had applied
exactly )/2. Displacing the feedpoint about 43in (11-4cm) from the similar thinking to his reactance method, opting for the
centre line was found to produce the desired directional pattern.
CF provides the reactance X which amounts to +21. Due to the in-phase connection. A large amount of algebra aided
short elements FR is halved, giving a value of 0-62 for g/q by a good deal of luck finally produced an explanation
corresponding to a gain of 5dB since the phase shift was found to be proportional
to X/(R —R,,) for the in-phase connection and
mismatch (though the operator may be unaware of this X/(R + R,,) for the out-of-phase connection. Hence for
because the swr meter is connected in the ‘““common”’ the first case adjustment was more critical in the ratio
lead where, as we have just seen, the two mismatches of (R + R,,)/(R — Rp), ie 137-5/8-5 or 16 times! Re-
may compensate each other). versing the connection to one element put matters
Due to the mismatches the “‘phasing-line” is more right, theory being further vindicated by reversal of the
likely to result in large current inequalities than it is to beam.
produce the desired phase shift, and a better method This arrangement was similar in principle to the one
makes use of the resonance principle based on equal illustrated in Fig 5.12 which serves to demonstrate the
and opposite reactances as illustrated in Fig 5.11. Given essentials of phase control in tightly-coupled systems
that wl = 1/wC, which is the standard condition for regardless of the method used. An essential precondi-
resonance, the generator sees equal but opposite impe- tion is overall resonance for the two elements plus the.
dances (+X) to the left and right, and the same current line joining them, so that off-centre connection of the
flows in each load R. A total phase difference @ is feeder, eg at AA’ or BB’, is equivalent to inserting a
obtained by putting tan (@/2) = X/R. If the loads are reactance +X for one direction and —X for the other.
unequal, equal currents can still be achieved by using Resonance is easy to achieve since if the system is tuned
suitable values of L and C, and any reactance seen by low this brings the shorter side closer to resonance so
the transmitter as a result of this operation can be tuned that more current flows, and vice versa. The current
out in the usual way. Fig 5.11(b) shows this idea applied can be checked with a probe (p237) or with a flash-lamp
directly to the W8JK beam for converting it into a bulb bridged across a few inches of conductor. Depend-
phased array [4], but at this point the argument has to ing on the result of this test, the loop CABC’ can be
be elaborated to take account of mutual impedance. lengthened or shortened as required.
If elements are widely separated so that Z,, = 0 the In view of the low radiation resistance and use of
phasing-line method has the advantage that any. resonant feeders the bandwidth is rather narrow as
amount of phase shift can be introduced without alter- explained in Chapter 8, so that for a practical system it
ing the impedance seen by the transmitter, though the was found better to use folded dipoles (p40) which
“+X” method is more suitable for providing precise incidentally increased the length AB (which can be
adjustment of phase over a limited range such as +45°, regarded as a phasing line) from about 16in (0-4m) to
a 2:1 mismatch being incurred for a 90° shift. In the Sft (1-5m). However, using inverted-V folded-dipole
 CLOSE-SPACED BEAMS 77

connected in parallel antiphase, the physical length of

the phasing line being }/8 and the feeder connected
directly to one element which is somewhat shorter than
the other. These arrangements, which include the
well-known “ZL Special’, are difficult to analyse,
though in at least one case the element details appear to
Beam direction when
be about right for normal parasitic operation and the
feeder is connected phasing line would most likely be playing a minor role.
to BB’
Another possibility consistent with some rather low
gain figures quoted for the “‘ZL Special” [5] can be
inferred from inspection of Fig 5.4, which shows that
with increasing @ one enters a region of operation
where gain is low but front/back ratio still reasonably
good, and since radiation resistance is high adjustment
should be relatively non-critical.
Electrical
|
centre of system
Some of these arrangements closely resemble two
adjacent elements of a log-periodic system as described
Fig 5.12. Two-element array with resonant (or mismatched) in Chapter 14 and, designated as “‘log cells’’, have been
feeders. If the elements are resonant the length CABC’ must be
an even number of half-wavelengths. AB can be regarded as “the used for the excitation of large arrays with a number of
phasing line” but is much less than )/8 if open-wire lines are used. additional parasitic elements such as the one described
Note that the feeders are crossed over as for the W8JK antenna on p170. In these cases the requirements for a symmet-
rical current distribution are presumably met and, in
elements spaced )/8 it was found that adjusting for considering phase shift, the log cell cannot be isolated
current equality in the centre of the 14MHz band from the system as a whole so further discussion here
resulted in a ratio of 0-7 at the band edges so that the would not be appropriate. Another type of phased
problem was still not fully resolved. The arrangement array using the resonance principle is the Swiss quad
shown in Fig 5.11(b) is somewhat worse than this, [7], in which the reactances are obtained by making one
radiation resistance being divided by eight whereas the loop larger than the other, though the same result plus
total wire length is only divided by four. the possibility of beam reversal can be obtained by
It is of interest to note that in this case the effective off-centre feed on the lines already described. How-
phasing line length is 9in (13cm), being equal to ever, quads appear to have acquired their reputation as
CF x 2. This was discovered experimentally in the first a result of the tighter coupling which makes driven
instance and later checked by calculation, thus casting Operation unnecessary provided the spacing is correct.
more than a little doubt on the usual assumption that a For parasitic operation the 0-1 spacing of the Swiss
length of 4/8 (or 8ft (2-4m) for 14MHz) is always quad would be much too close.
correct! Summing up, it can be stated as a broad conclusion
The tuning procedure described above has also been that to the extent that mismatches exist (as they usually
found usable with i/4 spacing despite the reactive will) it is essential to apply the principle of resonance-
coupling, though full analysis of the system under these plus-symmetry as exemplified by Fig 5.12, though this
conditions is rather complicated. In cases which have can probably be regarded as an extreme case.
been described of elements spaced 2/4 (so that the The above considerations, coupled with the need to
coupling includes negative reactance) and fed with make sure of correct operation in the case of very small
separate coaxial lines, most of the work will be done by beams, led the author to devise the lumped-circuit type
the feeder going to the forward element. With the other of phasing line shown in Fig 5.13. Exploiting the
feeder having, more or less, an “‘idling”’ role only, the resonance (+X) principle, the tuning control is used to
adjustment should not be unduly critical; however, as obtain amplitude balance with the coil taps providing
demonstrated by Figs 5.7 and 5.8, 4/4 spacing is too phase adjustment and beam reversal. This method,
wide. though particularly suitable for use with close-spaced
With 1/8 spacing the elements could each, in princi- beams, is of universal application and can be adapted
ple, be separately tuned and matched but it seems a by means of differential capacitors or ferrite tuning
pity not to take advantage of two feeders for beam slugs to provide a continuous fine adjustment of phase,
reversal. In this case one needs to start with elements though large phase changes result in a mismatch.
which are individually self-resonant and accept a fairly Despite these merits and the fact that the author still
high value of swr in each feeder as explained earlier in likes to ‘‘have it there just in case”’, the prototype line
this chapter. quickly found itself demoted to the role of remote tuner
A number of arrays have been developed in which for a parasitic reflector, this mode of operation proving
elements of approximate length A/2 and spaced 1/8 are simpler and equally good. Nevertheless, if the feeder
78 HF ANTENNAS FOR ALL LOCATIONS

To element (1) To element (2) are additive whereas for the reverse direction the phase
lead of the current is cancelled by the lag due to the
radiation now having further to travel. With two ele-
ments, if the currents and for some direction or other
L the two phase shifts are exactly equal, then for that
direction there must inevitably be zero signal as demon-
Beam reversing switch Beam direction With —_ pa
switch as shown strated by Fig 3.7.
To transmitter Reflectors operate on the same principle as directors
except for inversion of the phase relationship and
reversal of the direction of fire, the operation being
Fig 5.13. Phasing network for two-element beams. Coaxial outers
are bonded together. Elements of close-spaced arrays must be strictly akin in the case of two elements to the balancing
antiphase connected, so that if the inner or the coaxial feeder of an ac bridge. If balance is poor this suggests a flaw in
goes to the right-hand side of one element it goes to the left of the design, eg an insufficient range of adjustment has
the other. L and C are chosen to resonate at approximately the
frequency in use, but detuning in conjunction with adjustment of been provided.
tap positions can be used to obtain the required amplitude Until the comparatively recent emergence of anten-
balance combined with correct phase adjustment despite modest nas such as the VK2ABQ, the hands of antenna
errors in tuning or matching. Typically for 14MHz, L = 1-6uH and
2C = 200pF (maximum). Method can be adapted for any type of designers (including the author) were tied by the
feeder. Too high a value of L results in excessive rf voltages customary assumption that the right shape for antenna
elements (loops excepted) was “‘straight’’, despite the
length is unsuitable or the coupling badly out, such a rather obvious necessity demonstrated earlier in this
phasing device provides a convenient remedy, and if chapter for increasing the coupling between them.
switched into circuit for reception only can be highly It still remains for full advantage to be taken of
beneficial as a nulling device for interfering signals. controlled coupling. This is one of the main concerns of
Front/back ratios in the region of 25—30dB are read- Chapters 12 and 13 since it is felt there need no longer
ily obtainable with parasitic operation and, although be any excuse for continuation of the relatively poor
driven arrays can do better by providing the operator performance hitherto associated with two-element
with control over amplitude as well as phase, the parasitic dipole arrays. To summarize and interpret the
benefit from this is restricted for two reasons: conclusions from the early part of this chapter, it would
appear that each element needs to be resonated in the
(a) high rejection ratios demand a critical balance
band centre with the other element open-circuited. By
which holds only for a very small range of frequen-
adjustment of spacing between the adjacent ends to
cies or arrival angles;
give equal currents in the elements, it should then in
(b) interference is often from more than one source or,
theory be possible to place deep nulls in the 130°
if from a single source, the arrival angle may be
directions, though some slight retuning or change in
varying or there may be more than one arrival
the angles of the nulls may be needed in practice.
angle.
One other point needs to be cleared up. Reference
Adjustment for suppression of a particular interfer- has already been made to the non-reactive character of
ring signal is unlikely to yield maximum gain for the mutual impedance in the case of 4/2 dipoles spaced
transmission, and a fully engineered system needs to by 1/8, and this leads also to a symmetrical situation in
include separate phase and amplitude controls for the case of parasitic excitation, performance being
transmission and reception. identical irrespective of whether the parasitic element is
tuned as a director or reflector. This is a mathematical
identity which, in conjunction with the independence of
Parasitic arrays size demonstrated earlier, invalidates claims of better
These have in most cases only one driven element, the performance in the case of closely spaced directors.
others being energized from it by mutual coupling, and The origin of these can be traced back to what is
are commonly known as Yagi arrays after one of the evidently a slight error in the original 1936 paper by
pioneers in the field. Brown [6] who first drew attention to the possibility of
Parasitic elements take the form of single reflectors, obtaining gain with closely-spaced beams.
placed behind the driven element, or directors (of To confirm this Fig 5.14 was calculated from the data
which there can be many) placed in front. Starting from in Figs 5.2 and 5.3 for a two-element parasitic beam
the antiphase condition, radiation from a director is having a spacing of 0-05/, and it will be seen that for a
advanced in phase because (like the front element of a director the gain conforms very closely to the driven
driven array) it is “nearer the receiver”, and also element case from Fig 5.4. The depths of null,
because of detuning in the capacitive direction (by though reaching infinity only at one point, are very
shortening) which gives a phase lead to the current considerable.
flowing in it. For the forward direction the phase shifts The improvement in directivity compared with more
 CLOSE-SPACED BEAMS 79

<i be aaa
Bs a tine Gain (director) Fegilheed abel spacing, a small increase in R being due to the overcou-
pling. Such low values of R imply very close tolerances,
ET PT fecitrencetoo] || narrow bandwidth and possible losses as explained in
Zi Infinity ie
Chapter 3, so that spacings as close as this are not
at Pb)
=0-15 ed recommended.
Though admittedly feasible with careful design, there
(dB)
ratio
Front/back are usually better options available, in particular those
aestas
hes= A/20 spacing
iat
(director) based on the use of bent elements as discussed earlier in
(dB)
Gain
naresh of nulls with A/8 iio
Ty this chapter and illustrated in Chapter 12. It is never-
theless interesting to note that director operation with
i aesdits (reflector or director)
closer-than-usual spacing is one method of obtaining
“
~ ce
= S =
Al20
spacing i pe tighter coupling and improved directivity, a point clear-
A
(ohms)
(reflector) ahi eo Radiation Radiation
resistance
resistance ly brought out in Brown’s paper [6], notwithstanding the
slight error (+0-3dB) in the actual gain figures. This
was quickly put to use in the amateur field by Van
Roberts, W3CHO, who developed a two-element beam
Ratio of phase shift/spacing (¢/%,)
using a close-spaced (0-14) director and having 5dB
Fig 5.14. The effects of very close spacing; curves are calculated gain with 17dB front/back ratio [10].
from mutual impedance data, the director gain curve being
almost identical with that for equal currents, Fig 5.4. Failure to
achieve an infinite depth of nulls is due to slight overcoupling and
a small increase of spacing should correct this. The curves
Three-element arrays
demonstrate the superiority in all respects of director operation In principle, similar reasoning can be applied to close-
when mutual reactance is positive, though the gain difference is
small. The radiation resistance is almost identical for reflector spaced beams with additional elements but calculation
and director operation, the inferior performance of reflectors is much more difficult and it is simpler to rely on design
being due to too low a current relative to that in the driven charts such as Fig 5.16 and 5.17. It is however easy to
element. The dotted curve shows for comparison the depths of
null for a spacing of i/8 see that a three-element beam is capable of producing
an infinite front-to-back ratio in a manner analogous to
widely spaced parasitic elements (Fig 5.5) is due to that of a two-element beam if the outer elements are
tighter coupling and in part to the positive X,, term, each supplied with half the current of the centre
which operates as explained earlier to improve the element and subjected to 45° phase shifts in opposite
performance of a director and degrade that of a reflec- directions. This special case is easy to calculate and as
tor. For a director the coupling is slightly in excess of shown by Fig 5.15 provides a very considerable im-
critical, reaching a maximum value of 1-035 for X = 0, provement in rejection of unwanted signals, though the
corresponding to @/q@,) = 1:0. The radiation resistance gain is not impressive, being only 4-6dB, an increase of
curve follows fairly closely that of Fig 5.4, subject to 0-4dB compared with two elements tuned in the same
multiplying the latter by two to obtain the total resist- way.
ance and dividing by (2:5)* to allow for the reduced There is however no reason for supposing this to be
an optimum design, particularly in view of the wide
range of alternative options indicated by the charts
from which it will be evident that any attempt to
approach the maximum gain of 7-6dB, which is theoreti-
cally possible with equispaced parasitic elements [21],
results in a very low value of R, about 24Q only. From
inspection of Fig 3.10 it looks as if it should be possible
to keep the loss resistance down to something of the
order of 0-202, assuming the use of aluminium alloy
tubing with an average diameter of about 0-5—0-75in
(1:3-1-9cm). This gives an efficiency of 95 per cent and
an effective gain of 7-3dB, although 6dB or thereabouts
field
forward
of
Percentage
is the most that has normally been achieved in practice
90° ees 180° [21].
Degrees relative to beam heading ——>
This discrepancy is almost certainly due to the high
Fig 5.15. Comparison of two- and three-element arrays for rejec- price which has to be paid for maximum gain (in the
tion of interfering signals from the back direction when the phase form of reduced bandwidth), and the difficulty of
shift is equal to the spacing (q/q~g = 1). Currents adjusted to give keeping manufacturing, erection and maintenance
null at 180° in both cases. Corresponding gains are 4-2dB and
4-6dB. Average levels shown by dotted lines indicate 16dB tolerances within sufficiently close limits. It can be
improvement with three elements overcome to some extent by using additional feeders to
80 HF ANTENNAS FOR ALL LOCATIONS

overlooked that in general for.a given cost, and degree

of practical difficulty, the lighter the beam the higher it
can be erected; in terms of practical results this may be
worth a lot more than additional elements.
+20 The optimum tuning does not vary greatly for the
different spacings, and it is therefore possible to use a
simple practical design chart (Fig 5.17). An array made
with its element lengths falling in the shaded regions of
the diagram will normally give good performance with-
oO
out further adjustment. The length of the director may
be decided in advance and the reflector may then be
adjusted experimentally to improve the front-to-back
ratio. It will be seen that the radiator is somewhat
DIRECTOR
REACTANCE
—20 longer than a normal dipole; this is because the parasi-
tic elements have a detuning effect on it. The addition
of a second director also spaced 0-14 to 0-15A will not
materially affect the above recommendations but major
changes of shape such as bending the ends inwards will
-40 invalidate use of Fig 5.17.
It will be appreciated that with three elements, as
with two, the reactive coupling plays a major part in
—20 Oo +20 +40
REFLECTOR REACTANCE determining the relative currents and as a result the
experimenter aiming for a particular working point on a
Fig 5.16. Contour chart showing contours of gain (solid lines) and
input resistance (broken lines) as a function of the tuning of the diagram such as Fig 5.16 must expect to find difficulty
parasitic elements. This chart is for a spacing of 0-15) between unless he sticks to the use of straight elements spaced
elements, but is also typical of arrays using 0:2, + 0-1) spacings around 0-125—0-15\. He may do equally well but this
will depend on achieving rough equality of currents in
effect remote fine-tuning of the parasitic elements but, the parasitic elements, and because of low current in
although it might be thought possible to enlarge on the director no improvement in gain can be expected
such a facility to obtain instantaneous reversal of beam
direction, it has been found rather easy to incur sub-
stantial losses unless the range of adjustment is kept to
a minimum. geet I N GZ
Referring to Fig 5.16 with these problems in mind, OUUOU:
Viiimamunare
NSSOy ue Uualaluads mee
yy
ame
the lower right-hand corner of the chart is clearly the
TTT |asomron [TT
region to aim for, where the gain is still over 6dB while
yiaeeequeen
; RSSNY
7 =
RS Ne AUT
L > @oSo iy w gasere= Ade

the feed resistance approaches 20Q. This region of the Son

chart corresponds to rather long reflectors (+40 to
+50Q reactance) and rather shorter than optimum . eectanil
G44!
directors (—30 to —40Q), and also is found to be the
region of best front-to-back ratio.
Although the chart is for antennas with both spacings
equal to 0-15A, the resistance and bandwidth are both FACTOR
TUNING
somewhat improved, with the gain remaining over 6dB
excluding losses, if the reflector spacing is increased to
0-2 and the director spacing reduced to 0-1. Such an
array would operate satisfactorily over at least half the
28-29MHz band or the whole of any other band for
which it is likely to be constructed.
CONDUCTOR LENGTH/DIAMETER RATIO
The overall array length of 0-3A is practicable on all
bands from 14MHz upwards, though the 20ft (6m) Fig 5.17. Design chart for Yagi arrays, giving element lengths as a
boom required for a 14MHz array is rather heavy for a function of conductor length-to-diameter ratio. The tuning factor
Lis divided by the frequency in megahertz to give the lengths in
rotary array and there is a temptation to shorten it. feet. These curves are for arrays of overall length 0-3\, with
Spacings should not be reduced below 0-1A for either reflector reactance +40 to +60Q and director reactance —30 to
reflector or director as the tuning will again become too —40Q, and give arrays of input impedance between 15Q and 20Q.
Element lengths which fall within the shaded areas will give an
critical and the transmitter load unstable as the ele- array which can be used without further adjustment, though the
ments or the feeder move in the wind. It should not be front/back ratio may be improved by adjusting the reflector
 CLOSE-SPACED BEAMS 81

hana] ik RPO 90°. Thus in case (i), assuming a square shape and
putting @ = @p, there is 90° phase difference between
the element currents; referring to Fig 2.5 and allowing
(c) = (d)
for phase differences, relative fields for the forward and
endwise directions are given by 0-7 + 0-7 = 1-4 (for-
ward) and 0-15 V2 x V2 (endwise), ie a ratio of
1-4/0-3 or 13-4dB in favour of the forward field. Since
the bent-over ends constitute a W8JK-type pair, about
3dB gain must be assumed, so that the ratio of endwise
energy to forward energy is (0-3/14)7/1-4 or 3-3 per
cent, a loss of 0-14dB only. This figure escalates rather
rapidly if the bending is increased further. A pair of
V-shaped elements may be mounted so that they hang
* \Y | Sede down as inverted-V’s from a single spreader, coupling
being adjusted by bringing the ends together so that the
bird’s-eye view resembles Fig 5.18(k).
Fig 5.18(i) or (j) may be derived by taking a single
quad loop, Fig 5.18(c), tilting it over into a horizontal
position and placing insulators in the sides to make it
(j) & Zeeupies (k)
into two elements, Fig 5.18(j). This is the basis of the
VK2ABOQ antenna which is rapidly gaining in popular-
ity [11, 12]. The spacing as shown is wider than
Fig 5.18. Typical shapes of beam element; (a) to (g) are seen in
desirable but it has been found possible for a third
elevation, with the ends hanging downwards in cases (f) and (g); element to be fitted into the same space, thus yielding
(h) to (k) are plan views of V or U-shaped elements arranged in an efficient three-element design as described on p172.
pairs
Alternatively the square shape may be distorted into a
rectangle as in the modified version of the VK2ABQ
when a third element is added to the quad in the usual beam shown on p173.
manner. A similar conclusion may also be reached by
regarding the quad as a pair of Yagis stacked one above X-beams and slopers
the other but so close together that they almost merge
One omission from Fig 5.18 is the X-beam which has
into one as discussed in the next chapter.
been reserved for separate consideration in view of its
many attractions, combined alas with serious defects. It
Arrangement of elements has appeared in numerous forms including quad and
“sloper” varieties, but the main essentials are illus-
Typical shapes of elements are shown in Fig 5.18 and
trated in Fig 5.19(a) which also shows the radiation
there are few restrictions on choice, most types of
pattern for excitation as a W8JK type antenna. Though
element except ground-planes being suitable for use in
not normally used in practical designs of X-beam this
close-spaced beams. However, they must be properly
provides yet again a useful key to the understanding of
co-ordinated since it would not do to have, say, one
an important principle. The first point to observe is the
element vertical and the other horizontal. Although
basic symmetry which ensures that the antenna looks
loop and dipole elements are sometimes mixed in
identical when seen from any four directions 90° apart.
widely-spaced arrays, close spacing is difficult to define
Instead of the figure-of-eight pattern previously experi-
and almost certainly not advisable when one element is
enced with the W8JK connection there is now a clover
split into two portions separated by distances of the
leaf, ie two figures-of-eight at right-angles. If now a
order of 4/4, as is the case with the quad loops, Figs
phase shift is introduced between /, and —J,, there will
5.18(c) and (d), the delta loop (e) being roughly
no longer be complete nulls in the 45° directions and
equivalent.
the field in one direction will be given a boost at the
_ V-shaped elements may be mounted directly in line
expense of the opposite direction.
and facing the same way as in Fig 5.18(h), though the
Graphical integration with Fig 2.5 as a starting point
arrangements shown at (i) to (k) (or (f) and (g) on
has produced two results of interest:
which the ends are hanging down) are better since the
coupling between elements is readily adjustable for (a) for the initial clover-leaf pattern the width of each
meeting the optimum design conditions specified ear- lobe between “3dB points” is 42° compared with
lier. However, due to radiation off the ends it is 66° for the normal W8JK mode, but as there are
inadvisable for more than half the length to be bent four lobes instead of two one might expect a
over as in (g) or (1), or for the angle of the V to exceed decrease in gain of the order of 84/66 or 1-05dB.
82 HF ANTENNAS FOR ALL LOCATIONS

A Field increases in this

a
, direction
| a phase
if Iz is given
lag
or 1-15 for a single driven element, assuming that —
Lines of a po the coupling between elements is adjusted for
fire
equal currents. Since the centres are in close
proximity it should be possible to devise some form
(a) ies (
of inductive coupling for this purpose.
For the endwise directions the equivalent dipoles
almost constitute a conventional W8JK pair as
shown at the left of (b), but can be regarded more
accurately as a pair of “half W8JK’s” with a phase
difference @ which can (if one needs to be precise)
be taken into account by a small field-strength
correction factor, 0-92 for @ < q@ and 0-7 for
a hy
@ = 2p. The field for the endwise relative to the
forward directions, ignoring the correction, is given
by
\ Directions
a NS ES of
ra radiation
013A 0-352 NA
sin(@,/2)
an
(@=Phase lag
ae sin(P Po
Po)
relative to -I,)

and is plotted in Fig 5.19(c).

In Fig 5.19 we are looking down on what could be
any of several antenna systems. It could for example be
a pair of crossed i/2 dipoles cut in their centres and
re-joined to form a pair of V-shaped elements, though
by suitable feeding ‘‘plumber’s delight” construction
may be used and the cutting avoided. It could also be a
set of wires sloping down from the top of a mast, in
which case the four-way symmetry is an invitation to
(dB)
directions
indicated
use them for a beam system with directional switching.
Ratio
in
field
to
field
forward
of
It then becomes important to consider the worst direc-
O 0-5 1-0 1-5 2:0
(c) $1%o tions, ie those at 45° to the lobes shown in Fig 5.19(a).
Unfortunately there can be no front/back ratio for
Fig 5.19. (a) X-beam or “sloper’’ with elements in antiphase. (b)
shows (left to right) resolution of the cross into pairs of dipoles these directions, and any success one may have in filling
for calculation of front/side, front/back and front/45° ratios. (c) in the dips must inevitably lead to a worse radiation
shows variation of each of the rejection ratios with phase angle, pattern for the back directions. Nevertheless, practical
front/back ratio being identical with Fig 5.4
designs of X-beam include the use of sloping wires
which form part of the guy-wire system, and UA3IAR
[9] has devised a switched-beam quad operating on
(b) for evaluating performance in any pair of directions these lines.
the X may be replaced by equivalent dipoles as Unfortunately, looking at the right-hand side of Fig
shown by the dotted lines in the middle sketch of 5.19(b), it will be seen that for the 45° directions two of
Fig 5.19(b). The dipoles are assumed to support a the four half-elements have become invisible. The
sine-wave current distribution and their spacing is other two look exactly the same from either direction,
equal to 26 per cent of the length of a diagonal, ie this being proof that despite suggestions to the contrary
0-13 for a \/2 cross. When phase shift is introduced there cannot be any back/front ratio on these bearings,
the “boost” in the direction indicated means that the radiated field being proportional (for both direc-
endwise radiation is proportionately reduced and tions) to sin (q/2). It is given relative to the forward
most of the lost 1-05dB can be recovered. The field by
previous curves (Fig 5.4 et seq) are useable as a
fairly good approximation for obtaining gain, 0-7 Sin(@/o)
radiation resistance, and front-to-back ratio, pro-
vided q/q@p is greater than about 0-5.
sin(* 5 Ps)
Due to length reduction and a small correction
factor (0-88) due to the relative ineffectiveness of This is also plotted in Fig 5.19(c) which shows for
the centre portions of the elements arising from the comparison the front/back ratio based on Fig 5.4. To
close spacing, the radiation resistances in Fig 5.4 fill in the 45° nulls as much as possible it is clearly
must be divided by 2:3 if both elements are driven desirable to use as large a value of @ as possible, the
 CLOSE-SPACED BEAMS 83

front/back ratio for the UA3IAR quad being consistent Small beams
with a (q@/qo) of about 1-7. The gain from Fig 5.4 is The beam obtained by adding a third element to Fig
2:8dB, the radiation resistance for each of the two 5.18(j), which was mentioned earlier and is described in
driven elements being 50Q, ie 5702 read directly from detail in Chapter 12, occupies only half the area of a
Fig 5.4 divided by 2:3 for the X configuration but conventional “‘full-size’’ beam based on i/2 elements
multiplied by two as the elements are loops. These but differs only very slightly in performance. In effect,
figures are the basis for a viable switched-beam system the end halves of 4/2 dipole elements add relatively
provided that a ‘‘dipole’’ standard of performance little to the performance and therefore tend to waste
over four narrow arcs is acceptable. It seems difficult to space. If an even smaller beam is required it becomes
account for the reported drop of only 3dB for these impracticable to use three elements, except by com-
directions, particularly as it is depicted as being com- promises which lead to fairly rapid deterioration of
_ bined with reasonable front/back ratio. Possibly the performance with decreasing size. However, since the
results could be affected by vertically-polarized compo- gain advantage from the third element is normally only
nents of radiation which are likely to be appreciable in 1dB, restriction to two elements is not a serious loss,
the case of dipole slopers, though one would expect particularly as the necessity for two feeders, imposed
them to cancel for quad antennas provided a balun is by bandwidth problems, brings with it possibilities for
used! the “‘nulling out”’ of interfering signals as described in
To get X-beams further into perspective it is interest- Chapter 9.
ing to compare the crossed-dipole array with Fig It is reasonable to suppose the existence at all times
5.18(j), which uses identical dipoles but brings together of a standing requirement for beams to be as small as
the ends instead of the centres. This halves the space possible subject to meeting a given performance speci-
occupied, increases the radiation resistance by about 33 fication. This may well be set so high in terms, for
times and provides a much better radiation pattern. example, of operational convenience that it can only be
Alternatively the dotted lines in Fig 5.19(b) can be met by the use of full-sized monoband elements. On
replaced by dipoles of identical size and spacing, reson- the other hand, the need to retune the transmitter when
ated by means of bent-over ends; performance will be changing frequency within a band may be accepted as
almost identical but with elimination of the pattern part of the price for multiband operation or fitting an
defects and requiring only just over a third of the space, antenna into a small space. Often a rather modest
the radiation resistance being nearly doubled because reduction in size is accepted when a different method
the current distribution is rectangular rather than sine would have resulted in a much smaller structure provid-
wave. For “all-round switchable”’ directivity, vertically ing identical performance, or better performance for a
polarized arrays are more suitable and might be ex- given size.
pected as discussed in Chapter 10 to provide compara- It is not intended by this to suggest that a satisfactory
ble performance even when restricted in height. antenna should be scrapped merely because it could
To sum up, the very considerable attractions of have been better designed, and there are times when an
X-arrays derive from the single point of support and the “inferior” solution of a problem may result in so much
possibilities of directional switching; the price paid is saving of time or expense that it becomes the only
much larger size for the equivalent performance and a alternative to no solution at all! This is likely to apply
basically worse directional pattern. The reader may particularly in cases where it is desired to adapt an
well feel, particularly after perusal of Chapters 10, 12 existing antenna to operate on 10, 18 or 24MHz. A
and 13, that there are always better options available. reader conversant with the basic principles should have
“Sloper” arrays based on 2/2 dipoles sloping down no difficulty in working out such compromises if the
from a common point are becoming increasingly popu- need arises, and for present purposes the emphasis will
lar, particularly on the lower-frequency bands. Despite be on making beams as small as possible.
some similarity with X-beams, including some features Yet again the W8JK beam is a good point from which
of the directional patterns, points of maximum current to start, being still widely favoured despite a typical
are widely separated and they are thus exempt from feedpoint impedance of only just over 4Q for the two
criticisms relating to low values of radiation resistance. elements in parallel. Nevertheless, it has already been
Inevitably, however, they result in two modes of radia- shown that putting in a phase shift equal to the spacing
tion, acting both as low-height horizontal antennas apart from other advantages multiples the radiation
suitable for short-range high-angle contacts and vertical resistance by four, and the top curve of Fig 3.12 shows
radiators having a reasonable dx capability; it is clearly that the length of elements can be shortened to a mere
better for dx working to bring the lower ends back 30 per cent of i/2 if division of the radiation resistance
towards the mast as described in Chapter 13, p190, so by four is acceptable. This merely restores it to the
that as much power as possible goes into the vertical W8JK value, so that if satisfactory performance is
mode. A separate low-height inverted-V dipole can be assumed in the case of the W8JK it must follow that a
used to take care of any short-range requirements. phased beam having elements only 0-15A in length (10ft
84 HF ANTENNAS FOR ALL LOCATIONS

has been calculated [14] for 14swg (2mm) wire elements

12ft (3-66m) long, spaced 8ft (2-44m), but is equally
applicable to a 10ft (3-1m) square design. The upper
solid-line curve is part of the gain plot from Fig 5.4
reproduced for comparison, from which it can be seen
that the loss of gain is very slight, and in the light of
discussion in Chapter 10 would be more than offset if
advantage can be taken of the smaller size to gain a few
0-62 loss resistance
foreach element
extra feet of height.
Front|back
(dB)
ratio
Fig 5.21, taken from reference [14], shows various
V2a
(dB)
dipole
Gain
over ways which have been used for folding or shortening
antenna elements in order to fit them into smaller
spaces. Of these only (b) and (d) have much practical
0-4 0-6 0-8 1°0 1:2 1°4 1-5
relevance to the task of trying to make beams as small
Phase shift] spacing as possible without loss in performance, though in some
Fig 5.20. Gain and front/back ratio for two elements with and cases combination of (c) and (b) to effect a modest
without losses, as a function of phase angle. Note that losses degree of additional size reduction may be acceptable,
tend to increase the optimum value of phase shift. The lower
curve assumes a spacing of \/8 and a loss resistance for each
particularly if stubs are used in place of coils. There are
element equal to 4 per cent of its radiation resistance as two imperatives which follow from the discussion of
measured with the other element removed (Ham Radio) losses in Chapter 3 (p26) and which may be stated as
(3m) for 14MHz) will perform equally well, provided follows:
the reduction of length is carried out in the right way. (a) The current distribution in the radiating portion of
To illustrate the possibilities of size reduction with- the antenna must be uniform.
out appreciable loss of gain, the lower curve in Fig 5.20 (b) The antenna must be brought to resonance without

gly |
on ae
eS , SS"
Fig 5.21. This shows 13 different methods for loading short elements. The merits of each type are discussed in the text. (a) Capacitance
plates consisting or wire grids. (b) \/2 elements with folded ends. The length must be increased slightly to maintain resonance. (c) Small
capacitance hats—the effective capacitance is enhanced by near-resonance with the inductors. (d) A loop with capacitance hats. This is
equivalent to a stacked pair of (b)-type elements with their ends in contact. (e) Centre loading with an inductor. (f) A \/2 element with a
folded centre. This is similar to (e) with a stub instead of the coil. However, the R values from Fig 3.12 are transformed by the stub to give
an even lower value at the closed end of the stub. (g) Loop equivalent to (e). (h) Resonant feeders. (i) Version of a two-band element as
used by DL1FK. The capacitor tunes the inductance of the centre of the radiator to increase its effective value at the lower frequency.
Series resonance of the capacitor with its connections shortens the electrical length for the higher frequency. (j) Lumped-circuit
equivalent of (i) as used in one form of the G4ZU minibeam. (k) Alternative version of (d) using linear loading at voltage maxima.
(I) Capacitance connected between points of maximum voltage. (m) Linear loading at a point of maximum current (Ham Radio)
 CLOSE-SPACED BEAMS 85

small as possible there are advantages in larger size,

About 18in (49cm)
Deltamatch doubled back to- particularly in regard to bandwidth and multiband
tex, xX wards the centre operation which are topics for later chapters. At the
smallest possible size it becomes essential to use sepa-
rate feeders to each element to allow remote adjust-
ment and avoid bandwidth restrictions but, as already
stressed, important advantages follow in any case from
the use of two feeders.
Loop elements as in Fig 5.21(d) are not envisaged for
(a) “minimum size”’ arrays since the construction becomes
very complicated if taken to the limit. The great virtue
of this type of element is that it leads to the smallest
possible diameter of turning circle, though owing to
their relatively large vertical extent such beams may
Dimension shown are still be considered “‘large’’ compared with ones using
for 14MHz
10ft
type (b) elements. With this restriction on the use of
kan) 10ft
small loops it can be assumed that neutralization will
( b) Aluminium
tubing not be required.
16ft (4-9m) of Of the other arrangements shown in Fig 5.21, (a)
wire in each load
provides maximum efficiency but is inconvenient, and
it is not practical to accommodate enough loading if the
size is reduced to a minimum. Types (c), (e), (g) suffer
from losses in the loading coils, greatly aggravated in
cases (e), (g) due to the triangular current distribution.
(61cm)
(f) is similar to (e), though use of the stub instead of a
coil roughly halves the losses, and it is often a conve-
nient method for achieving a modest degree of length
5aN
2ft
_
reduction not exceeding about 20 per cent.
In the case of resonant feeders (h) the losses in the
Feeder connections are
shown dotted dipole are multiplied in proportion to the total length of
a is connected to d and wire in the system, or alternatively the feeder can be
c connected to b to provide reckoned as operating at an swr well in excess of 100
the neutralizing capacitances Cipdtiee are novshown
with consequent large losses. Methods (i) and (j) have
Fig 5.22. Miniature beam with neutralization; (a) based on flat
been used for multiband systems including miniature
construction, illustrates the principles; (b) shows loading wires beams but involve quite large circulating currents in the
folded in vertical plane to achieve further size reduction and resonators and therefore increased losses. Furthermore
improved construction. For further details see Fig 12.14
the loading is essentially inductive so that the current
distribution is triangular and the radiation resistance
the use of power-consuming devices; this rules out
very low. Method (k) was the first small loop to be tried
the use of loading coils except possibly in an
by the author and was quickly abandoned though it has
auxiliary role after most of the loading has been
since been described elsewhere. Because very little
provided by other means.
current flows in the loading section the inductive effect
It is impossible to meet these requirements at the size is small and it was found to be almost identical with (d)
limit set by gain considerations alone without incurring which is a lot simpler.
severe overcoupling between the elements. This is be- Another proposed method, (1), breaks the cardinal
cause the large “‘blobs” of capacitance needed at the rule that paths carrying large currents must not double
ends of the elements for tuning them to resonance also back; if they do the current “‘flowing the wrong way”’ is
couple them tightly together. The answer to this prob- subtracted from the effective total and this brings the
lem is neutralization of the excess capacitance as shown radiation resistance down with a big bump! In this case,
in Fig 5.22. The method is identical to that used for assuming a 12ft (3-7m) square loop, the current
push-pull amplifiers, the current in the neutralizing through the capacitor is roughly equal to the loop
circuits being too small to upset the radiated field, and current so that the signal is equal to that from one wire
for monoband operation there are few problems [13]. considered in isolation. This divides the radiation re-
The arguments just presented might seem to suggest sistance by four so that the current is doubled and the
that all beams should be as small as possible but this is total loss (ignoring the vertical wires) is multiplied by
not the intention since gain is only one aspect of six compared with the “‘top-hatted”’ loop, (d). Method
performance. If it is not necessary to make a beam as (m), known as linear loading, is basically the same as (f)
86 HF ANTENNAS FOR ALL LOCATIONS

but gets round the problem of how to dispose of the equal currents 0-707] flow in each of the two elements.
stub and can likewise be used to effect a modest size With /8 spacing and 135° phase difference (ie
reduction, particularly if end-loading is used in addi- @/po = 1) the field strength ‘“‘in the beam”’ is identical
tion. with that from a single element and is only a fraction
In general it is advisable to aim for a more or less R/(R;, + R) of that which would be produced by an
square overall shape since otherwise the diameter of efficient dipole, although the radiation pattern is the
turning circle for a given value of R is increased. Thus same as that of an efficient beam having the same
in going from 10ft (3m) square to a 14ft (4-3m) by 7ft spacing and phasing. However, 3dB improvement in
(2-1m) rectangle at 14MHz, R is unchanged but the the radiated signal may be obtained by driving both
diameter of turning circle is increased from 14ft (4-3m) elements in phase, which destroys the front-to-back
to 15-8ft (4-8m), and if 14ft is the element length it ratio. There is thus a conflict between transmitting and
becomes more difficult to accommodate the necessary receiving requirements.
loading. Going in the other direction, 14ft spacing is Fig 5.23 compares the variation of gain with phase
getting too close to the A/4 limit and gain is decreased angle for low-loss elements, the extreme lossy situation
partly from this cause but also because radiation from discussed above and two intermediate cases, the low-
the folded-over ends can no longer be ignored. Addi- loss curve being an extended version of that shown in ©
tional considerations applicable to multiband operation Fig 5.4. Front-to-back ratio is also plotted over the
are discussed in Chapter 7. range of main interest for the case of lossy elements. It
will be noticed that losses cause the maximum of the
gain curve to shift to the right, the loss of gain due to
High-loss beams “incorrect” phasing being more than offset by the
An interesting problem arises [3] if, for any reason, it is increase of efficiency due to higher radiation resistance
necessary to use inefficient elements, a situation likely as one moves further away from the antiphase condi-
to arise, for example, when there are restrictions on the tion. For the same reason it would be advantageous, in
erection of visible antennas. A thin wire trailing over a the intermediate cases only, to use wider spacings. It is
wet tiled roof (or even a thicker wire just inside the roof notable that with wide spacings, compared with a single
space) clearly breaks the rules of good antenna design, lossy element, 3dB gain always exists in some direc-
and resistance losses may be greatly augmented by tions, and these can be varied at will, depending on the
losses in brickwork or other poor-quality insulating phase shift.
materials. But the situation is by no means hopeless, Reference has been made already to the extreme
remembering that even a few watts of effective radia- unsuitability of the W8JK beam for use in lossy situa-
ted power is capable of providing regular dx contacts. tions, eg as an indoor antenna. For proof of this
For reception, owing to the high external (galactic) consider first, by way of example, a two-element driven
noise level a large number of decibels can be sacrificed array tuned for maximum nominal front/back ratio
without adverse effect on signal-to-noise ratios and, (~ = Gp) but having a loss of 3dB due to its surround-
however great the losses, unidirectional patterns iden- ings, leaving a net gain of 1-2dB over a dipole at the
tical with those of typical high-quality beams may be same height but in the open. From Fig 5.4 the radiation
achieved by the use of two or more suitably-phased resistance of each element is 28Q and by implication
elements, although a completely different set of design this has been assumed equal to the loss resistance. In
requirements arises. the W8JK configuration, however, the radiation resist-
These may be derived from calculations similar to ance drops to 8-5Q and the efficiency to 8-5/36-5 or 23
those on earlier pages by adding a relatively large loss per cent. This is a total loss of 6-4dB, being 3-4dB
resistance R; to the radiation resistance R, the phase worse than for the phased array and 2-2dB worse than a
shift due to a series reactance X being given approx- dipole. Transferred to the lossy location the dipole is
imately by X/R, which bears little relation to the degraded in the ratio 101/73 or 1-4dB only, so that in
previous formula X/R. Because of the assumed high going from an indoor dipole to the W8JK there is a loss
value of R, very little current flows in a parasitic of 0-8dB whereas the phased array provides 1-2dB
element so both elements must be driven. Much higher gain.
values of reactance X are required to produce a given For a large initial loss the extra loss for the W8JK
radiation pattern, but alternatively phasing lines may reaches a limiting value which is equal to the ratio of
be used without any of the complications described the two radiation resistances, ie 28/8-5 or 5-2dB. Since
earlier since the mutual impediances which give rise to the W8JK has no front/back ratio, less gain than a
them are negligible in comparison with the loss resist- dipole, and needs only trivial modifications (p25) to
ance. turn it into the phased array which was the starting
The mathematics are further simplified and it is easy point for this example, it is difficult to account for its
to divide the power equally between two elements, so continuing popularity as an indoor antenna, except on
that if a current J would flow in one element alone, the basis that little is expected from indoor antennas,
 CLOSE-SPACED BEAMS 87

R,_ = Total loss resistance of one element

Rp = Radiation resistance of one element

Fig 5.23. Gain and front/back

psd ies
ratio for a pair of lossy elements
spaced )/8. The elements are
assumed to be driven with equal
=
currents. Gain or loss relative to
a loss-free dipole is obtained by
subtracting the loss figure 10 log
(1+ R,/Rp) from the gains indi-
cated

(dB)
losses
similar
with
relative
Gain
element
single
to In phase

135 180
Phase-shift in degrees relative to the antiphase (8JK) condition

accurate measurements are difficult, it needs no more is accurate or not it is well known that vertical polariza-
than a watt or so of effective radiation to produce some tion is relatively better at the lower frequencies. It is
dx contacts, and there is usually no accurate way of nevertheless the author’s view that vertical beams have
measuring losses. been greatly underrated as dx antennas for the higher
If the loss resistance exceeds 40Q or so it would be frequencies. It is true that for equal performance in the
better in any event to use a dipole for transmission plus two cases a vertical array needs to be relatively elabo-
a separate short antenna for phasing out interfering rate, but this is less of a problem in “low height”
signals as discussed on p131. situations which allow the mechanical problems to be
resolved in various simple ways as described in Chapter
13.
It may be that some of the prejudice against vertical
Vertical arrays antennas has been engendered by that commonest of
It is necessary at this point to make another brief all excuses for a poor signal, “I am only using a ground
descent from free space. The majority of amateur plane’. The ground-plane antenna appears prone to
beams for the higher-frequency bands use horizontal serious snortcomings (p43) and the “Zepp” feed or
elements, probably due in large part to the mechanical J-match in its usual form, though it may seem the
problems of mounting vertical arrays. In the latter case, obvious method for feeding a vertical dipole, is open to
for a given mean height the total height is greater, the even stronger objections (p46). In practice at low height
lower ends of the elements get entangled with guy a vertical dipole can usually be centre-fed with a
wires, and there is coupling between the elements and balanced line, the balance (though inherently imper-
supporting structures which may result in losses, distor- fect) being usually good enough.
tion of the radiation pattern and tvi. The feeding of The use of coaxial cable plus a balun is also
vertical antennas can also cause problems. acceptable; alternatively a “‘balanced”’ Zepp feed (p46)
The relative merits of horizontal and vertical polar- or a tuning unit working against a small counterpoise
ization are discussed in Chapter 6 (p103) and Chapter 10 may be employed (p44). A further alternative is to use
(p133), from which it seems clear that the main virtue monopole elements in conjunction with short counter-
of vertical antennas lies in the possibility of obtaining poise earths as described on p190; these can be regarded
fairly good performance even when height is severely as a kind of dipole, being subject to exactly the same
restricted. A typical ‘“‘breakeven”’ height for equivalent rules, and they can be expected to provide almost the
dx performance is around 25-30ft (8-9m) at 14MHz same performance for the same height reckoned to the
and probably proportional to wavelength. Whether this “centre of gravity” of the current distribution.
88 HF ANTENNAS FOR ALL LOCATIONS

One particularly interesting aspect of vertical beams

is the ease with which they can be rotated electrically to
provide all-round coverage. Whereas horizontal beams
are wide angle in the vertical and narrow angle in the a
horizontal planes, the reverse holds for vertical beams
rs
and reasonably good azimuth coverage in the case of
two or three elements can be obtained with a minimum
of four switched directions. A driven element may be
surrounded by a ring of, say, four parasitic elements
and the beam rotated by using relays to make small
changes in the length of the parasitic elements (p189).
Alternatively a ring of four close-spaced elements may
be connected in various ways, and the author favours
an arrangement of four “‘half-loops” or the vertical
equivalent of inverted-V elements. These can be sup-
ported by a single pole with the top half of the elements (dB)
element
tosingle
respect
with
Gain

forming the top part of the guy-wire system as de-

scribed on p190. Optimum methods of connection and
phasing for three or four close-spaced vertical elements
have been the subject of a computer study [15] which Number of elements

has been used as a basis for recommendations in Arrays of A/2 dipoles. Upper curve Is supergain,
lower curve is additive gain
Chapter 13.
eee = Arrays of isotropic sources
—x——x—-x— Yagi arrays (from reference [19]
Practical beams and the Uzkov limit 0) Practical four-element supergain antenna (Bloch)
Reference has already been made (p28) to the Uzkov [*] Equal- amplitude elements with optimum spacing [18]
limit [16], according to which the maximum gain
a Typical three-element beam, “/8 spacing
theoretically obtainable from n elements is equal to n?.
This is a mathematical result based on isotropic sources Fig 5.24. Maximum end-fire gain of array of equispaced elements.
This figure is taken from the paper by Bloch et a/, with the two
and the gain figure is slightly reduced when these, lower curves and points added (Proc IEE)
together with the standard of reference, are replaced by
real radiators such as A/2 dipoles for which corrections Particular interest attaches to the circled point which
have been calculated by Bloch, Medhurst and Pool [17] was the end-product of an attempt to reach the Uzkov
as reproduced in Fig 5.24. performance limit in the case of four elements as
For an equally-spaced arrangement of dipoles or other described in the above reference. Mention has already
“point sources’ the gains indicated by these curves been made of the very rapid escalation of losses, and
cannot be exceeded under any circumstances, and from practical difficulties in general, as spacing is reduced.
investigations reported in the reference unequal spac- Bloch and his co-workers appear to have used the
ings appear to make matters worse rather than better. widest spacing consistent with their purpose, ie 0-2h,
Fig 5.24 also shows for comparison the gain limit (n) and to demonstrate operation of the supergain princi-
for additive systems. It should be noted that the n? ple with more than two elements the vector diagrams
curve assumes infinitesimal spacings, though it remains shown in Fig 5.25 have been constructed from data
reasonably accurate for spacings up to at least 0-2), given in their paper. The apex angles are accounted for
whereas the n curve requires spacings of not less than by the spacing plus or minus an 8° phase shift. Dia-
0-5) and therefore relates to a much larger type of grams on the right have been drawn for a typical
array. The n? curve is in fact correct within 0-15dB for two-element array (maximum gain) and demonstrate
four elements at spacings up to 0-2) but there is a drop the underlying similarity. Other data for the four-
of about 1-2dB in gain for spacings of 0-25. The element array are summarized in Table 5.1 and Fig
crosses represent the closest practical approaches to the 5.26.
theoretical gain for which there is reliable evidence. The measured gain (75MHz) was 8-7dB so that,
The figure for n = 2 is taken from Fig 5.4, those for adding 0-8dB for the known losses, a further 0-6dB
n = 3,n = 6, and upwards of n = 6, being taken from a remained to be accounted for by losses in the coils and
comprehensive experimental study of long Yagi arrays capacitors of matching networks. The measured band-
[19]; this contains one mistake insofar as a gain of only width was about 1-7MHz which is sufficient to suggest
2-6dB is recorded for a driven element plus reflector, the possibility of a workable design covering most of
but apart from this the figures are consistent with data any one of the hf bands, though the need for improve-
from other sources. ments in the feeder system is indicated in the reference.
 CLOSE-SPACED BEAMS 89

4 Elements 2 Elements ‘Orthodox' end-fire array having elements 4-Gibinent

in same position
as the 'super gain' ar j .
Pp Rat ga aay. super gain' array

=i
For ward
direction

With A
CAT
U7

Vi
Back

LAK
direction

PEO
'é

TAL
SSSoy LLL
Vl SP
60 50440) 30,20 0 2030 40 50 60
Fig 5.25. Phase and amplitude diagrams for optimized four- Fig 5.26. Radiation pattern of the experimental four-element
element supergain antenna. Based on details given in the paper supergain array compared SLi Sieh eee end-fire array (Proc
by Bloch et al, these diagrams demonstrate the mode of opera-
tion. Front/back ratio = AB/A’B’= 10-5dB. The marked angle is
““equal to the spacing” plus or minus a small phase shift (8°), this ensuring the correct current ratios follows from the fact
being mainly responsible for the difference in length of AB and that a maximum possible gain of 6-5dB only has been
A’B’. It will be seen that a very small phase error could make a
considerable difference to the final answer, thus demonstrating a calculated for a set of four elements carrying equal
characteristic feature of supergain arrays. Corresponding dia- currents with optimum phase differences [18].
grams for a two-element beam (shown on right) exhibit the same The geometrical principles of Chapter 3 (p22) apply
characteristic ‘‘pincer effect”
irrespective of the number of elements so that if, for
example, the spacings are divided by four the angles
An interesting problem arises from the negative between the vectors in Fig 5.25 must also be divided by
resistances listed in Table 5.1; these indicate that four. This means that the currents are multiplied by
elements 3 and 4 are returning power to the transmitter four and the radiation resistances divided by 16, be-
for re-issue to elements 1 and 2, and it is clear that coming 0-32, 0-425, —0-1 and —1-25Q@ respectively.
relative phases and amplitudes have to be carefully The reader has already been introduced to the difficul-
controlled, a matter likely to present some difficulty. It ties which arise when radiation resistances fall below
appears to the author that this might be achievable by 5Q or so, and these become further emphasized when
separate feeding of elements 2 and 3 with a phasing unit bandwidth is discussed in Chapter 8. In view of the
of the type shown on p77, elements 1 and 4 being inherent difficulty of reaching the gain limit for more
excited parasitically from elements 2 and 3 with careful than two elements it is evident that any such attempt
control of coupling, eg by bending the elements in- must be based on using the widest allowable spacings,
wards or outwards as required. The importance of eg 0-2) as in the design based on Table 5.1 and Fig 5.25.
This assessment is difficult to reconcile with gain figures
Table 5.1. Theoretical design data for four-element which have been quoted for log-Yagi arrays having
supergain array element spacings of only 0-05A, though these can
Element number provide impressive performance and designs such as
1 2 3 4 the one featured on p170 deserve consideration by
anyone looking for a ‘“‘best possible” monoband beam.
Relative magnitude of currents 8-155 18-386 18-386 8-155
Driving point resistances (() +5-1 468 +16 —20-1 The experimental points in Fig 5.24 need some
Input powers (relative) +340 +2,288 -539 —1,336 discussion since the measured gain in the “‘supergain”’
case is only slightly above the curve for Yagi arrays of
Power gain relative to \/2 dipole = 10-1dB
Current in i/2 dipole to give same field as array, in same units as more or less conventional design, and it might there-
the above currents = 10-3 fore appear as a negligible return for a lot of hard work.
Total radiated power (relative) = 753 It is however instructive to consider Fig 5.24 in con-
_ Ditto for reference dipole (for same field) = (10-3)? x 73 = 7,745
Hence power gain relative to A/2 dipole = 7,745/753 = 10-1dB junction with Fig 5.27 which shows the gain of Yagi
Theoretical maximum power gain for very close spacing (from Fig arrays as a function of length on the basis of two
5.26) = 10-3dB separate studies using different assumptions and
Half-power beamwidth = 52°
Front-to-back ratio = 10-9dB methods, the spacing between elements being 0-2A for
Calculated losses in antenna + feeder = 0-8dB the crosses and 0-26—0-3A for the circles. In the design
 90 HF ANTENNAS FOR ALL LOCATIONS

tae Experimental supergain point (Bloch) from Fig 5.26 eee

(dB)
dipole
short
to
relative
Gain

Boom length (A)

Fig 5.27. Variation of gain with boom length for Yagi arrays indicating comparison with supergain arrays. The dotted curve is an
attempted “best fit’’ based on the assumption of gain proportional to length; this has no practical significance for small antennas but the
relatively steep slope for large boom lengths suggests that parasitic excitation may not make best use of the available length

of Yagi arrays with large numbers of elements there are In general it might be concluded that spacing can be
so many variables that it is not possible to specify a increased with advantage as the number of elements is
“correct” method of design, but the points in Fig 5.24 increased. We return to this topic in Chapter 6.
are in fairly close agreement with each other and with It may be noted that sources of loss in the case of the
other comparable studies. General conclusions. to be supergain antenna described above are known and an
drawn include the following: improvement of 1dB looks possible. This would leave
(a) There is no difficulty in reaching the limit for two the shortfall several times greater in the case of the
elements but with increasing numbers the difficul- Yagi but the amount of this (2dB) is hardly enough to
ties escalate rapidly. justify the added complexity of an optimized design.
(b) With three or four elements it is still possible to get In choosing an antenna system the gain for a given
fairly close to the limit but only if the spacing is number of elements is only one of a number of criteria,
increased to the maximum allowable. The ‘‘typic- and reduction of boom length may be more important
al” figure for n = 3 is based on amateur beams than cutting down the number of elements.
with spacings of 4/8, whereas the Yagi curve Were it possible to overlook a few mundane practical
assumes spacings of V5 and demonstrates the likely details, the prospect of supergain antennas with many
improvement from the wider spacing at 28MHz elements operating at the Uzkov limit would be quite
when close-spaced “‘14MHz” elements are used on fascinating, and it is indeed not impossible that one day
both bands. This counters the widely held belief someone will build a 1,000-element beam the size of a
(justified, alas, for the two-element case) that if matchbox and having a gain of a million! By invoking
14MHz beam elements are used at 283MHz the the principle of superconductivity, whereby the resist-
Spacing is necessarily too wide for optimum per- ance of certain conductors falls to zero at temperatures
formance. slightly above absolute zero, the problem of efficiency
might be overcome but one would then alas be left with
At large n the difficulties escalate to the point where a bandwidth so narrow that it could take years to signal
the additive gain curve applies; thus for n =10 there is a the equivalent of a single morse dot!
shortfall of 6dB relative to the supergain curve, and for The nature of the difficulties can be anticipated from
a given number of elements only 2dB improvement a simple numerical example based on the square-law
compared with additive gain, though the latter would relationship (p70) between radiation resistance and the
demand an increase of array length from 2-2) to 4-5i. dimensions of an array; thus at 14¢MHz a W8JK beam
 CLOSE-SPACED BEAMS 91

with dimensions of 2in X 14in would have a radiation 5.16 can be recommended as a means of achieving a
resistance per element of considerable reduction in average interference levels.
Higher gains and longer boom lengths tend to result in
x polane a
8-5 x (
sex2) X (sxe 73) = 05 * 10-72 worse front/back ratios although this can be largely
offset by increasing the number of elements [21, 23].
It is not clear how such elements would be constructed The available data is not however directly applicable to
and resonated but even a normal full-sized element small beams with bent elements, and further design
develops a reactance of this order if the frequency is studies are needed for these cases.
shifted away from resonance by about 5 parts in 10°. On The maximum gain that can be expected in practice
this basis bandwidth (calculated as explained in Chap- from small beams varies only from 5dB relative to a
ter 8) comes to 6 X 10~-4Hz and the morse dot can now single dipole in the case of two elements spaced by 0-1
be signalled in about 1,000 seconds or less than half an to 0-184 up to 6 or 7dB for three elements with
hour. Unfortunately after all this (including provision boom lengths of 0-25—0-5A [19, 20, 21, 23]. Similar
of a supply of liquid helium) the gain is still a mere 4dB! figures apply to vertical arrays, including monopole
The reader who has studied the earlier part of this elements working against short loaded counterpoises at
chapter may hasten to point out that the W8JK is the heights upwards of 2 or 3ft (0-6—0-9m) from the ground,
worst possible antenna for such an application and the but the pros and cons of vertical versus horizontal
above time could be divided by four, but the prospects polarization can be quite complicated as Aiscussed in
are still not encouraging. Chapter 10 (p133).
Even if, for example, due to environmental con-
straints elements have high losses, there is usually some
Summary, conclusions and advantage in using more than one but it is essential, and
recommendations in this case possible, for all to be driven. Apart from this
Two-element amateur beams should consist of closely- or as a means of correcting small errors of design or
spaced sources of radiation, equal in amplitude and adjustment after erection, driven operation is not
correctly phased. The majority of designs fail to meet recommended.
these conditions, and suffer from more-or-less serious Two-element beams may be reduced in size to only
defects. This includes loop arrays such as the quad 10ft (3m) square at 14MHz, and pro rata, without losing
which, though satisfactory in terms of performance, more than 1dB of gain but for this purpose it is essential
should not be used for rotary beams since they are to use capacitive end-loading. This results in excess
heavy, unsightly, expensive, and due to high wind coupling between elements, an effect which can be
loading frequently blow down! In addition to providing counteracted by means of neutralization. Two feeders
optimum performance the recommended designs result should be used to effect beam reversal and some degree
in a minimum of weight, size and cost, besides being of remote tuning.
relatively easy to tune. The specified conditions are X-beams and a number of other element shapes are
easily realized in practice since they require only the use not recommended.
of reflector elements having their ends correctly spaced
from those of the driven element. References
Unfortunately, up to the time of writing, attempts to
~ equalize currents have usually relied on phasing lines [1] ““Supergain aerials’, L. Moxon, Radio Com-
which work correctly only if mutual coupling can be munication September 1972, p586.
neglected. As, however, close-spaced beams work sole- [2] ‘“Two-element driven arrays’, L. A.Moxon, OST
ly and entirely by virtue of such coupling the decline in July 1952, p28.
popularity of phasing-line systems is hardly surprising. [3] ‘‘Gains and losses in hf aerials”, L. A. Moxon,
However, although good performance is compatible Radio Communication December 1973 and Janu-
with only a narrow range of options (the same for all ary 1974.
equal-current two-element beams) and requires adher- [4] ““The Moxon beam’’, L. A. Moxon, RSGB Bulle-
ence to strict rules, it is quite difficult to achieve an tin, August 1953.
antenna that does not work at least after some fashion. [5] The ARRL Antenna Book, 12th edn, ARRL,
With a little assistance from the long arm of coinci- 1970. (The matter is not referred to in the 13th
dence, even driven arrays based on phasing lines have edition).
been known to produce topline performance; this road, [6] “Directional antennas”, G. H. Brown, Proceed-
however, is more likely to lead to disappointment. ings of the IRE, Vol 25, January 1937, p78.
Three-element beams are less amenable to calcula- [7] “The Swiss quad beam aerial’, R. A. Baumgart-
tion owing to the large number of variables and there is ner, HB9CV, RSGB Bulletin June 1964.
no clearly-recognizable ‘‘optimum”’ set of conditions, [8] The ARRL Antenna Book, 13th edn, ARRL,
but designs based on the right-hand lower corner of Fig 1974, p200.
92 HF ANTENNAS FOR ALL LOCATIONS

[9] ‘Quad with switchable polar diagram’’, L. Vsevol- aerial arrays’, Bloch, Medhurst and Pool, Pro-—
zhskii, UA3IAR, Radio (USSR) No 6, 1978, ceedings of the IEE, Part III, September 1953.
pp18-19. See also ‘“‘Technical Topics’, Radio [18] ‘“The gain of an idealized Yagi array”, D. G.
Communication October 1978 et seq. Reid, Journal of the IEE Vol 93, Part IITA, 1946,
[10] ‘‘The compact unidirectional array”, W. Van B. p546.
Roberts, W3CHO, Radio (USA) January 1938. [19] “Yagi antenna design’, P. P. Viezbicke, NBS
[11] F. Caton, VK2ABQ/G3ONC, Electronics Austra- Technical Note 688, US Dept of Commerce, De-
lia October 1973. cember 1976.
[12] ‘“Technical Topics”, Radio Communication March [20] ‘‘A design of Yagi-Uda antennas by non-linear
1980. See also numerous earlier references in optimization techniques’, M. Kominani and K.
‘Technical Topics’’ from January 1974 onwards. Rokushima, English text in Electronics and Com-
[13] ‘“‘Technical Topics”, Radio Communication munications in Japan, Vol 61-B, No 1, 1978,
February 1977 and June 1978. pp47-54. (See also ‘Technical Topics”, Radio
[14] ‘High-performance small beams”, L. A. Moxon, Communication March 1980).
Ham Radio March 1979. [21] Antenna Theory and Design, H. P. Williams, 2nd
[15] The ARRL Antenna Anthology, 1978, p114. edn, Pitman, p145. }
(Material originally presented in QST by J. L. [22] ‘“‘Theoretical treatment of short Yagi aerials’, W.
Lawson, W2PV). Walkinshaw, Journal of the IEE, Vol 93, Part
[16] ‘“An approach to the problem of optimum direc- III-A, 1946, p564.
tive antennae design’, A. I. Uzkov, Comptes [23] ‘“‘Evaluating aerial performance’’, L. A. Moxon,
Rendus de Il’Academie des Sciences de |’URSS, Vol Wireless World February and March 1959.
532 p35: [24] ‘“‘“Yagi antenna design’, J. L. Lawson, W2PV,
[17] ‘‘A new approach to the design of superdirective Ham Radio May 1980.
 CHAPTER 6

Arrays, long wires, and ground

reflections

Thus far, attention has been focussed mainly on anten-

nas which can be regarded as “point sources” of F, & sin (22sn8) (2)
radiation, or closely-spaced groups of sources which
If F, is the pattern of the individual units (eg cos @ in
when properly coordinated can be regarded perhaps as
the H-plane for a horizontal dipole) the overall pattern
“super point sources” since the overall dimensions
is given by F, X F,, ie for each direction the patterns Nice
remain small compared with a wavelength. To com-
and F, are obtained separately and then multiplied
plete the picture it is necessary now to consider arrays
together. Thus for a pair of in-phase dipoles spaced 4/2
of point sources such as:
the values of F, given by (1) above are multiplied by cos
(a) Long wires which, in effect, constitute ‘‘strings’’ of
8 to obtain the lower curve in Fig 3.9. A potentially
point sources.
difficult problem, or rather group of problems, can
(b) “‘Additive-gain’”’ arrays consisting of a number of
therefore be resolved by breaking them down into two
separate sources, adequately spaced, and co-
very simple problems though there is one important
ordinated by the use of such feeders and phasing
limitation to this approach. If the units are in any way
lines as may be necessary to cause the radiation to
dissimilar the array factor concept is not readily applic-
add up in-phase in a desired direction.
able and, despite some extra labour, it is easiest to
(c) The two-element array formed by an antenna of obtain each point on the overall pattern by the method
whatever size and its image in the ground. This is
the point of re-entry after the long sojourn in free of Fig 3.3, making due allowance for the relevant
space, and it is also the ‘moment of truth” since it amplitude and phase differences in each case.
is the.effect of the ground in conjunction with Fortunately the need to obtain complete directional
height restrictions which determines whether dx patterns for arrays rarely arises in amateur work,
performance will be good, bad or indifferent. In though an amateur with two beams (which may be
this wider context the term “‘ground”’ is to be different) may decide to operate them as a two-unit
understood as including structures erected on the array in which case he might like to know, for example,
ground such as buildings and power lines, as a how much the direction can be allowed to vary without
result of which it may be necessary to depart from having to re-adjust the phasing. Another problem of
the simple ‘‘antenna plus image”’ concept. this type arises as will be seen later in the case of a
vertical antenna and its image in the ground.
A rather more complicated problem arises if spacing
Array factor between units is insufficient so that the presence of one
Most of the above cases can be handled in the same
way, using an array factor which is merely the pattern
produced by isotropic sources similarly spaced and
phased as in Fig 6.1. This can be obtained very easily by
graphical addition as in Fig 3.3 or, for two like-phased
sources, more quickly from the expression

F, « cos (74 sin

a °) (1)

where F, is the array factor and the quantity in brackets Fields due D:
is half the phase difference, at the receiver, expressed in to (1) &(2) 2Fcos(5)
radians. A plot of this for \/2 spacing has been included
Fig 6.1. Derivation of array factor for point sources in-phase and
in Fig 3.9. Similarly for antiphased sources we have out-of-phase

93
94 HF ANTENNAS FOR ALL LOCATIONS

position is changed radically and there are two main

~---=
-- ies
oe

cases to be considered: Fig 6.2. In case (a) the wire is

centre fed and there are two in-phase concentrations of
current separated by /2. This is sufficient to provide
(a) some additive gain and, although radiation is still
concentrated at right-angles to the wire, the pattern
width is reduced from 90° to 50° for 3dB drop in signal.
As explained on p98 this provides a gain of just under
2dB, and the length can be further increased with
advantage up to about 5A/4. At this spacing, ie 3A/4
between the centres of the current loops, the mutual
resistance is zero and the power gain is therefore 3dB,
being exactly equal to the number of elements as
explained on p24.
(b) Strictly speaking, this calculation relates only to a
spaced pair of separate elements, but the centre-fed
54 wire (which is usually known as an “extended
Fig 6.2. Full-wave (1)) dipoles corresponding to Fig 6.1. (a) shows
double-Zepp’’) is more convenient, the current in the
two half-waves in phase, with the ends close together (collinear short 2 X 1/8 central portion being too small to affect
pair) and spaced apart (“extended double Zepp”). (b) shows a 1) the directional pattern appreciably.*
end-fed wire with out-of-phase current loops In contrast to this example Fig 6.2(b) shows an
end-fed 14 antenna which can be considered as a
unit modifies the currents and therefore the radiation transmission line using the ground as the return path.
pattern of another unit. In this case it is the modified Since there is now no discontinuity in the centre, the
unit pattern F', which must be multiplied by F,,; though current and voltage can only vary smoothly along the
this may be difficult in principle such effects can be wire according to ordinary transmission-line principles
corrected in the case of a vertical antenna and its image as explained in Chapter 4. Hence the two half-waves
in the ground. Fortunately, for the most part, amateur are of opposite phase, and radiation must therefore be
hf antenna systems can be divided up into single units, zero at right-angles to the wire, as well as in the
widely-spaced groupings of units, and a number of endwise direction. The radiation pattern for Fig 6.2(a)
reasonably simple cases in which coupling between and (b) may be obtained from the above array factors,
units, if not small enough to be ignored, can be allowed multiplying in each case by cos 0. :
for on the basis of mutual resistance data as explained The resulting patterns are shown in Fig 6.3 from
below. which maximum radiation in the 1) case (b) will be seen
Exceptions to this include long Yagi and log-periodic to occur at an angle of 50° to the wire. The radiation
arrays which are difficult to evaluate without a compu- resistance relative to a 4/2 dipole can be deduced with
ter or model range. However, the results of a number of the help of mutual impedance data as explained below
studies have been published and provide adequate and used to derive the gain which comes to 0-S5dB. Fig
guidelines, including gain figures for long Yagi arrays 6.3 applies to one quadrant only, but making use of
as already summarized in Fig 5.27 which is based on symmetry the solid line curve can be used to construct
several sources. An example of a log-periodic array and the well-known clover-leaf pattern of Fig 6.4(b).
a general explanation of its mode of operation will be Though Fig 6.4(b) is perhaps more pictorial, Fig 6.3
found in Chapter 14. tends to be more informative and this form of presenta-
tion has already been used in Chapters 3 and 5. —
Radiation patterns of full-wave antennas
So far the discussion of radiation pattern has been End-fed long wires
mainly confined to general principles and short conduc- As we have seen in Chapter 4, end-fed wires can be any
tors. It has been shown that going downwards from, length, and it will be noticed from Fig 6.5 that as length
say, A/2 there is little change in pattern, and such increases the four main lobes decrease in width and
antennas can therefore be considered as “point move closer to the endwise direction. With longer wires
sources”. In other words, as far as the pattern (and smaller lobes known as sidelobes appear in the radia-
gain) is concerned it is as if the radiation was coming tion pattern but, although these are often important in
from a source of infinitely small dimensions, but with
due regard of course for angles of inclination and * This current does subtract about 10 per cent from the total wanted
field for a given loop current but, since the directivity and therefore the
polarization. : gain is virtually unchanged, the only practical consequence is a 20 per
If the length of wire is increased, say, to 1A the cent decrease in R.
 ARRAYS, LONG WIRES, AND GROUND REFLECTIONS 95

BS Ae Nea
100 f(a) gamer

ENA A
input 7 \ fon [\ [/\ Ds eX {>> {\ !
i i |

A? eRe
80 (a) Terminating Bee o
if used

EPA ONTie.
eA Noite |AE
60

ce NS Wa
“jf anes Nb ee a
40

Bie SN Be hw
field
maximum
of
Percent

SESS
20
‘Extended =——"*
double Zepp'

slEARSEN
90 70 50 30 10 O
Angle relative to the wire (degrees)

Fig 6.3. Directional patterns corresponding to Fig 6.2 shown in

Cartesian form, one quadrant only being shown since the overall Fig 6.5. Current distribution in a long-wire antenna is shown at
(a). Standing wave pattern disappears if antenna is terminated in
pattern is symmetrical
a suitable load and average current is then the same at all points,
barring losses. At (b) the long wire of (a) is represented by point
commercial applications, it will be sufficient here to sources of alternate sign. Radiation in direction AC is zero when
AB — AC = 2/2 which occurs when the angle y = V(2//) where /is
note that they increase in number and decrease in width in wavelengths and y in radians. (c) is a typical radiation pattern
with increasing length of antenna, being mostly be- for an unterminated long-wire antenna, the null directions indi-
tween 5 and 10dB down on the main lobes. Further cated by arrows being those given by the above formula (Wire-
less World)
data for long single wires are given in Table 14.1.
Consider next a much longer wire; assuming this to shown in Fig 6.5(a) and the wire can be regarded as a
be unterminated a standing-wave pattern exists as string of dipoles of alternate sign as shown at (b). For
most directions each dipole can be paired off with one
90° 54°
of opposite sign, allowing where necessary for phase
50° differences resulting from different distances of travel.
- Ye31° However, nearly in line with the wire alternate dipoles
are separated by nearly \/2 and thus produce fields
which add in phase. The large number of dipoles makes
ho|> up for the relatively small field which each produces at
this angle, even to the extent of providing considerable
gain if the wire is long enough as shown by Table 14.1,
7 p198.
When the wire is terminated in its characteristic
impedance there is no standing wave pattern but taking
a ‘‘snapshot” at any given instant of time the appear-
ance would be very similar to Fig 6.5(a) and to a large
extent the same reasoning applies. This, however, is
now a travelling wave moving towards the termination,
and for the backward direction radiation from any
point on the antenna is cancelled by radiation of
opposite sign from a point )/2 further on. This has /2
less distance to travel and left the transmitter half a
cycle earlier, so there are three phase reversals in all,
Fig 6.4. Polar diagrams for end-fed wires up to 2). The angles of
main lobes and crevasses are shown, also the angles relative to resulting in cancellation. Hence the radiation pattern
the wire direction (ie 90°—6) at which the loss is 3dB in the main becomes unidirectional.
lobe. The lobes should be visualized as cones surrounding the Long wires such as these are usually grouped to form
wire. As length increases the lobes move closer to the wire
direction, so that in the 2) case ther2 is some radiation at useful V- or rhombic arrays, Figs 6.6 and 6.7, which can
wave angles in the endwise direction as shown by the dotted line provide high gain and are much used for commercial
96 HF ANTENNAS FOR ALL LOCATIONS

8
REINFORCED
RADIATION

(a)

™.
TO FURTHER POLE 'B'
7
7
7

des
MAXIMUM
Ra ety
(c) | (d)
Fig 6.8. Attempts to form arrays by bending wires into loops. The
k—4—>
4 B
quad loop at (a) has superficial resemblance to two separate
point sources but stacking distance ()/4) is insufficient to produce
significant difference in pattern compared with single source. The
same applies to the delta loop at (c) whereas the folded dipole
TERMINATING included at (d) for comparison has no pretensions to be other
RESISTOR
than the equivalent of a dipole. In contrast the bi-square loop at
(b) is equivalent to four elements; each current loop can be
represented by a pair of “equivalent dipoles” at right-angles
(Chapter 2, p10) but the vertical components cancel each other,
leaving the equivalent of four horizontal (as drawn) dipoles
(b)
point-to-point services. Such antennas may be termin-
ated or unterminated, and are wideband to the extent
Fig 6.6. V-beam derived from two long wires arranged at the that they can be designed to give reasonable perform-
appropriate angle for alignment of one pair of main lobes.
Addition of resistors as shown at (b) results in unidirectional ance over about two octaves. Optimum design requires
pattern that the lobes of the individual wires should be aligned
in the wanted direction and this dictates the shape of
the array. Although beamwidths are narrow and rota-
tion is not possible there are usually sufficient side-
MAXIMUM lobes, Fig 6.6(a), to provide some useful coverage in
RADIATION most directions. Very impressive performances have
been achieved by a few amateurs able to erect arrays of
this type, but in view of their limited interest for the
average reader further consideration, which is one of
the topics for Chapter 14, has been restricted to general
guidelines and some possibilities of obtaining increased
gain for a given area occupied.

MAXIMUM
RADIATION Loop antennas
Single wires are often bent into the form of a loop, the
most important examples being the quad, Fig 6.8(a),
and the bi-square, Fig 6.8(b). The quad can be dis-
torted into a variety of shapes such as triangular, being
then known as a delta loop, Fig 6.8(c), or even
squashed flat in which case it becomes a folded dipole,
Fig 6.8(d). The folded dipole is usually )/2 and has the ~
same radiation pattern as an ordinary dipole, although
Fig 6.7. Terminated and unterminated rhombic antennas if used with a resonant open-wire feedline it can give a
 ARRAYS, LONG WIRES, AND GROUND REFLECTIONS 97

- -_-7-=n
we poasecy (a) from the appropriate direction. The components which
Normal
dipole
full-wave
are aligned in any given direction can be directly added
if the dimensions are small, or added taking due
account of phase differences if the dimensions are
(b)
large. If the problem is three-dimensional each source
a meee ee
A Da) Po
St + eee
Ae
Inductive loading,
loops contracted has to be resolved into three components mutually at
right-angles, but fortunately many important practical
cases such as the “‘bi-square”’ antenna referred to above
-- emo. - ~ ——— Cc)
Tg
- = =~.

4s Capacitive loading,
involve only two dimensions.
Of PO pn fen to loops expanded

Fig 6.9. Centre-fed antenna (a) can be contracted by inductance as Centre-fed long wires
at (b) or stretched by capacitance as at (c)
One other case of practical interest is the centre-fed
long wire such as a 3-5MHz dipole which has been
good account of itself over a frequency range of at least pressed into service at 14, 21 or 283MHz. At 14MHz this
half an octave. becomes a pair of 1A radiators as in Fig 6.10, which
The quad loop is perhaps second in importance only compares the pattern with that of an end-fed wire of the
to the dipole. It usually consists of 1A of wire bent into a same length, illustrating the tendency for radiation to
square so that its largest dimension is )/4, and offers a be concentrated at a much larger angle to the wire
number of practical advantages other than the high gain when this is centre fed. These patterns were obtained
frequently claimed. This it does not provide, as ex- by taking the out-of-phase curve from Fig 6.3 and
plained on p100, though this is also self-evident from the multiplying it by the array factors appropriate to a
small size which rules out any appreciable amount of spacing of 1A.
additive gain, and links its radiation pattern closely
with that of an ordinary dipole. The bi-square, on the
other hand, has A/2 sides and thus achieves additive Gain, radiation resistance and mutual
gain, though usually at the expense of height. coupling
It will by now be evident to the reader that these three
Stretched wires properties of antennas are closely interrelated, and the
important role of both mutual resistance and mutual
It is possible to ‘“‘stretch”’ a A/2 current loop to fill the reactance in the case of close-spaced beams has been
whole of a wire of any length by the use of series demonstrated.
capacitors as in Fig 6.9(a). Uses of this principle range In contrast to this, additive-gain arrays have been
from the use of a single capacitor to enable, say, a
‘\ > 2 xA out of phase
14MHz dipole to be used at 21MHz, to the production (centre fed)
of a nearly uniform current distribution in a long wire 100
which then bears a close resemblance to the collinear
array of Fig 6.11. Inductance as in Fig 6.9(b) has the \U 2 xAin phase
opposite effect and is used for example as a loading (end fed)

device for mobile antennas.

Wires bent into odd shapes

Wires often have to be bent into “odd shapes’’ to fit
them into an available space, and in dealing with these
cases it is convenient to think of the wire as consisting
of a number of point sources, one to each straight
section or more than one if a section is much longer field
maximum
of
Percent

than 4/2. If the arrangement is two-dimensional (ie

capable of being drawn to scale on a flat piece of paper) eel
the radiation from a given source can then be resolved
into two components at right-angles, one in any desired oo
direction and the other at right-angles to it, as ex-
plained in Chapter 2 and illustrated in Fig 6.8(b). For
iseta neee
90 50 30 10 1e)
the moment it will be sufficient to note that the ‘“‘value”’ prae relative tothe wire (degrees)
to be assigned to each component is simply the average Fig 6.10. Comparison between centre-fed long wire (eg )/2 dipole
current times the apparent length of the wire as seen for 3-5MHz band used on 14MHz) with the same long wire end fed
98 HF ANTENNAS FOR ALL LOCATIONS

assumed to consist of independent elements which is a mechanical problems of actually doing it. From Fig 5.2
valid approximation for demonstrating the principles. (solid line) it is found that for dipoles in this configura-
This approximation is not strictly true in practice as tion R,, = 44@, and from the dotted curve R,3 = 182.
spacings are dictated by compromises, and mutual Knowing that gain does not depend on size of element,
resistances can be as high as R/10 out to spacings of it will be evident that the halving of R means that Rj,
several wavelengths. Although the amateur is usually and R,3 must also be halved.
not concerned with practical hf arrays involving this R,4 may present some difficulty unless it is realized
kind of spacing, it is instructive to consider the general that elements 1 and 4 are parallel full-length dipoles
case of n elements all coupled into each other by spaced A/2, for which the mutual resistance read directly
complex impedances. This might appear to pose a from Fig 5.2 is —10Q, half of which is attributable to the
rather formidable problem, but in practice most of the horizontal components. The driving point impedance
complication disappears if the currents are assumed to (assuming centre feed) for each dipole is therefore
be all in-phase or out-of-phase, this being the condition 36:5 +22+9—5 = 62:5Q* and for a given radiated
for full additive gain when elements are spaced by 4/2 power the current J relative to the current Jp in a X/2
which is typical of practical arrays. We then have at the dipole is given by
terminals of the first of n elements
Ip = V7BK4 X 62-5) = 0-56
E/l,=R+RmtRmtRmt..-+R —— mn
To obtain the field strength relative to that from a
where the subscripts denote the mutual resistance single dipole we have to multiply by the number of
couplings to each of the other elements and a negative dipoles but divide by V2 because of the 45° inclination
sign implies antiphase excitation. The reactance terms of the wire, so that the gain as a current ratio is
add in the same way but are automatically eliminated (4 x 0-56)
V2 = 1-53 or 3-7dB, in fair agreement with
by normal tuning processes. the 4dB figure usually quoted. The radiation resistance
This result applies equally to large commercial arrays referred to a current point such as the end of the i/4
and simple arrays of considerable importance to the stub in Fig 6.8(b) would be 250Q if Z) = Z,, but typical
amateur, such as the two half-waves in phase shown in values in practice are likely to be 600Q and 1,000Q
Fig 6.2(a). The gain of this has already been worked respectively. Transformer action as described on p35
out for the special case of R,,=0 which requires a reduces the impedance to 250 x (0-6)? = 90Q which
centre-to-centre spacing of 3/4. More commonly the provides a reasonable match to 75Q feeder.
ends are adjacent so that the spacing is only /2, and the
mutual resistance of 219 from Fig 5.2 is added to the
natural 73Q resistance of each current loop. Assuming The quad loop
some arrangement of feeders which applies the avail- One further example brings us back yet again to the
able power to the two resistances in series, the loop controversial topic of the gain of quads! It has already
currents are found to be less than those in a dipole in been pointed out that as a first approximation one can
the ratio 73/(2 x 94), ie 0-62, but as there are two loops regard the quad (like the dipole) as a point source of
the field produced exceeds that from a dipole by 1-24, radiation similar to all other point sources. If, however,
which is a gain of 1-98dB. one needs to be more precise, there is no objection to
regarding the quad loop, Fig 6.8(a), as a pair of stacked
The bi-square antenna M2 dipoles. It should however be noted that bending
The simplicity of this method of calculation can be over the ends to form a loop will reduce the field
appreciated by applying it to a much more complicated strength by a factor of 0-7 and therefore (for the same
example, the 2A loop commonly known as a bi-square gain) halve whatever the radiation resistance may
antenna and illustrated in Fig 6.8(b). Each side can be happen to be for the unbent dipoles.
resolved as just described into equal horizontal and From Fig 5.2 it is found that for the appropriate
vertical components, but it will be obvious that the spacing (\/4) the mutual resistance is 40Q so that the
vertical components cancel, leaving only the four hori- total radiation resistance will be 72 + 40 = 112Q. The
zontal point sources which are in phase for the direc- two dipoles are, in effect, connected in series so that
tions normal to the plane of the loop. Because the wires the current compared with that in a single dipole is
are inclined at 45° each produces for a given current given by 72/224 = 0-567. The total field from the two
only 70 per cent of the field from a A/2 dipole and this dipoles is therefore 2 x 0-567 = 1-134 times that from a
halves the radiation resistance. single 4/2 dipole, which is a gain of 1-09dB only, in
Mutual impedance data is not available for this agreement with data for stacked dipoles to be found in
arrangement as it stands, but the point sources can in
principle be replaced by A/2 dipoles since there is no * It may appear that coupling involving the vertical components has
been ignored, but since they do not contribute to the radiated field the
way this can have much effect on the radiation pattern, only role left to them is helping to bring about the stated values of
and for present purposes we are not faced with the mutual impedance.
 ARRAYS, LONG WIRES, AND GROUND REFLECTIONS 99

most standard textbooks. When the ends are bent over

to form a quad loop the gain is reduced due to radiation
from the bent-over ends in a high-angle endwise mode,
and the amount, though small, is certainly enough to
bring the gain down to under 1dB. Bending the ends
also results in loss of much of the 0-4dB by which the
gain of A/2 dipoles exceeds that of very short dipoles.
Figures in excess of this have been frequently quoted
COLLINEAR
in professional as well as amateur literature, but what-
ever their source can be disregarded, particularly as the
discrepancy has been allowed to pass without com-
ment. Moreover, in at least one case the excess gain is
stated to be irrespective of the number of elements. ee ‘ Bs
This assertion is incompatible with the basic principles
of stacking which dictate, as explained below, that in
multi-element arrays even the 1dB gain disappears. It is ww || ;
not intended by this necessarily to disparage the quad
since account must be taken of other advantages, such END-FIRE
as the ease with which optimum performance can be
obtained in practice, and the good back-to-front ratio Fig 6.11. Collinear, broadside (vertical) and end-fire arrays, illus-
resulting (subject to correct spacing) from near- trating the terms. The arrows indicate directions of maximum
radiation and the voltage standing waves on three of the anten-
equality of the currents in a driven element and in a nas show polarity and phase. End-fire arrays may be parasitic (ie
parasitic reflector as discussed on p74. Yagi type) as shown, or all elements can be driven with close or
wide spacing between elements. Broadside and collinear arrays
use the additive gain principle and must be widely spaced
Stacking
Gain may be obtained not only by stacking elements to azimuth. This is important if it is desired to cover a
form beam arrays but by stacking beams to form large geographical area such as North America or
“arrays of arrays’’, this being largely an extension of obtain all-round coverage by beam switching rather
the same principles, though close spacing is no longer than beam rotation.
applicable. Additive principles apply, and gain results So far as dx signal strength is concerned it makes no
only if the addition becomes significantly less effective difference whether gain is achieved by narrowing of the
within the beamwidth of the individual units. Otherwise horizontal or of the vertical beamwidth. It is often
there will be no narrowing of the directivity pattern and thought that because broadside (ie vertical) stacking of
therefore no gain. It is of interest to note that insuffi- horizontal antennas reduces high-angle radiation more
cient stacking distance must inevitably be reflected in low-angle radiation is produced, but the reverse is true
an increase of radiation resistance, since otherwise the for a total height less than 0-8. This is due to the
individual antenna currents would be only 3dB down reduction in mean height as compared with using the
on those for a single element and, with in-phase top antenna by itself. The reduction in high-angle
addition, 3dB gain would result as already explained. radiation may however have advantages for reception
Measurement of radiation resistance can indeed be as discussed in Chapter 9.
used to determine the effective gain of an antenna, End-fire arrays such as the Yagi operate by restric-
though this relates to the antenna at a particular height tion of beamwidth in both planes, and with increasing
and in a particular environment. gain need to be spaced further apart both horizontally
The required spacing for satisfactory stacking de- and vertically. Horizontally-polarized quad arrays can
pends on the patterns of the individual units and the be regarded as a pair of Yagi arrays stacked vertically
method of stacking. Fig 6.11 shows the three usual with a separation of A/4, the azimuthal pattern being
methods of stacking single elements, these being that of a single Yagi and the vertical plane pattern
known respectively as collinear, broadside and end-fire. obtainable by multiplying that of the Yagi by F, for a
It will be evident that different methods of stacking pair of sources spaced 1/4.
produce different radiation patterns; for example, the To illustrate this, the reverse process was applied to
collinear stacking of horizontal antennas tends to pro- one published directivity pattern for a four-element
duce a very narrow azimuthal beamwidth as illustrated quad. This was divided by the array factor for )/4
by the lower curve in Fig 3.9. In contrast to this, spacing to obtain the pattern for a single four-element
a broadside array consisting of horizontal antennas Yagi and, by comparing the 3dB beamwidths, the gain
stacked one above the other, or an end-fire array of difference was estimated to be 0-35dB only.
vertical elements, produces a much wider beam in In the same vein one recent study (reference [19] of
100 HF ANTENNAS FOR ALL LOCATIONS

Referring to Fig 5.4 we may decide to aim for

maximum gain, equating the phase shift (@) to “half the
spacing”. A spacing of 0-15 corresponds to 0-15 x
260° = 54° so that m = 27° and a = cos @ = 0-89. From
the previous example (single pair), R = 224Q so that
from equation (9) in Chapter 5 we have for. the driven
element
Ree = 224 — (0-89 x 183) = 61-5Q
Gain(dB) If the element currents J are all equal they can be
le—S—ei(>0-5A)
obtained relative to the equivalent “dipole” current [p
Collinear pairs
from the ratio of the resistances, ie
NIp = V73I2 X 61-5) = 0-77
ves Ls gett 5A)
The total phase difference between the fields produced
Becdoeaind al by the driven element and by the reflector is 54° plus
27°, ie 81°. Adding with due allowance for this phase
O-2 0-4 difference as explained on p12, the gain in field
Spacing Ntpee Fives a)
strength relative to a dipole is
Fig 6.12. Variation of gain with spacing for collinear and broad-
side pairs 4 X 0-77 sin ()
Chapter 5) predicts a vertical spacing requirement in
= 2-00 or 6dB
excess of A for a stacking gain of 2dB in the case of
seven-element horizontal Yagi arrays. At 0-4) spacing The gain of a single dipole plus reflector with equal
the gain is 0-8dB only and it is evident that it will be currents and @/q@,) = 0-5 (from Fig 5.4) is 5-2dB, so that
very close to zero at a spacing of 0-25), though the the stacking gain in this case is only 0-8dB compared
curve stops short of this point. Even greater spacing is with 1-1dB for the single pair of the previous example.
required with collinear stacking. Though the drop in this case is quite small the stacking
It is quite inconceivable that the quad should have gain disappears rapidly as the number of elements
been granted some special dispensation in regard to increases.
stacking gain, and it is suspected that the ““quad myth” Although this example approximates very closely to
has now gained such wide currency that measurements the case of the quad some gain is lost as discussed
incompatible with it are automatically disregarded earlier (p97) when the ends are bent over to form
while the remainder serve to reinforce the myth. Reli- loops. This is difficult to estimate accurately but a
able measurements are virtually impossible at hf (p236) reasonable allowance would be in the region of 0-1-
and even measurements at uhf with scale models are 0-3dB. This loss occurs also in other arrangements
liable to serious errors. using bent elements, such as the various versions of the
In contrast to this, calculation of gain in the case of VK2ABOQ array.
the simpler types of antenna usually presents little Stacking gains for collinear and broadside pairs are
difficulty as already illustrated by a number of exam- plotted in Fig 6.12, end-fire arrays being fully covered
ples. The following calculation is however particularly in Chapter 5. Fig 6.13 shows the performance to be
interesting, being a close-enough approximation to the expected from the phasing together of two separate
quad for further demolition of the ‘“‘quad myth”, as well antennas, assuming them to be far enough apart for
as serving the immediate purpose of demonstrating the mutual interaction to be neglected, in which case a
reduction in stacking gain when the gain of individual switched phasing line provides the most convenient
units is increased. method. The three-position switch S1 in Fig 6.14
Consider a driven element consisting of two i/2 provides phase shifts of 0°, 60° and 120°, and with the
dipoles stacked vertically and connected in series (as in addition of a phase reversing switch S2, shifts of 180°,
the last example), a similar pair being placed 0-15A 240° and 300° become available. This allows the phase
behind them to act as a reflector. We need to work out to be adjusted within 30° of the optimum value, the loss
first the spacing between the upper element of one pair in gain for this amount of phase error being only 0-1dB.
and the bottom one of the other, ie 0-252 + 0-15? = Unless corrected by a suitable transformer there will be
0-29. The value of R,, for use in equation (9), p74, is a 2:1 mismatch, assuming an swr of 1-0 for each of the
obtainable from Fig 5.2, allowing for two contributions separate antennas.
of 3192 corresponding to the 0-29) spacing and another Fig 6.13 shows the dependence of stacking gain on
two of 60-5Q2 from the 0-15A spacings. This gives a total relative performance of the individual antennas and the
of 183Q for R,,. way in which power is shared between them, and it is
 = ARRAYS, LONG WIRES, AND GROUND REFLECTIONS 101

|» Zz
12

»<
1
; (7

antenna
(1)
Equal performance,
unequal powers

Unequal
performance, To transmitter

(dB)
Gain
4:1 balun
Phase reversal
| Ea, te

Unequal performance, From transmitter To antenna

proportional power-sharing

Fig 6.14. Phasing switch. Two }/12 lengths of cable allow the
relative phase to be switched in 60° steps assuming both anten-
nas to be matched. With use of a phase-reversing switch (neces-
sary in any case for setting-up) this provides a total of six steps of
60°, the maximum phase error then being 30°. Receiving-type
oO 4 8 12 16 20 wafer switches (not miniature) are usually satisfactory. Both
Power or performance inequality (dB) poles must be switched. A 4:1 balun is wired to provide phase
reversal for a coaxial-line system as shown at (b), but with
Fig 6.13. Gain from combination of two antennas. When the
balanced lines it is necessary only to reverse one pair of feeder
antennas differ in performance the gain as read from the figure is
connections. More switch positions may be provided for obtain-
relative to the best antenna alone, and is maximum when power
ing sharper nulls on unwanted signals
is shared in proportion to the gain difference, with the better
antenna receiving the larger share. Both antennas must have the
same polarization at the angle 0 exists only by virtue of the difference BC
in path lengths. This produces a phase shift which can
interesting to note that if the antennas have identical be determined by measuring AB, converting it into
gain a power inequality of 10dB reduces the stacking fractions of a wavelength and, if the answer is required
gain only from 3dB to 2dB. On the other hand, a in degrees, multiplying by 360. If BC is equal to A/2 the
performance inequality of 6dB reduces the extra gain direct and reflected waves reinforce each other to
from addition of the worst antenna to 1-0dB only, even provide 6dB gain relative to free-space propagation,
this being subject to arranging an unequal division of but halving the height or the radiation angle reduces
power based on the gain ratio. this gain by only 3dB. For even lower heights the
It is useful to note that directivity in one plane does decrease in signal strength becomes more or less direct-
not affect stacking requirements in another, eg adding ly proportional to height as shown by Fig 6.16. To help
reflectors to a collinear array does not need an increase . in relating this to practical situations, the point marked
of end-to-end spacing. with a cross applies to a typical dx signal at 14MHz,
assuming an antenna height of 50ft (15m). This applies
equally to all types of horizontal antenna, thus invali-
Effect of ground dating claims that one antenna is better because “‘it
After our sojourn in free space it is now time to come lowers the angle of radiation’’.
down to earth, and by way of softening the impact it is On the face of it Fig 6.16 might appear to be of
helpful to ensure that initially our antennas are all doubtful value since one does not know the actual
horizontal, the ground flat, and all interfering struc- angles of radiation required in any given instance, and
tures removed. Provided the height is not too low, and AB is therefore also unknown. However, the author
assuming for the moment that we are interested mainly has found that predictions of relative average dx per-
in dx (which usually requires a low angle of radiation formance based on radiation angles of 6° are by and
relative to the earth’s surface), it becomes possible to large consistent with experience, although this is an
regard the earth as a perfect reflector. Putting this oversimplification (Chapter 5). There is no doubt
another way, where before there was one antenna there (p14) that much lower angles would often be useful so
are now in effect two, the second one being the mirror that it is usually best in any case to erect antennas as
image of the first. high as possible. Unfortunately, even in the absence of
These two “antennas” are in antiphase, as pictured planning restrictions or other constraints, an increase of
in Fig 6.15(a) from which it will be clear that radiation height beyond, say, 50—70ft (15—20m) usually leads to
102 HF ANTENNAS FOR ALL LOCATIONS

(a) (b)

Aerial

1 VERTICAL
] POLARIZATION [

Fig 6.15. Images of antennas in perfectly conducting ground. The horizontally polarized iimage is in antiphase and aitesls PSfadiation
along the earth’s surface. At appreciable wave angles the path length from theiimage is greater by the distance 2h sin 0, resulting in a
phase difference of 4h sin 6/) radians. At (b) the image of the vertical antenna is in phase and supports radiation along the surface

rather rapid escalation of costs and practical difficulties The ‘‘perfect earth’? assumption, despite its validity
for a relatively small advantage. for low-angle reflected signals, does not apply to vertical
There are other effects of the ground which (though incidence. At low heights, say less than 1/4, a horizon-
relatively trivial) have given rise to some confusion. For tal antenna experiences very considerable ground
example, as height is varied there is interaction be- losses which cause the radiation resistance, or perhaps
tween a dipole and its image in accordance with the one should say apparent radiation resistance, to depart
mutual impedance data for parallel dipoles, Fig 5.2, from the “perfect earth’ value as indicated by the
and this causes the radiation resistance to vary with dotted curve.
height as shown in Fig 6.17. It is often supposed from These losses are often ignored in the literature or
this that a height of 0-6A is specially advantageous since even dismissed as insignificant, in which case the
the lower radiation resistance implies a larger current consequences would be extremely interesting and in-
and therefore a bigger signal, as indicated by Fig 6.18. deed somewhat dramatic. This is because at low height
This effect is however too small to override the general above a perfect earth the radiation resistance varies as
desirability for dx working of putting the antenna as the square of the height.
high as possible as discussed in Chapter 10 and also Halving the height therefore doubles the current, so
demonstrated by Fig 6.16. The variation of resistance that the dx signal should stay constant from a height of |
with height is likely in any case to be negligible with about \/4 downwards until the antenna is almost lying
close-spaced beam antennas owing to the nearly anti- on the ground, by which time the currents and voltages
phase fields existing directly underneath the antenna.
Height of centre of vertical halfwave aerial
above ground
0-25 0-3 0-4 0-5 0-6 O-7 0-75
1
RELL ALL ee
90 ne a ee
BNE GE ARMNE Eo
80 aNNy SRE AGE ia
ti | PALazON tty
70 Pe [SSN a

iid Nts da
oOOo i, Vertical
Aerial
va
Typical beam antenna at 14MHz ai
(height = 50ft or 15*2m)
yy| ee ee
a! EESee ae
40 | Eee
ADEE
FHorizontal
= Aeriall | | [ | [ | | | |
Ratio
field
free-space
to
actual
of ECE Gt bee ee
30 ap
(ohms)
resistance
Radiation
Ff GS
PEPE CE HEE EEE RE
O Ot O-2 O83 O-4 O-5 O-6 O-7 OB O-D 1-0
1°5
Height (A) Height of horizontal halfwave aerial in wave-
lengths above ground
Fig 6.16. Variation of dx field strength with height of transmitting
or receiving antennas. Though drawn for a radiation angle of 5:7° Fig 6.17. Radiation resistance of )/2 horizontal and vertical
the curve is valid for other small angles if height is adjusted in dipoles with height above a perfect earth. Dotted curve indicates
inverse proportion to the radiation angle typical behaviour of horizontal antennas over actual earth
 ARRAYS, LONG WIRES, AND GROUND REFLECTIONS 103

be made for the variation of radiation resistance with

height as shown in Fig 6.17). In practice, however, one
is more interested in the real earth, in which case the
magnitude and phase of the ground reflection vary
between wide limits as discussed in Chapter 10. Below
a certain angle, analogous to the Brewster angle in
optics and about 8 to 15° depending on soil constants
for the majority of residential areas, the phase of the
reflection coefficient reverses and for very low angles
Straight-line its magnitude rises to unity (Fig 10.2) so that there is no
approximation
difference between vertical and horizontal polarization.
Dipole ov Over a range of useful angles, however, the reflection
perfect coefficient is usually in the region of 0-4—0-7 only, so
earth
that even at low heights cancellation of the direct wave
is incomplete and some low-angle radiation takes place.
The amount is too small to justify oft-repeated claims
that vertical antennas are “good” low-angle radiators
voltage)
(Relative
strength
Field
at
low-angle
given
any but it is enough to permit reliable dx communication
Height in wavelengths
even when height is severely restricted, particularly if a
number of such radiators can be assembled to form a
Fig 6.18. Effect of height on dx signal strength with horizontal beam. This is extremely valuable at the lower frequen-
dipoles. The ripples are attributable to variations in radiation
resistance caused by mutual impedance between the antenna cies where it is usually difficult with horizontal antennas
and its image and are greatly reduced in the case of close-spaced to achieve sufficient height for efficient low-angle
beams radiation. Guidelines to assist readers with problems in
this area and a number of related antenna designs will
would be very large. This incidentally is another ver- be found in later chapters.
sion of the finding (pp26 and 84) that the directional Taking into account the Brewster angle effect and
pattern of small arrays is independent of size, and it is a the rarity of situations in which full benefit can be
great pity, after drawing attention to such interesting obtained from a reflection coefficient of unity (Chapter
prospects for what is believed to be the first time, that 10, p139), it would be reasonable to conclude that
attention has to be directed also to the earth losses electrically there is little to choose between vertical and
which despite some references have received very little horizontal polarization. This turns out to be something
attention in earlier literature. of an over-simplification, the optimum choice depend-
The author was himself caught by this to the extent of ing largely on environmental factors so that it may have
much effort wasted, before discovering that the feed- to be resolved by experiment. It depends to a large
point resistance of a A/2 dipole levels out with reducing extent on unknown quantities and may be further
height at more or less the value indicated by Fig 6.17, complicated by constructional preferences. These
though the actual value is believed to vary considerably aspects are also discussed in later chapters. Vertical
from one location to another depending on ground polarization is however particularly useful over sea
constants. As a result the steady decrease of signal water, which is a special case since its conductivity is
strength with height shown in Fig 6.16 is modified to some two orders of magnitude better than that of other
the extent of 1dB or so. natural ground or even fresh water, and this reduces
Since the extra resistance represents ground losses it the Brewster angle to about 2-3°.
is not “radiation resistance” in the strict sense but Vertical antennas are often of monopole type such as
should be regarded rather as an attenuator, and the Marconi antenna (Fig 6.19) which can be regarded
although this distinction is not important for matching as one half of a dipole, the other half being the image in
it is highly significant for reception since the attenua- the ground. Alternatively it may be recalled that the
tion applies to noise as well as signals. It follows (as the ground-plane antenna, Fig 4.19, uses a ‘‘virtual earth”
author has found on many occasions) that when there is consisting usually of a set of three or four radial wires
a very low antenna at one end contacts are “non- arranged so that radiation from them cancels. The
reciprocal”, stations with low antennas being able to practical advantage of this is the elimination (almost) of
give much better signal reports than one is able to give losses in the earth connection (Fig 12.19) and the
them. possibility (in theory) of mounting at any height so that
In the case of vertical polarization over a hypothetic- (assuming perfect ground) up to 3dB “‘collinear”’ gain is
al perfect earth, Fig 6.15(b), the antenna and its image obtainable from the antenna and its image.
are in the same phase, and maximum radiation occurs There is no foundation for the belief that a vertical
at zero elevation (though some allowance may have to monopole has any special qualities in relation to
104 HF ANTENNAS FOR ALL LOCATIONS

Current Pad
Practical methods of increasing the electrical height
distribution ~ without undue increase of physical height include the
along radiator
use of short end-loaded dipole elements supported by
catenaries which also provide the end loading as in Fig
pe A 6.20. This also tends to eliminate earth losses.
a|>
A favoured height for vertical antennas having their
base at ground level is 54/8, corresponding to one half
of the “extended double-Zepp” described on p94. It
should be appreciated that, although a vertical antenna
at sufficient height with its image approximates to a
collinear pair, the pattern over a real ground has a
“nick out of the middle’”’ at low angles, ie just where
Current density inearth
one can least afford it, due to phase reversal of the
image for low-angle radiation. Nevertheless, the de-
crease in radiation resistance with increasing height as
Ri Rrad Re
shown in Fig 6.17 means that current increases and
whatever low-angle signal there is must therefore also
(b)
increase.
Thus far the ground has been assumed to be smooth,
I base
flat and uncomplicated. The effect of ground slopes is
largely predictable but that of power lines, buildings,
Fig 6.19. (a) The Marconi antenna. Current distribution on a short
vertical radiator over a plane earth. The return path for the trees, hills and other objects is mainly a matter of
antenna current to the generator is via the ground which has guesswork plus a few empirical guidelines. These mat-
relatively poor conductivity. The earth resistance is typically ters are the subject of Chapter 10.
around 1002. (b) Equivalent circuit for (a): A,_, ohmic losses
(usually small); R_e, effective earth resistance (usually large); Raq, One curious effect of the ground is to produce
radiation resistance vertically polarized radiation “off the end” of a hori-
zontal dipole. Thus if a dipole is disposed horizontally
low-angle radiation, and remote from ground it differs above an earth plane acting as a reflector, there will
from other short antennas only in the shape of the then be some radiation along the line of the antenna
current distribution. Like them it is a point source of
radiation but since it is a half-dipole the effective length
for use in the /./ formula (p8) is exactly half that of a
dipole, at the same effective height. This means that
the radiation resistance is exactly a quarter that of a
dipole, ie 18-392 in free space or at large heights.
At sufficient height above the ground the ground
plane or a vertical dipole with its image as in Fig 6.15(b)
constitutes a collinear pair, having the narrow beam-
width in the vertical plane corresponding to a gain of
3dB. In contrast, a 4/4 (or shorter) grounded monopole
(including a ground-plane antenna at ground level)
together with its image resembles a single dipole, and
to account for loss of the 3dB of collinear gain one must
suppose the radiation resistance to be doubled, thereby
dividing the current by V2. The additional 18Q can be
regarded alternatively as the mutual resistance between
the antenna and its image. More simply, though
perhaps less instructively, the total figure of 36Q can be
assumed on the basis that this is exactly one half of a
dipole.
The Marconi antenna has an important place in radio
history but, though sometimes used on the lower
frequency bands, is of doubtful merit for the majority
of amateur applications. This is due to (a) losses due to
earth currents, (b) the desirability if possible of raising
the antenna somewhat higher above ground level so as Fig 6.20. Alternatives to Fig 6.19: (a) dipole; (b) dipole or mono-
to obtain some or all of the potential gain of 3dB. pole? (see Chapter 4)
 ARRAYS, LONG WIRES, AND GROUND REFLECTIONS 105

wire. This may be appreciated by holding a pencil signals there will be considerable reduction of directiv-
parallel to the surface of a mirror and observing the ity. On the other hand it may often be possible to
image from the end-on direction at various heights increase the strength of a wanted signal, or reduce
above the mirror. Closing one eye, the pencil shrinks to interference, by switching to another antenna at a
a dot but the image, though perhaps greatly reduced in different height. It is useful to note that the dipole
length, remains visible. Neglecting ground losses the tends only to exhibit reasonable horizontal directivity
amount of radiation is determined by the usual cos 6 at the lower wave angles, and will tend to become
formula, and is therefore proportional to the apparent omnidirectional for the higher angles associated with
length of the image as viewed in this way. Since this short-skip propagation.
image appears to be in the vertical plane if the mirror is
horizontal, the radiation is vertically polarized.
Ground-wave propagation normal to the antenna
wire is of course horizontally polarized but for in- Summary and conclusions
termediate directions the plane of polarization will be The performance of any antenna system in free space
inclined, with completely vertical polarization off the can be analysed by considering it as a number of “‘point
ends of the antenna. sources” of radiation or any convenient grouping of
Of greater importance, apart from local working, is point sources. This was done in Chapter 5 for close-
the considerable high-angle sky-wave radiation “‘off the spaced sources and has now been extended to include
end” of a horizontal dipole. For example, applying the wide spacings, ground reflections and large arrays of
cos 8 formula it is evident that there will be only 6dB various types.
discrimination against signals arriving at an elevation It is important to appreciate the distinction between
angle of 30° from an endwise direction as compared close-spaced arrays, which rely on tight electrical cou-
with signals in the main lobe, neglecting the effect of pling for their operation, and wide-spaced arrays in-
the ground, which due to the higher angle is more likely cluding antennas and their images in the ground. Such
to favour short-skip interference than a wanted dx arrays, though basically independent of mutual cou-
signal. In between these two directions the plane of plings, usually experience some degree of interaction
polarization changes gradually from horizontal to one between their component parts and this may modify the
inclined at 30° to the vertical, remaining of course performance to an appreciable extent. There are some
always normal to the wave direction and in the plane of borderline cases, and wide-spaced groupings of close-
the wire. spaced sources are also possible. It is important to
Due to the varying effect of the ground as the ratio of appreciate that in general the greater the gain of
vertical to horizontal polarization changes, it is difficult individual groups the further apart these must be
to compute accurately the vertical plane radiation spaced; however, if gain is obtained entirely by restric-
pattern for directions well away from the main lobe. tion of the radiation pattern in the vertical plane the
These directions in general are of minor importance but required horizontal stacking distance is not increased,
it should be noted that for short-skip (ie high-angle) and conversely.
CHAPTER 7

Multiband antennas

It is generally essential to meet operational require- the directivity resembles that of a pair of antiphase
ments with the smallest possible number of antennas so sources (p94). At 7MHz the radiation resistance is very
that multibanding, ie making a single antenna operate low, about 72 only, but reasonable efficiency can be
on several different frequencies, is a subject of particu- achieved with care. The current distribution in the loop
lar importance. It is also perhaps the one which poses at 7, 14, 21 and 28MHz is shown in Fig 7.1, and by using
the largest number of interesting problems. There are a A/2 tuning stub as in Fig 7.2 simultaneous resonance is
two main essentials: it must be possible to feed the obtained in or close to each of these bands, the overall
antenna efficiently and the radiation pattern must be length of the system being A/2 at 7MHz and approxi-
acceptable on each band. mate multiples of this on the higher frequencies.
However, the harmonic relationship is not exact and
slight adjustments are desirable when changing bands.
The “odd bit of wire” It is thus advisable to have access to the lower end of
As explained in the last chapter any “odd bit of wire” the stub, in which case there is no problem in tuning the
can be used as an antenna by attaching one end to a loop also to the 10, 18 and 24MHz bands.
suitable tuning unit operating in conjunction with a For multibanding by the use of resonant feeders,
suitable earth return path. This latter need only be a loops have a big advantage compared with dipole
short counterpoise if the wire is near A/2 resonance but elements since they have much larger radiation resist-
has to be of low impedance if the wire is shorter than ances. A 14MHz quad loop has nearly twice the
about 5A/16. Subject to provision of these essentials, radiation resistance and, when used with a long re-
therefore, the ‘“‘odd bit of wire” is a multiband antenna, sonant feeder, almost twice the bandwidth of a similar-
though its efficiency falls off fairly rapidly as the length ly fed single-wire dipole. If “‘squashed flat’’ so that it
is reduced below about /4. turns into a folded dipole there is a further 2:1
improvement in this respect. If the loop is open-
circuited at the top for 283MHz operation there is a
Resonant-feeder systems further doubling of the radiation resistance and 2-4dB
Reference has already been made to the use of dipoles of extra gain (p98).
fed from long resonant lines but it will be evident from For rotary beams a square loop is preferred, and it
the previous chapter that the directional pattern separate feeders are used to each element it becomes
changes considerably with frequency. For example, for possible to switch the beam direction so that less than
a 3-SMHz dipole fed in the centre there is a maximum 180° rotation is needed; there is then no serious prob-
in the broadside direction at 3-5 and 7MHz, whereas at lem with twisting or short-circuiting even of open-wire
14MHz and higher frequencies there is little or no feeders. On the other hand, the folded dipole or delta
radiation in the broadside direction but four main lobes loop is convenient if for example an antenna is to be
at angles to the wire which decrease with increasing slung between two trees. The ability to reverse beam
frequency. Such dipoles can be employed as beam direction by means of a switch is in itself of great value.
elements but their use in this way is restricted to about
one octave due to the change in directional pattern and
to non-optimum spacings. Loops can be used similarly Optimization of spacing
except that some means must be used to open-circuit With two elements, whether dipoles or loops, if the
the top of 14MHz loops at 283MHz, otherwise operation spacing is optimized for one band it tends to be
as a beam is typically restricted to the 14, 18 and non-optimum for another; thus if the spacing is 1/8 at
21MHz bands only. Operation on 10MHz may be 14MHz it becomes 4/4 at 23MHz. This is too wide (p71)
feasible subject to some reduction in performance. so that some gain is lost and there is considerable
Even without the open-circuit provision, a single deterioration in front/back ratio. The problem is largely
14MHz quad loop can provide good performance on all resolved in the case of three elements since the boom
bands from 7 to 28MHz, although at 28MHz the length then becomes A/2 at 28MHz, anda substantial
polarization is changed from horizontal to vertical and amount of additive gain results as may be inferred from
106
 MULTIBAND ANTENNAS 107

Fig 7.1. Current distribution in

loops of various sizes at dif-
ferent frequencies. The arrow
lengths indicate very roughly
the difference in the relative
contribution to the total field
strength. Illustrations (a), (c), (d)
and (e) represent a 14MHz quad
loop excited at 7, 14, 21 and
28MHz respectively, while (b)
and (d) correspond to a slightly
oversized 21MHz loop used at 14
and 28MHz.. Illustration (f)
shows the effect at 28MHz of
open-circuiting the top corner of
a 14MHz loop, thereby turning it
into a “bi-square” element; by
using a linear resonator this
effect can be achieved without
disturbance of operation at 14 or
; 21MHz

Fig 5.27. There is still some deterioration of front/back to bend the elements of a two-element array inwards so
ratio but this may be found less important at the higher that the ends (or points of maximum voltage) are
frequencies, and should in any case be amenable to the brought towards each other. This improves perform-
use of “‘nulling”’ techniques as discussed in Chapter 9. ance at 14MHz as explained in Chapter 5, and at 23MHz
Due to the increased element length and spacing (these it brings the centres of gravity of the current distribu-
being measured in wavelengths) there is a considerable tions in the elements closer together as shown in Fig
increase in radiation resistance at the higher frequen- 7.3. In this way it should be possible to achieve
cies, and this allows operation closer to the optimum reasonable coverage of all bands (including 10, 18 and
condition, an advantage which can if desired be traded 24MHz) from 10-28MHz. Bandwidth is expected to be
for greater bandwidth. rather narrow at 1OMHz but the band itself is so narrow
If a shorter boom is required the problem can be that this is not a problem! Another method for the
resolved by use of a third element which operates on multibanding of loops is to feed them symmetrically as
the higher frequencies only. Yet another alternative is in Fig 7.4 which shows an arrangement developed by
DJ4VM [1]. By providing a symmetrical current dis-
tribution this eliminates unwanted modes which might

To loop To loop

TOT L1 L2
(21-9m)

(a) 3ft 4in

(101-5cm) (b)
es

d
d C3

Fig 7.2. A loop and stub arrangement with resonances close to 7, 14, 21 and 28MHz. A low-impedance feeder may be connected at point
(d) or a 600Q line at (e). Resonances are not in exact harmonic ratio so that slight changes in stub length may be needed; for 14/21MHz
this may be effected automatically by compensating circuits tuned to about 18MHz as shown at (b) where C1 and C2 are 500pF and C3 is
250pF. By alteration of stub length 18 and 24MHz may also be covered
108 HF ANTENNAS FOR ALL LOCATIONS

Current
distribution
28MHz Current
distribution
14MHz

Insulators
can be
~ polythene
cord

Plan view

Vs
Fig 7.3. End view of a pair of vertically polarized quad loops
“doubling” in the role of guy wires, the loops being pinched in
towards the mast at top and bottom. Current distributions for 28
and 14MHz are shown dotted on left and right respectively but
each applies to both loops. Effective spacings, ie between “’cen-
tres of gravity” of the current distributions, are S; and Sp at 28
and 14MHz respectively. For this version (unbroken loops) the
loops must be fed symmetrically at each of the side corners using
the DJ4VM method (Fig 7.4)

be expected to result in (a) extra interference at 18 and

Fig 7.4. Current distribution in centre-driven quad elements. The
21MHz and (b) a relatively poor pattern with reduced current loops are always symmetrical. The feedpoint irnpedance
gain at 24MHz. is variable and tends to be highly reactive so that resonant
Antennas based on these principles are described on feeders are needed for multiband operation (Ham Radio)
p191.
unless the apex angle exceeds about 120°. This is due to
(a) a big increase in relative strength of the vertically
Use of traps polarized end-wise mode of radiation and (b) reduction
The most usual method of multibanding, and in the in the effective mean height if the whole element is
author’s view the one with the least merit, is by the use used, as may be seen from inspection of the current
of traps as shown in Fig 7.5. These are tuned circuits distribution. Due to increased losses in conjunction
which act as insulators to cut a dipole, monopole or with the lower value of radiation resistance, traps are
beam element down to size at the higher frequencies, not recommended for beams using inverted-V ele-
thereby throwing away additive gain which is “worth” ments.
almost 2dB when a dipole is used at twice its fun-
damental frequency. In addition losses are significant,
at least in the case of beams; bandwidth is reduced due
to the high Q of the traps which is essential for keeping (a) (b)
losses as low as possible; elements have to be heavier or
wires thicker in order to support the traps; and the

ae"
traps are a frequent source of unreliability due to
ingress of moisture.
It is only fair perhaps to add that most proprietary
beams are of this type and many of them produce
outstanding dx signals; the loss in performance com-
pared with a monoband beam should according to the =
=e

author’s calculations be no more than just noticeable,

though enough to justify the preference of many lead- Fig 7.5. Trapped dipole. One half is shown at (a). The traps act as
insulators, the resonant lengths for f,; and f, being reduced
ing dx operators for monoband beams. somewhat by the inductive reactance of the included traps. When
Leaving aside the question of weight, however, there applied to a ground plane antenna this may consist of a self-
are sound arguments for the use of traps in the case of supporting tube as shown at (b), the tube being cut and then
rejoined using rod insulators over which the coils are wound.
inverted-V dipoles since these are somewhat unsatisfac- Separate sets of radials are needed for each band. Trap circuits
tory at harmonic frequencies unless cut down to size, or must be well protected against the weather
 MULTIBAND ANTENNAS 109

Trap losses can be readily estimated if the radiation current in the capacitor this has to be multiplied by
resistance is known; thus in the simple case of a roughly two. For a coil Q of 200 we thus have to reckon
two-band beam for 14/21MHz based on Fig 7.5(a), and a loss resistance of 570/200 x 2, ie 5-7@ in series with
aiming for the 6dB point on Fig 5.16, we have the radiation resistance. This is a loss of 100 log,
R = 20. Z, from Fig 4.13 is typically about 800Q so 22-7/17 = 1-26dB. Provided that for some odd reason
that the impedance as seen between the ends of the one would rather lose power at 14MHz than at 21MHz
driven element at 21MHz is 800/20 x 800 = 32,0002. (where the QRM is not so bad!) this result can be
The impedances of the two traps in series appear as a construed as a slight improvement. Note however that
loss resistance across this and are equal to the Q of the the use of rf speech processing could raise the mean
coil times its reactance or that of the tuning capaci- power (for 400W p.e.p) to 100W, in which case each
tance. trap will be dissipating 12-SW. This is enough to melt
On such occasions the author finds it saves time to the solder (unless the traps are heavily constructed, in
remember a rather approximate but nicely-rounded which case the mediaeval hazard of a shower of molten
figure of 8,000@ per picofarad at 21MHz, 12,000Q at lead will no doubt be replaced by one of falling
14MHz, and pro rata. Thus, having read somewhere or debris!).
other that 40pF is a typical value of trap capacitance, The losses calculated above are only part of the total,
and confident (however misguidedly) of one’s ability to since losses in the parasitic elements must be reckoned
achieve a coil Q of 200, the impedance of each trap separately. Assuming a typical current ratio of 0-6, a
comes to 8,000/40 x 200 = 40,00022, ie a total of fraction 2 X 0-6? or 72 per cent must be added to the
80,0002. The ratio of power lost to power radiated at above losses, bringing them over the 2dB mark, and
21MHz is the inverse of the impedance ratio, ie 32/80, some further loss will occur due to the extra traps in the
so that only 80/112 of the power is radiated, a loss of more usual case of three-band operation. These exam-
1-5dB or a quarter of the total gain. ples are further pursued in the next chapter, demon-
Critics, particularly those who feel that all calcula- strating the adverse effect of traps on bandwidth.
tions should go through a computer and be accurate to In concluding this discussion it is perhaps fair to
at least 10 places of decimals, will no doubt contest this mention that one design of commercial beam uses traps
figure, but to restore confidence in trapped beams in the driven element only (with separate parasitic
requires drastic measures and could generate even elements for each band) and has an enviable reputation
worse situations. For example one might reduce C to for performance.
10pF which consists largely of the coil self-capacitance
and probably limits the Q to about 140 though, even so,
the loss is reduced to about 0-6dB which is a substantial Stacking
improvement. At 14MHz however there is now a large Another common method of multibanding is “‘stack-
and lossy inductive reactance which may be obtained by ing” in its various forms. Fig 7.6(a) shows )/2 dipoles
dividing 21/14 = 1-5 into the 21MHz reactance figure for different frequencies connected in parallel, in which
of 8,000/10 = 8002, ie 5708. This is shunted by case only the one for the frequency in use presents a
capacitive reactance of 1,200@ and, using the normal low impedance to the feeder. The others are highly
rules for impedance in parallel, we have a net inductive reactive with impedances of several hundred ohms,
reactance of except that a 7MHz dipole, if included, does duty also
on 21MHz. The same principle may also be applied to
—1,200(570) = 1,086 radial or counterpoise systems for the grounding of
570 — 1,200 vertical antennas. Some systems have been described in
or 2,172 for the two in series. This is added to a which dipoles for adjacent frequencies are formed from
reactance 1-7Z, = 1,3609 seen looking inwards from 300Q feeder so that they are stacked within a few
the traps and tunes with about 3pF which is provided by millimetres of each other. This is, to put it mildly,
the extension of the dipole beyond the traps. difficult to reconcile with the usual stipulation thai
The length of this extension can be found from the when beam antennas for different frequencies are
Smith chart, p62; the reactance, being equal to 5-4Z,, is mounted on the same mast the separation must be at
equivalent to a “‘line length” of 0-028) or 1-9ft (0-58m). least 6ft (1-8m).
These extensions carry a triangular current distribution The usual method of resolving the conflict is
and thus add relatively little to the radiation resistance apparently to make sure that both statements do not
at 14MHz which, with some help from Fig 3.12, is appear in the same chapter, but there could be a simple
found to be about 17Q, ie a reduction of 15 per cent explanation since both statements are evidently based
only. on experience, and it is likely that when one beam is
So far not too bad, but what about the coil losses? being tested the feeder of the other one is pushed aside
The current at this point in the element is about 0-5 with careless abandon. There will then be a significant
times that in the centre but because of the circulating chance that the other beam plus its unspecified length of
110 HF ANTENNAS FOR ALL LOCATIONS

Table 7.1 Typical trap dimension for 2in (5cm)

spacing and 14swg (2mm) wire as shown in Fig
7.6(b)
(a)
Frequency (MHz) Length (in) Length (cm) Capacitance (pF)

28-5 16 40-6 50
20 16-3 41-4 68
21 20 50:8 68
18 24 61-0 80
14 30 76-2 128

(b) recommended when the frequency separation is much

Zero current here less than a half-octave. As an alternative the use of
Fig 7.6. (a) Multiband )/2 dipoles. The dipoles are cut to the relays as below has many advantages.
appropriate length for each band and the centres joined by short The so-called ‘‘multiband quad”’ consists of separate
lengths of low-impedance feeder. (b) Traps for suppression of arrays for 14, 21 and 28MHz stacked inside each other. -
currents induced from neighbouring dipoles. The object is to
prevent any disturbance to an element AB as a result of currents This is another instance of “‘cutting down to size’’, but
which it may induce in CD. The trap lengths are not critical, the even worse than before since using a quad loop at twice
required capacitance being inversely proportional to length. See
Table 7.1 for dimensions
its fundamental frequency provides 3-4dB of additive -
gain (p98), ie nearly twice as much as in the case of a
dipole element. It seems a great pity to throw this
feeder will resonate near enough to the wanted frequen- away, besides being quite unnecessary since there are
cy to cause considerable mutual coupling. several alternatives, two of which are described below.
Adding 6ft (1-8m) to the mast is one possible remedy In the case of stacked arrays, instances of interaction
though it would be (a) simpler (b) better (c) cheaper to have been reported and this could be due either to
short-circuit the unused feeder or, if already short- unused feeders of unsuitable lengths or to higher-
circuited, to try an open-circuit instead. This is little frequency currents flowing in lower-frequency ele-
more than conjecture but some evidence that the ments as discussed above.
trouble arises from the driven element only and is
therefore controllable from the shack can be inferred
from the commercial practice of mounting parasitic Multiband operation by retuning
elements for different bands in fairly close proximity on An antenna intended for one frequency may, in princi-
the same boom, and the author has even stacked a ple, be used on any other simply by retuning and
21MHz quad a few inches inside a miniature 14MHz making any necessary changes to the matching arrange-
quad with negligible interaction. ments. For example, starting with a i/2 dipole for
On the other hand, in the case of the modified 14MHz, the length becomes 3)/4 at 21MHz, ie the
VK2ABQ-type beam (p173) serious disturbance of half-length increases from A/4 to 3A/8 and, remember-
28MHz operation was found to occur due to 283MHz ing that the reactance of a 1/8 length of line is always —
current flowing in adjacent 21MHz elements; this re- equal to Zp, this is the amount that has to be removed
sults in critical adjustments and narrow bandwidth but to restore resonance. Since the reactance of the anten-
except in extreme cases can be cured by the use of na is inductive (Fig 4.7), capacitance must be inserted
simple linear traps inserted each side of the centre of and for a typical wire dipole (Z, = 1,000) the re-
the 21MHz element, Fig 7.6(b). Each trap consists of a quired value comes to 8pF. At 28MHz an insulator is
single capacitor plus a long lead, whereby it is con- required and this can take the form of a tuned circuit
nected across part of the existing element to form a which is usable also as a matching device. Fig 7.7 shows
resonator. It has only a modest task to perform and how band-switching might be effected using, say, a
there is negligible disturbance of normal operation on Ledex switch mounted in a waterproof box, though an
either band. Use of traps in this way whenever the need equivalent arrangement can be worked out using re-
arises should in principle allow the use of parallel lays. At least one such arrangement has been described
dipoles to be extended to include coverage of 10, 18 in the literature but objections on grounds of reliability
and 24MHz. may have some validity, unless there is easy access to
Unfortunately interaction can take subtle forms and the antenna by climbing or lowering the mast. This is
may not be obvious, as for example when it alters the because components specifically ‘“designed for the job”
coupling factor between driven and parasitic elements. are not available at the time of writing except at high
Unless and until such effects have been more fully prices. Nevertheless some relays such as RS type
investigated, or unless the operator has plenty of time, 349-658 have been found suitable; these are small, light
patience, and experience, this method cannot be and relatively inexpensive, contacts being rated at 5A,
 MULTIBAND ANTENNAS 111

a suitable fraction of L. Another method, Fig 7.8(d),

U

Wa1wiizfeeder
uses capacitors C, at the higher-frequency current
nodes where they have no effect; at a lower frequency
they tune out the inductive reactance of the system,
thereby providing two-band operation.

Ledex-type switch
The linear resonator
A disadvantage in each of the above cases is the need to
feeder split the element in the centre which is no great
2
hardship in the case of, say, a A/2 dipole since the
insulator need not be of particularly high electrical
eves
r\ feeder
quality. Unfortunately at 283MHz the dipole is full-wave
so that high rf voltages exist across it and the situation is
Fig 7.7. Centre tuning of 14MHz 1/2 dipole; C, C,,,, C resonate with entirely different. It was frustration due to delay in the
L at 28MHz, a low-impedance feeder being coupled into L as
shown, though a high-impedance feeder tapped across L may
delivery of a “better’’ insulator that caused the author
also be used. At 21MHz the capacitors are series resonant with to look for a different solution. This took the form of
the excess inductance of the antenna and C,, is chosen to match using part of the element itself for the inductance L as
any desired feeder impedance. At 14MHz the element is series fed
in the normal way with low-impedance line, the shunting effect
shown in Fig 7.9. This allows the element to be
of C,,, being negligible. A common low-impedance feeder may be supported in the centre and is compatible with “‘plum-
used via two extra poles on the switch. This is only one of many ber’s delight”? mechanical construction.
possible arrangements
An unexpected bonus was provided by the accidental
discovery that, due to a highly asymmetrical type of
250V ac. A number of measures as follows may be resonance, the addition of C had very little influence at
taken to ensure reliability: 14MHz. In effect L virtually disappeared at 14MHz by
1. Use relays preferably in conjunction with other reverting to its normal role as the ‘‘middle bit of the
methods such as close stacking of multiband ele- antenna’’, so that normal monoband dimensions could
ments, or the linear resonators described below. be used and performance was virtually unaffected by
This simplifies the switching as compared with
more basic methods involving switched networks
as in Fig 7.7.
2. Take precautions to ensure that switching opera-
tions are never carried out while rf power is
applied.
3. Test thoroughly on full power before erection. C2

hey
Referring again to Fig 7.7, another alternative is to G4

use “‘passive”’ switching by suitable electrical networks (b)

i ‘ Sy
which provide appropriate reactances at each
frequency; Fig 7.8(a) shows one of several arrange-
ments developed by VK2AOU and DJ2UT [2], and Fig
7.8(b) a somewhat similar scheme developed by the C2

author. In Fig 7.8(a) one circuit could be tuned to act as

an insulator at 28MHz, and the other detuned to Ese 3
ea
provide the required capacitance at 21MHz. At 14MHz Vc}
there will be some added inductance which can be eas % Ayia
compensated by adjustment of element length.
Three-band matching is achieved with the aid of
auxiliary driven elements for 21 and 28MHz, and it is
CH
possible to think of many variations on this theme (d)
which has also been applied to quad antennas. In Fig egrets tee hie hy pom
eaters
at
7.8(b) the circuit L1, C1 is resonant at 283MHz and C2 CE GE

provides the additional capacitance required for

21MHz. C2 can be switched by means of a relay or as Fig 7.8. Multibanding networks with passive switching: (a) as
developed by VK2AOU; (b) semi-active network (G6XN); and (c)
illustrated by the circuit L3, C3, which is also resonant passive version of (b); (d) L, C,, act as an insulator for the highest
at 28MHz and acts as a passive switch. Matching was frequency, C; tunes out the inductive reactance at a lower
achieved on all three bands by tapping the feeder across frequency
112 HF ANTENNAS FOR ALL LOCATIONS

ability to isolate or detune any metal structure without

jee]
cutting into the structure as would be necessary with
ordinary resonators. .
In linear resonator systems three-band matching may
be achieved by connection of the feeder across an
appropriate fraction of L as in the case of the inductive
stub, Fig 7.8(b). The principle is illustrated in Fig 7.10
where (a) represents any \/2 element with a parallel
C2
(b) feed, such as a T or delta match. Selecting points A and
B at random, the wire joining them can be regarded as
G3 C3
an inductance L tuned by the capacitance C,, of the
outer portions, as shown on the right. As an approx-
Fig 7.9. Linear resonator. An electrical conductor AB having an imation, provided L is not too large or too small, the
inductance L is bridged by a second conductor which includes a
capacitor C. At (b), capacitors C3 form additional resonators current-loop radiation resistance R can be regarded as
which act as passive switches to remove C2 at 28MHz being in series with it, giving rise to an equivalent
parallel resistance w?L?/R between A and B.
To use the same antenna at 283MHz we need only
multibanding. With C adjusted for 21MHz operation connect a small capacitor Cy. across L, thereby tuning it
there was a downward shift of only 13 per cent in the to resonance so that it acts as a centre insulator and the
design centre frequency at 14MHz but this increased to dipole becomes two half-waves in phase (Fig 7.10(c)).
about 5 per cent in a later design as a result of increased By “borrowing” some of the dipole to act as an
diameters and spacings of conductors. inductance, the half-waves have been shortened, ie
For three-band operation without relays the neces- capacitance has been subtracted, but it is easily re-
sary reduction in capacitance at 283MHz is effected by placed by a slight increase in the value of C,. The loop
additional linear resonators formed by C2, C3 as shown
Rp
in Fig 7.9(b), an arrangement which has also been
applied to quad antennas (see p158). A detailed ex- E A B D
planation and analysis of linear resonators will be found (a) 14MHz ia ry
in an appendix to this chapter and antennas using this x x
principle are described on pp160, 171-2 and 218.
Although two-band operation is easy to achieve with X X
this method and three-band operation not unduly diffi-
C21 Rp
cult, the inclusion of further bands would require either
a servo-controlled variable capacitor at the centre of | le D
each element or banks of preset capacitors selected by (b) 21MHz BEnEGmlisl Pan Gia Ge
relays or a Ledex-type switch. The method currently
favoured by the author would use two-band elements
according to Fig 7.8(d) or 7.9(a) selected by relays
x ge
which also ensure that any element likely to cause x X
problems on a given band is open-circuited or detuned.
To avoid the need for protecting coils against the
E a) 3 =a
weather (Fig 7.8(d)), turns must be of large diameter, (c) 28MHz
self-supporting with at least 4in spacing between turns;
alternatively open-wire stubs may be used and can if
necessary be folded into the form of a four-wire cage
(p249).
Fig 7.10. Principle of multiband matching (not to scale). (a)
The linear resonator was first employed by DLIFK inductance (L) of AB is tuned by the capacitance of AE, BD. Ap is
as a means for obtaining multiple resonances of small the radiation resistance referred to the terminals of L which acts
beam elements [2]; in this case, at the lowest frequency, as a step-down autotransformer. Value of feedpoint impedance
depends only on the distance XX so that A, B are arbitrary points
the resonator is acting as a loading device and there is which can be chosen to coincide with the terminals of the linear
inevitably a large circulating current which increases resonator of (b) or (c). (b) Inductance L is roughly equal to the
losses and restricts the amount of size reduction which inductive reactance of AE, BD which appears in parallel with it.
C2, tunes both reactances, and the radiation resistance can again
can be achieved without serious inefficiency. In con- be represented by a parallel resistance R,,. (c) L and Cog form a
trast to this, at the higher frequencies the radiation parallel-resonant circuit to act as an insulator between the two
resistance is much higher and losses should be negligi- half-waves and allow them to be energized in phase. It also acts
as a transformer to match the radiation resistance A, into the
ble. Other applications of linear resonators are de- feedline. To compensate for “borrowing” part of the antenna to
scribed in later chapters, an important feature being the form the inductance, the value of C2 is slightly increased
 MULTIBAND ANTENNAS 113

Table 7.2
parasitic elements were used for 21MHz, and for
Band R R Rp Rp 28MHz it was found possible to incorporate linear traps
(MHz) (dipole) (beam) wlorZ, (dipole) (beam) in the 14MHz elements without detriment to their
14 70 20 200 570 2,000 performance. Despite some good signal reports, all
21 200 100 300 450 900 adjustments proved to be rather critical, bandwidth was
28 200 150 600 1,800 2,400 narrow and the feeling developed that this was not an
ideal system, one problem being excessive coupling on
radiation resistance is stepped up by the antenna itself 21MHz via the 14MHz elements. This required sepa-
acting as a A/4 transformer, so that it appears as a rate neutralization and the complications were felt to
_tesistance Z,*/R across L, where Z, is the characteristic be excessive.
impedance of the antenna.
On 21MHz the antenna, looking outwards from AB, 21MHz loading
mM
is an inductance L, in parallel with L as shown in Fig
7.10(b). It is convenient to choose the length AB so that 28MHz trap

L and L, are roughly equal, and a radiation resistance > ——

R measured at a current antinode becomes roughly 2R
when the reference point is altered to the centre of the
element. If J is the current in the element at this point
the total circulating current in C,, (Fig 7.10(c)) is 2/,
and the parallel impedance R, is again given by w?L/R. 28MHz loading (a) (b)

Typical values for these quantities are given in ohms for Fig 7.11. Multiband capacitive hats for miniature beams. Note use
dipoles and beams by Table 7.2. of linear trap in (a) to keep 28MHz current out of the 21MHz
For an swr better than two, which is adequate for loading wires. Traps used in this way do not experience high
voltages or contribute appreciably to losses
most purposes, an impedance variation of 4:1 is
allowed, so the above impedance figures provide a
reasonable starting point for the design of a matched
feeder system. In principle, all that is needed to use the A two-band arrangement on the lines of Fig 7.11(a)
linear resonator itself as the matching device is to tap is more promising, being very similar to the stacked-
the feeder across the appropriate fraction of L, given dipole arrangement described earlier except that the
by VZ)/R, where Z, is the feeder impedance. For dipoles share a common centre portion. This, though of
example, ifa 300Q feeder is tapped across 45 per cent low impedance, is responsible for the whole of the
of L a 5:1 impedance ratio is obtained and the worst radiation. Up to the time of writing this project has not
swr in the case of the beam will be 1-7. been completed, though it has been established that
This example is based on arbitrary assumptions and linear resonators are necessary, as in the case of Fig
in fact does no more than establish a “tendency to 7.6(b), to keep 28MHz current out of the 21MHz
match’’, but with this as a starting point it is usually loading wires, and it can be assumed that neutralization
possible to obtain much lower values of swr than those along the lines of Fig 5.22(b) will be essential. Band-
indicated above. width has not been checked but any difficulties should
be overcome by using two separate beams stacked
concentrically.
Miniature beams
Both the lumped circuit, Fig 7.8(b), and linear resona- Advantages of multiband operation
tor methods were first applied to the parasitic elements In the case of Yagi beams currently available on the
of centre-loaded small beams, the loading being pro- commercial market, the universal use of traps for
vided by the resonators [3, 4]. Centre loading has multibanding means that best dx performance is being
already been criticized (p28) because of very low values achieved by those with monoband antennas to the
of R and the situation is made a lot worse by multiband- disadvantage of those unable to erect separate antennas
ing since the circulating currents in the resonators are for each band. However, multibanding if properly
much greater than the element currents. On the other executed results in better than the usual monoband
hand, with short elements there is no question of performance at the highest frequency since in effect
“additive” gain and therefore no objection in principle one ends up with two beams stacked side by side. In the
to stacking. case of three-element beams the advantage should be
In one case a small 14MHz beam similar in principle greater than the figures quoted, due to a threefold
to Fig 5.22 was used in conjunction with a separate increase in radiation resistance at 21 and 283MHz which
centrally-placed driven element, having a linear resona- enables the design parameters to approximate more
tor to give 21/28MHz operation. Separate (stacked) closely to the theoretical maximum gain condition.
114 HF ANTENNAS FOR ALL LOCATIONS

Use of high-frequency antennas at lower To antenna

frequencies
Another aspect of multiband operation arises if after,
say, giving priority to dx working at the higher frequen- Sie
cies, there is a desire to operate also on lower frequen- Switchbox
cy bands and possibly “‘have a go” at some dx on these eer LF |
bands, as well as using them for shorter ranges. Vertical Radial | ———_ Radial

antennas, provided they are used at both ends of a r-~e@1D

is: tal
| : Corl can be
contact, are best for ground-wave communication and a Radial Radial ! inserted here
-i- if necessary,
vertical antenna is usually best also for dx (although in Toe E but forthe LF
the case of 7MHz, if a fixed wire beam can be erected at position only

a good height, say 5Oft (15m) or so, this may provide Yyy, From
transmitter
the best choice for a particular direction). A vertical
Fig 7.12. Use of 14MHz beam as top-loaded ground-plane
antenna is poor for the shorter ranges beyond ground- antenna for 3-5MHz. Feeder should be brought down well clear of -
wave range, due to the lack of high-angle radiation, and the mast if possible, but in the case of a metal mast it should be
in this case it may be best to use a horizontal antenna possible to achieve isolation by using a linear resonator DE. For If
and bounce the signals off the ionosphere at about operation the feeder could probably then be tapped onto DE but
illustration assumes either a wooden mast (preferably dry!) or
vertical incidence. In this case very little height is adequate clearance between feeder and a non-resonant metal
needed, ranges of up to several hundred miles being mast, the beam element being insulated from the boom. C or L
consistently achieved with powers of a few watts and can be inserted as shown in order to resonate the vertical
structure
antennas only a few feet off the ground.
Fig 7.12 provides an instructive example, typical of
situations which frequently arise in practice and which dipoles and, provided the reader is fully conversant
can be exploited very effectively. AC is a 14MHz dipole with the earthing requirements (p44) and measures for
or beam element with a feeder BDEF supported by a prevention of feeder radiation (p51), there should be
wooden mast. A break at D with, say, a plug-and- no difficulty in achieving a satisfactory compromise.
socket connector allows BD to be used as a vertical A/4 Metal masts and towers can themselves be used in
antenna top-loaded by AC on 3-5MHz. This is operat- many cases as vertical radiators on low-frequency
ing against a ‘‘two-radial ground plane’’, or more bands as shown in Fig 7.13 [5]. For this it is usually
precisely the centre of a horizontal dipole which serves assumed that the tower height should not be much
as an artificial earth, connection to the feeder being greater than 5)/8 since otherwise the current distribu-
made as shown. If the length of the radiator is not tion contains two loops of opposite phase and low-angle
exactly resonant, it may be tuned by a series inductance radiation tends to be cancelled, though it should be
or capacitance as indicated, and the “radials” may be possible to overcome this limitation by use of a linear
bent or pruned as necessary to fit them into the resonator near the mast centre as shown in Fig 7.14.
available space. In view of the advantages of being able Given the availability of, say, a tall mast supporting a tv —
to switch polarization, provision may be made to use antenna for reception in a fringe area, this could be
the radial wires as a horizontal dipole for short-range pressed into service as an all-band vertical and Fig 7.14
working. is a suggested method of achieving this. Such an
Using baluns as necessary there is little difficulty in arrangement breaks new ground and has not been tried
adapting this idea for any type of feeder, a 4:1 balun up to the time of writing, though its feasibility has been
being recommended for 300 to 600@ line feeder, indicated by the author’s experiments with linear re-
although an atu will then almost certainly be required sonators.
in the shack unless the feeder length does not differ On the other hand, towers and masts supporting hf
greatly from )/2. For use as horizontal radiators the beams have often been used as low-angle radiators for
“radials” can be made from 300Q ribbon and con- the longer wavebands and are capable of giving excel-
nected as a folded dipole. These and similar vertical lent dx performance. Cables attached to hf beams
ground-plane arrangements are capable of excellent dx should preferably be run down inside the metal struc-
performance on 3:5MHz but may be adversely affected ture and earthed for rf at the base of the support. This
by wet weather unless the top is insulated and the is to ensure that rf currents do not stray along unpre-
feeder brought down clear of the mast. These precau- dictable paths such as via the mains wiring to ground. If
tions may also allow a metal mast to be used provided it the cables have to be run down externally they should
is non-resonant. be screened, the screens being bonded to the structure
To recommend such an arrangement in view of the at several points if this is possible, though usually this is
remarks about ground planes on p145 may seem incon- not essential. The structure should be checked for dec
sistent, but it is difficult to convert such structures into continuity, any junctions liable to develop high
 MULTIBAND ANTENNAS 115

Connection to tower

Equivalent
1-8MHz feed wire circuit
No8 copper wire <(
spaced 30in (O-76m)
maximum from
tower x

Fig 7.14. Use of linear resonator to convert a tall mast into a

X...indicates collinear vertical array for a higher-frequency band. The tuned
stand-off insulator circuit acts as an insulator and phase inverter but for this purpose
the centre must be earthed. The impedance to earth need only be
small compared with the end impedance of the collinear ele-
ments and can easily be provided by portions of the guy-wire
system

— 3-5MHz feed wire

No 8 aluminium wire
spaced 15in (O-38m)
Due to the difference in their requirements and in
from tower particular the nature of their frequency allocations,
commercial operators have hitherto been more in-
terested in extensive frequency coverage (wide band-
2 x 150pF variable width) than in gain or small size, but the allocation of
Abe capacitors 10, 18 and 24MHz to amateurs tends to bring their
needs more closely into line with commercial ones.
250pF variable ———__ From the initial reactions of many amateurs to these
capacitor
frequency allocations the log-periodic antenna may
well have a start over its competitors. This may in some
instances be an acceptable compromise but lest there
Coaxial cable
feeders
Buried radials should be a “sales drive” to this effect it seems
connected to
= base of tower appropriate to point out some of the disadvantages.
/
WA
/
/ |
|
\
\
\
\
N
The design of log-periodic arrays is one of the topics
for Chapter 14 and reference may be made to Fig 14.8
Fig 7.13. Details of shunt-fed tower as used at W5RTQ. The
1:-8MHz band feed on the left side connects to the top of the which shows a typical log-periodic antenna designed to
tower through a horizontal arm of 1in (25cm) diameter alumi- cover a frequency range of 10-30MHz. This uses 11
nium tubing. The other arms have stand-off insulators at their elements increasing progressively in length by about 4
outer ends made of 1ft (30cm) lengths of plastic water-pipe. The
connection for the 3‘5MHz band, right, is made similarly at 28ft to 1, the overall length of the array being in the region
(8-5m) but two variable capacitors are used to permit adjustment of 40ft (12m). The elements are connected in alternate
of matching with large changes of frequency (ARRL Antenna phase so that, looking at two elements only, they bear
Anthology)
some resemblance to the driven arrays of Chapter 5.
resistance due to rust or corrosion being bridged if The rear element of the pair is ““delayed”’ by an amount
possible, though this is not feasible with telescopic “comparable with the spacing between the elements”’
towers or masts. so that radiation takes place in the direction of the
shorter elements. At any one frequency only the three
or four elements nearest to resonance are doing useful
The broadband approach work, so that from the point of view of performance a
The methods described thus far have achieved opera- rough equivalence with a monoband close-spaced Yagi
tion on several bands by providing multiple resonances, might be expected, conversely a monoband beam with
some or all of which may be quite sharp. There are a 40ft (12m) boom length would provide several de-
however a number of ‘‘broadband” antenna systems cibels of additional gain at 14MHz.
capable of providing continuous coverage over a band The calculated gain for the example in Fig 14.8 is
of one octave or more, thereby meeting all or most of 6-5dB and another typical design [6], using 12 elements
the user’s multiband requirements for dx communica- on a 26-5ft (8-1m) boom to cover 14-28MHz and having
tion. The snag about this approach lies in the impossi- a longest-element length of 38ft (11-6m), also provides
bility of combining high gain with large bandwidth and about the same gain. Reasons which currently cause
small size. many amateurs to opt for monoband beams rather than
116 HF ANTENNAS FOR ALL LOCATIONS

multiband beams using traps would therefore apply in tend to be ‘‘easy’’ for two-band operation, ‘“‘difficult”’
this case a fortiori. Moreover, quite apart from size, the for three-band operation, and “impossible” for more.
extra weight and windage of the large number of Stacked elements may be feasible for six bands
elements must be included in the reckoning. It is most subject to the methods described but, though three
unlikely that a log-periodic antenna with a frequency bands are ‘‘easy’’, a six-band system may well turn out
range of 10-30MHz (or even 14-30MHz) could be built to be rather elaborate. This seems to indicate a clear-
which would be comparable in size as well as gain with, cut decision in favour of stacks of two-band elements. It
say a two-element quad having resonant feeders as in is true that coupling and spacing can be “optimum”
the DJ4VM design. only at one frequency, but from discussion in Chapter 5
The root of the matter is that with the log periodic the spacing needs only to be kept within the limits of
antenna one is paying for continuous coverage with no about 0-12-0-19X for two elements with a possible
gaps. This is a major facility which, even with 10, 18 increase to 0-25A for three elements; with driven opera-
and 24MHz taken into account, bears little relation to tion coupling errors can be adequately compensated
amateur needs; one rarely gets something for nothing over a sufficient range.
and in this case the price in terms of size and weight, It is important, in view of the earlier emphasis on
performance, or both is considerable. On the other making full use of the available aperture, to realize that ~
hand, the log-periodic antenna dispenses with tuning “stacking” does not mean that elements have to be cut
units and band-switching, thus making for maximum down to size, even though this happens to be the usual
operational convenience, and some amateurs able to practice. Elements such as Fig 7.8(d) can be arranged
face up to the mechanical engineering involved and free to fill the whole of the available space and the “‘capaci-
from planning constraints may feel the price worth tive stretching” makes them particularly suitable for
paying. Also, for those not too cramped for space, an stacking [9].
arrangement of three or four vertical log-periodic The linear resonator holds little promise in its pre-
arrays suspended from a single main support as sug- sent passive form for operation on more than three
gested on p194 has considerable attractions. bands but, subject to the development of servo-
The broadband log-periodic antennas discussed controlled capacitors, it is believed that it could enable
above are not to be confused with those intended for three-element beams to be tuned continuously over the
monoband operation and discussed in the next chapter. range 14-28MHz with possible extension (at slightly
Arrays using various combinations of long wires, reduced efficiency) to 1OMHz. Some difficulty would
(pp199-202), though in general optimized for a particu- arise in matching the driven element at 18MHz, where
lar frequency, are essentially broadband. They usually L is short-circuited by the capacitive channel, and at
provide high gain over at least an octave, though in the 10MHz where the circulating current in L, C will be
case of single wires provision of suitable earths presents large, resulting in a high impedance at the normal
something of a problem. The author believes however feedpoint. However, at worst, the driven element could
that it should be possible without affecting gain to erect support a separate thin-wire stretched dipole for
a very long wire as an inverted-L with one end brought 18MHz and the feeder could either be switched or
into a tuning unit in the shack. allowed to operate as a resonant line for 1OMHz. The
beam would be reversible at all frequencies but a
programming device would be needed to enable set-
tings, once determined, to be memorized and repro-
Six-band beams duced. There seems little doubt that given the demand
In conclusion it may be useful to summarize the above such a system could be developed for an economic price
discussion in the context of the more recent frequency but this would require considerable development effort
allocations, with a no-compromise beam covering all of a kind currently in short supply.
bands from 10-28MHz as the main objective. Short- There are however a number of possibilities involv-
listing the possibilities we have: ing the use of relays; for example two resonators in
parallel each with its own independently-operated relay
(a) Loops or dipoles with resonant feeders
would provide a total of four higher-frequency bands,
(b) Stacked loops or dipoles, matched feeders
ie five bands in all. It should normally be possible to
(c) Multiband elements
arrange that the resonator not in use on one of its
(d) Various combinations of (a), (b) and (c)
frequencies either assists or does not impede the action
Ignoring for the moment resonant-feeder systems, the of the other resonator. —
possible combinations are almost unlimited, but multi- The requirement can also be met by the use of
band elements (including trapped elements) are not resonant feeders provided these are short or appropri-
practicable for coverage of more than three bands. ate matching stubs switched into circuit as required.
Traps and log-periodic arrays are both ruled out by the Again, unless the specification is relaxed the use
“no compromise” clause, whereas other alternatives of three elements is essential because of spacing
 MULTIBAND ANTENNAS 117

requirements, though the employment of at least two

stacked beams is a possible alternative. With loops fed
at the bottom in the usual way it appears difficult with
a beam to achieve efficient coverage of 25MHz at
the same time as the other bands, but the DJ4VM
technique (though less neat) offers a possible solution.
Additional possibilities based on Fig 7.3 exist in the Fig 7.15. Lumped circuit equivalent of linear resonator
case of vertical loop arrays.

Conclusions and recommendations risk of tvi etc which may result from bringing a “‘hot”’
wire too close to the mains wiring.
Multiband operation is usually regarded as a comprom-
Loop elements (including folded dipoles) are much
ise and it is true that in some cases practical advantages
more suitable than ordinary dipoles for use with re-
can be realized in return for accepting a reduction in
sonant-feeder multiband systems, particularly if com-
bandwidth. However, most existing multiband systems
bined with some relays which can be used to open-
waste energy quite unnecessarily in the form of heat,
circuit the top of a loop at 23MHz or connect matching
besides failing to realize the improvement in gain which
stubs as required.
should result from the increased aperture at the higher
frequencies. There is felt to be no excuse for this, and it
is unfortunate that so many amateurs have no choice Appendix—the linear resonator
but to purchase commercial antennas which have
For all its apparent simplicity the linear resonator is a
hitherto, for the most part, followed the dictates of
complex device. Its main features are summarized in
fashion rather than the principles of sound engineering.
Fig 7.15, which shows the equivalent “lumped circuit”’,
The best choice of method usually involves a delicate
and Fig 7.16, which illustrates the frequency response
balance of conflicting factors. For the amateur con-
of typical resonators assuming conductors of equal
structor it is believed to lie in most cases between the
diameter and varying degrees of coupling between
use of resonant feeders, stacked elements or some form
-them. To assist interpretation Fig 7.17 gives a rough
of central tuning unit with band switching. Switching
idea of the way the coupling factor varies with spacing
may be active (eg using relays) or passive, such as Fig
of the conductors. For the more advanced amateur who
7.9(b). In general two-band operation involves fewer may wish to develop his own designs a number of basic
problems of design and adjustment, and passive switch- formulae will now be given. If r is the turns ratio, k the
ing is likely to be difficult for more than three bands.
coupling factor and y =wLC (where w = 2x X frequen-
Due to mutual interaction a similar restriction tends to
cy) we obtain
apply in the case of closely stacked dipoles; on the
other hand, switching by relays becomes rapidly less x= off eet
attractive as the number of “‘switch positions” increases wee 2k yo
due to the rapid escalation in the number of relays
Parallel resonance (X = ©) is given by
required. This adds up in the author’s case to prefer-
ence for the use of stacked two-band elements selected > 1
by one or two relays. If more than three bands are Yp ~ 2 (1 — k)
required there is felt to be no sensible alternative to
The series resonance (X = 0) is given by
this, except in cases where the advantages of open-wire
resonant lines may be felt to outweigh the restriction of 2 1
bandwidth or other inconveniences. ys oR (1 — k)
Horizontal antennas for the higher frequencies, plus The frequency at which J; = 0 is given by
feeders, can frequently be used as top-loaded verticals
for the lower frequencies, though for optimum cover- 1
age of shorter ranges a horizontal antenna is required; Yo = 72 = kr
this can be at a low height. The use of horizontal For resonators with equal-diameter conductors (r = 1)
low-frequency antennas, centre-fed, at the higher fre- the frequency ratios are
quencies is common practice but not generally recom-
- mended by the author in view of the narrow angular fo/fp = V2 (ie half an octave)
coverage which, in association with Murphy’s law,
ensures that wanted signals are always in an unfavour-
able direction. This is felt to outweigh any slight gain
hilfe =r)
there may be in the favoured directions. Long wires
may be end fed on all bands, apart from the increased fille =)
118 HF ANTENNAS FOR ALL LOCATIONS

Parallel resonances
coupling is exactly half an octave below the series
2:0
tt 4 resonance, so that the short-circuit at 21MHz would be
expected to result in an open-circuit at the lower
frequency. By increasing the coupling factor so as to
bring the resonances closer together the idea appeared
feasible, and was even tried with some success—
although interaction between the tuning of the series
circuit on 21MHz and the length of element required
for 14MHz was sufficient to make the adjustment
extremely difficult.
Unfortunately Fig 7.16 does not tell the whole story,
and it could be argued that bringing the resonances
together does not get rid of the problem but merely |
brushes it under the carpet. This is because series
resonance might be defined in two ways—the frequen- —
cy at which the reactance between terminals is zero,
and the frequency at which the series circuit can be
resonance, 7 regarded as an effective short-circuit so that no current
Sera ah
flows through the inductive path. Normally with
lumped circuits these frequencies would be identical,
but this does not apply when mutual coupling is pre-
sent. In this case the parallel resonance and the short-
Parallel resonance,
conventional circuit
circuit frequency retain their half-octave separation,
whereas at the zero-reactance frequency there is no
short-circuit and at 21MHz there could well be a large
and undesirable circulating current, causing appreci-
able losses and perhaps a reduction in bandwidth. This
depends on the dimensions of the ‘‘trombone’’, Fig 7.9,
Series
tH
resonances
with close spacing resulting in high current but con-
siderable enhancement of the “disappearing induct-
Fig 7.16. Reactance curves for linear resonators showing effect of ance” effect. Typically, this shift in the 14MHz
mutual coupling. Frequency scale is relative. X/wL is the ratio of
reactance between the terminals of the resonator to the natural
reactance of the conductor. Mutual coupling shifts the curves to
the right and brings the series and parallel resonances closer Spacing (inches)
together. Dotted curves show for comparison the characteristics
of conventional resonators

Current ratios are given by

I = I + f.

ijteall Cate ee
TI. y(1 — kr)
which reduces to —kr at series resonance. Practical
values for linear resonators are k = 0-2-0-5, r = 1-0-
i335;
In Fig 7.16 it will be seen that one effect of coupling (r?)
ratio
Inductance (k)
factor
Coupling
is to bring the series and parallel resonances closer
together, so that if, for example, this type of resonator
is used to generate the series capacitance needed for
bringing a 14MHz dipole to resonance on 21MHz, the
effect at lower frequencies is reduced. There are sever- 25 0-5 0:75 1°O
re diameter (inches)
al ways of using such a device to achieve multiband
resonances, and perhaps the most obvious would be to Fig 7.17. Variation of coupling factor with spacing, and induct-
use the series resonance for “shorting out” about one ance ratio with diameter of smaller conductor. Figures based on
third of a 14MHz dipole and thus bring it to resonance length of 7ft (2:1m) and large conductor diameter of 1in (2-5cm)
on 21MHz. However, this tends to create difficulties at but these are not critical. Values for k assume 1in diameter for
both conductors but decrease by only 11 per cent if one conduc-
14MHz since the parallel resonance in the absence of tor is reduced to 0-1in (2-5mm). Use scales indicated by arrows.
 MULTIBAND ANTENNAS 119

Mast may provide

resonance as a result of adding the 21MHz capacitance virtual‘earth Brea ee
varies from about 1 per cent to 7 per cent as the spacing
increases from lin (2-Scm) to 10in (25cm). However,
due to the increased circulating current, bandwidth
reduction and the onset of appreciable losses, spacings
less than Sin (12:5cm) are not recommended for ‘“‘main-
stream’”’ applications, though suitable for some pur-
poses such as those covered by Fig 7.6(b).
It is useful to be aware of what happens as the length ee
ee ee ee

of the “‘trombone” is varied. At 23MHz this is acting as

an insulator and the voltage across it tends to stay Gamma match

constant so that both rf current and losses are inversely

proportional to length. The current is about four times
the loop current of the dipole, so that for comparison
with the radiation resistance any losses must be multi-
plied by 16. For a 5ft (1:5m) trombone a typical Fig 7.18. Excitation of 1, mode by unbalanced feed. Centre of
calculated value of loss resistance is 0-158 which element is virtually isolated from earth so an incoming wave
could set up the current distribution shown dotted. Voltage can
translates into a negligible power loss of only 1-6 per therefore be developed across the feeder. By reciprocity this
cent. At 21MHz the situation is more complicated, but mode can also be radiated, but not with a balanced feed. Arrows
with the recommended dimensions the current /, indicate current flow and the feeder length may be any odd
number of quarter-waves. The linear resonator capacitor is not
through the capacitor divides into roughly equal pro- shown as it has no effect on this mode
portions between L and the inductance L, of the outer
portions of the dipole, each of these being somewhat
inductance of the series circuit; this is not subject to the
less than the current /p associated with the radiation
same restrictions, though it is important to ensure that
resistance.
As a rough guide, J, should be roughly equal to 2Jp, losses remain negligible in comparison with the radia-
and anything more than 3/p is definitely excessive. The tion resistance. Some increase of inductance is possible
ratio increases rapidly as the length is decreased, but as by decreasing the conductor diameter, but it is difficult
the length is increased it gradually falls to unity so that to achieve a ratio greater than about 1-5 without
all the current is flowing in the capacitor and none loading coils, and in general any diameter from about
0-lin (2:-Smm) to at least 0-5in (13mm) is electrically
through L, which can no longer be used as a matching
device. This of course is the “short-circuit” series acceptable, with mechanical and bandwidth considera-
tions tending to favour the higher value.
resonance and is somewhat lower than the true series
resonance as defined by “‘zero reactance’’. The gradual Spurious resonances
transition means that the usual distinction between Under certain “‘fault’’ conditions, or when adjustments
series and parallel resonance has become somewhat are badly out, it is possible to excite spurious modes of
blurred and there are matching complications but still radiation in elements using linear resonators. The most
the basis for a workable system; the 21MHz radiation obvious one is the full-wave resonance involving two
resistance being so large the capacitive arm of the half-waves in antiphase. DLIFK [4] warns against this
trombone can be routed directly via the balun terminals to the extent of ruling out the use of full-wave ele-
and the element becomes series-fed. However, the ments, but this mode cannot in fact be excited with a
low-frequency resonance has now been moved con- symmetrical feed system and is easily detected. It is
siderably lower, due partly to the larger trombone and predictable that it could be excited quite efficiently with
partly to the ‘‘capacitive end-loading” effect produced an unbalanced feed under certain extreme conditions
at 14MHz by moving the ends of the trombone out- (Fig 7.18). Due to normal tolerances its presence has
wards. This was found to reduce the required element been detected when using a balun, but only in a very
length by about 14 per cent. weak form, and for this purpose the resonator had to be
Referring to Fig 7.9, any increase of C beyond the badly detuned, since otherwise the unwanted mode was
value required for 21MHz will have an appreciable and completely swamped by the wanted one. The unwanted
eventually catastrophic effect at 14MHz, though under mode is associated with current flow in the boom, and if
proper control it can be used to convert a director into a both modes are present there will be unequal currents
reflector. On a reflector it has the opposite effect, in the two halves of the radiator. If the full-wave mode
turning it into a director. This may at first sight be is present alone, a pick-up loop moved along the
surprising but the explanation will be obvious from radiator indicates a null in the centre. It is likely that
careful inspection of Fig 7.16. under some conditions this mode could cause confusion
As well as by use of mutual coupling, the resonant when looking for resonances with a gdo.
frequencies may be brought closer by increasing the It is possible, when using a gdo or with sufficient
120 HF ANTENNAS FOR ALL LOCATIONS

mistuning, to excite a mode in which current is wholly the capacitor must of necessity be equal to the sum of
or mainly confined to the trombone section and the two inductive currents, provided the point of
there is little or no radiation from the antenna. This is observation is close to the junction of the conductors.
most likely to be experienced at 21MHz and a low value If a gdo is coupled in off-centre, many spurious
of swr is no guarantee that it is not occurring. Conclu- modes will be observed. This happens with any “‘plum-
sive evidence of its presence is provided by the follow- ber’s delight’? beam, the modes corresponding with the
ing symptoms: many different ways in which a wave can wrap itself
round the structure as discussed in Chapter 18.
(a) Equal currents in the inductive and capacitive
branches, little or none in the remainder of the
elements. References
(b) Inability to tune the parasitic elements or obtain
[1] ““A new multiband quad antenna’, W. Boldt,
useful field strength readings.
DJ4VM, Ham Radio August 1969.
(c) Smoke rising from the balun!
[2] ‘“‘A different multiband aerial system’’, H. F. Ruck-
If capacitance is being increased from a low value this is ert, VK2AOU, Amateur Radio April 1978.
the first mode encountered. The remedy is to increase [3] ‘“More about the Minibeam’’, G. A. Bird, G4ZU, -
the capacitance until a reasonable ratio of inward-to- RSGB Bulletin October 1957.
outward-flowing inductive current is achieved. This is [4] R. Auerbach, DLIFK, DL-QTC July 1960.
typically about 0-3 to 0-6. [5] The ARRL Antenna Anthology, p34.
Measurements of this kind need only be very rough. [6] The ARRL Antenna Book, pp208 and 161.
Fig 18.12 shows a simple “relative current” meter [7] ‘““‘The ‘disappearing inductance’—a new trick and
which may be calibrated against any other rf indicator some better beams’’, L. A. Moxon, Radio Com-
of known characteristics, such as an oscilloscope. Small munication April/May 1977.
size is important for avoiding capacitance pick-up, and [8 ‘High-frequency antennas and propagation modes
Med

the meter should be held at a constant distance such as in relation to the amateur service’, L. A. Moxon,
0-S5in or lem (which can be judged roughly by eye) IEE International Conference on Antennas and
from the conductors carrying the current to be esti- Propagation, 28-30 November 1978. See also ré-
mated. The distance should be slightly increased for sumé in Radio Communication May 1979.
conductors of smaller diameter. It will be found useful [9
=
‘““‘Loaded wire aerials’, F. C. Charman, G6CJ,
to remember that at 21MHz the current flowing into RSGB Bulletin July 1961.
CHAPTER 8

Bandwidth

The term “bandwidth” when applied to antenna power) a risk of voltage or current breakdown in the
systems can have a number of different meanings. For feeder system. Poor swr bandwidth is often accompa-
example, in commercial systems an antenna may be nied by poor directivity bandwidth which takes the
required to operate without re-adjustment over a band form of a drop in front-to-back ratio or gain, or both, as
of several octaves, perhaps even simultaneously on a the band-edges are approached. These effects are not
number of frequencies disposed at random. In such directly connected and may manifest themselves quite
cases the antenna can be accurately described as independently.
“broadband”, though it is certain to be inefficient
unless based on some massive structure or the use of
very long wires. In contrast, the amateur requires Directivity bandwidth
coverage only of a small number of narrow bands well A large front-to-back ratio is the result of a very
separated in frequency and, although in principle this accurate balancing out of signals from the rear direction
object can be achieved by broadband operation, in the and is therefore upset by even a small change in the
above sense this is usually neither feasible nor desir- relative phase or amplitude of the quantities being
able. balanced. Thus in the case of two elements it is readily
It is comparatively easy to design an antenna to shown that a phase error of 1/10 radian (5-7°) limits the
operate on several of these bands, this being the front-to-back ratio to a maximum of 20dB, regardless
process known as multibanding which formed the sub- of whether the initial adjustment achieved 30dB, 40dB,
ject of the last chapter. It needs to be distinguished or even infinite rejection. This drop however does not
from the broadbanding typical of commercial systems necessarily mean that the directivity is worse; it may
even though in some cases these may be alternative merely be different, as in the case of Fig 3.6(c) where
ways of achieving the same objectives. To be fully the null instead of disappearing has split into two in
“broadband” in the amateur context, an antenna only different directions. In the case of most current antenna
needs to operate satisfactorily over the whole of each of designs nothing can be done in the shack to correct a
the frequency bands for which it is intended without drop in performance at the band edges. This is because
requiring re-adjustment of the transmitter. This needs the directivity is obtained by means of elements which
to be achieved if possible without significant change in are parasitically excited or connected in some way
gain or directivity, and these conditions are fairly easy which likewise predetermines the phase and amplitude
to meet in the case of monoband dipoles or loops, long relationships. On the other hand, by using separate
wires, and some of the methods of multibanding discus- feeders to each element or the use of some mechanical
sed in Chapter 7. In the case of close-spaced beams, form of remote tuning, this limitation can be overcome
short dipoles and a number of multiband systems and arrangements will be described in a later chapter
compromise is unavoidable, and it is necessary to which enable an antenna to be adjusted from the shack
distinguish between swr bandwidth and directivity for peak performance on any given frequency.
bandwidth. In each of these cases it may also be
necessary to distinguish between the bandwidth for
satisfactory operation without retuning, and the band- Operating bandwidth
width over which acceptable performance can be By the use of resonant feeders as described in Chapter
achieved subject to adjustments capable of being car- 7, antennas can be made capable of operation over a
ried out by the operator. very wide band such as two or three octaves subject to
retuning by the operator, the price paid being very
narrow bandwidth in the absence of such re-adjustment.
The swr bandwidth A wide operating bandwidth may also be achieved by
This is a property of the antenna plus feeder system remote tuning, eg the use of servo-controlled capacitors
and, although a high swr can always be corrected so far as suggested in the last chapter.
as loading of the transmitter is concerned, it may be Log-periodic antennas, together with rhombics and
accompanied by appreciable extra losses and (with high (to a lesser extent) some monopoles or dipoles using
121
122 HF ANTENNAS FOR ALL LOCATIONS

R
(a) on p36. Applying this result to a )/2 dipole having a
radiation resistance of 73Q and Z, equal to 1,000Q,
Cc Bahdwianiier we have
L panel's |iS to
uae8h8
ZOf = tae eer
TET OO ges 0-093
a This is a bandwidth of 9-3 per cent and it is clear that in
4
ae R/2 | (b) the case of a simple dipole the bandwidth is proportion-
R 4 al directly to R and inversely to Z). The “3dB down”
Zo Bandwidth Sea
Zo Se
Ht
points correspond to an swr of 2:6 which may be found
excessive, but the figures thus obtained need only be
Rl2 multiplied by 0-75 to obtain the bandwidth for an swr of
Fig 8.1. Comparison of lines and tuned circuits. The two express- 2-0. Alternatively the reactance at the centre of a \/2
ions for bandwidth are comparable except that at (b) 72/4 has dipole near resonance may be obtained from Fig 8.2
replaced wl
which shows also the variation in resonant length with
diameter. Shorter lengths imply some reduction in R ©
multiple wires, can claim to be “broadband” in all three which is roughly proportional to L?.
of the respects listed above. HF receiving antennas
tend to be broadband in the operating sense, since as
explained in Chapter 9 “efficiency” is not important.

Estimation of swr bandwidth

In the case of antennas other than beams it is usually
not difficult to estimate swr bandwidth using a simple
tuned circuit, Fig 8.1(a), as a model. The bandwidth of
a tuned circuit between “3dB down” points is found
by equating the reactance
1
oL- cE

to the resistance R in which case

R = 2wL X (f/f)
where (Of/f) is the fractional frequency change. This
can also be expressed in the form:
Bandwidth = 26f = R/(aL).
Carrying out a similar operation for a resonant line or
antenna, Fig 8.1(b), the reactance (from p37) is equal
to Z)2nAl/A, and for a 4 line or 2 dipole
Al = (N4) x (f/f). The resistance is therefore equal
cael °

Reactance and end- bei

to the reactance when correction of half- >
wave dipoles

R=Z. Z . (8fif) = 2 . 2Z,(8f/f)

and the bandwidth 26f is given by
Bie
Lo. K Reactance (Ohms)
Fig 8.2. Tuning and reactance chart for )/2 dipoles. Resonant
Compared with the result for the tuned circuit, 27L (ie lengths are slightly shorter than the physical length and decrease
wL/f) has been replaced by Z,/f and, although there is with increasing diameter of conductor. For practical hf dipoles L/d
a small additional factor 4/m or 1-27, it is evident that varies between about 300 and 50,000 so that the length correc-
tion can vary between about 2:4 and 4:5 per cent with R (at
the wl of the tuned circuit has been more or less resonance) varying as the square of the resonant length. These
replaced by the Z, of the line. This result is to be corrections vary with height and other environmental factors and
expected in view of the equivalence between lines and are not applicable to beams or folded elements. Bandwidth is
given by the per cent detuning for a total reactance change equal
tuned circuits when used as transformers, as explained ‘to 2R
 BANDWIDTH 123

If a wire is bent in some complicated manner it is polarized endwise ‘““W8JK” mode is only 2dB down on
difficult to predict bandwidth exactly, but a rough idea a dipole.
is sufficient for most purposes and the lack of a formula These figures are by no means useless but, to get to
for Z) which takes shape fully into account is not the point, what has happened to the bandwidth? The
important, as will be appreciated by looking at Fig 4.13 resonant feeder is an unsuitable length, and to cover all
and visualizing the major changes which have to be bands it is necessary for matching to use an atu (which
made in order to alter Zy significantly. complicates any arithmetic) or “play around” with
There is no difficulty in making a rough estimate if feeder lengths and matching transformers whenever the
each section of wire is resolved into orthogonal compo- band is changed. If however the system length can be
nents as explained on p10 and illustrated in Fig 6.8(c), reduced from 7O0ft (21-3m) to S5Oft (15-2m) this will
which shows a right-angled delta loop fed at the improve performance at 10MHz and simplifies the
right-angle. example. Because the “‘line length”’ is 4/2 instead of 4/4
A practical application of this is shown in Fig the bandwidth for a given Z, is half that given by the
11.11(f), p158, where dimensions are given for a previous formula. However, because half the system
21MHz loop attached to a resonant feeder, and it is consists of 600@ line the average Z, should be only
interesting as an example of the method to consider about 800, ie
whether this might be used on the 10MHz band. To
simplify the examples it will be assumed that the Bandwidth = 26f = 2 Af
dimensions are adjusted slightly so that a_half-
wavelength fits exactly round the loop at 10-1MHz.
bol f
~ 4002
The long side of the triangle is 0-36 of the total
length. Referring to Fig 2.5 it can be seen that the or about 0-5 per cent of the frequency. This is 50kHz,
middle 36 per cent of the current distribution yields a which is equal to the width of the band, so that the
square count of 408 compared with the total of 760, so system is viable, if somewhat marginal. Restoration of
that if the sides of the triangle could be ignored the the length to 70ft can be expected to increase the swr on
radiation resistance would be (408/760)? x 73Q. the transmitter side of whatever matching device is
However, each side contributes 176 ‘‘squares” and each used to about 3 or 4 at the band edges, assuming correct
of the orthogonal components in Fig 6.8(c) is equiva- matching at band centre. This calculation is typical also
lent to 176 + V2 or 126 squares. Two of these represent of a 14MHz loop used at 7MHz.
current flow in the wrong direction so that (126 x 2) In the important case of short end-loaded elements
must be subtracted from the 408, which is more than a accurate calculation is difficult but R can be obtained
little unfortunate since there are only 156 left and the with the help of Fig 3.10 or 3.11 and, as a very rough
radiation resistance has come down with a big bump, ie approximation, a Z, of 1,000Q2 can be assumed. Thus
to (156/760)? x 73 or an insignificant 3-1Q. for a 10ft (3m) element at 14MHz R = 14 and
The vertical components have been ignored, but to assuming the same cross-section and total length of
make sure this is a reasonable assumption they can be wire a bandwidth reduction of 14/73 compared with a
considered as a ‘““W8JK pair” with an average separa- )/2 dipole can be expected; this comes to 250kHz which
tion of around 1/4, corresponding to a 90° phase is adequate for coverage of the phone band.
difference. The field they produce corresponds to Where the antenna or antenna-plus-feeder system
126 x V2 squares but, allowing 3dB gain for the W8JK, consists of sections of known Z, it is a simple matter to
the current needed to produce it is reduced by V2, work out, for example, swr at the band centre and band
leaving 126 squares as the reckonable contribution for edges by making use of the Smith chart as explained on
working out the radiation resistance. This yields p60. The necessary reactances can be obtained from the
another 2Q, so that the assumption was somewhat short-line formula Z)2z // as discussed on p37.
inaccurate. One further example concerns the use of traps and is
Clutching at this further straw we have a total of complementary to the investigation of trap losses in the
5-1Q and there are still some losses to be reckoned. last chapter, p109. In view of the difficulty in allowing
The total wire length, including the resonant feeder, is for ‘coupled circuit” effects the example relates to a
1-5) at 10MHz so that, referring to Fig 3.11 and single-wire dipole but a 14/21MHz element and a
dividing by two for the more or less sinusoidal current typical trap capacitance of 40pF are again assumed.
distribution, the loss resistance (assuming 14swg (2mm) Taking a value of 1,000@ for Z,, the impedance be-
copper wire) is found to be 1Q. tween the dipole ends at 21MHz comes to (1,000/
Though nothing to do with bandwidth, it is perhaps 73) X 1,000 = 13,700Q2. This appears across the two
interesting to note in passing that 3-1/6-1 of the total traps in series, these being equivalent to a single tuned
power goes into a horizontally polarized wave at right- circuit of 20pF, ie X ~ 400Q giving a Q of 13,700/
angles to the wire. For this mode, therefore, we have a 400 = 34 which corresponds to a bandwidth of 624kHz.
3dB loss with respect to a dipole, whereas the vertically This is adequate for band coverage at 21MHz but drops
124 HF ANTENNAS FOR ALL LOCATIONS

to only 171kHz for a radiation resistance of 200, (q@/q@p = 1-5), at which point the front-to-back ratio is —
corresponding to that of the beam but taking no still quite high. Alternatively one might, for example,
account of the coupling which improves matters. be interested in the bandwidth for better than 20dB
However, even if this is doubled it remains inadequate, front-to-back ratio for which, from Figs 5.4 and 5.6, the
and the answer probably is to reduce the trap capaci- allowable reactance change is 36 per cent of the value
tance. which would produce a phase change equal to @, ie
14 x 1-25 x 0-36 = 6:3Q. This translates into a band-
width of about 60kHz in good agreement with mea-
sured values.
This example follows usual practice in relating front/
Estimation of directivity bandwidth back ratio to a single bearing off the back instead of
The same methods may be applied to parasitic elements taking an average based on all back directions which
in order to obtain the directivity bandwidth, except that would be more meaningful but also a lot more difficult. —
in this case there is no recognized definition of band- In some cases what appears to be a worsening of the
width. For a three-element beam this could be based on front/back ratio is merely a shift in direction of the nulls.
keeping within the shaded areas of Fig 5.17, although The difficulty is accentuated if one tries to compare
the antenna will work with only slightly reduced per- beams with different numbers of elements. For exam-
formance well outside these limits. ple reference [23] of Chapter 5 presents large numbers
For two-element beams it will be noted from Figs 5.4 of curves showing the variation of front/back ratio with
and 5.6 that much larger phase shifts can be tolerated if frequency for various three- and four-element beams,
the main interest is reduction of interference rather and these show quite dramatically the advantage in
than maximum gain. The latter condition is much more terms of directivity bandwidth of operating in the lower
critical from the point of view of matching, and results right-hand corner of Fig 5.16.
in greatly increased interference from some directions What they fail to demonstrate is any significant
as well as a possible increase in average interference advantage compared with this author’s results for criti-
levels. cally-coupled pairs of elements. Being for one direction
A reasonable compromise for the band limits, refer- only, they fail to show improvements in average front/
ring to Fig 5.4, might be @, = @ and q, = q,/2, in back ratio as illustrated for example by Fig 5.15, and to
which case for \/8 spacing the reactance can be allowed obtain the full story one needs to plot the complete
to change from 36-5Q to 73Q. For a typical wire beam directivity patterns which has been done (though only
(Z) = 1,00092) we have for a single frequency) in reference [21] of Chapter 5.
Al/h. = 36-5/(1,000 X 2x) = 0-0058 Between them these two very valuable references
provide a fairly complete but rather complicated pic-
To convert this into a percentage change in the length ture, from which an attempt has been made to distill
of a 4 line there are multiplying factors of 4 and the aspects likely to be of most concern to the reader.
100, ie the bandwidth as a percentage is given by
0-0058 x 400 = 2:32. This amounts to 329kHz at
14-2MHz so that reasonable band coverage is possible,
though no account has been taken of possible ampli- Other factors affecting bandwidth
tude changes which could affect the depth of nulls. The swr bandwidth of a coupled pair of elements may
With tubing elements the bandwidth can be in- be greater than that of a single element, this being
creased by at least 50 per cent which could be important closely analogous to broadening the bandwidth of
if one has strong views as to the choice of working-point amplifiers by the use of coupled pairs of circuits. The
on Fig 5.4, particularly if this choice veers in the reduction in impedance of a primary circuit as a result
direction of maximum gain. However, an even more of detuning is offset due to a decrease in the impedance
important consideration may well be the need for transferred from the secondary, and this effect was
lightweight structures leading to wire beams with direc- most noticeable in experiments with small slightly
tional switching (pp179, 221) or compact rotary beams overcoupled beams, the swr bandwidth being much
such as those described on p172. greater than the pattern bandwidth.
For a 14MHz beam folded into a 10ft (3m) square The line-reactance formulae indicate that if a re-
(p177) the directivity bandwidth is reduced relative to sonant system such as an antenna or antenna-plus-
that for full-sized elements by the same factor 14/73 as feeder contains several half-wavelengths, the reactance
the swr bandwidth of the individual elements, but arising from any given departure from the resonant
compared with the previous example there is a 25 per frequency will be increased in proportion to the total
cent increase due to the wider spacing. This results in a length, assuming a constant value of Z). This results in
bandwidth of 77kHz which can be roughly doubled if a proportionate decrease in bandwidth as well as
the gain is allowed to drop by an additional 1dB increased losses due to the greater length of wire. Such
 BANDWIDTH 125

effects may or may not be important depending on the circuit effect’ which tends to improve matters as indi-
value of radiation resistance, the possibility or other- cated earlier.
wise of retuning from the shack, and the importance or
otherwise attached to rapid changes of operating fre-
quency. In the case of resonant-feeder systems a rough Discussion
estimate of bandwidth may be obtained by assuming an Having read this chapter, the reader will appreciate
average value of Z,) between 600 and 1,000Q using the that the term “bandwidth” can mean a lot of different
formulae already given, and dividing by the total num- things. It tends to be used rather loosely but the
ber of quarter-wavelengths measured from the furthest meaning can usually be inferred from the context and it
point on the antenna to the transmitter end of the would be awkward to have to define or qualify it on all
feeder. occasions. In general, better directivity bandwidth can
It is important to remember that bandwidth is a be traded for gain, particularly if there are a lot of
property of the antenna plus feeder system and could elements. SWR bandwidth, though often used as the
well involve several changes of Z). Where large trans- main criterion of “bandwidth’’, is of little intrinsic
formation ratios are involved it is usually best to importance which is a pity in view of the impressive
perform the operation in two or more stages, and if it is figures which can be achieved with overcoupled ele-
necessary to use two different feeder impedances the ments. It is true however that a poor swr bandwidth
respective lengths should be chosen to minimize any may have undesirable practical consequences, especial-
mismatches in accordance with advice on p39. ly in terms of convenience for the operator who likes to
It has been found that changes in the matching move around the band a lot. In some cases it may be
arrangements can have a considerable effect on swr necessary to reckon that an antenna is ‘“‘unusable”’
bandwidth. In one case a big improvement resulted in when swr exceeds some specified value; this defines the
going from a gamma match to a balanced system; edges of the usable band of frequencies and therefore
conversely it was found that a large ‘delta’, in which the “‘useful bandwidth’’.
matching was controlled by choice of wire diameter, Since directivity and gain are closely linked it has not
resulted in less than half the bandwidth obtained with a been thought necessary to define a separate ‘“‘gain”’
smaller series-tuned delta match which in effect con- bandwidth, but it may be noted that maximum gain
nected the feeder directly to points of the correct does not in general coincide with best directivity; at one
impedance. Unfortunately there seem to be no guide- edge of the useful band gain may be falling off rapidly,
lines available in regard to this aspect of matching, and although the directivity remains good, whereas at the
the author can do no more than invite the reader to other edge directivity tends to become very poor before
share his own misgivings about gamma matches! the gain has dropped significantly. In this case there can
In the case of a beam the radiation resistance cannot be only a subjective definition of bandwidth depending
be related directly to bandwidth, owing to the ‘“‘coupled on what the individual user regards as acceptable.
 CHAPTER 9

Antenna design for reception

Capture area chapter stacking distance was linked with the directivity
As might be expected, a good transmitting antenna can of the separate beam units and not their size. These are
be expected to do well as a receiving antenna and, in merely different ways of looking at the same problem.
terms of signal power delivered to a matched receiver, Capture area is formally defined by the expression
there is full reciprocity between the transmitting and
receiving properties of an antenna. This may be a good
Ao An
G5
point at which to explain the term antenna aperture
since this will be encountered quite frequently in the where G, is the power gain (as an arithmetical ratio)
literature. It is easy to fall into the trap of thinking relative to an isotropic antenna. For a dipole this
“there is no substitute for it’’, a frequent claim which becomes
distracts attention from the possibilities of “‘supergain”
(subtractive gain) arrays, to which class most amateur A = 1-64
=
4n
2 = 0-130
hf beam antennas belong as explained in Chapter 5.
A large antenna system can be loosely regarded as The shape of the capture area can be deduced from
intercepting most of the signal energy passing through inspection of Fig 5.2, bearing in mind that apertures
the space occupied; this space is defined as the aperture can be regarded as “just touching’? only when the
or capture area, and doubling it can be expected to mutual resistance is small. For the collinear pairs in
double the amount of signal energy collected. This ties Chapter 6 it was found that the mutual resistance falls
in with the fact that in the case of large beams, as to zero at a spacing between centres of 0-71), the
already explained, doubling the size doubles the trans- corresponding separation for the broadside pair being
mitting gain, thereby endorsing the principle of recip- 0-44 so that pairs of elements having these spacings
rocity. will provide exactly 3dB gain. The approximate ratio of
As atray size is reduced, eventually a point is reached two-thirds for equivalent couplings is applicable to
where the additive gain principle on which large arrays closer spacings as well. Assuming an elliptical shape, it
depend is no longer operative, because if fields are follows that the capture area as defined by the formula
additive for one direction they must be additive for all above has a major axis of 0-51) along the line of the
directions, the dimension being too small to generate wire and 0-34) at right-angles to it as shown in Fig 9.1.
significant phase differences. The antenna system is The corresponding spacings provide gains of 2-:2dB
now small enough to allow use of the subtractive only, so that the capture areas as defined above are
principle as explained in Chapter 3, so that as size is somewhat imprecise and to some extent overlapping.
further reduced “directivity gain’ stays constant but Nevertheless the ‘‘just touching”’ rule tends to provide
the bandwidth shrinks instead. The antenna now has a
high Q and, just as tuned circuits couple together more
tightly if the Q is increased, so the smaller the antenna
the more tightly it couples into the surrounding space. 0°34, (Ws)
If it is desired to retain the concept of aperture this is
. still possible provided a short dipole, for example, is
regarded as effectively occupying the same space as a
2 dipole. This is important to remember in the
context of stacking, since if the dipoles of individual
beam-units of a large array are replaced by smaller
units the separation required between them remains
unchanged. The apertures of the individual units Fig 9.1. Capture area of dipole; this is most easily conceived as a
rectangle but is in fact elliptical, being given more accurately by
should be not quite touching so that the respective the area WW, of the rectangle multiplied by 1/4. Equating it also
collecting areas are completely separate and not par- to 0-132 and making use of the two-thirds shape factor as
tially shared. It will be recalled also that in an earlier explained in the text enables WW, to be determined

126
 ANTENNA DESIGN FOR RECEPTION 127

a reasonably good compromise between gain and over- for transmitting and receiving. It can however be
all size. misleading if an antenna is being designed for receiving
In most treatments of antenna gain, this book being only; it can happen that given two equally good anten-
no exception, there is an implicit assumption that the nas for transmission one may be better for reception.
transmitted (or received) energy is concentrated within This is because in reception one is not concerned with
a “beamwidth’’, within which there is no discrimina- signal strength alone but with the ratio of signals to
tion, but which embraces the whole of the transmitted noise or interference. HF receivers usually have, and
energy* apart perhaps from that contained in a few can always (unless the antenna is very small) be de-
small “‘sidelobes’’. This simplifies explanations but falls signed to have, much lower noise levels than the
short of being an accurate description of real-life external noise normally arriving from outer space (cos-
situations. Fortunately, for the most part, no great mic noise), even if atmospheric noise which is common-
precision is needed and the definition of beamwidth as ly regarded as the major source is ignored. Any losses
the “width between half-power points” usually leads to in the antenna system therefore decrease the signal and
acceptable answers. the noise equally, leaving reception unaffected, thus
The above figures for gain versus separation distance allowing full benefit to be obtained from whatever
may also be obtained from the array factors plotted in directivity is available.
Fig 3.9, thus demonstrating the identity between re- Fig 9.2 shows the ratio of external noise levels to the
ceiving and transmitting gain expressed as a power internal noise level of a low-noise (3dB noise figure)
ratio. receiver, the allowable reduction in antenna efficiency
In putting this into figures it is easiest to assume a for less than 1dB drop in signal-to-noise ratio being
rectangular capture area A, in which case (Fig 9.1) only 6dB less than the noise levels given by the curves.
This demonstrates the possibility in the case of “‘receiv-
A,
= =~ 2 = WW, = we er only” applications of exploiting superdirectivity to a
OQ much greater extent, any bandwidth problems being
where G is the gain over an isotropic antenna (or 1-64 overcome by remote tuning using variable-capacitance
times the gain over a dipole) and a.,/a, is the ratio of the diodes or the use of short “‘active’ antennas as de-
beam angles. It will be seen that if the capture area is scribed later in this chapter.
held constant the gain increases (beamwidth decreases) Antennas with strong high-angle lobes can be ex-
as 1/A, ie as the square of the frequency. pected to pick up more cosmic noise since the iono-
In view of conflicting figures elsewhere in the litera- sphere is transparent at high angles—this is consistent
ture some readers may be a bit confused at this point, with the author’s own experience in a ‘“‘quiet”’ location
but any difficulty can be resolved by realizing that the of lower noise levels when using vertical antennas.
ratio of the axes of the ellipse is also the ratio of the Unfortunately vertical antennas are usually worse in
radiation-pattern widths in the E and H planes. A long regard to pick-up of man-made noise, and there must
thin ellipse means the beam angle is much narrower in be many cases where the ability to switch polarization
one plane than the other, and it is impossible to have a would be useful for reception. Using horizontal anten-
narrow angle without gain which must in fact be at least nas, maximum discrimination against noise arriving at
equal to the ratio of the axes. It is no coincidence that vertical incidence occurs for heights which are a multi-
the ratio in Fig 9.1 is almost the same as the gain of a ple of 4/2, and in theory there might for reception be
dipole over an isotropic radiator. A ratio of three, for some advantage in using these heights. However, at the
example, would involve a gain of three and could not higher frequencies the ionosphere will usually be trans-
be achieved with a single dipole. Because of this the parent to cosmic noise over a range of angles extending
apertures of the dipoles comprising the top and bottom downwards well into the vertical patterns of most
parts of a quad loop are mainly overlapped, and the antennas.
gain is only 1dB compared with a dipole. Additional With long-wire antennas and some large arrays the
elements increase the apertures so that there is more situation is rather different since high-angle lobes are
overlap and the stacking gain is even less. greatly reduced. Terminated long-wire antennas such
as rhombics can be expected to have lower noise levels
since the resistance (ideally) absorbs half the noise
Signal-to-noise ratio power. It also of course absorbs half the transmitter
The principle of reciprocity is useful up to a point since power, but this is energy which would otherwise be
in most cases it is a mistake to use different antennas radiated in the back direction and leaves the forward
signal unaffected.
* There is a close analogy between beamwidth of antennas and A reduction in noise level, particularly if it is
bandwidth or receivers. In the case of receivers “energy bandwidth”
tends to be nearly equal to “3dB down bandwidth”, a single tuned
achieved by the use of low-efficiency superdirective
circuit being the main exception. Similarly with antennas the approx- antennas, will demand greater receiver sensitivity. This
imation is least accurate in the case of a single element. could involve a clash of interests since some receiver
128 HF ANTENNAS FOR ALL LOCATIONS

100

80

o oO
Fig 9.2. Variation of external noise level with frequen-
cy, indicating the cosmic noise level which provides a
realistic basis for the design of receivers and antenna
systems from about 10 or 14MHz upwards. The man-
40 made level, though typical of some suburban loca-
tions, varies between wide limits and the atmospheric
noise levels, though typical of night-time extremes,
are also very variable
(dB)
temperature
room
20

thermal
at
noise
to
relative
level
Noise

1°75 35 7 14. 28 56 112

Frequency (MHz)

designers have wisely taken advantage of the high antennas, the ability to reduce sidelobes is limited by
external noise levels to combat the overloading effects the fact that the number of possible nulls (which is
of strong signals. The measures adopted, such as the loosely related to the number of elements) is quite
omission in some cases of rf stages, degrade the sensi- small. Increasing the depth of a null does not affect its
tivity so that there may be no reserve for exploiting the width, and for benefit to be obtained from what is in
virtues of some types of low-noise antennas. In the case effect better balancing out of an unwanted signal, this
of lossy antennas strong signals are of course less of a has to fit into a relatively narrow slot in the pattern.
problem and there could be some value in providing a Given this situation it may be possible to rotate the
wider range of front-end options in receivers. This beam so that the interference drops into the slot. This
would also allow advantage to be taken of abnormally may however reduce the wanted signal and it is no help
low noise levels such as can occur in the early stages of if there is more than one signal or, as is often the case
recovery from an sid (sudden ionospheric disturbance). (particularly with short skip distances), more than one
angle of arrival.
Using a computer to control phases and amplitudes,
Interfering signals there are possibilities in the case of n elements of
In most of the examples in this book the emphasis is on nulling out n — 1 signals, though such a system would
transmitting gain, and usually if, say, 1dB improvement normally be beyond amateur resources. However, in a
is obtained there is no need to worry about where this later chapter the use is suggested of multiple-feeder
comes from, ie whether it is due to less back, side or systems whereby the antenna directivity and gain is
high-angle radiation. Indeed if noise and interference under the control of the operator in the shack, and in
were to arrive uniformly from all directions then the this case it should not be difficult to arrange relays so
performance in reception, in terms of mean ratio of that a different phasing network such as that in Fig 5.13
wanted-to-unwanted signal or noise power, would by is brought into use for reception. A single source of
definition be identical with the transmitting gain. This interference can then be nulled out without too much
point is usually overlooked, greater importance being difficulty provided it is not in the same direction as the
attached to front/back ratio defined for an arbitrary wanted signal.
angle (the reciprocal of the beam heading). As will be The sidelobe problem is an important one for desig-
seen, however, it usually makes more sense, even from ners of large arrays, particularly for reception since
the point of view of interference reduction, to aim for often sidelobes can be greatly reduced by sacrificing a
somewhat greater gain. certain amount of gain, but does not greatly concern
All antennas radiate some energy in nearly all direc- the average amateur operating on the hf bands.
tions and likewise receive from these directions. The Amateurs will of course be anxious not only to
radiation pattern may contain in addition to the wanted reduce interference in reception but also to cause as
“main lobe” a number of smaller ones known as little interference as possible, so that if a separate
sidelobes (p95) which may account for most of the receiving network is used as suggested above, it must
unwanted signals received. With conventional low-gain be switched out when checking whether or not a
 ANTENNA DESIGN FOR RECEPTION 129

15
signal strength for the back directions in the case of
three elements being clearly demonstrated. The for-
ward gains are 4-6dB and 4-2dB respectively, so that in
terms of transmitting gain the larger beam is slightly
down even compared with the best that can be achieved
with two elements, and it is unlikely that much im-
provement in this respect can be affected without loss
of front/back ratio.
It will be recalled that the ‘‘best possible” (for gain)
four-element design (p89) is quite poor in this respect
Percentage
of
forward
field
so that there seems to be here a particularly strong case
90° 135)
Degrees relative to beam heading ——>
180° for the use of separate tuning units for transmission and
reception. With two elements a deep null can be
Fig 9.3. Comparison of two- and three-element arrays for rejec- created in any given direction, but this is in no way
tion of signals from the back direction, currents being adjusted to
give a null at 180° in both cases and the gains being 4:2dB and equivalent to the kind of rejection demonstrated in Fig
4-6dB respectively. The dotted lines indicate “average” levels 9.3 since it has to be carried out afresh for each case of
interference.
channel is already occupied. Similarly if different Nulling is seen nevertheless as a very useful addition-
antennas are used for transmission and reception it is al facility, equally applicable to two or three elements
important to switch over to the transmitting antenna for and capable of improving discrimination by a further
channel checking since a null in one pattern may well order of magnitude. In the process it takes care of
coincide with a sidelobe of the other. reflections from other antennas and structures, which
Another point to be noted is the possibility of can otherwise seriously compromise the interference-
interference by one antenna, or even overhead wires, rejecting properties of both two- and three-element
with the pattern of another antenna. An antenna can of beams. Whether the three-element performance indi-
course act as a reflector of interfering signals, rather in cated by Fig 9.3 is achieved in practice will depend on
the manner of a radar reflector, and it is important to whether the mutual couplings happen to be such as to
realize that from this point of view its radiation pattern give the required current ratios, and this will certainly
is not directly related to the normal pattern which is not happen in the case of a quad because as already
derived in terms of voltages delivered to the feeder explained (p80) the current in a director is very small.
system. The reflector of a beam, for example, will in This does not necessarily exclude the possibility of
general respond fully to signals coming from either finding some other combination of amplitudes and
direction and re-radiate them in both directions. A phases which is equally favourable, but it does imply
degree of coupling between two antennas sufficiently the need for a computer and probably a degree in
low not to affect the transmitting gain of the one in use advanced mathematics in order to be sure of arriving at
may nevertheless result in an unacceptable level of an optimum result!
interference due to re-radiation of signals from the At this point it needs to be explained that the
back direction. performance indicated by Fig 9.3, good though it is, is
certainly not optimum. In both cases gain can be
increased and the null in the —180° direction then splits
Advantages of three or more elements into two so that one might expect the overall perform-
It has been shown in earlier chapters, especially on ance to be improved. As pointed out earlier (p128), in
p88, that, in practice, increasing the number of the case of omnidirectional interference, eg noise or
elements from two to three or four with close spacing large numbers of signals arriving from all angles with
has comparatively little effect on transmitting gain. In equal probability, maximum gain coincides by defini-
contrast to this, it might be conjectured that quite a tion with minimum interference but this is based on
small current flowing in a third element would suffice energy considerations and is no more realistic than the
for cancelling out the residual response of two elements usual practice of considering only a single arrival angle.
in the back direction, which is likely to be at least 10dB In practice, with maximum gain the nulls have moved
down on the main lobe. round towards the end-on directions for which there is
Calculations are difficult and laborious for more than already considerable rejection.
two elements except in a few simple cases such as the A more useful measure of discrimination against
example worked out in Chapter 5, p79, which it is interference is the number of degrees in the back
useful to recall at this point, noting in particular Fig direction over which the signal strength exceeds some
-3.15 which has been reproduced as Fig 9.3. This given percentage of the forward field strength. This can
compares backward radiation for two-and_three- be interpreted in terms of the probability of a given
element arrangements, the big reduction in average level of interference as has been done for the
130 HF ANTENNAS FOR ALL LOCATIONS

Wanted]unwanted
signal ratio

oe
[unwanted
(dB)
equalled
signal
time
the
of
25%
for
exceeded
or
fel
eS |
tow
Paine eee Nese cae
a
= (degrees)
beam
the
to
nulls
of
Direction
relative
heading
we

or 0-5
wanted/unwanted
(dB)
Ratio
equalled
of
25%
for
time
signal
exceeded
the 0-6 0-7 0-8 0-9 1-0
— a)
Ratio
wanted
of Rates of Spaeth signal voltages tier cent) assuming cr. field strengths (b) Ratio phase angle| spacing (4/2)

Fig 9.4. Discrimination of two-element beams against interfering signals, in terms of the probability of given ratios of unwanted to
wanted signal voltages. Interference is assumed to be coming in on the back of the beams from a single continuous source which moves
around, so that all directions over a 180° arc are equally likely, the unwanted signal being equal in field strength to the wanted signal.
Note that minimum interference requires ¢/qg = 0:78, corresponding to a gain of 4-6dB (as marked on the curve) and, from (b), the
placing of nulls in directions of 140° relative to the beam heading. The solid curve in (b) is the effective front/back ratio based on the
stated interference probability. It shows an improvement of 3-5dB, coupled with 0-5dB increase of gain, compared with the condition
© = Po implicit in the usual procedure of tuning for “maximum front/back ratio”

two-element case in Fig 9.4 based on Fig 5.6. It will be with the poor strong-signal performance of most re-
evident that minimum interference requires the beam ceivers. A monoband beam is likely to be better in this
to be tuned for nulls at 140° and 220°, not the usual 180° respect and remote tuning (p174) can also help.
which results in slightly more interference and 0-5dB
less gain.
On the other hand, tuning for maximum gain results Nulling of interference
in much greater likelihood of interference, a substantial The nulling out of an interfering signal is a critical
drop in radiation resistance which could have consequ- balancing operation involving both phase and ampli-
ences such as restriction of bandwidth, some operation- tude as shown schematically in Fig 9.5, being closely
al inconvenience and perhaps (after taking due account analogous to the use of an ac bridge. It requires two
of side effects) a net reduction in gain, all for a knobs and a sensitive indicator, the human ear being
theoretical increase in gain of only another 0-5dB. If more or less adequate provided the agc is not operat-
therefore the same adjustment has to serve for both ing. A good S-meter is preferable, and in this case also
reception and transmission, the 140° null condition is the agc should be disconnected or the rf gain turned
easily the best compromise, and this applies equally to
a quad, bi-square, or two-element dipole array regard-
less of whether elements are driven or parasitic. Possi- Phase
ble exceptions include extreme cases of miniaturization
and other “‘lossy”’ situations (eg some indoor antennas)
Signal
for which a rather larger value of phase shift may be Variable Variable (1)
beneficial as discussed on p86. Similar considera- gain phase
Subtract|
(amp] atten) control Add
tions apply in principle to three- and four-element
arrays, though actual figures are not available.

To receiver
Out-of-band signals
Fig 9.5. Use of “two knobs” for balancing out a given signal from
A point that needs to be watched is the risk of two sources subject to phase and amplitude differences. The high
interference from strong out-of-band signals, due to the noise levels prevailing in the hf band allow the use, if necessary,
of considerable attenuation without degradation of signal/noise
narrow pattern bandwidth of most beams combined ratio
 ANTENNA DESIGN FOR RECEPTION 131

Any dipole- this one may have been bent to fit into a roof space
of short-skip interference from Europe. By using two
separate antennas a useful degree of discrimination
should however be possible, assuming bearing separa-
tions in excess of some 30-5S0°, depending on beam-
widths. Ideally for this application the antennas should
Signal (2) be beams separated by about 40—SOft (12-15m); the
attractions of such an arrangement as a means of
Short active antenna obtaining increased gain and/or directional switching
Broad band amplifier where discussed in Chapter 6 and one antenna rotated
so that it has a null in the direction of the wanted signal
could be used for the nulling of interference from other
directions.
Signal (1) for Fig 9.5 Nulling is difficult and may be useless when more
Fig 9.6. Equivalent, for reception only, of a two-element beam. than one interfering signal is involved but it follows
This is suggested for use where space is insufficient for a normal from Fig 9.3 that the likelihood of two or more
beam. Reference to Fig 9.2 indicates the number of decibels of interfering signals being experienced together or in
allowable losses in the second signal path, including any losses in
the networks of Fig 9.5. The broadband amplifier will need quick succession is greatly reduced if the starting point
protection during transmission but, depending on the space is a three-element rather than a two-element beam. In
available and the operating frequency, may not be needed in all this case the attractiveness of nulling as a technique is
cases
considerably enhanced but, even if the advantages are
not fully realized with only two elements, the possibili-
well back. With a little practice, by simultaneous ties of (a) separate optimization of beam patterns for
two-handed operation of phase and amplitude knobs a reception and transmission, and (b) creating the effect
null may be quickly obtained on a single source of of a beam for reception when it is not possible to erect a
interference, provided the arrival angle is constant or beam for transmission should not be overlooked. This
the two antennas respond in the same way to changes in is a relatively new area which offers much interesting
arrival angle, a condition most likely to be secured if scope for the experimenter.
the two antennas are close together and have similar
radiation patterns.
In a system recently developed [1] the normal con- Conclusion
nection to a beam antenna provides one signal and the Interference levels in the hf bands are likely to in-
other is obtained by means of a separate connection to crease. It will be necessary to develop optimum
the reflector. In principle the idea is applicable to any methods for reducing average levels of interference as
type of antenna subject to availability of the nulling well as nulling out individual sources of interference.
signal, which does not require an efficient pick-up There is enormous scope for new developments and in
device since “receive only” conditions apply (p127). this chapter it has been possible only to suggest a few
However, low efficiency may have to be made up by tentative guidelines. Nevertheless, the desirability of
additional gain in the nulling channel; thus a small separaté feeders run back into the shack from each
active “‘probe”’ associated with a dipole as in Fig 9.6 can element of a beam for the improvement of transmitting
provide, for reception only, the exact equivalent of a performance has been featured in other chapters, and is
two-element beam plus the advantages of (a) instan- a starting point for the development of systems fully
taneous beam reversal and (b) greater flexibility, so and separately optimized for transmission and recep-
that the direction of a null can be altered in a few tion.
moments.
Nulling is of course not possible if the unwanted
signal is in the same direction as the wanted signal, and Reference
for a British station maximum benefit is likely to accrue [1] ““Technical Topics”, Radio Communication May
when trying to work to the west from the UK at times 1980, p489.
CHAPTER 10

The antenna and Its environment

It is often remarked that “‘it’s the antenna that counts’’, obtainable. Let us therefore start with the simplest
though from the point of view of dx performance the case, that of a flat unobstructed site.
author’s first priority would be to have a good location,
preferably as illustrated in Fig 10.1. Matching the
antenna system to this environment, half-a-dozen sim- Flat ground with no obstructions
ple inverted-V dipoles, possibly with reflectors but each The average amateur may not recognize this as relevant
requiring nothing more than a clothes-prop for support, to his own circumstances. What about, for example, all
could be expected to provide unbeatable performance the telephone wires and television antennas, to say
in any direction at the flick of a switch. Unfortunately, nothing of the house itself, its neighbours and the tree
unless one is able to acquire a castle in Spain or is at the end of the garden? It would appear in fact that
“operating portable’’, nothing much can be done about these items can usually be disregarded though there are
this and usually, far from boosting the signal, ground important exceptions and reservations as discussed
reflections are the main obstacle to good performance. later. The first point to note is that in most cases a wave
At a more practical level we are faced by the fact that leaving a horizontal antenna and heading upwards
some locations are good, others bad and, although in towards the ionosphere with a typical take-off angle of
many cases the reasons are obvious, there are others a few degrees will clear all or most of any surrounding
that appear to defy rational explanation. Often conclu- obstructions, and is ““unaware”’ of the ground, except
sions are reached from faulty reasoning which usually indirectly to the extent that the radiation resistance
by implication invests soil conductivity or some freak of (hence the antenna current and the corresponding field
antenna construction with magical properties. Though strength) varies slightly with height (p103). The ground
much is known, there are still large areas of ignorance for its part, is a virtually perfect reflector at low angles of
as well as grey areas which invite speculation. The incidence and this is true for homogeneous flat ground,
author proposes to indulge freely in these, hoping that whether this be desert sand, agricultural land, lakes or
proofs will be forthcoming in due course; in the mean- seawater.
time there are many cases where even on the basis of In the case of horizontal polarization and low angles,
existing knowledge better performance should be therefore, the situation is very simple as already discus-
sed in Chapter 6. The direct wave and the reflected
Shack wave are of equal amplitude, and combine in phase to
12dB gain at 1° elevation give a gain of 6dB relative to the free-space condition
ee
when the angle of radiation 6 (in radians) is equal to
M4h. At 14MHz for a typical radiation angle of 6° this
corresponds to a height of 170ft (52m), the gain being
Ring of six multiband reduced to 3dB for a height of 85ft (26m); below this
\
\ inverted-V dipoles at height low-angle field strengths can be reckoned as more or less
Reflection Se of 10 to 25ft (3 to 8m)
directly proportional to height. It makes virtually no
<---
difference whether the ground is electrically good, bad or
indifferent, as may be judged from the reflection
coefficients listed in Table 10.1. There is some uncertain-
ty as to the true effective height since this is influenced by
Reflection
the depth of penetration of the reflected wave into the
ground and is likely to be greater in the case of poor soil.
However, the author has found in his own case (heavy
clay soil) that the linear relationship between height and
signal strength is more or less valid on the basis of the
Fig 10.1. Artist’s impression of the ideal location. This provides
high effective gain at low angles due to reflections from the actual physical height, and others have reported similar
foreground and the far distance. Polarization must be horizontal findings.
132
 THE ANTENNA AND ITS ENVIRONMENT 133

30° 0-65,
180170 0-26,
0-5, 176 0-05,
180

10° 180
0-2,90
0-1, 10
0-25, 0
0-45,

16
3, coefficient
Reflection

6° 0: 0
0:6, eflection

[1,
references
in
2] coefficient

iven 3
0-1,0-5,
90 7
0-38,5 (degrees)
reflected
of
lag
Phase
wave

(0) 5 10 15 20 25 30
Angle of radiation (degrees)

Reflection
coefficient
angle
phase
and(degrees)
for 170
175
0-95,
0-85,
160
0-75,
150
0-5,
0-4,80
polarization
vertical
at stated
angles
12 4
0-5,
0-7,1:5
0-89,0 0
0-8,
0:92,0 Fig 10.2. Typical variation in magnitude and phase of the reflec-
tion coefficient above and below the Brewster angle. At vertical
incidence the reflection coefficient approximates to the values
given for horizontal polarization in the 90° column of Table 10.1
90° 0-97 0-55 0-25

One other factor to be taken into account in the case

10° 0:97 0-90 0-80
of single-element antennas is the variation of radiation
resistance with height due to interaction between the
6° 0-98 0-94 0-88 antenna and its image; the effect of this on field
strength was illustrated in Fig 6.18.
3° 0:97 0-93
Vertical polarization
at for 0-98 0-997
0-998
0-999
1:000
polarization
horizontal
angles
coefficient
Reflection
stated 0-94 0-99
0-968
0-987
0-995
0-99 0-97 In the case of vertical polarization and perfectly con-
the
Based
3-30MHz.
data
gon
range
1°

ducting ground, the antenna and its image are in phase

so that maximum radiation takes place at zero angle.
in
ies Over real ground the position is very different, as
illustrated by Fig 10.2. This shows how with increasing
angle above the horizon the reflection coefficient drops
(approx
metres, 0:14-0:04
penetration 16 90 rapidly to reach a low value at the “‘pseudo-Brewster
angle’, thereafter increasing again until at high angles
1
17 28
it becomes equal in magnitude to the coefficient for
(approx
angle3-30MHz)
degrees)
[12] horizontal polarization. Meanwhile the phase angle
starting at —180° as for the horizontal case also de-
creases, being —90° at the Brewster angle and there-
after dropping rapidly to a low value. In other words
phase reversal takes place around the Brewster angle,
[12] 4-5
(mho/m) 10-3
1072
to1073
5to
x AO
2%1074
5to
x 1073
5x 107°
5to
x so that at high angles the antenna and image tend to be
in phase; this is in fact the situation for al/ angles over
perfectly conducting ground.
3
This phase reversal means that there would be 6dB
Brewster
of
Depth
Conductivity,
Dielectric
constant 80
typical 84 15 gain at very low angles but for (a) the imperfections of
the ground and (b) mutual impedance, which in the
ideal case would reduce the antenna current by 3dB, an
effect which also shows itself as a broadening of the
radiation pattern in the vertical plane: Fig 10.3(a). Fig
10.3 compares typical radiation patterns for horizontal
and vertical antennas while Fig 10.4 shows how in the
reflection
and
of
penetration
depth
angles,
Brewster
conductivity,
dielectric
values
Typical
Table
10.1.
constant,
wave
frequenc
ground
of
various
for
coefficients
at
types
of ground
Type Sea
waterconductivity)
(best
Fresh
waterground
moist
Very
30 Average
ground (lowest
of range)
end
ground
dry
Very conductivity case of vertical polarization the patterns are modified
by the presence of the ground.
Inspection of Table 10.1 shows that for the majority
134 HF ANTENNAS FOR ALL LOCATIONS

Ls
2
h
iN

(e)
Fig 10.3. Typical radiation patterns in the vertical plane. (a, b) Vertical antennas over perfect earth. (c, d) Horizontal dipoles and collinear
arrays. (e) Broadside horizontal arrays. (f) End-fire horizontal systems. Only half of each pattern is shown, the antennas being
symmetrical about the vertical axis unless reflectors are used. Diagrams (a) and (b) hold for all azimuth directions, the remainder only for
the direction of maximum radiation, ie broadside. Beamwidth in azimuth depends on the length of the arrays

of practical cases the “‘pseudo-Brewster angle’”’ varies ground” which makes for maximum convenience and
from about 1° for sea water to 28° for very dry soil and, ease of erection. The mean height of A/4 is a reasonable
because the magnitude of the reflection coefficient is approximation for most antennas erected to this
less than one in all cases, it can be deduced that there is
some residual radiation even at very low angles and Perfect Above At At low
earth 0B $B angles
over the worst ground. This is shown in Figs 10.5—10.7
which compare vertical and horizontal antennas at Antenna t
14MHz for various types of ground and radiation angles
between 1 and 10°, demonstrating that height depend-
ence is much less in the case of vertical antennas,
particularly at the relatively low heights typical of most
amateur antenna systems. Though calculated for
14MHz these curves are valid as a rough approximation
for other frequencies in the hf band, and it will be seen
that vertical antennas are much superior at low heights.
Moreover low-height vertical beams such as those
described in Chapter 13 are very much cheaper and
easier to erect than the usual horizontal rotary beam,
which at 14MHz needs to be at a considerable height to Brewster
justify the extra cost, though it has to be remembered angle %B
that the site under discussion is flat and unobstructed. ROUND YY
The lower curve in Fig 10.8 shows the break-even Vp BHP OS. 4447 YY

height for vertical and horizontal polarization for aver- Fig 10.4. Vertical-plane radiation pattern for a short vertical
monopole. The dotted curve shows the effect of imperfect
age ground and a wide range of elevation angles, the ground. Simplified vector diagrams at inset show trends of
vertical antenna being assumed to “‘have its feet on the behaviour above and below the Brewster angle
 THE ANTENNA AND ITS ENVIRONMENT 135

I
vias
\

field]
Ratio
actual
of
free
field
space field|/
actual
of
Ratio
field
free
space

0-75 1-0 1°5 2-0 0-75 1 1°5 2-0

Height (A) Height (A)
Fig 10.5. Dependence of field strength on antenna height and Fig 10.7. As Figs 10.5 and 10.6 but showing the very different
angle of radiation for average ground. Solid lines are vertical, situation which exists in the case of sea water. The solid lines are
dashed lines horizontal polarization. Points above the horizontal based on the assumption of constant antenna current, and the
dotted line illustrate a gain relative to free-space propagation. dotted lines allow for the decrease in this current (rise in radia-
Dotted curve links points of equal performance for both polariza- tion resistance) over a perfect earth
tions, horizontal being best when the indicated heights are
exceeded and vice versa ground are also indicated and again the differences
are not very great.
specification. It will be seen that in this case the Fig 10.9 shows the calculated improvement in signal
break-even height is not critically dependent on angle strength for a horizontal antenna at a height of 0-75A
of radiation, which is rather fortunate since this is together with experimental averages obtained for
normally unknown and in any case likely to vary Several antennas used over a nine-month period. It will
between wide limits. For 6° elevation the break-even be seen that a horizontal antenna needs to be erected at
heights taken from Fig 10.6 for very dry and very moist a height of about 50ft (15m) to gain an advantage of
Ubi

eee ee eae
‘ : ee ae Vertical polarization
EA Vertical polarization
|oistson |

Polit hewn Fig 10.6. As Fig 10.5 for a radiation angle of

amEo 7 es Bale
6° but showing the effect of extremes of
Average
da |
soil conditions

Angle of radiation =6°

field]
field
Ratio
free
actual
of
space

O 0-25 0-5 0-75 10

Height (A)
136 HF ANTENNAS FOR ALL LOCATIONS

best be resolved by experiments with temporary anten-

nas before making a final choice of polarization. It is
AVERAGE GROUND
considered that in most cases the logical choice lies
between vertical antennas at low height and horizontal
antennas at heights upwards of about 1/2.
The effective mean height of a vertical antenna may
be reduced below 2/4 by the use of short end-loaded
elements or monopoles. Elsewhere (p44) it is recom-
mended that monopoles should be operated against a

oo
Oo

Pama
short inductively-loaded counterpoise, not a buried-
earth or \/4 radial system, in which case it can be
regarded as an asymmetrical dipole, the equivalent
(A)
Break-even
height “height above ground” of a 4/4 monopole being about
ara Vertical at mean
12 above its base. However, the term
height’’ when applied to a ground-based monopole
“effective

height of 4/4 “—~Very dry soil usually refers to its actual physical height.
A low-height vertical antenna (A/4 or less) with its
image closely resembles a single dipole, in contrast to a
O 2 a 6 8 10 vertical dipole with its image which resembles a col-
Angle of radiation (degrees) linear pair, so that there is a broadening of the vertical
Fig 10.8. Break-even height for horizontal and vertical polariza- pattern with (in the case of perfect ground) a loss of
tion. Lower curve shows the height required for a horizontal 3dB. It is therefore to be expected that, although
beam to achieve the same performance as a vertical beam at a reducing the height of the centre of the current distribu-
fixed height of \/4. Top curve assumes both antennas at the same
height, and implies that in theory for angles below 2° or so it tion to 4/4 has no adverse effect on gain, there will be
would usually be an advantage if the elements of the average some loss, say 2dB, with further reduction in height
horizontal beam could be rotated into the vertical position. including the use of near-ground-based monopoles.
However, reference to Fig 10.5 shows the improvement to be
small. The effect of extremes of soil conditions is indicated for a Something of this order has been observed by the
6° angle, the horizontal being always better if its height exceeds writer although good dx performance is still achievable.
High vertical antennas tend to be unattractive for
several reasons:
about | S-unit at 14MHz; it would however be nearly as
(a) The relatively small return for the extra height.
good (and often much easier) to use two low-height
(b) Elements project both above and below the top of
vertical beams suitably spaced and phased.
the mast; the extra height tends to be unwelcome
The theoretical results show that the signal strength
from a vertical antenna should increase with height
and, although the increase is much less steep than in the
horizontal case, the break-even height becomes much (dB)

greater if both antennas are assumed to be at the same

I
height, as shown by the top curve in Fig 10.8. In the
es
author’s case, however, tests on the air failed to Estimated
establish any height advantage at all when this was spread of
i
Pe
actual reports t
increased from 0:25 to 0-6\. If anything there was a 1
I
drop in signal level, the same result being obtained with i
two different antennas although theory predicted an ! : iy
increase of about 2dB. Apart from this one anomalous I Vaverage report
1 (very approx)
result good agreement has been found between theory !
at 4/4
34/4 of
height
at
Vertical
Horizontal
of
height I
and practice, and there is no reason to believe the i
1
results are not typical for the specified site conditions i
1
which are met at the author’s location for the long path 1
1
to VK/ZL, over which all the tests were conducted. !
i
Even so, it should perhaps be stressed that perform- field
Relative Y
ance is much less predictable in the case of vertical
polarization, and especially so when receiving perform- Angle of radiation ( degrees)
ance is taken into account as discussed in the Fig 10.9. Comparison between theory and observed performance
last chapter. It will be found later, moreover, when of vertical beam at low height ()/4) relative to horizontal beam at
3/4. The theoretical curve assumes equal free-space perform-
considering the effect of obstructions that the issue ance for the two beams. Note the good qualitative agreement,
becomes even more complicated and this situation can though precise measurements were not feasible
 THE ANTENNA AND ITS ENVIRONMENT 137

should be partly neutralized as a consequence of the

decrease in radiation resistance; the dotted curves in
Fig 10.7 illustrate this for a /4 grounded vertical
antenna over perfect ground. Sea water provides the
best approximation likely to be available, but the effect
will be less in the case of a beam and over ‘“‘normal”’
ground.
Fig 10.7 demonstrates the possibility of achieving
very low angles of radiation by using vertical polariza-
tion over sea water but, as we shall see, occupants of
seaside cottages may well find themselves disappointed
ro)
rs)
jo
in this. An alternative is indicated by the horizontal
dotted lines marked ‘‘no reflection” in Figs 10.5 and
(dB)
Performance
loss
@ 10.6. Without reflections, free-space field strengths
would be obtained at all times regardless of height,
polarization or angle of radiation and, although most of
the recent discussion has hinged on the fact that
reflections are inherent features of an earthbound
environment, readers may by now be cautious over
accepting appearances at their face value. At this point,
O-1 O3 0.5 1-0 3-0 5-0 10
however, we find ourselves moving away from the
Angle of radiation (degrees) initial assumption of a _ flat, unobstructed,
Fig 10.10. Comparison of short horizontal and vertical radiators at homogeneous site, and before the discussion can con-
7MHz assuming flat, open country. “Zero loss” occurs with tinue the reader needs to be conversant with Fresnel
in-phase addition of the direct wave and a reflected wave of equal
amplitude. Antenna heights are indicated in wavelengths for
zones.
horizontal polarization. Low height assumed for vertical
' radiators, the effect of the ground on radiation resistance being
neglected in both cases Fresnel zones
Thus far the discussion has been simplified by assuming
and there are usually problems due to the lower
reflection to take place at that point on the earth’s
ends getting entangled with the guy-wire system!
surface which gives the shortest distance of travel for
(c) Interaction between the elements and supporting
the reflected ray. Looking at this ‘‘point” more closely,
structures. Methods of getting round this (eg p156)
it is found to be not a point at all, but an area—defined
are more easily implemented for low heights.
by the fact that reflections from all parts of it add up
When antenna height in wavelengths is translated into more or less in phase. For low angles and large heights,
actual height the balance tends to swing strongly in such as can sometimes be achieved in the course of
favour of vertical polarization for 7 and 3-5MHz, portable operation, the zone can be a very large area
particularly for fixed arrays which can, for example, use which may well be broken up by hills and valleys or
short elements supported by catenaries and providing even disappear over the horizon. At the other extreme,
end loading as on p193. Guidelines for 7MHz are important when considering sloping ground in the next
provided by Fig 10.10, from which it may be deduced section, complications may arise because the zone
that a mean height of at least 60ft (18m) and more (though relatively small) is nevertheless too large to be
likely 100ft (30m) is required before a horizontal beam accommodated on the available slope.
becomes competitive with a good vertical array, though To obtain the distances to the near and far edges of
this needs further experimental confirmation. the zone for any given angle of radiation [3] the
Going up in frequency to 283MHz, horizontal anten- antenna height h may be multiplied by the factors given
nas should be easy winners to the extent that the in Fig 10.11, the shape of the zone being an ellipse
break-even height drops to 12-15ft (3-7-4-6m); for having a width of 5-66h. As a compromise dy may be
normal heights about 6dB additional gain over a verti- increased by 1-6 times, dp reduced by 40 per cent and
cal antenna can be expected. However, a different the width by 30 per cent for a reduction of 2dB in the
picture emerges if it is assumed that the 14MHz array is amplitude of the reflected wave. The height of obstruc-
conventional and the same size of element (or the same tions should not exceed h/4 from dy to 0-25d,, h/2 from
element) is used at the higher frequency. A vertical )/2 0-2-0-6d, and h thereafter, though larger obstacles may
dipole, for example, can then be used as a collinear pair be tolerated if they block less than 5 per cent of the
at 28MHz and, if theoretical height gains are realized, horizon; these figures also being a compromise. Some
the advantage of the horizontal antenna at 28MHz values of d; and dy are listed in Table 10.2, based on
compared with 14MHz is reduced to 2dB. Even this the above compromises and an antenna height of SOft
138 HF ANTENNAS FOR ALL LOCATIONS

cosa
2-828

Fig 10.11. Factors for estimating position of

near and far edges of Fresnel zones. The
distances are equal to the antenna height
multiplied by the factors shown

CS pee
Bs

hbase
wpsotcolam
Patdat
(s
Distance
edge,
fornear
factor
cota
~ (3
Distance
far
for
factor
+cota
edge,

(15m). It should be noted that these are ‘‘flat earth” Sloping ground
figures, d; being reduced considerably due to earth
curvature for low angles and large heights. Fig 10.12 shows two environments which are very
The higher angles are included due to their relevance to different despite the similarity of the foregrounds. In
the case of sloping ground, but it should be noted that if Fig 10.12(a) the antenna with its reflection constitutes a
half the beamwidth is less than the angle of radiation two-element array with the hill acting as a very tall
the zone will not be fully illuminated and, since the mast. For a very low angle relative to the earth’s
reflection if any would be additive, some loss of gain surface a gain of 6dB in the direction of the slope is
can be expected. This effect can be reduced by angling obtained from the foreground reflection provided
the beam halfway between the direct ray and the fa rept
reflected ray. The negative distances mean that the & 4sin q,
zone extends behind the antenna and explain the
experimental fact that it is better to place the antenna where q, is the angle of slope, and another 6dB (ie
on the side of a hill than at the top. 12dB in all) from the distant reflection from the sea or a
The conditions specified above for the Fresnel zones flat plain, provided 2h, sin 8 = 4/2. As an example this
imply the possibilities in some cases of losing at least requires a height of 1,000ft (300m) for a radiation angle
part of the reflected wave, thereby achieving good of 1° at 14MHz, though the heights can be divided by
results in what might appear to be unlikely circum- 1-5 for a loss of just over 1dB in each case. The further
stances. edge of the foreground zone is about one-sixth of the
way down the slope for a 20° slope, this fraction tending
to be constant with slope angle, since doubling the
Table 10.2. Fresnel zones angle halves the optimum height and hence the distance
to the far edge.
Angle of radiation Near edge Far edge In this type of situation the near edge of the distant
(degrees) (ft) (m) (ft) (m)
Fresnel zone could get cut off by the slope but in the
1 800 244 10,290 3,136 present instance it is nearly three miles (Skm) out and
3 272 83 3,500 1,067 requires a slope of only about 3° for clearance. On the
6 120 36 1,660 506
10 59 18 1,002 305 other hand, the far edge is at 40 miles (64km) (ie
20 0 0 460 140 beyond the horizon) so that simple geometry is no
30 (—21) (—6) 320 97 longer applicable. However, if very low heights or
45 (—50) (—15) 210 64
angles are involved the ideal solution is to get rid of the
 THE ANTENNA AND ITS ENVIRONMENT 139

explained by assuming an average angle of 6° in the case

of the long path between Europe and Australasia. One
such contact was achieved with only 1-5W p.e.p. from a
location such that lower angles were firmly blocked off
by a mountain ridge. On the other hand, the evidence
available [5, 6, 7] indicates considerable advantage
from the use of angles of 1° or less. These findings can
be reconciled on the assumption that the 6° angle,
which is representative of what can be achieved with
good antenna systems in a typical environment, is low
enough to provide good propagation, but that an angle
of 1° or less results in extra focussing gain (p16) or
reduced attenuation to the extent of at least 10-15dB.
To realize this fully it would be necessary to involve
a wave exceptional circumstances such as those now under
py deg)
discussion.
(ft)
Altitude The type of location shown in Fig 10.12(b) is in
general much easier to find than case (a) and is
oO 2000 4000 6000 8000 representative of some home locations; here we look
(b) Distance (ft) out across mountain ranges or rolling hilly country
Fig 10.12. (a) Ground sloping down to the sea or flat plain. The which occupies the near distance and fails to meet the
direct wave and three reflected waves all add in phase if h, = Fresnel zone requirements specified earlier. As a re-
M(4 sin (6 + @,)) and h, = (4 sin 8), these being the heights of
A above ground and sea respectively. Alternatively the antenna
sult, the distant reflecting area is broken up, but there
with its image can be regarded as a two-element array at a height is still a gain of 6dB from the foreground reflection and
h,. At (b) is shown a typical ground profile for mountainous comparatively little height is needed. In terms of the
country; distant low-angle reflections are non-existent for trans- previous example (assuming 20° slope) we can in effect
mitter at B and probably unimportant (due to break up of Fresnel
zones) for transmitter at A. In both cases low-angle reflections remove the top one-sixth of the slope complete with
(not shown) are obtained from the foreground antenna and put it down wherever we like, provided
the ground in front of it is broken up but without
distant reflection in which case the disappearance of disturbing the horizon as seen by the antenna.
part of the zone over the horizon should be helpful. The author has found it quite easy to achieve long-
This example has considerable bearing on require- path contacts from many such locations with very low
ments discussed in Chapter 2 (p17) and is indicative of powers, in the range 0-5-3-0W p.e.p. (ssb) and in-
exciting prospects for anyone prepared to pack port- verted-V dipole antennas with apex heights of 10—24ft
able gear into a rucksack and take to the hills. More- (3-0-7-3m) and slopes of 20°-50°, though other loca-
over, though not directly applicable to the average tions approximating more closely to Fig 10.12(a) have
home environment, it focusses attention on several been markedly better. Even if angles below 6° are cut
important principles. off, very good results can still be expected, particularly
The main unknown in the equation is the optimum if the reflecting area is bowl-shaped so that it acts like a
angle of radiation; the author has pointed out else- concave mirror to focus the signal in the desired
where [4] that most experimental results can be direction, Fig 10.13(a). Some relatively poor results
(including total failures) have been attributed to convex
Source Source ground which disperses the wave as shown at (b).
As we have seen there is an optimum antenna height
for down-slope working which is in complete contrast
to the flat ground case, and also in conflict with
requirements for other directions for which the antenna
must be as high as possible. Working against the slope
is usually difficult but in some cases one might expect to
Yi find an optimum height for the up-slope direction as in
Fig 10.14, where the reflected wave is blocked off by
(a) Concave surface (b) Convex surface the top of the hill. Although one attempt by the author
Fig 10.13. This shows how rays are dispersed by reflection from a to prove this point ended in failure this is not regarded
convex surface but converge after reflection from a concave as conclusive and the idea might well repay further
surface. In accordance with the laws of optics the angle of study, bearing in mind that antenna position will be
reflection of each ray is equal to the angle of incidence. The
surfaces are assumed to be spherical and suitably placed be- critical. A fixed wire beam for the down-slope direction
tween the source and the observer is easily erected since little height is needed and a
140 HF ANTENNAS FOR ALL LOCATIONS

Dire ct Reflected consistently reporting the signals as about two S-units

Wave wave
stronger than any other Australian stations. Later (in the
Y Hilltop Antenna course of portable activities described on p214), contact
was held with VK3MO down to less than 25mW of
>

radiated power, thus establishing the two-way nature of

the effect. Further evidence of something unusual was
provided by measurements using an aircraft as reported
by VK3MO and VKSTT [11]; these revealed a lobe at 4°
Fig 10.14. Demonstration of possible launch of low-angle wave by in the vertical plane as compared with 10° which would
antenna on “wrong side” of a hill. Note that height and place-
ment of antenna is critical, which may account for poor results have been expected on the usual assumptions of plane
usually obtained homogeneous ground.
The present author’s own estimate of the signal
conventional rotary beam could be used for coverage of enhancement [7] was at least 12dB, making it necessary
other directions. This would also provide a very valu- to account not merely for apparent ray-bending at the
able beam-switching facility. ground but excitation thereby of a propagation mode
Vertical polarization tends to be unattractive when involving considerable ionospheric focussing. Since
the ground is sloping. Fig 10.15 illustrates the situation rays are capable of penetrating fairly deeply into some
in the case of a steep slope, the image being tilted back types of ground (p133), it seems to the author that the
so that it makes little or no contribution to the wanted possibility of ray bending analogous to that in the
field strength. If on the other hand ‘“‘vertical”’ is inter- ionosphere and requiring some form of inhomogeneity
preted as “normal to the slope’’, the situation is repre- in the ground might be a possibility. Amateurs, by
sented by the dotted lines which are inclined to the virtue of numbers and variety of locations, are well-
wanted direction by the angle q,, the signal being placed for adding to knowledge in this area. It is
directed down the slope instead of towards the horizon. perhaps also worth noting the discrepancy between the
A similar situation applies to high-gain end-fire hori- standard assumption that horizontally polarized waves
zontal beams and a maximum of three close-spaced are fully reflected at low angles, even in the case of
elements is recommended, with collinear stacking if “worst ground” including city industrial areas, and the
more gain is needed. basic requirement for Fresnel zones (p137) which in-
With gentle slopes these arguments are less cogent clude conditions for full reflection that are certainly not
and choice of polarization may be more difficult, met in all cases.
particularly when allowance is made for ignorance of
the angles of radiation actually required. The surest
way to resolve the matter is probably to try both, but At the seaside
the break-even height for horizontal polarization will The behaviour of vertically polarized waves over sea
be considerably reduced even for a slope of only 3—5°. water as illustrated in Fig 10.7 allows efficient low-
At 14MHz and above this will almost certainly make a angle radiation to be achieved very easily in the case of
horizontal antenna the more attractive proposition. maritime mobile operation, but is somewhat more
difficult to exploit from a shore base. There will usually
be a stretch of foreground or foreshore separating the
Exceptional locations antenna from the sea as shown for example in Fig
Relative performance for flat unscreened sites with 10.16. From inspection of Table 10.2 it seems likely
properly designed and adjusted antenna systems tends that for angles of radiation of a few degrees the
to be predictable (at least for horizontal polarization) reflecting zone will lie partly on land, partly on water,
but anomalies are not unusual. In cases of abnormally
bad performance it is clearly important to eliminate
faults in equipment or antenna design before looking
for more subtle explanations, and the author is not
aware of any case histories relevant to the present Directions of
discussion. propagation

On the other hand, reference has already been made

to the possibility that good performance from an
apparently poor location might be explained by absorp-
tion of the ground reflection, and one case of exceptional Fig 10.15. Uselessness of vertical polarization for the exploitation
performance from a good location deserves mention as it of steep ground slopes. Image is tilted back into the slope so that
is well documented and witnessed. The high signal levels it is normal to the direction of propagation and therefore wasted;
it is in any case attenuated due to the reflection coefficient being
in the UK from VK3MO are well known and, even when less than unity. Situation is even worse with antenna normal to
this station was using only a dipole, the author was slope as shown dotted
 THE ANTENNA AND ITS ENVIRONMENT 141

———————————
Direc tion of radiation

Far edge
for 6°
Fig 10.16. Typical seaside location 3230ft
showing position of Fresnel zones (985m) Ue nee for
relative to the land-sea boundary for gooft 9500ft es
g
3° and 1°
radiation angles of 1° to 6° (243m) (2895m)
6° = 3°
Near edges for the indicated
angles of radiation

so that two reflected waves, each of somewhat reduced Houses including their internal wiring and plumb-
amplitude, can be expected. This is a complicated ing
situation which is difficult to analyse, though it can be Power lines and supporting structures
assumed that for some very low angle (possibly well Telephone lines and supporting structures
under 1°) even the near edge of the zone will be pushed Trolley bus wires etc
out to sea. This suggests dramatic possibilities but Lightning conductors
(from the lack of published information) either these Lamp posts
have not yet been explored or there is a hidden catch Guttering (metal)
somewhere. It is hoped that readers in suitable loca- Other antennas (amateur and tv)
tions will be encouraged to experiment on these lines. Antenna supporting structures (towers, masts)
Gasometers
Tower blocks.
The effect of obstructions Some of these items can be grouped together and a few
Any objects in the vicinity of an antenna are potential general rules can be identified.
absorbers or reflectors of radio waves but unless they
are very close or else comparable in dimensions with
Trees
the wavelength the effect is usually negligible. There As potential hazards these are usually much overrated.
are two possible ways in which an obstacle can affect Horizontally polarized antennas can usually be more or
the wanted radiation from an antenna. less buried in trees with little or no ill effect, though to
(a) If a non-conducting object is subjected to an be sure of this it is best to use full-sized quad or delta
electric field it acts like the dielectric in a capacitor and loops due to their relatively high values of radiation
can be regarded as having a current flowing through it. resistance. The use of resonant feeders or stubs
If, therefore, the object is of electrically poor quality, ie reaching down to ground level from such loops, apart
if the power-factor of the dielectric is high, losses will from the uses discussed in Chapter 7, allows the
ensue; in any event there will be some distortion of the performance to be monitored by comparing the ele-
electric field in the vicinity. Such effects are confined to ment currents which should be about equal for both
objects in very close proximity to the antenna. directions of fire; this is the same as for quad loops out
(b) If the object is a conductor, a current is induced in the clear and proves they are not being affected by
in it and this in turn sets up its own field so that energy the tree since any losses would show up in the form of
is re-radiated. Conceivably some of this energy may be undercoupling (p86). The sizes even of inductively-
radiated in the desired direction and the correct phase loaded loops have been reduced to 12ft (3-7m) square
but, on average, it represents wasted power from the for 14MHz for an average loss of less than 3dB which
transmitter. The conductor will also have some losses might have been due to losses in the tree, the radiation
so that part of the power is in any case wasted as heat. resistance being reduced to one-quarter. This can be
However, normal rules are applicable so that very little translated back into a probable loss of less than 0-75dB
current flows in the object unless it is fairly close to for full-sized elements.
resonance. A very different situation existed when two 14MHz
Common objects likely to have a significant effect loops with their feeders, one each side of the tree-trunk,
are listed below: were tuned asA/2 vertical elements spaced A/10 for 7MHz;
Trees in this case it appeared that the tree trunk was acting as a
142 HF ANTENNAS FOR ALL LOCATIONS

Direct wave

Fig 10.17. Low-angle radiation

with vertical polarization by the
use of trees for absorption of the
reflected wave
Reflected wave |
(attenuated) ji
f

highly efficient attenuator, though further experiment is marked screening effect or even a slight ‘““boost” could
needed to establish whether or not a tree can be used to be exerted up to distances from the transmitter antenna
support a vertically polarized array of the type shown in of 50yd (46m) or so. This figure is based on observa- -
Fig 13.9. A single row of trees spaced some 30—-40ft tions of the effect one antenna can have on another; if
(9-12m) from a vertical antenna was found to be however objects responsible for screening are accessi-
innocuous but 50yd (46m) or so of thick woodland caused ble, detuning may be possible by the use of linear
several decibels of attenuation at 23MHz. This would be resonators as described on p163.
sufficient to absorb most of the reflected wave at angles In general the influence of vertical or horizontal
suitable for dx working. conductors tends to be confined to radiation from
Under the same conditions no attenuation was vertical or horizontal antennas respectively except at
observed in the horizontal case, and there would very close ranges, though a long inclined conductor
certainly be no appreciable attenuation of the reflected even if nearly horizontal will respond to vertical radia-
wave over the kind of distances that this might have to tion. A loop of wire such as a quad element does not
travel through flat woodland at a radiation angle of, however have a ‘‘way up” but responds equally to both
say, 6°. The range available for these measurements polarizations. As a kind of informed “‘guesstimate”’ it is
was not sufficient to establish whether or not horizon- believed that the screening effect at SOyd (46m) could
tally polarized rays at very low angles such as 1° would be as much as one S-unit, though in many cases it is a
be absorbed or not. However, it seems at the very least lot less. Screening in excess of two S-units has been
highly probable that, at least in the case of vertical observed at about 30yd (27m) range in the case of a
polarization, trees can be used to achieve low-angle second beam in the line of fire of the one in use. It
radiation as in Fig 10.17 by attenuating the reflected might be thought that this could be avoided by ensuring
wave so that it no longer interferes with the direct wave that the second beam has a good front-to-back ratio,
from an antenna located at tree-top level. but in practice this is no help at all because the reflector
of the second beam reflects backwards the signal from
Obstructions other than overhead lines the first.
The effect of houses and their contents is difficult to Back-to-front ratio applies only to voltages at the
predict. In some data tables [8] city industrial areas are terminals of a beam, which is of course where it is
listed as having a dielectric constant of 3-5 and an normally wanted, and does not relate to the properties of
effective conductivity of 10-* to 10-4mho/m which puts the beam as a reflector of other signals. If the beam
them in almost the same category as (though slightly which is causing the screening is rotated to an end-on
better than) the “very dry soil” of earlier examples in position the effect on the radiated signal should be
this chapter, ie they are nearly perfect reflectors of negligible, but in reception an unwanted signal from
horizontally polarized waves at low angles and require the back direction reflected from the second beam
no special consideration. could still be greatly in excess of the signal received
Though doubtless satisfactory as an approximation directly on the back of the first. This means that
in dealing with medium-wave propagation from reflections from metal objects of all kinds (but particu-
high antennas, it seems obvious that for amateur hf larly other antennas tuned to the same wavelength) can
communication tall buildings must inevitably infringe very seriously degrade the interference-rejecting prop-
the Fresnel zone specifications quoted earlier. This is erties of the beam in use. :
particularly so as buildings include wiring, plumbing, In a built-up area it is reasonable to expect that the
lightning conductors and metal guttering, all having reflected wave will be to some extent absorbed by
resonances in the hf band. Some of these resonances brickwork; 2dB has been quoted as the effect of a brick
may be sufficiently damped to be ineffective at least in wall [9] and the author has observed some 3—4dB loss at
relation to outside antennas, but in other cases a the “wrong side of a house’’. Part of the wave will be
 THE ANTENNA AND ITS ENVIRONMENT 143

used in setting up currents in loops of electric wiring, i a _ yp B

conduit, plumbing, lamp posts etc and the aggregate of = == aa = Powerline —

such effects must inevitably reduce the reflection coeffi-

cient to some extent. This will be beneficial if, at the
required angle of radiation, the reflected wave is
opposing the direct wave and conversely if the waves
are additive the signal amplitude will be reduced. The
direct wave can be expected to escape into space
without encountering these obstructions.
As discussed earlier the “‘additive” case is exception-
al and it might be expected that (other things being
equal) those in built-up areas would be better off than
those in the country but, although examples could be
cited in support of this, other evidence points in the
opposite direction. Considering the two extremes, if
the reflected wave is not absorbed signal strength
should be proportional to height (which is a fairly VW Pole or pylon
common experience), whereas with complete absorp-
tion, which leaves only the direct wave, height should Nes (b)
make no difference to the signal strength. However,
raising the height may also have the effect of clearing a Fig 10.18. Re-radiation by power lines. At (a) some fraction of the
power in the wave A is re-radiated in directions B, C, D and E. For
small number of objects in the immediate vicinity, and the wanted direction A, radiation may be enhanced or opposed
this almost certainly accounts for a number of known by the lobe B. At (b) it is assumed that energy travels outwards
cases in which the improvement from raising the height along the power line in both directions from the corner, becom-
ing gradually attenuated without reflection. If reflection does
was disproportionately large. take place, eg at insulators, there will be radiation also in the
reciprocal directions of B, C, D and E.
Overhead wires
There is very little information on the effect of over- since the waves propagated along it will gradually
head wires on signal strength, though power lines are a diminish and eventually die out before being reflected.
frequent source of interference in reception. Neverthe-
less the resemblance of such wires to long-wire anten- High-rise buildings
nas suggests that if, say, 10 per cent of the radiated If it is possible to get a horizontal beam onto the roof of
power were to find its way into such a conductor the a tall building, say a block of flats, the dx prospects are
re-radiated field could be comparable with the direct excellent. In the case of full-sized elements there is no
field which it might assist or oppose. Referring to the need to have much clearance above the roof provided
dotted curve in Fig 5.3, the mutual resistance between any detuning effects are corrected, though there seem
2 dipoles spaced 3-25A can be as much as 10 per cent to be no useful figures available. Clearances would
of their self-resistance, and if 72 X 1-1W are fed into probably need to be at least double in the case of a
one of them the currents will be respectively 1A and miniature beam such as Fig 12.3 on account of the
0-32A. larger voltages and currents. There is no certainty that
Although the calculation gets more complicated, it horizontal polarization will give best results in all cases,
takes little imagination with this as a starting point to though this is the most likely situation. Severe screen-
imagine tight coupling between an antenna and neigh- ing can be expected in the case of waves propagated
bouring power wires, even to the extent of achieving through buildings.
power gains in directions nearly in line with the wires as
in Fig 10.18(a) This would be accompanied by reduced Portable operation
signals in other directions. The power line could there- The exploitation of sloping ground as described earlier
fore be expected to distort the radiation pattern in ways (p139) for dx contacts using very low power and simple
which might be helpful along the direction of the line antennas such as the inverted-V dipole is a fascinating
but would be harmful for other directions. Use of hobby in itself, but suitable locations are not always
vertical polarization would be the most obvious way of ready to hand. Finding the best ones demands intensive
avoiding this problem. study of maps, much travelling, probably a long climb
Fig 10.18(b) illustrates other likely aspects of the on foot in the early hours of the morning to a height of
situation by assuming a sharp bend in the power line. If 1,000ft (300m) or so, and a lot of luck with the weather.
the line is very long in each direction and the capaci- Good results can however usually be expected from a
tance of insulators can be ignored, each leg can be cliff top or steep ground with a drop of 100ft (30m) or
thought of as a separate terminated long-wire (p198) so to the sea, and one exceptionally good location
144 HF ANTENNAS FOR ALL LOCATIONS

DX working is not normally possible from a flat site

with powers of 1-2W ssb and the sort of antenna
heights (less than 20-30ft (6-9m)) which are easy to
Reflected wave
from A1 diverted achieve. On the other hand, at 14MHz single-hop
upwards
100ft
working over distances of 800-1,600 miles (1,300—
Reflected
eflec wave (30m) 2,600km) has usually been found possible, in many
from A2 cancels Al i cases with an antenna height of only 8-10ft (2:5—3-0m).
the direct wave a
a ZA Yi The indoor environment
Aq A. Gas angle <1° Indoor antennas tend to be coupled rather closely into
their surroundings and this is liable to generate a lot of
300 ft
AJes00%
problems. Factors to be taken into consideration in-
gia clude:
(2560m) (a) Absorption in brickwork or wet (ie external)
Fig 10.19. Geometrical model of situation giving better low-angle woodwork;
propagation from lower antenna assuming a ground slope less (b) Losses and distortion of radiation patterns by ~
than the angle of radiation. For radiation angles around 1° there is
no reflected wave from A, in the same direction as the direct currents induced in mains wiring, plumbing, tele-
wave. Higher angles are cancelled by the foreground reflection, phone wires etc; |
lower angles by the distant reflection. Horizontal scale is longer (c) TVI and other interference due to the same cause;
than the vertical, the angles being multiplied in proportion. For a
typical 6° angle the situation is reproduced if the horizontal scale (d) Difficulty in accommodating full-sized elements,
is divided by six, and good low-angle propagation can then be and aggravation of the above effects in the case of
expected over the range 4° to 6° approximately loaded elements and particularly small beams;
(e) Earthing problems, closely related to (b);
found by the author was near the top of a steep-sided (f) Height limitations;
railway cutting which appeared to be acting as some (g) Shortage of engineering guidelines;
kind of horn radiator! In general, steep bowl-shaped (h) The prevalence of unsound design practices.
contours are usually effective but (as explained earlier) To minimize the difficulties it is important to keep
convex foregrounds should be avoided. the radiation resistance as high as possible unless large
In the case of a cliff top the centre of the dipole clearances can be arranged, and (except for reception)
should be pushed out horizontally from the edge using, there may be no advantage in using a beam. Particular
say, a pair of 8ft (2:-4m) garden canes tied together, and attention must be given to earthing, and it is believed
the ends can be strung back to tufts of grass, points on a the best answer in most cases is likely to be a short
guard rail or anything else available. Polarization must counterpoise tuned with a series inductance, arranging
be horizontal and a typical set-up is illustrated in Fig the antenna feedpoint and the counterpoise to be as far
16.9. There is still much to be learned, however, and away as possible from wiring or plumbing. All other
the less-enthusiastic experimenter content merely to try conductors near the antenna need to be checked for
the best locations that happen to be handy in the course possible in- or near-band resonances with a grid dip
of, say, a holiday camping trip may be in for a few oscillator having a coil of large area (p231), and any
surprises. that are found should be removed, eg by means of
On one occasion a very gentle ground slope provided inductively-coupled traps (p163).
the author with three UK contacts from Tasmania with
1-5W p.e.p., although the dipole was hanging from a The earth connection {
tree branch at only 20ft (6m). After much puzzling an The ground underneath and around the antenna has a
explanation was conjectured on the lines of Fig 10.19, number of functions. Its role as a reflector of signals has
the lack of any gain from reflections being compensated just been discussed at length but it is often required to
by the low-angle focussing gain. This is a fairly common provide all or part of the return path for the antenna
situation which might apply to some home locations, current, ie the rf circuit comprising the antenna system
and it is interesting to note that if the antenna height is is completed via current paths in the ground as in Fig
doubled the reflected wave clears the distant hill and 6.19. This illustrates the Marconi antenna, the lower
cancellation takes place. If this explanation is correct side of the generator (eg the outer conductor of a
success may have to be attributed on this occasion to coaxial feeder) being connected to a buried earth which
the author’s lack of prowess in aiming for a higher tree in its simplest form is a metal spike driven into, the
branch! ground. The efficiency of such a system, neglecting
To exploit vertical polarization over sea water the losses in the antenna wire, is given by R/(R + R,)
same portable antenna could be used as a top-loaded where R is the radiation resistance (369 for a A/4
ground-plane operating against a small counterpoise as radiator) and R, the resistance of the ground con-
discussed in Chapter 16, p215. nection—typically about 100, giving a loss of 6dB.
 THE ANTENNA AND ITS ENVIRONMENT 145

The more usual way of avoiding the need for an earth

mat is the ground plane (p43), a variant of the counter-
poise but possessing fewer advantages. Though deman-
ding less space than the earth mat, this likewise has an
environmental impact and also encourages the feeder
to radiate.
h=4
Summary
For working dx from a flat unobstructed site at 14MHz
and above the choice lies between a horizontal beam
(which should be as high as possible) and a vertical
Fig 10.20. Typical radial earth system for short vertical antenna
beam, which can be almost ‘“‘sitting on the ground’,
though there is some advantage if its electrical centre
Even to get it as low as this may involve a lot of hard can be placed at a height of about 4/4 to 5A/16. The
work such as driving a 2in (Scm) diameter 6ft (1-8m) relative merits of horizontal and vertical polarization
rod into heavy clay soil, a procedure which the author depend on the height available, ground constants and
strongly deprecates and has no intention of repeating. possible obstructions; rough tests may be advisable
To overcome this problem the method usually re- using temporary antennas before proceeding with a
commended is the burial of a large number of radial ’ permanent installation.
wires around the base of the antenna as in Fig 10.20, Advantage can be taken of sloping ground to obtain
typically 50-100 wires of lengths up to 3A/2. In the optimum performance in down-slope directions using
amateur context this involves a somewhat mind- horizontal antennas at low heights. Obstructions on the
boggling expenditure of time, money and effort which other hand may result in an optimum position, height
confers no significant benefit. One of the main attrac- or polarization which in general needs to be found by
tions of vertical antennas is the possibility of achieving experiment, though some guidelines have been sug-
good performance without using any appreciable gested.
amount of horizontal space, let alone the large amount On simplistic assumptions ground conductivity is of
of real estate required for an efficient buried radial no significance for low-angle propagation of horizontal-
system. Neither is it necessary to incur the wrath of the ly polarized signals but there is evidence in some cases
xyl by the digging up of flower beds, or risk the lethal of “bending” of the reflected wave or some equivalent
potential gradients that must almost certainly exist at effect, with important practical consequences.
the ground surface in the event of a lightning strike. In portable operation advantage can be taken of
It would be a different matter if one could surround steep ground slopes to achieve dx contacts with very
the antenna with a high-conductivity ground system low power and a simple horizontal antenna at low
extending to the outer limit of the Fresnel zone, height.
thereby realizing not merely a low earth resistance but Short-range working may involve high-angle
the enormous benefits of efficient radiation at very low
angles. This has indeed been achieved with impressive
results [10] but such attempts at controlling the en-
vironment are incompatible with amateur resources.
Fortunately the earth resistance can be held to a low
value by using the rather simple type of counterpoise
earth shown in Fig 6.20(b), which requires very little
space and, in effect, converts a monopole into a dipole.
In this case the main source of loss is the resistance of
the loading coil which, even for a counterpoise of only a
few feet in diameter, could be as low as that of one of
the ‘mind-boggling earth-mats” described earlier. In
many cases the counterpoise need consist of no more As.

than a short rod, though if the ground clearance is a

~ ANA
hi. ww

insufficient there will be some extra loss due to earth

currents. The main disadvantages are reduced band-
width and multiband problems which require similar
solutions to those for the elements themselves. Often
the vertical monopole can be re-arranged as a dipole
with some advantage (p156). “the ground plane... this likewise has an environmental impact”
146 HF ANTENNAS FOR ALL LOCATIONS

reflections from the ionosphere in which case vertical [6] ‘““A comparison of long-distance hf radio signal
antennas are relatively inefficient and horizontal reception at high and low receiving sites”, M. R.
antennas are preferable; in this case however there is Epstein, et al, Radio Science Vol 1, 1966, pp751-
no need for height, which can be as little as a few feet 762.
even at the lower frequencies. [7] ‘High-frequency antennae and _ propagation
modes in relation to the amateur service’’, L. A.
Moxon, IEE International Conference on Anten-
References nae and Propagation, 28-30 November 1978.
[1] Radio Engineers Handbook, F. E. Terman, [8] Radio Engineers Handbook, F. E. Terman,
McGraw-Hill, 1943, pp700-707. McGraw-Hill, 1943, p709.
[2] Radio Communication Handbook, 5th edn, [9] ‘Attenuation through trees and brickwork”, K.
RSGB, 1977, p11.9. Bullington, Bell System Technical Journal May
[3] “‘Siting criteria for hf communication centres”, W. 1957, p593.
F. Utlaut, NBS Technical Note 139, US Dept of [10] ‘“‘A low delta surface-wave interferometer array
Commerce, April 1962. for hf radio communication’’, J. F. Ward, Nature,
[4] ‘“‘Low-angle radiation”, L. A. Moxon, Wireless 205, 13 March 1965.
World April 1970. [11] ‘“The effect of ground on the directional pattern of
[5] “Effect of antenna radiation angles upon hf radio a 14MHz antenna’, A. G. Bolton and I. J.
signals propagated over long distances’’, W. F. Williams, Amateur Radio July 1977.
Utlaut, Journal of Research of the National Bureau [12] Radio Communication Handbook, 5th edn,
of Standards Vol 65D, No 2, March-April 1961. RSGB, 1977, Chapter 11.
CHAPTER 11

Single-element antennas

For reliable long-distance communication a beam A straight dipole such as Fig 11.1(a) or (b) is mainly
(though desirable) is far from essential, while for of interest when at least one support (such as a tree or
shorter ranges a single element has some advantages chimney) is already available and it is not feasible to
since all directions tend to be of equal interest. The use this for supporting the centre of a loop or a short
likelihood of replies to a CQ call tends to be increased extension for an inverted-V. A tree may be difficult to
more by the absence of directional discrimination than climb so that it is impossible to use it for anything more
by a slightly bigger signal limited to a single direction. complicated than the attachment of one end of a
The operator is moreover spared the time-wastage and dipole, and in any case a single element of any kind
inconvenience of waiting for a rotary beam to rotate! supported at its centre might overhang a neighbouring
In general a single element is much easier to con- property. Straight dipoles may be made rigid enough to
struct and erect than a beam, relatively inexpensive and require supporting only at their centres in which form
less likely to lead to problems of planning consent and they are most commonly met with as beam elements.
trouble with neighbours. The decrease in signal They may be useful on their own and, for example, it
strength compared with a beam is no worse than results may be found that they can be erected in a tree without
from switching out the linear amplifier, and is small climbing it (eg by throwing a rope over a branch) or at a
compared with average differences between one signal substantially greater effective mean height than would
and another. The signal is admittedly less competitive, be possible for a loop or inverted-V. For such applica-
it is more difficult to find a clear channel, and interfer- tions it is also possible to construct a bamboo “bow”
ence is also much more likely, but (except in this last with a wire dipole as the bowstring.
respect) the owner of the single-element antenna is not The folded dipole, Fig 11.1(b), provides almost twice
seriously handicapped. Moreover, current develop- the bandwidth (and for a given weight of copper has
ments should enable interference to be nulled out with only half the losses) of a single-wire dipole, though for
the help of a relatively small antenna as described in most purposes this is adequate and slightly simpler.
Chapter 9, and in a few favoured cases both forward Taking mechanical considerations into account, one
gain and screening against signals from the back direc- can arrive at a complicated set of trade-offs between
tion may be provided by the environment itself as single and two-wire dipoles, though the pros and cons
described in Chapter 10. are mostly trivial. The author tends to favour folded
Much of the alleged superiority of beam antennas is dipoles using, say, 18swg (1:-2mm) hard-drawn copper
thought to have psychological roots, the acquisition of a wire with a central gap between conductors of about
beam being accompanied by a natural desire to do it /30 for all bands from 7—28MHz, but further discussion
justice. Thus it arrives complete with mast and (at twice of the mechanical aspects will be found later in this
the height of the dipole) provides a big improvement, chapter (p150).
most of which might have been achieved more easily There is no need for insulators as such, recom-
and cheaply by putting the dipole up at, say, three mended practice being the use of thin polythene or
times its original height. nylon cord (1—2mm diameter) since this is light, strong,
cheap, an excellent insulator and affords minimum
“visual impact’’. However, so as not to extend this last
Half-wave horizontal dipoles feature to the diagrams, insulators are included to
Fig 11.1 illustrates those types of horizontally polarized indicate ends of wires throughout this and other chap-
2 dipoles likely to be of interest to the reader who has ters.
digested the arguments in Part 1. Some of these are Even when no insulators are used, the resonant
important building blocks for beam antennas to be length of a dipole is slightly less than the free-space
described in the next chapter, and have been included half-wavelength, depending on the ratio of length to
partly or mainly for this reason. In the first instance the diameter but typically 2-4 per cent for wire elements
discussion is confined to monoband operation. rising to 4-7 per cent for lin (2:-5cm) diameter tubing at

147
148 HF ANTENNAS FOR ALL LOCATIONS

Spacers if needed 14MHz or half this at 283MHz. Tapering both in its

usual form and as illustrated in Fig 11.1(b) will require
some slight increase of length, but neglecting this the
Delta match
(preferred) or correct formula to use for bare wire dipoles without
centre-fed
7582 twin
with Thin open-wire line
or 30082 twin
insulators is
Length (ft) = 478/frequency (MHz)
Length (m) = 145-7/frequency (MHz)
It is however advisable if possible to check resonance or
swr before erection (p231).
The inverted-V, Fig 11.1(c), has important advan-
tages since the single support also carries the feeder
5082 coaxial cable 4 which blends in with it so that there is less visual
(or use 6002
with delta match
line impact. Moreover, since there is no feeder drag on the
at X, X) wire, this can be much thinner and the antenna
perhaps even rendered completely invisible. The effec-
tive mean height with 90° apex angle is about 5ft (1-5m)
less than the height of the support, but this is largely
offset by the absence of sag and since there is very
Delta match much less pull only a flimsy support is needed. It should
or centre-fed
with 7582 twin Thin open-wire
usually be possible therefore to achieve considerably
line or 30082 greater height, and the risk of serious damage in the
event of breakage is much reduced.
The ends may be attached to relatively short supports
depending on the space available, the base-line
required being not less than 2(h,—h,) where h, is the
apex height and h, that of the auxiliary poles or other
available anchoring points such as a garage roof or tall
bush. If necessary the ends of the dipole may be bent
over slightly. As an alternative to additional poles two
short lightweight spider arms (eg bamboo) may be fixed
to the mast, the arm lengths and their distance from the |
top of the mast being at least 12ft (3-7m) for a 14MHz
dipole. The apex angle should be as large as possible
and not less than 90°, for which angle the radiation
resistance and bandwidth are roughly halved.
In those cases where the antenna is required to be
j invisible the wire size can be reduced to about 28swg
(0-4mm) [1], the main problem being to render it
des eA hod Ae rsdal ete gd
areas at Pest pten ani sort invisible to neighbours without becoming a navigation-

pet psig
al hazard for birds! It is essential, however, to ensure
that in the event of breakage there is no risk of
accidental contact with power lines or hazard to pedes-
trians or traffic. An inverted-V with a 90° apex angle
has only half the radiation resistance of a straight
Spacers Arrows indicate current flow
2 =~
dipole, so that if wire size is also reduced there may be

Fig 11.1. Dipole elements. (a) Conventional )/2 dipole. (b) Folded-
n dipole or “squashed delta loop”. (c) Inverted-V dipole; coaxial
feeder plus balun may be used as the mast takes the weight
—» directly. (d) Folded inverted-V dipole. (e) Inverted-U dipole. (f)
Folded inverted-U dipole. (g) End-loaded dipole; this is suitable
ae | 0-51 as a low-height vertical or, in its horizontal form, fitting into an
attic. (h), (i), (j) are as (g) but allow further length reduction, (j)
being suitable for suspension horizontally between two poles or
{ OSI use as a beam element. (k) Centre-loading. (I) Coil loading. (m)
———
Loading by bending the ends over. Due to excessive current flow
in the reverse direction the folded-back dipole (n) and equivalent
loop (o) are not recommended
 SINGLE-ELEMENT ANTENNAS 149

a case for using a folded construction as in Fig 11.1(d)

in order to maintain efficiency and bandwidth.
The U shape, Figs 11.1(e) and (f), is closely equiva-
lent to the V and has a number of applications. Suppose
that operation is required on 7MHz, given two supports
only just far enough apart for a 14MHz dipole, Fig
11.1(a) or (b). All that is needed is to bend the ends
over to form a U-shaped element. For fitting dipoles
into an even smaller space, including indoor locations,
any of the arrangements (g) to (i) may be used depend-

| |
ing on the space available; (g) to (I) are preferred on 502 cable to trans-
cable to
grounds of symmetry but (j) provides a somewhat transmitter
mitter, any length
(for straight dipole
lower resonant frequency for a given area. Some use 7582)
bending over of ends is allowable if necessary but it is
essential that any wires running parallel with the centre
of the element should be very short or carrying only ie, Gueteuee Cantious.
very small currents (p85). In case (g) roughly 3ft (0-9m)
(total) of extra wire is required at the ends for each 2ft
(0-6m) reduction in overall length.
0 (d)
\752 twin
It is usually more convenient to fold the centre of the
element as in Fig 11.1(k) but the radiation resistance is y atu or

then halved for a reduction in length of only 20 per cent Q

V | 4:4 balun
risa ee eee ae
which should not be exceeded in the case of a beam : | cable to
4:1 balun cable to transmitter | transmitter
element. Reasonable efficiency is possible in the case (any length but iShack
of, say, a 14MHz heavy-gauge single dipole used in this preferaby less than 4/2)
way at 7MHz, the radiation resistance being 722, loss Se ee ow oe oe ow ew

resistance typically about 192 and the bandwidth about

1 per cent which is adequate for operation over the (f )
European phone or cw bands. The efficiency in this bP 600 82 line
(any length but
case is 85 per cent, ie there is a loss of 0-6dB. Similar use bh preferably <A)
of 28 or 7MHz dipoles is more restricted since the 3002 twin
14MHz and 3-5MHz bands are relatively wide. Tuned 1
'Shack
|
feeders if used in these cases must be very short; j atuor
paet balun
alternatively a matching stub should be used as close as !
possible to the antenna with some means of switching it | cable to | cable to
j transmitter
j transmitter
out of circuit for multiband operation. iShack
Fig 11.1(1) shows another common method of
shortening, with suggested lengths. For the same 20 per Fig 11.2. Methods of feeding horizontal dipoles. (a) Delta match
cent reduction in length the radiation resistance is to 6002 line. (b) Coaxial feeder with balun. (c) Combination of
open-wire and coaxial feed. (d), (e) Balanced twin feeder. (f)
reduced to 52. Thus the loss is only 0-4dB, this being Resonant-line system for multiband operation
increased to 1-:2dB if the end sections are halved,
reducing the length by 40 per cent in all. The method convenience and may be unsuitable if the feeder has to
shown at (m) is equivalent to (1) in terms of radiation travel any distance inside a building or very close to it.
resistance, but preferable since it cuts out the coil losses Fig 11.2(b) shows the best way of using coaxial feeder,
and can be used for fitting an inverted-V into a smaller and this is recommended in particular for use with
space. multiband inverted-V dipoles since direct connection of
The folded-back dipole, Fig 11.1(n), and equivalent an open-wire line as in Fig 11.1(a), (c) or (e) is not
loop (0) are included as examples to be avoided practicable whereas coaxial line can be fixed to the
although they will be encountered from time to time in support, there is no drag problem and the visual impact
the literature. As shown by the arrows, current flows in is nil. On the other hand, in the case of straight dipoles,
both directions, thereby greatly decreasing the radia- . Figs 11.1(a) and (b), its use imposes penalties in the
tion resistance. For a half-size dipole or a loop as form of extra weight, windage, and visual impact. The
illustrated the radiation resistance is divided by four extra losses (and cost) of coaxial feedline, though not of
approximately compared with end-loaded elements major importance for lengths up to 100ft (30m) or so,
such as Fig 11.1(b) or 11.4(d). are to be avoided if possible and the arrangement of Fig
Fig 11.2 shows alternative methods of feeding these 11.2(c) seeks to make the best of both worlds. For
antennas; (a) is the most efficient but scores lower on feeder lengths exceeding A/2, open-wire line is used in
150 HF ANTENNAS FOR ALL LOCATIONS

Angle of sag Resonant open-wire line as at (f) allows an antenna

to be erected quickly without matching adjustments or
iF~E _» T accurate measurement of length, but requires an atu
Tension T and is mainly applicable to multiband systems as de-
Total weight W
~ 26 = 2Tsin@g scribed later. Alternatively, for long feeder runs the line
where @ is
in radians can be operated in the resonant condition as far as
some convenient point at ground level, where use of a
stub as explained on p41 allows the rest of the feeder to
operate as a matched line. Alternatively the arrange-
ment at (c) allows a dipole to be centre-fed with
open-wire line operating at a relatively low value of
swr, eg 2 or 3.
Sag 8 of equivalent simple dipole

=
Mechanical aspects of horizontal wires
t
i In the case of horizontal wire dipoles, Fig 11.1(a) or |
Folded dipole (b), there will inevitably be some sag resulting from the
as fig 11.1(b)
weight of the feeder; referring to Fig 11.3(a) this
translates into tension in the horizontal portion accord-
ing to the relation
W = 2Tsin 8
The weight of the antenna wire should be added to that
of the feeder but in addition there is wind pressure
which can be reckoned as producing a sideways force of
up to 20-30lb/sq ft (950-1,400N/m?) depending on
geographical area [9]. For a 14MHz dipole with a 4/2
open-wire (20swg or 0-9mm) feeder and a wind speed
Tension=W of 100mph (160km/h) we obtain from Table 11.1 a drag
of 66 X 0-7 = 5lb (2-3kg); if the feeder is allowed to
billow out away from the wind direction to the extent of
15° and anchored at its lower end this turns into an
(c) endwise pull of 10lb (4-5kg) which is the effective value
of W for use in the above expression. For a sag of 3ft
(0-9m) the tension in the antenna wire comes to 27-5Ib
(12-5kg) which from Table 11.2 requires the use of
Fig 11.3. (a) Tensions in centre-fed horizontal wires; advantage
can be taken of the folded dipole (b) to reduce tension by 16swg (1-6mm) hard-drawn copper.
reduction of W and increase of 0. (c) Method of ensuring constant A counterweight of 291b (13-2kg) can be used as in Fig
tension in the antenna wire. The counterweight is equal to the 11.3(c) to ensure that this figure is not exceeded. The
allowable tension T
windage is roughly doubled for coaxial feeder of 4in
multiples of A/2 so that the antenna impedance is (6mm) diameter but this can probably be accommo-
“repeated” at the terminals of the balun. The spacing dated by allowing more sag and billowing! Fortunately
of the delta and the precise length of the open-wire line wind speeds of 100mph (160km/h) are rare, and unlike-
should be adjusted for minimum swr in the shack. In the ly in the UK to coincide with conditions of severe icing
case of the inverted-V the open-wire feeder needs to be at
least 6in (15cm) clear of the support and (b) may well be Table 11.1 (from reference [8]). Calculated wind
preferred on grounds of appearance. drag at 100mph (160km/h) on tubes and taut wires
The use of matched twin feeder as at (d) and (e) is with the wind perpendicular to the axis of the tube
the simplest and neatest method for centre feeding of or wire
horizontal wires, but suffers from wet-weather prob- Diameter Drag
lems with the types readily available and a further (in) (mm) (Ib per ft length) (kg per metre)
drawback in case (e) is the additional windage. On no
account should the transparent plastic type of 300Q2 line re 51 5:8 8-6
1 25 2:9 43
be used since it can deteriorate very rapidly but, 0-5 13 1:47 2:2
provided phase-stability is not required and runs are 0-128 (10swg) 3:25 0:29 0:43
fairly short, other types may be used if given a coating of 0:08 (14swg) 2:0 0:17 0-25
0-0 36 (20swg) 0-9 0-07 0-10
silicone grease.
 SINGLE-ELEMENT ANTENNAS 151

Table 11.2 (adapted from reference [2]). Maximum and the half-waves out of phase, provided the require-
recommended tensions for hard-drawn copper ments for efficient end feeding (p46) can be met. The
based on 15 per cent of the breaking strain 1, antenna on the other hand is the best choice, given a
special interest in a particular direction, since it pro-
Wire size Tension
(swg) (mm) (Ib) (kg) vides useful gain and reduces interference from un-
wanted directions, assuming it can be aligned properly.
10 3-3 126 57 Using spreaders it can be backed by reflectors to
12 2-6 78 35
14 2:0 48 22 provide a high-gain beam system.
16 1-6 30 14
18 1:2 16 7:3
20 0-9 10 4-5 Loop antennas
Loops have two main advantages over /2 dipoles: they
for which no allowance has been made. The counter- fit into a much narrower space and have a higher value
weight is essential if trees are used for support because of radiation resistance (p98), the width required for
a very small movment of the tree represents a large resonant loops being only 0-5-0-7 times the length of a
increase of tension; the length occupied by the dipole is straight A/2 dipole, assuming no loading in either case.
given by A/(2 cos 8) and if, one end being fixed, the The loop at its fundamental resonance contains 1) of
other moves away by only 4in (10cm), @ is reduced wire, in which respect it resembles the folded dipole,
from 10° to 5° and the tension is doubled. Fig 11.1(b), which can also be considered as a loop
The use of hard-drawn rather than soft copper wire is “squashed flat’’.
frequently stated to be mandatory and this is confirmed Fig 11.4(a) and (b) shows the two most usual forms
by the above example for the case of Fig 11.3(a). In of the quad loop. These are strictly equivalent though
case (b), the sag is doubled for a given change in mean one or the other may be more convenient, (a) being
height and the dipole can even be allowed to turn into a more usual in the case of rotary beam elements,
delta loop (p153) without much ill effect, since whereas (b) is more convenient for fixed antennas,
reduction in mean height is largely offset by the slight allowing them to be suspended from a single support as
gain (1dB) of the single loop compared with a dipole. shown. -A spreader can be used in lieu of the cords if
The feeder wire size cannot be reduced since halving space is restricted. By virtue of its size, shape and high
the diameter only halves the drag but reduces the value of radiation resistance this type of loop antenna is
breaking strain to a quarter. particularly suitable for erection in trees, losses usually
The above examples serve to highlight the problems being negligible so that if a suitable tree is available
and provide basic engineering guidelines, though the high performance can be attained with a more or less
rules are commonly disregarded without serious reper- invisible antenna.
cussions. Antennas like those under discussion are For more than 20 years the author relied for most of
often constructed from “‘any odd length” of soft copper his dx contacts on pairs of loops wholly or partly buried
wire out of the junk box, maybe off some old mains in trees and operating as fixed but reversible beam
transformer, the occasional breakage being easily antennas. Although much more stringent conditions
mended. Moreover, in most of the other antennas apply in the case of a beam no adverse effects were
described in this book, wires are subjected to relatively observed except on rare occasions due to extreme
little tension and annealed copper which is more readily deposits of snow or ice. In the preferred arrangement
available and easier to handle is usually satisfactory. It the top corners of the loops were held up by a Oft
is of course vital, in any case where danger or serious (2:7m) spreader attached to a short extension pole
embarrassment might be caused by failure, to stick to secured to the main tree trunk, and the corners were
the rules and make sure of an adequate safety margin. pushed into position by stout bamboos secured to tree
branches with occasional help from 2in by lin (Scm by
2-5em) wooden struts. There should be at least 6in
Full-wave antennas (15cm) of polythene or nylon cord between the corners
Use of 4/2 dipoles at their second harmonic, ie as of the loops and the bamboo supports because wet
full-wave (1A) dipoles, is discussed later in the context bamboo is a very bad insulator.
of multiband operation but the topic may also arise in On the other hand, there is no need to worry unduly
respect of monoband systems. Given an available span if at times a few leaves (even wet ones) come into
of 1X the options include two half-waves in phase (the contact with the wires. An 11ft (3-4m) long gin (16mm)
1) dipole) and two half-waves out of phase, in addition diameter aluminium alloy tube with a hook at one end
to the basic A/2 dipole. The question may arise as to was used for threading wires over tree branches etc,
which of these is the best arrangement. None provides and a pruning saw tied to the end of the same rod
all-round coverage and if all directions are of equal was used for removing remote branches that could
interest there is little to choose between the A/2 dipole not be avoided. Bamboo rods, though slightly less
152 HF ANTENNAS FOR ALL LOCATIONS

May be attached to
a mast ortreetop
pari, or pase/ re spider arms

Fig 11.4. Loop antennas. (a), (b)

are quad loops, (a) being most
Open-wire line suitable for rotary beams and (b)
(a) more convenient for fixed
Open-wire line
arrays. At (c) the top of the loop
is open-circuited to provide op-
eration at twice the frequency
Short radial extensions with 3dB gain. At (d) loading by ~
if needed
\\~ 50 or 752 cable means of capacitive “hats”
allows considerable reduction in
size. Delta loops are shown at .
(e) and (f) with dotted lines to
10 to 12ft
(3-O0-3:7m)
illustrate the effective mean
for 14MHz tt Saami height; this is much lower in
height case (f)

Open-wire line

ianoletcon)

satisfactory, can also be used. Needless to say, this kind with very little adverse effect by the use of end loading
of operation should not be undertaken without climb- as in Fig 11.4(d). A 12ft (3-7m) square loop and a 1:1
ing experience or adequate advice and ample safety balun as shown present a good match to 75Q coaxial
precautions. It is essential for the constructor to be tied feeder, but further size reduction is possible with about
securely to the main tree trunk or a very thick branch Oft (2-7m) as the likely minimum, the swr then being of
for all such operations and trees should not be climbed the order of 1-5 using 50Q cable. This is recommended
in wet weather or with inadequate footgear. as a means of fitting a horizontally polarized antenna
The basic quad loop contains 1 of wire and the into a very short span without significant drop in
methods of feed shown at Fig 11.2(a), (b), (c) and (f) performance. For a 12ft (3-7m) square loop the drop in
are equally applicable in this case. However, the points gain, treating it as a broadside pair and referring to Fig
of connection for matched open-wire lines are some- 6.12, is found to be 0-5dB and, after allowing (say)
what further apart and require to be found by experi-
ment, the spacings indicated being rough guidelines
only. In case (a) or (b) a 4:1 balun and 50Q feeder may
be used; the swr is about 1-5 but can if desired be Lower half of loop.
improved by using a T-match as shown in Fig 11.5. The
required wire length is given very approximately in feet
by 990/f, assuming single-strand bare wire.
By open-circuiting the top of the loop as at (c) it
becomes a bi-square antenna, having its main reso-
nance at twice the frequency and a gain of 3~4dB as
explained on p98. There remains a resonance at the
lower frequency where the loop now operates as a kind
of vertically polarized W8JK antenna with the beam
direction rotated 90°. The dimensions are however
non-optimum; this is not normally used and has little to
recommend it beyond the “‘fact that it is there’? which Fig 11.5. Use of T-match for improved matching between a
may nevertheless be worth noting. Unlike the quad the single-loop antenna and a coaxial feeder. A 1:1 balun or a delta
connection may be substituted, dimensions being found by
bi-square is essentially monoband, though it will work experiment. Typical values for 14MHz are PQ 6—10ft (2-3m); C,
as a single element on other frequencies if various odd 70-150pF, though C is not required if the loop size is reduced
mixtures of polarization are accepted. somewhat. Alternatively adjustment of PQ and C can be used to
compensate for small variations in loop size. The method is
As explained in Chapter 5 (p84) the standard quad applicable in most instances where the radiation resistance of an
loop, Fig 11.4(a), can be considerably reduced in size antenna is not suited to available values of feeder impedance
 SINGLE-ELEMENT ANTENNAS 153

14swg than the dipole. Performance is equal at 80ft (24m) but

the delta loop is 1-7dB worse for a height of 20ft (6m).
It will be seen that Figs 11.1(b) and 11.4(e) are
Tubing tapers from
1%in to V2in (32 to 13mm) similar except for the angle between the wires which in
(for 21MHz elements) the first case is very small. An intermediate shape may
be preferred and a triangle with a right-angle at its base
is probably about optimum for this application, though
likely differences are no more than small fractions of a
Dimensions exaggerated to decibel. This form of delta loop is good also in respect
show construction
of visual impact, though because of the need for thicker
Tubing 2in dia (51mm)
(for 21MHz)
wire it is not quite equal in this respect to the inverted-
V.
In view of its historical importance, widespread use
and significance as a building block for more complex
systems it would have been difficult to ignore the
Fig 11.6. Self-supporting delta loop. Typically for 21MHz the sides simple horizontal dipole, Fig 11.1(a), but, except for
are made from aluminium alloy tubing, tapering from 1tin
(32mm) to in (13mm) diameter while the top may consist of 14 or
low-height applications, it is doubtful whether its inclu-
16swg (2 or 1-6mm) copper wire. The total length of conductor is sion in this chapter could be justified on grounds of
given in feet by 1,005/f(MHz) [4]. Feed (not shown) as in Fig 11.5 utility, bearing in mind the advantages of the delta
loop. Other versions of the delta loop, Figs 11.6 and
another 0-2dB for the short-dipole effect (p21), there is 11.7, require only a single support and these are even
still a slight gain compared with a i/2 dipole. more attractive assuming some increase in visibility to
Loops can be any shape. Figs 11.4(e) and (f) show be acceptable.
the popular delta loop which is very similar to the quad, In Fig 11.6 the sides of the delta loop can be regarded
although the gain is about 0-5dB less; the two sloping as a kind of mast extension and the effective height is
sides between them contribute one third of the radiated roughly 1/4 greater than the height of the support. The
field, the effective heights being given approximately feasibility of this will depend on the strength of the
by (h — 2/12) in case (e) and (h — 2/4) in case (f). The support and acceptance of an appreciable degree of
delta loop (e) can be suspended between two supports visual impact. Fig 11.7 is somewhat less ambitious, the
as in Fig 11.1(b); it is mechanically advantageous since mean height being roughly equal to that of the support.
the downward pull is not communicated to the horizon- The particular interest of this arrangement, which was
tal section and use can be made of a much shorter span. evolved by ZLIOI, lies in the method of erection
At 14MHz and low angles, assuming 40ft (12m) sup- which requires only a halyard over a pulley fixed to the
ports and an effective mean sag of 2ft (0-6m) for the top of the support.
horizontal dipole, the delta loop is only 0-5dB worse The delta loop with its apex upwards, Fig 11.4(f),
though commonly used at lower frequencies, suffers
Hoist up on halliard
i
from a big reduction in height and is much inferior for
dx communication to the top portion on its own,
connected as an inverted-V.
The radiation resistance of a delta loop is slightly less
than that of a quad but the same methods of feed can be
used.

End-fed wire antennas

Arrangements such as those illustrated in Fig 11.8
f_——— Low- impedance line
(though at one time commonplace) are now largely
obsolescent; nevertheless they still have important vir-
tues, particularly in regard to simplicity of construction
Sere connected in and the ease with which they can be improvised or
Aluminium tube extended parallel and fed here with
to near ground level 50 or 7582 coaxial cable
concealed. Virtually any bit of wire bent into any shape
can be used though, if possible, the length should be at
least 4/2, it should be as high as possible, and if it has to
be fitted into a small space the rules on p84 are
applicable.
Fig 11.7. Delta loop for 21/28MHz; this light, simple and unobtru- The feeding of this type of antenna is discussed in
sive antenna is hoisted into position by a halyard (CQ Magazine) Chapter 4 (p43) and is simplified by avoiding lengths
154 HF ANTENNAS FOR ALL LOCATIONS

in general effectiveness to a \/2 or A dipole at the same

height. In some cases there could be an increased risk
of picking up mains-borne interference or causing tvi.
At the higher frequencies performance is poor at
right-angles to the wire but there is likely to be some
gain in near-endwise directions, eg 2dB for a straight
133ft (40-5m) wire at 23MHz. Due to loss of energy by
radiation (or in the case of a thin wire, dissipation) as
the wave travels outwards along a single wire this is
partially terminated (p95), resulting in a somewhat
stronger signal at the higher frequencies in directions
away from the feedpoint.
Earlier remarks about insulation are applicable and it
is generally possible to use thin wire supported by
equally thin nylon or polythene cord; in addition to this
Variable LorC
there is no feeder to be concealed and for meeting ©
“invisible antenna’? requirements without loss of per-
formance the end-fed wire and the inverted-V are the
only serious contenders. The inverted-V may also be
Resonates with __ end-fed, though care should be exercised in such cases
counterpoise to ensure that as far as possible the current maximum is
at the highest point in the system.

|, At least A __|

Fig 11.8. End-fed single wire. Arrangement (a) is suitable for Vertical antennas
lengths of m/2+2/8 but (b) should be used if the length
approaches an odd multiple of 1/4. Suitable values for case (a) are Earlier discussion of vertical antennas (p134) led to
L = 85/f (MHz) and C (max) = 450/f (MHz), the units being mic- the conclusion that these work nearly as well with their
rohenrys and picofarads respectively. The loading inductance at
(b) tunes out the capacitive reactance if the antenna is short and “feet on the ground” as when raised up to the sort of
the capacitance removes the inductive reactance if the antenna is height that is commonly feasible. After exhausting the
long, reckoning (roughly) 300Q for a length error of \/16, and pro possibilities of vertical antennas at low height the only
rata. The counterpoise must present a reactance small compared
with the end-impedance of the antenna and can be very small (3 way to obtain better performance is a change to
or 4ft, approx 1m) if the antenna length is near n)/2. If the antenna horizontal antennas at a height of at least 30ft (9m) and
is near an odd multiple of )/4 the counterpoise should be at least perhaps considerably more; the situation may however
8, centre-connected and series-tuned to resonance with a suit-
able inductance become more complicated if it is not possible to
accommodate the span required for a horizontal anten-
which are near to 4/4, odd multiples of 4/4, or less than na or if the vertical antenna has to be raised to clear
\/4 since in these cases much more attention must be local obstructions or sources of interference. There was
given to the design of an efficient counterpoise. This also one slightly speculative idea (p142) requiring a
has to be tuned which is an added complexity in the vertical antenna to be erected at tree-top level. Despite
case of multiband operation. If a length of 133ft these exceptions it seems reasonable to suppose that in
(40-5m) can be achieved, this is ideal since the desired most cases the lower end of a vertical antenna will be
condition is met (nearly enough) for all bands (includ- more or less accessible and the recommendations below
ing 10, 18 and 24MHz) from 3-5MHz upwards. There is reflect some bias in this direction.
one drawback to this insofar as the antenna is brought Fig 11.9(a) illustrates a conventional vertical mono-
into the shack at a point of high rf voltage, which in a pole with a somewhat unconventional earth connection
few cases may lead to increased losses and coupling into based on the arguments presented in Chapter 4 (p43);
tv antennas or mains wiring. However, such couplings the shorter the antenna the more important it is to have
will be cancelled, at least in part, by equal voltages of a low-resistance earth connection. This determines the
opposite sign existing on a short counterpoise and the allowable size of the loading inductance which must
size or positioning of this may be important. have a loss resistance small compared with the radia-
Every case under this heading is different, the dimen- tion resistance of the antenna, and the value of induct-
sions and values given in Fig 11.8 being intended only ance in turn fixes a minimum size for the counterpoise.
as rough guidelines, though at least in case (a) they are VK3AM has been using a basically similar idea (often
not at all critical. In most cases, except perhaps those using small helices as counterpoises to short loaded
involving a lot of bends and close proximity to lossy verticals) in a successful project aimed at developing
objects, such antennas in their best directions are equal efficient antennas for use in confined spaces, including
 SINGLE-ELEMENT ANTENNAS 155

Aluminium
tube
(self-supporting) Aluminium tube
(self-supporting)

To shack

600 or 30082 balanced

or 4:1 balun into
coaxial cable
\

Balanced bo Pe

(a) (b) (co (d)

To terminal 'E'
of generator in
(a) or (b) above

Loading coil
600 or 3002 balanced
or 4:1 balun into
coaxial cable
\ Vertical
counterpoise

Witt: Conventional ground-plane

(e) (g) (h)

To'E' (j ) 2-radial ground-plane—

at low-frequencies use
© also asa dipole for high-
| TAC | angle short-skip contacts
2
To'E'

To*E*
(m) ips )
(k) To'E’
Ground-plane with shortened
radials

; (eA 02)
(i) 8 |

Fig 11.9. Vertical antennas. (a) Monopole with counterpoise; as antenna length is increased the size of counterpoise may be reduced as
for other end-fed antennas. (b) Top-loaded monopole; the “generator” represents some combination of transmitter, feeder and
matching system. (c), (d) Dipoles. (e), (f) Two forms of quad, insulators being added at points of zero current, which leaves the operation
unaffected but helps to keep high voltages away from the mast and, in the case of beam elements, is needed for directional switching;
(e) covers all bands 10-30MHz, (f) is monoband. (g) to (n) illustrate alternative forms of counterpoise for monopole elements with (I), (m),
and (n) requiring least space

small boats. When the length exceeds about 3/8 the close to d/2 the modified Zepp feed, Fig 11.10(d) or (e),
loading coil shown in Fig 11.9(a) becomes unnecessary may be used.
and the arrangement shown in Fig 11.8(a) may be used, Top loading as in Fig 11.9(b) allows much shorter
though this has the disadvantage of requiring a radiators to be used but the loading wire needs to be
weatherproof box. Alternatively if the length is very supported. Typically, for 14MHz a 6ft (1-8m) radiator
156 HF ANTENNAS FOR ALL LOCATIONS

somewhat arbitrary. Pursuing this line of thought a

stage further, Fig 11.9(c) and (d) will be instantly
recognized as dipoles but differ from (a) plus (g) only to
the extent of being more symmetrical. For a given
vertical extent without top loading, symmetry doubles
the radiation resistance, and (c) or (d) should therefore
be used in preference to (b) wherever possible; the
limitation here is the need for longer loading wires
when a whole dipole instead of a half-dipole has to be
fitted into the same vertical distance.
Apart from possible difficulty in finding room for the
longer wires there could be appreciable radiation from
them in a high-angle mode; this can be checked with
the help of the formula on p188. Fig 11.9(d) includes a
form of gamma match; the outer of the cable should be ~
connected to the electrical centre of the system which is
not quite the same as the physical centre, though the
difference is not significant provided the feeder is
non-resonant.
The main application for (a) and (c) is in the
important case of self-supporting antennas when top-
loading is not possible, (a) being applicable to A/4 and
(c) to 4/2 radiators. Minimum sizes for counterpoises
have not been fully investigated and the dimensions for
(1) and (n) are tentative.
Vertical radiators are frequently supported by stand-
off insulators mounted on wet wood which is a recipe
for large losses. A much better method of supporting a
vertical antenna from a wooden (or even a non-
resonant metal) mast is to use two wires which “billow
out”’ from the mast as in Fig 11.9(e) or (f); this may be
LMA ALE LLL LOE ALLL AALS IE
Zepp-fed as at (f) or symmetrically as at (e) which is
a derived from the DJ4VM quad [7], these providing the
basis for beam and multiband antennas described later.
spine

As it stands, this arrangement has some slight directiv-

Asymmetrical dipole developed by VK3AM for marine use. Only ity to the extent of a 2dB loss in the plane of the loop
2m high, this can be mounted on the cabin roof of a small boat. and 1dB gain at right-angles to this relative to a short
As set up in the photo and using only 60W p.e.p., good signal
reports were obtained from Europe dipole. In case (e) the feeder should be brought away at
right-angles if possible, otherwise it crosses a point of
could be hung from the centre of a 15ft (4-6m) horizon- high rf voltage, in which case it must be checked for
tal wire which makes no significant contribution to the balance and the measures described in Chapter 4 (p49)
radiation (p188); the radiation resistance is about 9Q so put into force if necessary.
that a loading coil resistance of 2-39 would cause a loss Fig 11.10 shows a number of different feeder
of only 1dB. For a Q of 150 at 14MHz this gives an arrangements, (a) being the standard method for a \/4
inductance of 4uH and a required length for the vertical antenna while (b) shows the use of a ferrite
counterpoise of about 16ft (4-9m). This example, in transformer to obtain a better match when the antenna
conjunction with the discussion of vertical polarization and feeder impedances are widely different as happens:
on pp133—-137, suggests the possibility of efficient dx for lengths other than /4. In the case of a A/4 vertical
communication even when height is severely restricted, antenna at a considerable height, a 4:1 impedance
though it should be as great as possible to achieve step-up ratio provides a good match to 72Q feeder.
maximum efficiency, possibly some height gain, and A method of connection to 300 or 600 lines is
shorter lengths for the top load and the counterpoise. shown at (c) while (d) and (e) are two versions of the
Alternative forms of counterpoise are shown in Fig modified Zepp feed (p46). Depending on personal
11.9(g) to (n) and it will be evident that a short vertical preference, either a balanced line (300 or 600Q) or a
one as in Fig 11.9(g) can be regarded alternatively as coaxial feeder with 4:1 balun may be connected to the
one half of a shortened dipole, thereby demonstrating points indicated. Mutual coupling (e) is preferred to
that the distrinction between monopole and dipole is direct connection as it reduces the extent to which any
 SINGLE-ELEMENT ANTENNAS 157

imbalance in the open-wire stubs can be communicated

to the feeder; the coupling is close to unity so that the
required length of stub is increased by at most a few
(a) (b)
inches. The feeder may also be mutually coupled as at
(f) into the counterpoise loading coil, instead of the
direct connection shown at (a); an example of this will Tuning reactance
eg as Fig 11.8(b)
be found in Chapter 13 (Fig 13.3) and another at the To
end of this chapter. transmitter

Alternative
tap Trifilar winding giving 9:4
Multiband antennas based on resonant impedance ratio but some
lines Loaded counterpoise
other ratios are possibie
(see Chap 4)
or A/4 radials-— or
The principles applicable to multiband operation were anything in between!
outlined in Chapter 7 with numerous examples. The
task here is to make a selection geared to the solution
of particular practical problems.
Although many different designs can be based on the
(c)
use of resonant feeders the best known is probably the
GSRV antenna, Fig 11.11(a) and (b). This derives from
the fact that a length of 102ft (31-1m) has been found
by many amateurs to be a good compromise for 4:1 balun
covering all hf bands, though this depends largely on 300 or
whether the lobes at the higher frequencies coincide 6008
balanced
with directions of particular interest. line

The GSRV antenna can be fed in either of the ways

illustrated in Fig 11.11. In the left-hand diagram, tuned | (e)
feeders (300 to 600Q2) are used all the way: on the right, 600 line or
Fig 11.11(b), the high-impedance feeder is 34ft (10-4m) 4:1 balun into
coaxial cable
long and is connected into a 72Q twin or coaxial cable.
At this junction, the antenna impedance is low on most
bands, as can be checked with the aid of Fig 4.32 for a
length of 34 + 51 = 85ft (25-9m).
It has previously been suggested that there is an
optimum height for this antenna, ie )/2 or 1A above
ground, but, although in some cases these heights may 50 or 752
provide better matching, the GSRV is no exception to or 4:1 balun =
the general rule that antennas should normally be into high-impedance
balanced line
erected as high as possible. Counterpoise
On 1-8MHz the two feeder wires are connected
Fig 11.10. End-feeding of vertical antennas. (a) }/4 monopole. (b)
together at the transmitter end, or the inner and outer of Addition of transformers for impedance matching. (c) Monopole
the coaxial cable joined, and the top plus “‘feeder”’ used with balanced line. (d), (e) End-feeding of dipole with unbalanced
as a Marconi antenna with a series-tuned coupling or balanced line. (f) Lumped-circuit equivalent of (d). (e) is
equivalent to Fig 11.9(c) but uses an indirect feeder connection, a
circuit and a good earth connection. loop of insulated wire (typically 3—4ft long) taped onto the end of
On the 3-5MHz band, the electrical centre of the the stub
antenna commences about 15ft (4-6m) down the open
line (in other words, the middle 30ft (9-1m) of the lowering the antenna slightly while observing the swr
dipole is folded up). The antenna functions as two on the 72Q twin-lead or coaxial feeder by means of an
half-waves in phase on 7MHz with a portion “folded” swr bridge, an excellent impedance match may be
at the centre. On these bands the termination is highly obtained on this band. If however low-angle radiation
reactive and the atu must of course be able to take care is required, height is all important and, as most cables
of this if the antenna is to load satisfactorily and radiate will withstand an swr of 2 or greater, any temptation to
effectively. improve the swr by lowering the antenna should be
At 14MHz the antenna functions as a 3A/2 antenna. resisted.
Since the impedance at the centre is about 100Q, a On 21MHz, the antenna works as a slightly extended
satisfactory match to the 72Q feeder is obtained via the 2X system or two full-waves in phase and is capable of
34ft (10-4m) of 4/2 stub. By making the height 1/2 very good results, especially if open-wire feeders are
or 14 above ground at 14MHz and then raising and used to reduce loss. On 28MHz it consists of two 1-5A
158 HF ANTENNAS FOR ALL LOCATIONS

5ift Sift | Sift >|

Sift 1 |
' ' (15:5m ) peat
1 (15-5m) aS

Matching stub 34ft

open- wire feeder
Convenient length of open-wire
Suitable feeder taps feeder (preferably an even or
or 29ft 6in
selected for optimum of 300 ribbon
odd multiple of quarter
loading on each Sa wavelength at 14MHz)

To transmitter via
7282 coaxial cable antenna tuning unit
to transmitter (b )
(a) Any length of 7582 twin-lead
(up to maximum of 100ft)
or 8082 coaxial cable

Linear resonator open-circuits top at

28MHz but 'disappears' at 21MHz
and below

7ft
v4 YY

(c) 3 to4in
12ft to17ft
(3-7 to 5:2m) (7-5 to10cm)
spacing

Stub
: switched- Line matched at lowest
in by relay frequency by
switching inthe
stub
For 28MHz move
(213m) feeder with balun to
this point and short-
circuit the lower end
(e) of the resonant line

0-2 toO-6A

Feed here
with 60082 line
: or here with
4
50 line and 4:1
balun (swr =1-5)

2Oft :
(6m)
Resonant “
feed line

(h)

(g)

60082 line

Fig 11.11. Multiband systems based on resonant lines. (a) and (b) are two versions of the G5RV antenna. (c) When an antenna is used well
below its normal resonance or if it is a long way from the shack, losses in a resonant feeder tend to be excessive and the bandwidth
narrows; to overcome this a matching stub may be switched in as shown. (d) Quad loops with resonant feeder, L = 70ft (21-3m), can be
used on all bands from 7-28MHz, though some additional tuning will be required for 10, 18 and 24MHz, and 7MHz is only practicable with
the larger loop size. (e) Addition of linear resonator to (d) to provide bi-square operation with extra gain at 28MHz; C ~8pF. (f)
Compromise” delta loop. (g) 14/21MHz folded-dipole with coil to tune out the capacitive reactance of the two half-dipoles at 21MHz. (h)
Horizontally polarized (original) version of DJ4VM quad loop
 SINGLE-ELEMENT ANTENNAS 159

in-line antennas fed in phase. Here again, results are particularly good multiband resonator as explained in
better with a tuned feeder to minimize losses, although Chapter 7, p106, and good results have been obtained
satisfactory results have been claimed for the 34ft on the 7, 14, 21 and 23MHz bands using the arrangement
(10-4m) stub and 72 feeder. shown in Fig 11.11(d), though swr (not measured) could
When using tuned feeders, it is recommended that have been fairly high on 7MHz and 28MHz. Neverthe-
the feeder taps should be adjusted experimentally to less, results included 28MHz ssb contacts with North
obtain optimum loading on each band using separate America using only 300mW on each of three occasions
plug-in or switched coils. Connection from the atu to when it was attempted and, with full power on 7MHz,
the transmitter should be made with 72Q coaxial cable S8 reports from Australia. There are however a num-
in which a tvi suppression (low-pass) filter may be ber of points to be observed:
inserted. (a) Resonances do not quite coincide for all bands and
When using tuned feeders there is of course no compensating circuits, Fig 7.2(b), may be used to
particular merit in the length of 102ft (31-1m). It should achieve this for 14 and 21MHz.
also be noted that the radiation pattern is of the general (b) Because of the very low radiation resistance on
long-wire type and the position of lobes and nulls will 7MHz the wire gauge should be as heavy as
vary with length and frequency. possible if operation is required on this band, and
A length of 102ft requiring two supports is not particular care must be taken to avoid any high-
always convenient, particularly if due regard is paid resistance connections. In the case of a 14/21MHz
to the mechanical considerations on p150. In principle beam (p179) both loops should be connected in
the GSRV antenna could be erected as an inverted-V parallel for use as a single loop on 7MHz, efficient
but in view of its length the ends will then be close to operation as a beam being out of the question.
the ground, the effective height reduced and dx per- (c) The line of fire for a single 14MHz loop used on
formance impaired, except that the slope of the arms 28MHz is in the plane of the loop with a basic
will result in vertically polarized radiation for some end-fire gain of 3dB, from which must be sub-
directions and frequencies. In particular there will be a tracted 3dB due to radiation of half the energy in a
vertically polarized ‘““W8JK mode” in the endwise vertically-upwards mode, leaving the gain roughly
direction on 7MHz which could be quite useful, and on equal to that of a dipole.
14MHz probably about half the power will be radiated (d) By using resonant feeders, preferably with match-
in a two-element wide-spaced vertical end-fire mode. ing stubs as in Fig 11.11(c), operation can be
For operation on all bands inclusive of 10, 18 and extended to include 10, 18 and 24MHz.
24MHz open-wire feeder should be used throughout. (e Operation of the loop plus the 4/2 resonant feeder
—

In this case the top can be reduced to about 60-70ft as a grounded i/4 vertical for 3-SMHz is feasible in
(18-21m) without serious drop in efficiency at 3-SMHz this case also.
provided heavy-gauge wire is used. However, unless (f) Operation as a bi-square beam with 3dB gain is
plentyof height or sloping ground is available it will possible at 283MHz by open-circuiting the top of the
certainly be much better for dx only on the lower loop. This has been done using a linear resonator,
frequencies to excite the feeder as a top-loaded vertical Fig 11.11(e), with little disturbance of operation at
radiator operating against a suitable artificial ground the other frequencies but for connection to a
(Fig 11.9) or, if near )/2 resonance, voltage-fed on the matched line the feeder system at the lower end
lines of Fig 11.10(e) or (f) or 11.9(f). It may be noted must be altered as shown in the figure. The band-
that with a length of 102ft the pattern at 1OMHz will width is fairly narrow, about 300—400kHz; if this is
consist of narrow lobes at right-angles to the wire, with insufficient it may be possible to dispense with the
a gain of 2dB. At higher frequencies the pattern tends resonator and use a relay. This would however
to favour the endwise directions with perhaps 2—3dB require thorough testing at full power prior to
gain. erection, with the loop laid out horizontally a few
If the open-wire feeder is a long one it may be feet from the ground.
necessary or at least desirable to connect appropriate (g) The loop can be reduced in size to about 12ft
matching stubs for one or more bands as in Fig (3:7m) square, keeping the total electrical length
11.11(c). This can if desired be done from the shack by (Fig 11.11(b)) constant with little or no loss at
means of relays. In this case only the lengths of line 14MHz. Even as a beam element buried in a tree,
from the stubs to the antenna are resonant, bandwidth good results have been obtained with such loops,
may be greatly improved, and losses can be held to an though a slight loss was suspected compared with
acceptable level for line length up to 200yd (180m) or the use of a full-size element. In this case most of
more. Relays as shown have to handle large currents at the radiation comes from the top half of the loop
the lowest frequency, high voltages at higher frequen- and in theory for typical heights and in the clear
cies, and must have low capacitance. this should at least neutralize any slight drop in
Contrary to general belief the quad loop makes a gain.
160 HF ANTENNAS FOR ALL LOCATIONS

The delta loop may be treated in a similar manner, — Dimensions as Fig 11.11 (e)
but up to the time of writing has not been studied in the for 28MHz
C is roughly doubled
same depth and there may be appreciable differences. for 21MHz
The “compromise” delta loop, Fig 11.11(f), (despite
some fairly strong end-wise radiation) radiates most of 17ft (5°2m)
its power in the normal direction over the whole for 14MHz

frequency range from 14-28MHz and could also be

used at 10MHz on the same basis as the standard
14MHz quad loop at 7MHz. The bandwidth and effi-
ciency :or this mode has been calculated in Chapter 8,
p123. The radiation resistance at 14MHz is about 70Q.
In view of its smaller size it could be attractive as an 300 or 60082 line or
alternative version of the loop in Fig 11.6, though for 4:1 balun plus coaxial cable

monoband 14MHz operation into a matched line the

Fig 11.12. The application of linear resonators to provide two-
loop needs to be loaded by an open-wire stub about band operation of loops. This can be adapted for any shape of .
12ft (3-7m) in length. resonant loop
In the case of a folded dipole or “squashed delta’,
Fig 11.3(b), there is virtually no radiation at the second however also explained that when the element is folded
harmonic unless the top side is open-circuited, eg by or otherwise modified to fit it into a smaller space any
the same method as that used in Fig 11.11(e). A 14MHz attempt to use the whole of it results in a lot of energy
folded dipole can, however, be used efficiently on being radiated in unwanted directions.
21MHz where it is not far from resonance, its radiation So far in this chapter comparatively little use has
pattern being much the same as at 14MHz. For a total been found for the straight dipole, but reference has
electrical length of 14 at 14MHz, as defined by Fig already been made to the use of linear resonators to
11.11(d) or (e), it is necessary to shorten the system by increase the versatility of loops and, by using two such
about 4ft (1-2m) to retain resonance at 21MHz, due to resonators as in Fig 11.12, two-band operation can be
the capacitive reactance of the two halves of the dipole achieved with a non-resonant line. A more elaborate
acting in series. This can alternatively be tuned out by form of resonator (p111 and Fig 7.9) allows three-band
an inductance across YY of about 7-S5uH (Fig 11.11(f)), operation but it is difficult to see this being extended to
but an exact value cannot be given since it is rather cover additional frequencies and a more elegant solu-
large and influenced by its own self-capacitance, which tion would make use of variable capacitors. Given the
in turn depends on precise constructional details; these genius of Heath Robinson plus a well-stocked junk box
are not available, the author’s experiments having got this should present no difficulty, but until some manu-
no further than establishing the principle. This induct- facturer is enterprising enough to market suitable ser-
ance has no effect at the fundamental resonance. vo-controlled variable capacitors most of us will have to
An alternative method [7] of multibanding the quad be content with something inferior. Nevertheless, the
loop, due to DJ4VM, is shown in Fig 11.11(h) which is availability of very small and light relays with short
basically similar to Fig 11.9(e). This relies on resonant leads and 5A contacts makes it possible to switch
feeders but has the advantage of ensuring optimum capacitances and add one band per relay per element as
current distribution and a “‘clean”’ radiation pattern at discussed on Chapter 7, p116. There a preference is
all frequencies, the antenna being completely symmet- expressed for hybrid options which combine stacking
rical. All bands (including 10, 18 and 24MHz) from with the use of linear resonators and relays.
7MHz upwards can be covered as previously de- Objections to the use of traps were recorded in
scribed, but in this case the loop size can be increased Chapter 7 but, though not favoured in general, they
with advantage to 21ft (6-4m) square; this increases were shown to have some merit for the multibanding of
gain slightly at the higher frequencies and, at the lower inverted-V dipoles. A very common situation is illus-
frequencies, improves the bandwidth and allows the trated in Fig 11.13(a), which shows how a 3-5MHz
use of somewhat thinner wire without detriment to the inverted-V dipole can be adapted as a dx antenna on
efficiency. one or more higher-frequency bands. Isolating the top
portion in this way increases the effective height and —
provides a simple solution to the matching problem.
Multiband operation without resonant
Traps of the type illustrated are available as proprietary
lines items, and in view of the problems of waterproofing
Earlier discussion (Chapter 7, p117) has emphasized the plus the need to maintain the highest possible value of
importance, in the case of, say, a conventional full- Q at all times, purchase is recommended rather than
sized dipole or quad beam element for 14MHzZ, of using construction.
the whole of it at the higher frequencies. It was An unusual form of trap using linear resonators is
 SINGLE-ELEMENT ANTENNAS 161

illustrated in Fig 11.13(b) and avoids the need for

weatherproofing, except that depending on their con- Typically 120ft (36-6m)
struction it may be necessary to give the capacitors a depending
design
on trap

protective coating of Araldite. It is also light and low

loss but the bandwidth is somewhat narrower than with
conventional traps. Exact design data is not available
but guidelines are indicated and the idea may commend
itself to experimenters.
The multiband dipole shown in Fig 7.6 has the
advantage of eliminating trap losses as well as being
easy to construct; moreover there is, in principle, no
definite limit to the number of dipoles which may be
connected in parallel so that it could be of particular
interest for adding 10, 18 and 24MHz band coverage.
_ Nevertheless, as pointed out in Chapter 7, interactions
130ft (40m) approx
between dipoles can occur, particularly in the form of
28MHz current flowing in the 21MHz element. With
more dipoles more such effects will manifest them-
selves but up to a point these can be controlled by the
use of capacitors as illustrated in Fig 7.6(b).

10ft (3m) long

Multiband vertical antennas spaced 1 ft (30cm)
Assuming access to the lower end of a vertical mono-
pole antenna as previously discussed it is possible to
tune it over a frequency range of at least 3 to 1 Fig 11.13. Inverted-V dipoles for 14/3-5MHz. (a) Conventional
without incurring significant losses or seriously upset- traps. (b) Linear traps. The principle can be extended to provide
coverage of some additional wavebands; estimated losses are
ting the radiation pattern in the vertical plane. All that 0-7dB at 14MHz, virtually nil at 3-5MHz
is required therefore is a suitable tuning box and an
appropriate earth connection for each frequency. Since amplitude. Capacitive stretching results in the equiva-
all the recommended “earths” are frequency selective lent of a collinear array and hence extra gain at the
the choice lies between a single earth, the tuning of higher frequencies.
which is linked with that of the antenna, Fig 11.14(a), Tuning units as in (a) may be switched remotely by
or a set of earths individually resonated to include all relays or Ledex switches but for two- or three-band
the desired bands, one common version of this being operation within a range of just over one octave the
the multiple ground plane shown in Fig 11.14(b). A linear resonator can be used as in (b) or (c), subject to
rather similar option is the multiple counterpoise of Fig acceptance of some reduction in bandwidth at the
11.14(c). These also illustrate the use of a linear highest frequency.
resonator as a tuning unit, though this is not recom- Some readers will no doubt be prepared to accept
mended for a tuning range much in excess of one sacrifices in return for the extra convenience of trapped
octave. verticals which are available as proprietary items cover-
The most usual method, and probably the easiest, for ing all or most of the hf bands. In comparison the
coverage of a frequency range greater than 23 or 3 to 1 stretched or collinear vertical antenna produces much
is the use of traps as in Fig 11.14(d), but there are less radiation at higher angles, so that in addition to
interesting possibilities for the experimenter based on some extra gain for dx there will be less short-skip
the use of capacitive stretching (p199) of the antenna at interference and less cosmic noise. Conversely the
the highest frequencies by means of linear resonators as trapped vertical gives better performance for medium
in Fig 11.14(e). These should preferably be located range working, though as previously emphasized
within 4/8 of points of maximum current. The required (p145) all vertical antennas (unless the feeder is radiat-
design procedure is to lay the antenna out horizontally ing) result in poor performance over shorter distances.
within reach of the ground and check the current
distribution along it as C is varied. The value of J,
which is equal for the resonator sections to ([c — [;), Indoor antennas
should be as constant as possible. However, except at The choice of a single element or beam for an indoor
the end of the antenna there must not be a zero since antenna depends on the interplay of a number of
this would imply a phase reversal and the object of the influences which are different in each case and can
exercise is the maintenance of constant phase and probably be resolved only by experiment, though some
162 HF ANTENNAS FOR ALL LOCATIONS

This can be
Fig 11.9(a) or (b) (b)
(a) in conjunction Approx one linear resonator
with Fig 11.10 Ic per half wavelength
Tuning (b), (c) or (e)
box and
matching 3 to4ft (1m)
unit for 14/28MHz I=Ic-l?@

Coaxial Coaxial cable

cable to transmitter
to transmitter
Ic
3-band radial system
(showing 2 only of the
3 or 4 radials)

To separate tuners
and radial systems for
lower frequencies

25 SESS = Coaxial cable to

ahead transmitter
(c)

Matching
Short counterpoise

(e)

To multiple radials or
counterpoises as at (b) or(c)

Fig 11.14. Multiband vertical antennas. (a) Use of variable L and C for covering a frequency range of 1-2 octaves. (b) Ground plane uses
switched tuning unit based on linear resonator with a separate set of radials for each band. (c) Equivalent arrangement using
counterpoises. (d) Use of traps. (e) Use of linear resonators to provide capacitive stretching whereby the whole of the antenna is used at
the highest frequency; a 7MHz monopole can be used as two half-waves in phase at 28MHz, operation on 7MHz being unaffected

guidelines will be indicated in the following pages. An any adjacent “‘lossy’’ material will introduce severe
indoor antenna is usually a last resort, but unless one is losses. It is possible that in some cases a beam such as
unfortunate enough to live in a steel-framed building it Fig 12.3 may give good results, at least in dry weather,
should be possible to radiate a useful and even in some but the experimenter with limited time at his disposal
cases a fully competitive signal. For best results the would probably do better to use a single element
guidelines in Chapter 10 (p144) should be strictly fol- for transmission with perhaps (when such techniques
lowed, each case being treated on its merits and are more fully developed) a separate small ‘‘active”
ready-made solutions regarded with suspicion. In the antenna for nulling out interference in reception (p131).
case of a building with three or more storeys and an This advice could, however, be at fault since there
accessible roof space (not too severely cluttered by seems to be no body of reliable experience on which to
plumbing and electric wiring), a horizontal dipole with draw. An ideal single element is probably a small or
end-loading as necessary in accordance with Fig 11.1(e) rectangular “‘quad loop” lying down in a horizontal
to (j) is likely to provide the best answer. position; this may be ‘“‘end-loaded”’ capacitively as in
In some very favourable cases it may be possible to Fig 11.4(d); the radiation resistance is fairly high, being
accommodate a fixed but reversible beam (even, in the 72Q at 14MHz for a loop 12ft (3-7m) square, and rf
case of a few very large houses, a rotary beam for 21, 24 voltages are correspondingly low. It therefore looks
or 28MHz) but good results are not in general to be like an ideal antenna for fitting into a small roof space
expected if a small beam is fitted into a space only just but spacing between the sides should not exceed 12ft
large enough to accommodate it. This is because very otherwise too much energy is directed upwards, to the
high rf voltages exist at the ends of the elements and extent that with 17ft (5-2m) spacing the gain is 2dB
 SINGLE-ELEMENT ANTENNAS 163

Part of house wiring which has already shown up in many different forms.

Pee
ice ore ety
Although in principle this could perhaps be favourably
exploited in a few cases, to do so would be decidedly
Preferably at least Insulated loop tuned to
tricky and possibly even involve some fire risks. The
6 to12in (15 to 30cm) PEGMENICY, 145 only sound rule is to ensure that the rf is confined to the
(a) | The ‘longer the better’ | antenna system. Unfortunately in the case of a typical
—up toat least 4/4 small home any resonant wire or pipe, unless very close
to the ground (eg under the downstair floorboards) or
Part of plumbing
running at right-angles, can be expected to show some
coupling, and in consequence current will flow in what
(b) | Vad | is likely to be a very lossy system. As well as being
dimensions as (a) dissipated, however, power will almost certainly be
radiated in unwanted directions and some of it carried
Fig 11.15. (a) Use of linear trap to suppress an unwanted reso- to tv sets in adjoining houses or other undesirable
nance. (b) How this result may be achieved without making any
; electrical connection to the conductor destinations. A short list of such conductors includes:
Mains wiring
down on a dipole. With additional folds or inductive
Metal conduit enclosing mains wires
loading it could be reduced in size to 8ft (2-4m) square
Central heating pipes
or even less, though if this is the only space available
Hot and cold water pipes
some loss in performance is almost certain to ensue.
Telephone wires
The “odd bit of wire” type of antenna cannot be
TV antenna feeders
ruled out even for indoor applications, and in the case
Bell wires
of a large house a long wire wandering through the
various roof spaces and ending up if necessary in a Extension speaker leads
Gas pipes
number of folds to provide end loading may be a good
choice, at least for the low bands, if rf can be kept Resonances can usually be located with a gdo having
out of the mains wiring, eg with the help of the a coil of sufficiently large area as described in Chapter
measures discussed below. 18; this should be held against two or three well-
In good conditions some dx communication should separated points on each of the conductors under
be possible with nothing more than a dipole in the roof suspicion, the reason for more than one point of
space of a bungalow and, if the house or bungalow is on application being failure of resonances to show up on
a steep hillside, outstanding results in the downhill the gdo if the test is made at a current node. Some
direction can be expected with no more than this. amateurs have apparently had success in removing
Failing such a slope or a tall house, the advice to try resonances by the use of plug-in capacitors at various
vertical polarization if possible before finalizing the mains outlets [5], but although this may shift the
antenna system holds for the indoor case also. frequency it seems likely in many cases to have no
Absorption effects may be worse in the vertical case, effect. This is because wires going to an outlet tend to
but on the other hand a vertical antenna in an attic be effectively in parallel, whereas plugging in an ap-
space, even if it has to project down through the ceiling pliance having a suitable length of lead, or just an odd
into an upstairs room, is less likely to find itself tightly bit of lead ending in a terminal block, must inevitably
coupled into the electric wiring system. A vertical at have some effect on resonances.
ground level may experience more absorption but Another useful trick is to put a trap in the wire which
otherwise should be nearly as good as one in the attic. is causing the trouble, but it is not necessary to cut into
Methods of shortening vertical antennas are discussed the, wire in order to do this. Instead, a linear-resonator
on p154 and the possibility of a phased pair of fairly type of trap can be made from a short length of
widely-spaced verticals may be worth considering. One open-wire line tuned to the required frequency, Fig
antenna particularly favoured by constructors of indoor 11.15(a). One wire of this line is then taped onto the
antennas is the W8JK beam, but this is in fact the conductor in question but without breaking into the
“worst possible” choice. This is because for its size the insulation. In the case of a pipe (not gas pipe), how-
W8JK has (p86) the /owest radiation resistance which it ever, direct contact should be made to it (eg with
is physically possible to achieve, and correspondingly earthing clips) so that it becomes one conductor of the
high voltages. In contrast the single-element loop dis- line, the other conductor of which should preferably
cussed above minimizes the losses. also be tubing, though the diameters need not be equal
This brings us to what is perhaps the real crux of (Fig 11.15(b)). This idea has many applications.
designing a good indoor antenna system: keeping rf It is particularly important in the case of indoor
currents out of other wiring. We have here yet another antennas to make sure that any earth current associated
example of the problem of coupling between wires with the antenna system is provided with a suitable
164 HF ANTENNAS FOR ALL LOCATIONS

return path. /t must not go through the mains wiring, for the lower frequencies as shown for example in
and this includes the mains earth. The problem is Fig 7.12, in which case a loaded counterpoise such
basically the same as in the case of an outside end-fed as Fig 11.9 (i) to (n) may be substituted for the
antenna and arises also to a greater or lesser extent radials. Though not always feasible in practice,
whenever there is any loss of balance in the antenna there is no technical objection to mounting two
system. An indoor antenna, even if of a balanced type such multiband dipoles on the same mast. If
and fed with a balun, may become unbalanced due to possible, before erection the dipoles should be
proximity to the other conductors in the house, which laid out horizontally within reach of the ground
may still affect balance even though resonance has been (but as high as possible) and checked for reso-
averted. Whatever earthing system is adopted must be nance, matching and adequacy of band coverage.
regarded as part of the antenna and cannot be consi- If this is restricted the fault is probably due to
dered in isolation. interaction involving the next lower frequency
elements which may be open-circuited by a relay
or as in Fig 7.6(b) by a linear resonator. Relays
Conclusions and recommendations may also be used to switch in series tuning capaci-
This chapter has provided a wide choice of single- | tance allowing the 14 and 21MHz dipoles to
element ideas from which it is hoped that readers will operate on 18 and 24MHz.
be able to select those best suited to their needs or by . For mounting in trees, use a quad loop with a
which they may perhaps be inspired to think of better resonant feedline extending at least to within
ones. It is felt nevertheless that some readers will be reach from the ground, along the lines of Fig
confused by the amount of choice available and may be 11.11(d) or (e), but consider the possibility of
helped by the author’s personal selection of “best replacing the linear resonator by a relay as discus-
buys’’. sed on p159. Note that this also can be excited as a
1. The end-fed single wire has fewest problems of top-loaded vertical for lower frequency bands. By
erection and, subject to a few reservations (see suitable tuning provisions at ground level 10; 18
earlier text), can give a very good account of and 24MHz can be covered.
itself. Typically, one fixes a tv-type antenna mast . For suspending between trees or masts use a
to the chimney, runs a wire up to this from the ‘“squashed”’ delta loop as in Fig. 11.1(b), (f), or
shack window and then out to any convenient 11.11(g) noting that the multibanding arrange-
support as far away as possible, 133ft (40m) ments described above apply equally to 11.1(b)
being a desirable length. With a similar arrange- and (f). Fig 11.11(g) provides only for two-band
ment during the mid-’thirties, using all bands from operation unless the inductance is switched out of
3-5 to 28MHz, the author almost equalled the circuit or capacitors switched across it as required.
previous year’s winning score in the W/VE con- The delta loop in its familiar equilateral form is
test. Though not remarkable in view of improved slightly down in performance due to the reduc-
conditions, this result was good enough to inspire tion in mean height; on the other hand, when
considerable respect for this type of antenna, compressed into one of its more orthodox
particularly as a W8JK beam at the same mean folded-dipole forms it becomes less
height (about 35ft or 11m) and a two-element attractive mechanically.
vertical beam failed to produce any improvement . A vertical monopole based on Fig. 11.10(a) as
in signal reports. A lot depends of course on shown in Fig. 11.16 which includes dimensions
whether the wire runs in an “‘interesting”’ direc- found by experiment, though mutual coupling of
tion, and a tall tree at the mid-point would the feeder (see p44) may be preferable. This may
probably be an improvement over the tv mast. If be erected at ground level but an improvement of
tvi is a problem several feet of the mains cord at least half an S-unit can be expected if the base
should be wound around as much ferrite rod as height is raised to about 0-2 or some 14ft (4m) at
possible and either a short loaded counterpoise or 14MHz. In the author’s case no further improve-
a single “‘A/4 radial” wire used as an earth connec- ment was found for heights up to 35ft (10-7m),
tion for the rig. nor would much be expected, and at this height
2. The multiband inverted-V dipole, Fig 11.2(b). improvement of about one S-unit in dx signal
This is usually the way to achieve the greatest strength resulted from the use of horizontal polar-
possible effective height on the most important dx ization as discussed in Chapter 10. Electrically,
bands. Although, as usually pointed out, a 7MHz despite the lack of symmetry, this antenna can be
dipole may be used on 21MHz, this is a comprom- regarded as a dipole. However, the ‘“‘monopole”’
ise and for best results on 21MHz it is necessary to description is appropriate insofar as it has the
omit 7MHz. However, the feeder of a 14-28MHz physical appearance and practical advantages of a
system may be operated as a top-loaded vertical monopole, such as ease of erection and ready
 SINGLE-ELEMENT ANTENNAS 165

other antennas qualify for nomination as “‘best buys’’.

These include the delta-matched inverted-V; the two-
band delta loops of Fig 11.7; short end-loaded elements
such as Fig 11.1(j) or (m); symmetrically end-loaded
Dural tube tapering from approx -/4 to Ya dia
elements such as Fig 11.1(g) or (i) in wire form
mounted vertically and slung between poles; and the
A|>
“top-hat loaded” quad loop of Fig 11.4(d), which
allows an antenna to be accommodated in the shortest
Plastic tube 'plugs-in' tothe top possible span without significant loss of efficiency or
of a ground stake
bandwidth.
Linear loading inductance is approx
11ft of 14swg wire for driven element
8"spacers

References
[1] ‘Invisible antennas”, T. J. Gordon, W6RVQ, OST
$$ $$$ 7#t ——____________+ November 1965, p87.
Fig 11.16. Constructional details of monopole with short induc- [2] The ARRL Antenna Book, 13th edn, ARRL, 1974,
tively-loaded counterpoise. The antenna may be fed with 500 p265.
coaxial cable at point X (outer to counterpoise) but do not expect
an swr of less than about two unless additional impedance
[3] H. R. Habig, K8ANV, The ARRL Antenna
transformation is used. Alternatively the feeder may be con- Anthology, ARRL, 1978, p84.
nected through a balun to a series-tuned triangular coupling coil [4] CQ May 1976.
placed in the angle at X
[5] Wire Antennas, W. I. Orr, W6SAI, Radio Publica-
access to the feedpoint. Up to the time of writing, tions Inc, 1972, p127.
multiband operation has not been tried but, along [6] ““Shunt feeding towers for operation on the lower
the lines of Fig. 11.14(c), should present little amateur frequencies”, B. A. Boothe, W9UCW,
difficulty for coverage of all bands from 14 to The ARRL Antenna Anthology, ARRL, 1978, p34.
28MHz and probably also the 10MHz band. In [7] “‘A new multiband quad antenna”, W. Boldt,
setting up this type of antenna there are important DJ4VM, Ham Radio August 1969.
practical points to be noted and particular care [8] ‘Aerial masts and rotation systems”, R. Thornton
must be taken to avoid feeder radiation (Chapter and W. H. Allen, Radio Communication August
4, p49). 1972.
[9] The ARRL Antenna Book, 12th edn, ARRL, 1970,
If operation is restricted to one or two bands some p276.
CHAPTER 12

Horizontal beams

The choice of horizontal beams is almost unlimited but Under this heading there is a choice of (a)
two types of rotary beam, the three-element trapped two-element Yagi arrays or (b) loops which
Yagi and the two- or three-element quad, seem to have provide the smallest possible diameter of turning
a clear lead in the popularity stakes. Despite this, the circle together with slightly improved gain and
reader who has persevered with Chapters 5 and 7 will bandwidth. This is at the price of much larger
not expect further consideration of trapped beams, and vertical dimensions which in some cases may
the quad does not emerge as the wisest choice for a aggravate mechanical and visual-impact difficul-
rotary beam since it poses major problems of weight, ties.
windage and visual impact. Nevertheless it must be . A “best possible”’ beam needs to be instan-
admitted that in terms of signal strength the shortcom- taneously reversible to reduce time wasted in
ings of trapped beams add up to something less than the beam rotation and for added convenience in net
average differences between locations, by no means all operation.
the big dx signals come from beams without traps, and . Size can be reduced if arrangements are made to
for large numbers of amateurs who need to buy a allow fine tuning of elements from the shack.
multiband antenna “off the shelf’ the only current . It should ideally be possible to establish different
alternative is the three-band quad. Some traps and directional patterns for transmission and recep-
some types of trapped beam appear to be much more tion according to the needs of the moment.
reliable than others and if it is essential to purchase a . Keeping the size as small as possible may enable
proprietary article it would be wise to seek the experi- height to be increased, thereby obtaining better
ence of other amateurs at club meetings or over the air. performance than would be possible with a lar-
The selection of beams for inclusion in this chapter is ger beam.
based on the following considerations: . The quad (at least in its usual rotary form) is
1. Erection of a “‘full-sized”’ three- or four-element considerably worse than a three-element Yagi in
rotary Yagi has proved feasible for many terms of windage and provides no more gain.
amateurs, but with increasing size the difficulties Three elements are needed to achieve the receiv-
escalate rather rapidly. The use of very large ing advantages illustrated in Fig 5.15, but the
beams tends to be limited to cases where there is third element of the quad is largely ineffective
plenty of space available combined with a special both in this respect and as regards gain because
interest in one (or a small number) of fixed of the relatively low current as explained in
directions; these are further discussed in Chapter Chapter 5. On the other hand, it adds even more
14, to the windage and, although the electrical de-
2. Full-sized elements can be reduced in length by ficiency might be remedied eg by the use of an
about one-third with almost no change in any extra feeder, suitable designs are not currently
aspect of performance. The only real advantage available and would add to the complexity.
of retaining the full length lies in the possibility . There are obvious advantages, including the
of extra gain at the second-harmonic frequency possibility of large cost savings, from the use of
provided no traps are used. beam-direction switching rather than rotation.
3. Requirements for “smallest possible’? beams For all-round coverage (albeit with reduced per-
arise because of limited space, or in order to formance in some directions) at least two reversi-
reduce the visual impact of supporting structures ble beams are required. Fixed beams ten also to
which need to be sturdily built to withstand the be more easily concealed.
weight and windage of large beams. ‘“‘Smallest 10. In this chapter the lower-frequency bands are
possible” is interpreted here as permitting a drop not specifically considered since in most cases
in gain of not more than 2dB relative to a requirements will best be met by vertical arrays
full-size three-element monoband Yagi, more for dx with horizontal dipoles for shorter ranges.
loss than this being regarded as unacceptable. However, the smaller beams described here can

166
 HORIZONTAL BEAMS 167

c Element lengths given

podb
7ft
(a) in Figs 5.16 plus 5.17 (2-1m)
About 10pF
SS (c) 1 for 28MHz ! }

Aluminium alloy boom about || | ||E10in (25-4ern)

2in (5cm) dia A typical

x x
pe:SR
(1-5m)
34 ft
Feed with delta match (10-4m)
at points x-x attachment torotator- see (d)
for method of fixing

Simple device for obtaining operation at one higher frequency

band with extra gain and without affecting performance at the
lower frequency 1/4 in (6mm) thick
plate

(b) | eey | Boom

X-A..... 36in (91cm)

cm Cm.... 150pF
> (not critical)
A
4:1 balun
Serrated yokes
Mast
50 feeder

Fig 12.1. Full-size three-element beam based on Figs 5.16 and 5.17. For two-band operation with increased gain at the higher frequency
linear resonators may be used as shown dotted. Typical designs for 14MHz use 1-5in (38mm) diameter alloy tubing, tapering in three or
four sections to 0-5in (13mm) but smaller diameters may be used for higher frequencies or with cord bracing. Open-wire feeder is
recommended but (b) shows a delta match suitable for use with 50Q coaxial line. Points of connection for 600Q may be slightly further
apart or the hybrid system (p44) may be used. Dimensions of linear resonator for 14/28MHz are shown at (c); the capacitor has no effect
at 14MHz but for 21MHz it has to be roughly doubled and element lengths must be reduced 2-5 per cent. For arrays without bracing the
boom should have an outside diameter of about 2in (5cm) and wall thickness of 0-2in (5mm). One method of attaching elements to the
boom and the boom to the mast is shown at (d), taken from the ARAL Antenna Book

be readily adapted for use on 7 or 10MHz, and close approach to infinite front/back ratio. Two
‘should give a good account of themselves at feeders are advised to permit precise tuning and
heights upwards of 40ft (12m) or over sloping instantaneous beam reversal from the shack.
ground. Despite the greatly reduced space occupied by
most of these arrays the first two are equal in gain
The beams to be described are of six main types:
and only slightly inferior in bandwidth to full-sized
(a) The Yagi of Fig 12.1(a), having three full-sized (or two-element monoband beams which are therefore
equivalent) elements based on Figs 5.16 and 5.17, not advocated. Additional bands can be accomo-
but with the suggested addition of capacitors to dated with the assistance of relays as suggested in
form linear resonators (shown dotted) to provide Chapter 7, p110. One feeder can be omitted if
operation on at least one extra band [1, 2]. beam reversal and the other advantages outlined
(b) Multiband dipole arrays derived from Fig 7.6; on p110 are not required.
these can be erected as inverted-V’s suspended (c) The miniature two-element monoband (or possibly
horizontally between two points (Fig 12.2(a)) or two-band) Yagi using end-loading by folding, Fig
for rotary use altered from V to U shape as in Fig 5.22, with neutralization of the excess coupling
12.2(b) and affixed to spider arms, in which case produced by loading, and in this case also the use
they can be closely identified with the VK2ABQ of two feeders for remote fine tuning and beam
array shown in its original three-band 14/21/28MHz reversal [5, 6].
form by Fig 12.3. This shows its derivation from a (d) Three-element compact beams with end-loaded
single quad loop and is included largely for its and/or bent elements.
historical interest in view of later improvements. (e) The miniature top-hat loaded quad as devised by
These are all basically similar and by carefully G3YDX, Fig 12.16. Two-band operation is possi-
controlled spacing between the ends it is possible to ble by the use of traps as patented by G3IMX, but
achieve critical coupling, equal currents, and a possible advantages in terms of bandwidth and cost
168 HF ANTENNAS FOR ALL LOCATIONS

Insulators are shown tomark _— See (b) Small Blastte insulators

junctions between wire
and plastic cord but
should not be used 72 twin plastic
lead (donot
transpose

It is desirable to use Cords to ground level

simple 28MHz traps or to 12 to15ft
(Fig 7.6b) at points (3-7 to 455m) long
marked T (a ) spreader arms
mounted lower down
the mast

Low-impedance twin-lead _

248
Approx
freq

Fig 12.3. The VK2ABQ beam for 14/21/28 MHz seen looking down
on the array. The dimensions are attributable to G3FRB
(b)
Fig 12.2. Multiband dipole arrays. Elements are connected in weight, windage and visual impact. The design aims are
parallel using low-impedance twin-lead, and spacing between as follows:
adjacent ends is adjusted to ensure equal currents in conjunction
with correct phasing; (a) can be erected as an inverted-V and (b) (a) to arrive at a reasonably stable mechanical struc-
is basically similar to the VK2ABQ array, Fig 12.3. Total wire
lengths for 14MHz are approximately 35ft Gin (10-8m) with a
ture based as far as possible on the beam elements
spacing between ends of 12-15in (30—-38cm), these dimensions themselves, although this is not practicable in the
being scaled down for higher frequencies in the appropriate ratio. case of the VK2ABQ and its derivatives;
(b) may be fed at X with 50Q cable and 1:1 balun. Use similar (b) To obtain multiband operation without loss of
feeders for each set of elements. Preferred system uses 1:4
bal-bal transformer into 300 or 600 line (not plastic); open-wire effective aperture at the higher frequencies or
line is used in multiples of 20m, terminating in a 4:1 balun plus having to complicate the physical structure; to this
coaxial line for the remainder of the feeder run
end use is made of additional wire elements draped
(less copper) are seen in the close stacking of loops
over the cord ties. ,
If the basic structure is that of a three-element beam
all of the same size as illustrated in Fig 12.4, this
being a suggestion for the experimenter. The indi- 14swg orthicker 22swg or thicker 26swg or thicker
cated wire gauges are for a loss of less than 0-3dB. ) 10pF

(f) Fixed arrays using loops with resonant lines in

accordance with Figs 11.11(d), (e) or (f) or the
DJ4VM quad: Fig 11.11(h).
21MHz 28MHz
As an alternative to conventional methods of con-
struction the author favours very light structures braced
with polythene or nylon cord (about 1mm diam) as in
Fig 12.5. This idea can be used for a wide variety of =mee
beams including the VK2ABQ array [3] which in any
case uses a somewhat similar mechanical principle.
Suitable materials include bamboo, glassfibre and alu- Fig 12.4. Set of 12ft (3-7m) square loops for 14, 21 and 28MHz; a
minium alloy tubing, depending on whether a particu- loop size of 10—13ft (3-4m) is suitable for all three bands but
lar structural member is or can conveniently be used larger sizes can be used for 14MHz and smaller ones for the
higher frequencies. Element spacings are not critical but ideally
also as part of the radiating system. This tends to be a should increase from about )/8 at 14MHz to 1/6 at 28MHz for 12ft
desirable goal from the point of view of economy, (3:7m) loops
 HORIZONTAL BEAMS 169

Linear resonator increases electrical length of reflector and can be 33ft (10m)
switched for beam reversal. This is only one of many options | (not critical) |
SSS ass eS, [Son mae ears: (a)
8pF 8pF *

Any feeder can be connected here —

with coaxial cable use 4:1 balun.
x x
Alternatively cable plus balun can
Y beconnected at y asin(c) below

8pF 8pF
End loading
(b)

x x :
y

Coaxial cable
Balun is 1:1 or 4:1
depending on degree of
folding.
An swr <2 can be expected
Fig 12.5. Lightweight three-element array braced by polythene or in most cases
nylon cord as shown dotted. The diagonals are secured to a light
3ft (1m) mast extension and the boom is extended slightly to (x)
provide “bowstring” anchorage for the outer elements. Addition-
al sets of ties can be used if necessary. The cords can support Stub folded into
very thin wire (eg 19swg/1mm) elements for the higher frequen- a 4-wire cage to
cies, this being acceptable in view of the relatively high values of reduce the physical
radiation resistance. In this way full use can be made at the length A
higher frequencies of the available aperture. Element lengths are
as for Fig 12.1 except that for beam reversal both parasitic
elements are tuned as directors, C being then switched in to tune Enlarged end view to show
one or the other as a reflector B arrangement of wires in cage

(d)
for the lowest frequency it becomes possible to use long D
end-loaded wire elements for the higher frequencies,
and because these have relatively high radiation resist- Fig 12.6. Two-band (21/28MHz) element suitable for use in con-
junction with Fig 12.5. In the case of 28MHz the capacitors are
ances the wire can be thin so that it is easily supported located at a current node and thus have no effect, but at 21MHz
by the cords and visual impact is minimal. Fig 12.6(a) they are suitably placed for tuning out the inductive reactance of
shows a suitable wire element resonating at 28 and the shortened 1) element. The basic element is shown at (a); (b)
and (c) show various ways in which this can be end loaded and/or
21MHz; ignoring the capacitor, the length of wire is folded for fitting into a smaller space. At (c) the fold forms a
3/2 on 28MHz, the middle i/2 being folded so that four-wire cage 4ft (1:2m) long; this is shown end-on, looking
there is little radiation from it. In some cases the other towards the element, at (d)
two half-waves may be optimally spaced, giving a gain but it is probably best to aim for the lower right-hand
of 3dB for a single element, though this will be corner of Fig 5.16 which provides a gain of just over
somewhat reduced in the case of a beam or if the 6dB and a radiation resistance of around 20. This
element has to be folded to fit into a smaller space as region of the chart corresponds to rather long reflectors
for example in Fig 12.6(b). The capacitor is added at a (+402 to +50Q reactance) and directors which are
point where the current is zero on 28MHz so that rather shorter than optimum (—30Q to —40Q) from the
operation on this band is unaffected, its value being point of view of gain. It results in reasonable bandwidth
adjusted to tune out the inductive reactance of the and constructional tolerances as well as good back-to-
outer arms of the dipole together with that of the stub front ratio. As explained earlier (p79) it is very difficult
at 21MHz. in practice to realize much of the additional 1-5dB gain
What has thus far been described is a basic set of which is theoretically possible with three elements.
building blocks from which the design of beam anten- For multiband operation there is no reason in princi-
nas can be tailored to fit individual requirements. ple why several beams should not be stacked on the
same boom, provided precautions (p110) are taken to
prevent the elements for one band resonating in or near
Full-sized beams other bands. Optimum performance at the higher
For one- or two-band operation the recommended frequencies will be obtained by making the elements as
design is in accordance with Fig 12.1, sizes of tubing for long as possible in conjunction with suitable tuning
self-supporting elements being as given in Fig 19.1. devices. Unfortunately, with self-supporting elements
There are various options based on Figs 5.16 and 5.17 this adds greatly to the weight which is almost
170 HF ANTENNAS FOR ALL LOCATIONS

proportional to the number of bands. Linear resonators Reflector

as shown dotted in Fig 12.1 overcome this to the extent _———

of allowing each element to operate on two bands, and if

the requirement is for two bands only this is a preferred
method (p112). Adjustment of capacitance is critic-
al, being roughly equivalent to the more usual length
adjustment which it replaces. Three-band operation by
this means requires the more elaborate resonator
shown in Fig 7.9(b), though as an alternative relays
may be used to switch the capacitors [2].
On the other hand, for the amateur prepared to
embark on relatively uncharted waters the cord techni-
que just described opens up interesting prospects for
much lighter and cheaper arrays and, by the use of the 1:1 balun

capacitively-stretched two-band wire element (Fig

12.6), a relatively simple and foolproof answer to the
problem of three-band operation. Moreover it should
not be difficult to add a further set of wire elements to 4 Coaxial

cover the 18 and 24MHz bands. It would be acceptable

to retain the linear-resonator system for two-band
operation and use the wire elements for one band only,
except that insertion of capacitors in the wire element is
much easier than construction of the linear resonator.
It also has the advantage of not exacting a penalty in
terms of bandwidth, which is appreciable in the case of tee ~

linear resonators at the second harmonic frequency

~ ~
~
~ ~
~ ee

and, in the case of traps, at all of the frequencies. ~

It is recommended that elements should be insulated

from the boom since only a small amount of asymmetry
is required for enough current to flow along the boom
to upset the radiation pattern. Due to the longer boom
the effect is worse with three elements than two, and
the problem is further exaggerated by harmonic opera- Fig 12.7. Layout of the log-Yagi array (ARRL Antenna Anthology) |
tion of elements owing to the presence of relatively
log-periodic array such as the one described in the next
large rf voltages in the vicinity of the boom. Insulation
chapter and provide additional gain as well as greater
from the boom is not necessarily a cure since some
bandwidth.
capacitive coupling to it is unavoidable, and it remains
important to ensure symmetry which is one reason for
not recommending the gamma match. Another is an Three-element compact beams
experimental finding, admittedly not fully explained,
Fig 12.8(a) shows one method of reducing the length of
that bandwidth was improved by some two or three
times when a delta match was substituted for a gamma an element with virtually no effect on operation at its
match. This was an isolated incident and may not apply fundamental resonance. Each tip of the element AB is
removed and replaced by a slightly longer vertical
generally, but the asymmetry of the gamma match is
obvious to the eye and it is possible to imagine cir-
length CD, about 23 to 3 vertical feet (0-8-0-9m) being
cumstances in which this definitely would cause trouble
required for each 2ft (0-6m) removed. The length of a
as described in Chapter 4, p50. 14MHz driven element can be reduced from 33ft (10m)
For those requiring the best possible monoband
to 24ft (7-3m) by using 4in (6mm) diameter vertical rods
beam and able to contemplate the use of a long boom Table 12.1. Log-Yagi array dimensions [4]
of the order of 0-4), the log- Yagi [4] of Fig 12.7 may be
a better choice than the usual addition of another Element Length (ft) Spacing (ft)
director to an array such as Fig 12.1(a). Dimensional Reflector 36-4 6:0 (ref to /,)
details are given in Table 12.1 for tubing elements of iF 35-14 3:51 (d42)
33-27 3°32 (d23)
about gin (22mm) average diameter and will probably
need some modification for use with wire elements. ls 29-81 10:57 (/, to dir)
The feedpoint impedance is 37Q. The four close-spaced Director 32-2
elements or “‘log cell” are in effect a short segment of a 1ft = 0-3048m
 HORIZONTAL BEAMS 171

a
ed 84in 99in
| Sie | (213-4cm) (251°5
(10m) cm)

xn
Ca
6 to 7ft
(a) (1:8 to 2-1m)

C3
Aus alloy Vo in (1-3cm) S/A6in (O-8cm)
(a) Omi dle wire dia tubing dia tubing
(b) 6 to7ft spaced 1-1/2in (3-8cm) from tubing
4Bin
(1-2m)

17ft _@x x
(5*2m)
Coils are approximately 7 turns, 5in(13cm) long, 5india

Fig 12.8. End-loaded elements for compact beams. Element

lengths can be reduced as shown at (a) without significant effect
on any aspect of performance. Further reduction is possible as
shown at (b), subject to a loss in gain of about 1dB and some
48in
reduction in bandwidth
(1-2 m)
52-7ft (1-7-2-:1m) in length, depending on diameters of (b)
tubing used for the radiating portion and effects due to
linear resonators or other tuning and matching devices
which may be required in the centre of the element.
With this type of loading the radiation resistance is only
slightly reduced, from 73Q to 60 (Fig 3.12) which is
barely significant. The gain of one element at the 10°8m (very approx)
second harmonic frequency is reduced from 1-9dB to Directors
about 1-4dB, so even in this respect the effect is quite
small. The process can be carried slightly further but
with increasing length of the loading rods the structure
rapidly becomes heavy and cumbersome.
Depending on the length of the loading rods a small
amount of inductive loading can be accepted without
much drop in performance; how far one should be
prepared to go in this direction is a matter of opinion.
Fig 12.8(b) is recommended as a lower size limit for this Bamboo
Aluminium
alloy
type of loading since there is already a drop of about tubing

1dB in effective gain, and if it is necessary to reduce 11*-2m

size still further equally good or better results can be (c) (very approx)

obtained from two elements. On the receiving side the Reflectors

greater ease of nulling with two elements can be set off
against a lower average level of interference with three,
though some experimenters might like to try their
hand at the addition of a third small element to Fig 12.9. Compact three-element beam. Driven element is con-
two-element beams for receiving only (p131). structed from 14in (32mm) diam tubing, tapering to 0-5in (13mm).
A better method [6] of arriving at a size of 17ft The feeder is connected at XX using delta match as shown at B.
The end-loading rods shown at A are also used to provide
(5:2m) square is shown in Fig 12.9, though in this anchoring points x, y, z for the ends of the parasitic elements.
case there is a fairly critical lower limit to the size and “Bowstrings” attached to the loading rods provide additional
fewer alternative options. In this case the full 24ft loading and counteract the pull of the parasitic elements. The
plan view, C, shows the layout of the parasitic elements. Approxi-
(7-3m) length is retained for the driven element which mate values of capacitors C1, C2, C3 are 10pF, 20pF and 12pF
forms the diagonal of the square, parasitic elements for respectively, short “whiskers” of open-wire line being connected
each band being hung round it. across them to provide fine tuning. To avoid overcoupling it was
found necessary to use relatively wide spacing in the case of the
Use of a separate thin-wire folded dipole “hung in 14MHz reflector. Approximate wire lengths are indicated for the
the rigging” for the 283MHz driven element is relatively parasitic elements (Ham Radio)
172 HF ANTENNAS FOR ALL LOCATIONS

Guy these points to mast extension 24ft (7-3m) diameter Fig 12.10. Reversible version of
turning circle three-element compact three-band
beam. With relay contacts closed a
as <5 Use ‘off’ ~~s_ he short parasitic element acts as a
ae contacts 14 MHz parasitic elements
use 14swg wire director for 21MHz; with the con-
tacts open it is series-tuned to oper-
ate as a director for 28MHz. Com-
Cord ties pared with Fig 12.9, fuller use of the
available turning circle provides
better bandwidth and a closer
Attach wire approach to optimum design on
here to cord Use 'on' 14MHz. R denotes relays; on 14MHz
= ues contacts
one of the reflector stubs is shorted.
Cord tie On 21MHz director operation is
obtained by shorting out a capaci-
24ft (7-3m) long
tor, on 28MHz an unshorted capaci-
driven element tor tunes the element as a director
32 Cord ties tomast and shorting a capacitor provides a
— extension as fig 12.9 or see text
slight reflector action. Back contacts
used on one pair of relays, front on
Use'off' ~~. the other; beam direction is re-
contacts
versed on 21 relative to 28MHz. All
relays are energized in parallel.
19ft (5*8m) long Total boom length is 19ft (5-8m).
O-5in dia insulated at centre Wire length total for 14MHz parasi-
(13mm) Requires about 1ft (30cm) tic elements is 32ft (9-8m). Wire
extension at each end or length total for 14MHz parasitic ele-
equivalent loading rods ments is 32ft (9:8m) plus 40in (1m)
Use'on' Wk for the reflector stubs. Value of C is
~~ -—
approx 15pF (calculated) inclusive of
- >
~~ eee insulator which needs to be of high
quality; length of dural angle carry-
simple and avoids the bandwidth limitations of the ing four stand-off insulators sug-
gested
three-band linear resonator, but this needs either to
have its own feeder or to be switched by a relay. equally acceptable, and on this basis it seemed unneces-
Various types and combinations of relay-switched two- sary to bother with reflectors. The parasitic elements
band elements would appear to be feasible for coverage were series-tunéd for operation as directors at 23MHz,
of additional bands, thus affording the experimenter having first adjusted their length to provide director
plenty of scope for innovation. operation on 21MHz when the relay contacts were
Fig 12.10 shows an experimental antenna which, closed. At any given time one pair only of relay
though fitting within a 12ft (3-7m) radius turning circle, contacts is closed and the beam fires in one direction on
provided the most facilities and ‘‘equal best” perform- 21MHz and the opposite way on 28MHz. To some
ance of any of the author’s antennas so far until blown extent the 21MHz director operates as a reflector on
down in a severe storm. Based on Fig 12.9, it makes 28MHz but, as in the case of a director fitted to a quad,
fuller use of the available turning circle and provides the benefit from this is marginal.
the additional (though optional) feature of instan-
taneous beam reversal by means of relays. Mechani- Two-element rotary beams
cally the design follows the general features of Fig 12.5 A two-element beam may be constructed on the lines of
but with a driven element identical to that of Fig 12.9. Fig 12.1 merely by omitting one element, in which case
The other tubing elements, which operate respectively the somewhat unconventional measure of bending the
as directors for 21 and 28MHz, can be loaded by ends inwards to achieve critical coupling and improved
vertical rods to keep them within the 24ft (7-3m) front/back ratio should be adopted as explained in
diameter of turning circle or allowed to project slightly Chapter 5. For monoband operation this offers the
outside it. The 14MHz reflector and director are wire advantages of improved bandwidth, simpler construc-
elements hung around the perimeter of the turning tion and perhaps neater appearance than the arrange-
circle and since their radiation resistance is relatively ments shown in Fig 12.2, though it is considerably
low the wire should not be thinner than 14swg (2mm). heavier and more expensive. Two-band operation is
Beam reversal is effected by means of short-circuited achieved easily and three-band operation is also possi-
stubs in the case of 14MHz, but an interesting situation ble by the use of linear resonators as already described.
arose in the case of 21 and 283MHz—it was found that Nevertheless, if the user is prepared to go to a structure
whereas large currents flowed in directors, reflector of this size a three-element beam offers better receiving
elements had little effect. This is the exact opposite of performance. If this is felt to be too ambitious the
what happens in the case of the quad, being due modified VK2ABQ array described below can be
presumably to the collinear as opposed to broadside recommended since it is almost as good, extremely light
configuration of the driven element. It was judged to be and inexpensive, and achieves multiband operation
 HORIZONTAL BEAMS 173

f
Wire secured to cord by binding with thin wire at intervals
e » to the side cords in the positions indicated. The wire
sets of radial was secured along the cord at frequent intervals by
Pe ties wrapping with a few turns of thin wire. This stopped
about 8in (20cm) short of the ends of the bamboo, the

al
slack at the corners being taken up by bending it into a
U shape and attaching it to the radial cords going to the

Ny
top of the mast. Spacers made of strips cut from plastic

eteety
tubing were found desirable about a foot from the end
of the U. The dimensions indicated in the diagram are
not critical but shape differences can result in small

| yi
her: ae
f
changes in the overall length required, and it is advis-
(See able to allow a bit of extra length with the ends bent

‘\
Vi
back as necessary and the wires only loosely secured
until adjustments have been made. If desired the cord
2a ty
can be removed, leaving only the radial ties and the
short insulating sections at the corners. The 28MHz
20ft (6-09m) elements can be added without the need for additional
Bamboo booms with aluminium centre-piece and 3ft light mast
extension at Z sothat cord slopes downwards cords except for insulation at the corners where the U
------------ Polythene cord, 90lb breaking-strain, <1mm dia sections (being relatively short) can be allowed to hang
14swg copper wire down. The use of 28MHz traps in the 21MHz ele-
Fig 12.11. Two-element beam based on Fig 12.2(b). Spider arms ments as shown in Fig 7.6 has been found necessary to
may consist of 4ft (1:2m) lengths of dural tube extended by overcome critical tuning, poor front-to-back ratio and
carefully-selected 8ft (2-4m) bamboo garden canes bound with
insulating tape. Traps as Fig 7.6(b) are desirable in the 21MHz narrow bandwidth on 28MHz. If more bands are
elements. See Fig 12.2(b) for details of feed. Note the method of added, more traps will be needed, and the require-
attaching the corners of the elements to the polythene cords, ments may be ascertained by exciting the antenna with
avoiding direct contact with the bamboo. The amount of wire in
the corner folds is not critical and these can be used for taking in each frequency in turn and checking all the other
any slack. The total length of each wire is about 35ft (10-7m) for elements with a current probe (p237). If currents in
14MHz, 23ft (7m) for 21MHz, and 17ft (5:2m) for 28MHz but can other elements are comparable with those for the
vary with the size and shape of the corner folds
elements in use, traps are needed.
with maximum simplicity and fewest problems. Alter- Values are not critical but traps should be tuned by
natively the constructional principles of Fig 12.5 may be applying a few watts of power at the frequency to be
applied to shortened end-loaded elements to produce trapped and adjusting the spacing between the wires for
the electrical equivalent of the modified VK2ABQ. equal currents, or (since the currents flow in opposite
Referring to Fig 12.2(b) there are many different directions) for zero current when the probe is held
constructional possibilities, but the author was fortun- half-way between the wires. The mid-position can be
ate in obtaining a rotary clothes line of rectangular judged well enough by eye and a check of the 21MHz
shape based on 4ft (1:2m) tubes which were extended elements just beyond the traps should now indicate
by 8ft (2-4m) bamboo garden canes to the required zero current. If one feels uneasy about possible losses
length of 12ft (3-7m) and wired as shown in Fig 12.11. It the relative currents in the traps and the 28MHz
is important that the aluminium tubing should be short element can be roughly estimated using the probe—the
enough not to have any resonance in the wanted range increase in total loss as a percentage can then be
of frequencies, and the bamboos should be given calculated, making use if necessary of the usual /?R
several protective coats of polyurethane varnish or formula to obtain the relative powers. An example of
wrapped with insulating tape. It will be noticed that the this procedure will be found on p238 and in the present
shape is rectangular in contrast to the square shape of case the extra loss was an estimated 0-016dB only.
the VK2ABQ array, Fig 12.2(c). This was done to With the close spacings involved in covering 10, 18
arrive at the optimum spacing and required no increase and 24MHz, interactions between dipoles become quite
in length of the spider arms, the perimeter of a complicated, and the alternative method of using relays
rectangle being only very slightly greater than that of a to open-circuit elements as required (p110) can be
square for side ratios up to about two; any excess wire recommended with greater confidence. The author has
was easily accommodated by tucking in the corners as however successfully replaced the 21/28MHz elements
described below. The author used two sets of cord ties, of Fig 12.11 by a pair of two-band elements on the lines
one linking the tips of the bamboos and also guying of Fig 12.6(c), both sets of elements being matched
them back to a light 4ft (1-2m) mast extension and a directly to 60082. A common feeder is used with pairs of
duplicate set in the position of the 21MHz elements. 14MHz traps to isolate the higher-frequency elements
The centres of the elements were attached to an when operating at 14MHz. A pair of atus plus baluns in
additional support (not shown) and the ends were fixed the shack provide remote tuning of reflectors and
174 HF ANTENNAS FOR ALL LOCATIONS

Two-element beam based on Fig

12:11 for 14MHz but using
stretched and folded elements
similar to Fig 12.6(c) for 21/
28MHz. Note the very close
spacing between adjacent ends
of the higher-frequency ele-
ments. Old tv line-output tubes
were used as high-Q (vacuum)
capacitors but later measure-
ments suggest that short
lengths of coaxial cable should
be satisfactory if adequately
sealed. The 14 MHz elements are
delta-matched to 6002 lines
which also feed the 21/28MHz
elements via pairs of 14MHz
traps; the method of matching
at 21/28MHz is similar to that of
Fig 11.10(e) with a stub length of
22in (55cm). A separate photo
shows enlarged details of the
corners of the 14MHz elements

matching to the transmitter, beam reversal being

effected by interchanging the feeders. At a height of
48ft (14-6m) this is currently providing slightly better
performance than the quad at 38ft (11-6m) which it has
replaced but, had they been available at the time, the
relays described in Chapter 7 (p110) would have been
used in lieu of the four traps which are bulky and
suspected of responsibility for a slight ‘““wet weather”
problem.
One practical point to have emerged from this and
one other experiment has been the need for very tight
coupling between elements of the type pictured in Fig
12.6, due presumably to the relatively large value of
radiation resistance. In the case of (c) a compromise
was possible covering both bands; this required
spacings of only an inch (2-5cm) or so over a length of
about 8in (20cm). In the case of elements of the type
shown in Fig 12.6(a) a spacing of about 6in (15cm) was
required over lengths of some 2-3ft (0-6-0-9m) at
each end. These arrangements avoid the 21/283MHz
interaction problem, improve bandwidth and can be
recommended to experimenters. Enlarged view to illustrate the tucking-in of the corners of the
As explained in Chapter 5 (p75) the use of two 14MHz elements of the array shown in the adjoining photo-
graph. The folds provide loading and are readily adjustable, thus
feeders, one of which is used for remote tuning of a providing considerable freedom as regards element sizes, shapes
parasitic element, results in a 50 per cent increase in the and spacings. Spacers obtained by lengthwise cutting of short
total feeder loss. The situation is similar in the case of a lengths of plastic pipe are used to maintain the shape ofthe folds
which (as may be inferred from the picture) is not critical! This
driven array with highly reactive coupling between the method allows the wire to be kept well away from wet bamboo
elements, since in this case also one feeder is matched but it remains nevertheless advisable to give the bamboo several
and handles most of the power, the maximum currents coats of polyurethane varnish. The same method has many other
applications; in particular it allows the tubing elements of typical
in both feeders being comparable. With straight ele- hf beams to support folded-wire elements for lower-frequency
ments spaced 1/8 and matched feeders, each handles bands as suggested in Chapte’ 17.
 HORIZONTAL BEAMS 175

half the power and there is no additional loss but beam

reversal from the shack results in a large mismatch
(p75) so we are no better off! It is felt that, provided the
basic feeder loss is kept as low as possible, the slight Effective mean height

increase in loss is likely to be more than made up by the

ability to tune the beam in situ and maintain peak Elevation
performance at all times. In view of the enormous
practical benefits from instantaneous beam reversal it
seems a great pity not to take advantage of this option.
There remains the problem of deciding whether to
use driven operation based on Fig 5.13(b) or parasitic
operation but there is no need for hasty decisions since
usually the same components can be used for both
purposes. In practice the parasitic mode has proved
more convenient with near-critical coupling and just as
good, though there may be trouble with some feeder
lengths unless the remote tuning arrangements are
adapted to suit. In the driven case it is important to
connect the elements in antiphase as with the W8JK Co pe

array, so that the recommended phase difference of

157° is obtained by a 23° shift from the 180° condition.
Whichever method is used adjustment should prefer- Anchor to
ably be for a null on a bearing of about 140° and not the cord but not
boom
reciprocal of the beam heading (p130).
Plan
Fig 12.12 illustrates the inverted-V equivalent of Fig (335m)
atitt
ae
| |
12.11; note that only two arms lower down plus one
fe 26ft (could be 24ft if ends insulated) |
spreader at the top are needed. Careful disposition of (7-92m)
guys etc allows sufficient rotation by means of two Fig 12.12. Inverted-V equivalent for Fig 12.11
cords.
under 2dB though a close approach to 3dB is obtain-
able for nearly all directions. These phasing steps are
Phasing of separate beams best achieved in the form of three positions with phase
Assuming beams in accordance with Figs 12.1, 12.5 or reversals as shown, the method of use being to find the
12.10, the easiest method of obtaining additional gain is position giving least signal and then reverse the phase
probably to erect two beams and phase them together of one antenna. For all-band 14~28MHz operation at
rather than use a longer boom with more elements. least six steps are required, since 60° steps at 283MHz
This provides a gain of 3dB whereas gain increases only become 30° only at 14MHz.
very slowly with increasing boom length, and with
larger and heavier beams it may be necessary to settle
for reduced heights. The technical problems are dis- Very small beams
cussed in Chapter 6, p100, and Fig 6.14 shows the Requirements for very small beams arise from such
recommended method of phasing. Alternatively, the obvious reasons as lack of space or the difficulty of
resonance method (Fig 5.13) can be used to provide providing an adequate support for large ones. There is
continuous adjustment of phase, but it has the dis- also the need for small size in order to minimize visual
advantage that large phase shifts may be needed and impact and hence the risk of objections by neighbours;
this results in variations of load impedance which need under this heading there is also the possibility that if the
to be compensated. beam is made small enough and the best choice is made
The objections previously raised to the use of lines as from the available design options the uninitiated will
in Fig 6.14 arose from mutual coupling between beam be unable to distinguish it from a television antenna!
elements, whereas the scheme now under discussion The best shape in this case depends more on the
rests on the assumption that the beams are far enough frequency and polarization of the local television trans-
apart for mutual coupling to be ignored so that the _ missions than on the hf design requirements which are
antennas constitute a pair of independent matched remarkably flexible. However, if the local tv is uhf
loads. The phasing requirements in this case can be only, there are problems and one may be restricted to
readily calculated from the line lengths and velocity the 21 or even the 283MHz band. In the latter case acute
factor, steps of 60° being small enough since this results frustration will be experienced during sunspot mini-
in a worst error of 30° only; this reduces the gain to just mum periods and a monoband beam for the 24MHz
176 HF ANTENNAS FOR ALL LOCATIONS

142 in
(360-7cm) |
exhibiting the flat response typical of slightly overcou-
16in 45in
pled circuits. A useful bandwidth of about 200kHz was
(40-6cm) (114:3cm) | achieved, the gain being 1dB down at one edge and the
i
Poe
front-to-back ratio down to 8dB at the other [6].
— With two feeders either driven or parasitic operation
70in
can be used exactly as described above in the context of
Fig 12.2. In this case the reflector should be made

| Fe
(177 8cm)
identical with the driven element and interchangeabil-
ity of the elements should be verified before erection.
30082 line
Due to the highly capacitive nature of the coupling it is
eo ew
essential to tune a parasitic element as a reflector, not a
(3-8cm)—
director, and with driven operation swr is unimportant
cM)
{ Real 22in spacingneutralizing _. A 22in in the case of the reflecting element.
(55-9
(55-9
cm)
capacitances Eqn This method of construction is inexpensive, light and
simple, and because the beam is reversible about 130°
Fig 12.13. Practical design details for small beam based on Fig of rotation is adequate. This can be achieved by the ©
5.22(a). The elements are identical with a total wire length of
37-3ft (11:37m). For initial adjustment one element without its means of two cords attached to points about 2-3ft
feeder may be tuned as a parasitic reflector. The driven element (0-6-0-9m) each side of centre on suitable spider arms.
can then be adjusted for an optimum match and the reflector
made identical with it, thereafter being tuned remotely using its
Some reinforcement such as a short stiff piece of wood
own feeder, though a check should be made to ensure this does or aluminium alloy tubing may be needed between the
not degrade the performance. The elements and neutralizing points of attachment.
capacitors may use 14-16swg (1:6-2mm) bare copper wire, but a
much lighter gauge can be used for the delta match and the
Fig 12.14 shows a neater version [6] using metal
neutralizing wires. The spider projects beyond the elements tubing for the radiating portions of the elements, the
which are suspended by short lengths of polythene cord (Ham vertical rods at the ends acting mainly as brackets for
Radio) supporting the wire zigzags. About 16ft (4-9m) of wire
band, despite its narrow width, may prove to be the was needed in each zigzag, and to avoid the need for
best option. This could be as little as 6ft (1-8m) square, heavy-gauge elements the ends were guyed back bow-
based on Fig 12.14 but using for the end loading string-fashion to short extensions of the boom. Two
whatever combination of vertical or horizontal rods and open-wire feedlines were used and Fig 12.15 shows
thin wires was judged least likely to offend the eyes of typical performance with parasitic operation of the
possible objectors. In Japan tv antennas are often gaily reflector, a front-to-back ratio better than 20dB being
painted in a variety of colours, an idea which might be obtainable over a bandwidth of 50kHz. This could be
worth importing—but consult the neighbours first! shifted to any part of the band by using the phasing-line
Fig 12.13 gives design details of the author’s first components re-arranged as a tuned circuit. Originally
serious venture into the minibeam field, the overall 30082 feeders were used with 4:1 baluns in the shack to
dimensions being slightly greater than might appear bring the impedance down to 75Q but, due to wet
from the diagram since the elements were tensioned weather and weathering effects (which proved dis-
between the six arms of a spider which projected astrous), a change was made to 600Q lines of resonant
beyond the corners and included two additional arms to length, no other alterations being made in the system.
counter the inward pull along the centre line. This Further reduction in the area occupied by the ele-
beam at 55ft (16-8m) was at least as good as a quad at ments would be possible by extending the vertical rods,
45ft (13-7m), but could only be properly tested in dry increasing the length of loading wires and accepting
weather—this was due to a tuning shift of about some reduction in performance, but it is doubtful
200kHz when it was wet which took it below the whether in practical terms much would be gained by
14MHz band, the bamboo being untreated. At this this and, if the main need is to reduce the diameter of
time only a single feeder was in use and so the tuning the turning circle, the capacitively-loaded quad (p178)
error could not be corrected. is a better proposition. For significantly reduced visual
Adjustment was carried out at a height of about 8ft impact it is necessary either to go to a higher frequency
(2-4m), tuning the reflector for maximum back-to-front or use loading coils with inevitable penalties in terms of
ratio and comparing the currents in the two elements. efficiency and bandwidth.
The current in the reflector being higher than that in A third design of minibeam [6] used 11ft (3-4m)
the driven element, the neutralizing capacitor was elements for 14MHz and a third central element for
increased in small steps, retuning the reflector each 21/28MHz which helped to support the loading wires,
time by bending the wire ends, and eventually achiev- three-band operation being achieved by a combination
ing current equality and sharp nulls. The tuning of the of linear-resonator type traps and stacking of additional
driven element was adjusted as necessary to maintain a parasitic elements. Performance on 14MHz was at least
reasonable value of swr but was fairly non-critical, as good as with the earlier designs and good dx contacts
 HORIZONTAL BEAMS 177

12ft a
: Approaching the stacking idea from a slightly dif-
\ (304-8cm)
ferent angle, there are possibilities of considerable
Boom
2ft
reduction in size of the arrangement shown in Fig 12.11
(61cm) which can in this case revert to the original square
V2in (1*3cm) dia
tubing ‘ shape. However, using folded-over ends as in Fig 12.13,
tubing
some extra trapping may be needed, if only of the 14MHz
(a) elements against 21MHz currents, and it would be
desirable if possible to replace the bamboo by glassfibre.
Gradually, however, with reducing size it becomes
necessary to choose between increased losses or a
neutralized system as in Fig 12.13(a) which is rather
complicated for three-band operation.

Relative gain

1 16 ft (4.9m)
1 of 16mm dia wire Neutralizing (b)
! ~ connections
1 O
me
W
=
(dB)
gain
Relative

F/B
(dB)
ratio

14:0 14-1 14-2 14:3

Frequency (MHz)

Fig 12.15. Typical performance of the 10ft (3m) square beam.

Measurements (solid line) were made with the beam at a height
of 7ft (2:1m) and the reflector tuned for 14-2MHz. The dotted
curve shows the improvement in swr after adding 4in (10cm) to
each end of both elements and retuning the reflector for best
front/back ratio at each frequency. Note that the gain curve
ott shows decibels down relative to best performance (Ham Radio)
Ome Lo OO ES
(ge NC
(61cm)
At this point it is useful to bring a number of threads
(c)
together. The parallel-dipole idea has already been
A fave inches of wire can be bent
inwards here if more lengthis needed helpful in the search for efficient methods of multi-
Fig 12.14. Based on Fig 5.22(b), this shows construction details of
banding, and closely resembles the multiband counter-
a 10ft (3m) square beam for 14MHz. The front, plan and side poise used for the grounding of vertical antennas (p161),
views are shown as (a), (b), and (c) respectively. In (b) the except that in the latter case resonating and radiating
“bowstring” principle is used to counter the pull of the loading functions have been kept separate. However, this is
wires and allow the use of lightweight elements. Each neutraliz-
ing capacitor, as shown in (c), is about 1:9ft (58cm) long. If more almost what has happened in the case of Fig 12.14! The
loading is required, additional wire can be added as indicated loading wires are very near self-resonance and the
in (c) (Ham Radio) radiator, being shorter and of much larger diameter
were obtained on 21 and 28MHz, but there were than the wires, possesses a relatively small impedance.
several bandwidth and coupling problems and in over- In other words, the tuning and radiating functions are
coming them the design became more and more very nearly separated in this case also. In this way one
complicated. Although none of the difficulties were can use either two sets of 4/4 wires or a pair of
insuperable it had ultimately to be accepted that effi- multiband counterpoises (as in Fig 11.14) connected by
cient multibanding is far from easy when beam size is a short length of larger-diameter tubing which supplies
reduced to the minimum, despite the fact that the the radiating function.
frequencies to be added are higher than the fun- This appears also to be the operative principle of
damental so that one might expect size to be less of a some commercial designs including mobile antennas;
problem. As a monoband design this arrangement has these commercial designs suffer from the criticisms
possible merit to the extent that there are no vertical which have been levelled against the use of small
projections, and there is no problem in using the capacitive hats working in conjunction with large in-
central member with some end loading as a dipole for ductances (p84). However, this fault is unavoidable in
21, 24 or 28MHz. the case of mobile antennas (p210) and the objections do
178 HF ANTENNAS FOR ALL LOCATIONS

not apply to /arge hats working with small inductances,

or to A/4 resonators which are more or less equivalent.
This argument reveals a unifying principle relating to
diverse systems and central to the multibanding of
small beams. Though not in itself solving all the
problems it signposts a possible line of approach, and
Fig 7.11(a) shows how it can be applied to a two-band
beam covering 21 and 28MHz. The arrangement is
similar to Fig 12.14 except that there are now two sets
of loading wires; only one corner of the array is shown (a)
in Fig 7.11. As might be expected, the interaction
problem which required the addition of traps in Fig
12.2 applies in this case also, two-band operation being
impossible without addition of the capacitors.
This design is still in the experimental stage at the
time of writing but three-band operation seems unlikely
to be feasible without some increase in size to allow
better separation of loading wires. The possibility of 10ft Sin
useful operation at 14MHz by insertion of inductances l" (3+17m)
in the form of radial.open-wire stubs at the point of aw se fet

attachment of the loading wires as indicated on the bi aaa

diagram has been considered. It has not been entirely (96-5cm)

excluded since a gain of 2-3dB appears theoretically

possible over a bandwidth of 10-20kHz after allowing
for losses. In this case it would be best to use a single
main feeder with a loosely coupled second feeder to
provide a small amount of remote tuning, and the
neutralizing adjustment will be extremely critical. The
idea could be worth trying during sunspot minimum
years, assuming easy access and plenty of patience.

The capacity hat loaded mini-quad

This is in a slightly different category to the other small Fig 12.16. Miniature quad array with “top-hat” loading due to
beams, being of much greater extent in the vertical G3YDxX. (a) Starting point for design; (b) final version
plane, but -if this is acceptable it offers the most the plane of the element. This null could only cope with a
satisfactory specification and provides the smallest small proportion of the inevitable interference on the band.
possible diameter of turning circle. The recommended When the element was raised to its operating height the
design is due to G3 YDX [7] and his description follows. resonant frequency increased by some 150kHz. Constructors
are advised to bear this effect in mind, as it will vary with the
“For spreaders, 8ft (2-4m) bamboo canes liberally coated surroundings of the antenna. In the author’s case the element
with varnish were used and these were attached to the boom in was built parallel to, and about 8ft (2:-4m) from the ground.
the usual double-X arrangement. A 6ft (1-8m) length of 1 by 1 When raised, the boom height was 30ft (9-1m).
by gin (25 by 25 by 3mm) dural angle was sawn into 18in The next step was to build a parasitic reflector of the same
(46cm) lengths, drilled to accept lin (2-Scm) car exhaust size as the driven element, but with a stub added; the boom
clamps, and the canes firmly attached with adhesive tape was an 8ft 6in (2-6m) length of lin (2:5cm) outside diameter
reinforced with light wire to prevent it from being undone by 16swg (1:6mm) alloy tube obtained to match. Because of the
rain and wind. Jubilee clips could be used for this application reduction in impedance expected due to the smaller loop size,
but they are expensive and tend to crush the bamboo if not to some two-fifths the impedance of the full-sized quad, the
carefully fitted. The spiders thus produced could accommo- addition of a parasitic element would result in a change of
date about 10ft 6in (3-2m) of wire per side, compared with 11ft feedpoint impedance to below limits where acceptable direct
8in (3-55m) per side for a full-sized 21MHz quad. match to coaxial feeder could be achieved. A gamma match
Trials were undertaken to build one element for resonance was used to match the line to the array; 16swg (1-6mm) wire
tests. Because the sides are a lot shorter than 12ft (3-7m), the with about 1-5in (3-8cm) spacing, and a receiving-type 100pF
arrangement shown in Fig 12.16(a) gave resonance at too high variable capacitor being used. The length of the gamma
a frequency, but eventually after cutting and trying, and some section was adjusted for 14,050kHz with the array near to the
bending (a gdo helps), the sizes shown in Fig 12.16(b) resulted ground to take into account the correction factor found
in resonance with the element in the air. Tested against a necessary in the single-element tests. A gamma section length
ground plane, the single-element loop did not appear to have of 3ft (0-9m) was about right.
any significant advantage, apart from possessing a deep null in The mast was walked up on the back of the house, and the
 HORIZONTAL BEAMS 179

2:1
particular direction that it quickly absorbs the whole of
their interest.
In
= 1-521
7) Loop arrays
| The author’s antenna experiments have usually re-
14100 14150 14200 14250 14300 14350 sulted in possession of at least one beam, rotary or
Frequency (kHz)
otherwise, capable of meeting normal operating re-
Fig 12.17. Miniature quad—variation of swr with frequency quirements, but it must be confessed that the construc-
tional needs of an ‘‘experimental’’ antenna are not
always best for ensuring survival in severe weather
resonant frequency was found to be about 14,190kHz. The
next job was the stub tuning. Due to a minor constructional conditions. Repeatedly over 25 years he has found
difference in the reflector, no stub was required for what himself returning to the fixed reversible quad based on
seemed to be the best front/back ratio. The author suggests two or three loops similar to that illustrated in Fig
that anyone constructing this array should test both elements 11.11(d) and more recently Figs 11.11(e) and 11.12.
for resonance rather than copy the dimensions from one to the Originally this approach was adopted as providing the
other. In this small design the odd half-inch is critical. best type of beam for mounting in trees as discussed
There are, however, some provisos which should be men-
tioned. earlier (pp141, 151), although in this case there was no
need for concealment—it just happened to be the
1. Because of the reduction in size, the swr bandwidth of easiest way of obtaining the necessary height. Latterly
the array is reduced. This is illustrated in Fig 12.17. It
should be noted that the array has a 2:1 swr bandwidth of
other supports have been used but many of the advan-
about 200kHz, which is a distinct improvement over tages remain.
some well-known miniature beams available commer- In view of earlier discussion the reader may be
cially at what seem astounding prices. The array will surprised by the reference to three loops, but it was
work on the cw end of the band with a reduction of difficult to shake off the feeling that there ought to and
directivity, and an atu is recommended because on just possibly might be some advantage, and it makes
14,025kHz the swr is above 3:1. Obviously bandwidth
beam reversal somewhat simpler. Two reed relays were
will increase as loop size increases.
2. The gain will not be as good as that of a full-sized quad used to make the changes from reflector to director,
due to the reduced element size, but in fact miniaturiza- taking advantage of the relatively low current in the
tion will only reduce it by some 0-5dB, which is hardly case of quad directors. Another advantage of three
noticeable on the air. Front/back ratio seems to equal elements when each has its own feeder—if one breaks
that of a full-sized quad; this suggests that an even there are two left and subject to minor changes at
smaller element may be useful for receive only.
3. It is important that the elements are tuned very carefully ground level there is little effect on performance!
or disappointing results may ensue.” In the case of fixed beams the mechanical considera-
tions are relatively less unfavourable to the quad since
there is no need for a heavy spider. The top corners can
Fixed beams be carried by a lightweight spreader and when a mast is
As a new amateur the author was equally interested in used as a support the corners can be pulled out in much
all available frequency bands and all directions, but the same way as the ends of an inverted-V.
after a time the interest in collecting OSL cards, With a reversible quad centred on VK2 good results
working new countries, and taking part in world-wide were obtained in most directions excluding polar paths
dx contests wore off. Instead of rarely working the
same station twice, most contacts are now with old
friends in a small number of directions, concentrating
on one or two frequencies and times of day which
happen to be best for this purpose. Exceptions include
a certain amount of antenna testing but it has for the
most part been possible to combine even this with
regular operation over the long path to Australasia.
This occupies as much time as can readily be spared for
operating and it is clear that many other amateurs find
themselves in a similar position, in which case needs are
usually best served by fixed arrays with their many
advantages such as lower cost, less visual impact on the
neighbourhood, and in some cases higher gain. Others
may find their hands forced in this direction by neces-
“it must be confessed that constructional requirements for ex-
sity or, if they happen for example to be living on a perimental purposes are not always best for ensuring survival in
steep hillside, may find that they do so well in one severe weather conditions”
180 HF ANTENNAS FOR ALL LOCATIONS

and the greater part of Africa. As a “‘fill in” for these of the high value of radiation resistance typical of
directions a single delta loop had for several years given collinear arrays. This example is illuminating for any-
a good account of itself on 14 and 21MHz, in spite of one lucky enough to have steeply sloping ground since
being undersized (60ft (18-3m) or wire in the loop) and it demonstrates how easily this can be exploited to
almost completely buried in a tree. On the strength of achieve exciting results.
this a two-element full-sized version was erected as a Arrays such as this over flat ground are of little use
second antenna for Australia. without several very tall masts and thus of limited
Though solidly constructed, well clear of branches interest to the amateur. It is in any case better to split a
and intended to be permanent, this had a very short life 1 collinear array into two separate )/2 arrays since this
due to unusual wintry conditions which turned the top allows the beam to be swung through a fairly wide angle
conductor into a thick rope of ice that proved too heavy by appropriate phasing of two halves, otherwise the
for supporting arms. This was before the delta loop had azimuth coverage is very restricted (Fig 6.3).
been officially invented and the corner supports must
have been considerably stronger than the tubing nor- Large Yagi arrays
mally specified. Conditions were very exceptional for In the case of a long narrow garden sloping in a wanted
the UK but this does seem to indicate a serious direction a long Yagi array probably makes the best use ©
weakness of 14MHz delta loops for areas liable to of the opportunities, and unless there are plenty of
experience severe winters. trees around the best method of construction might be
Prior to this disaster the delta loop had been success- as a set of inverted-V elements suspended from a
fully phased with the “‘tree quad”’ to form a high-gain catenary. One or two intermediate supports may be
array, the two beams being identical in performance. needed. Data for long Yagi arrays with thin-wire
Loop spacing in such systems should be about 11ft elements at hf is in short supply but director lengths of
(3-4m) for operation on 14 and 21MHz, or 8ft (2-4m) if 0-45 and spacings of 0-3-0-35) would appear to be a
it is required in addition to operate the system as a good starting point. A length of 0-49) and spacing of
bi-square at 28MHz, although this results in some 0-2 is suitable for the reflector. Adjustment of direc-
overcoupling at 14MHz if there are no losses. There is tors in 6in (15cm) steps based on dx reports with
however no need to worry if (as tends to happen with another station acting as a yardstick is suggested.
tree mounting) it is difficult to achieve accurate sym- Performance data will be found in Chapter 5, p90. The
metry. log-periodic array (p194) may also be worth considering
The use of resonant feeders as described in the last in this context. Alternatively, in the case of flat ground
chapter allows useful radiation to be obtained over all and restricted height, Yagi arrays using short loaded
of the hf bands except that for 1-83MHz and 3-5MHz it is vertical elements (Chapter 11) offer the best prospects.
necessary to operate loops plus feeders as a vertical
system. For 7MHz there is a choice of (a) operating Two-element dipole arrays
loops plus feeders as a vertical dipole, (b) using the Two-element arrays based on Fig 12.2(a) have many
outer elements of a three-element array as a two- advantages as detailed below and will often be pre-
element vertical beam, or (c) using the loops (prefer- ferred to quad arrays, being relatively inexpensive and
ably in parallel) as horizontal radiators. Beam opera- easy to erect. In their inverted-V form they represent
tion at harmonic frequencies is discussed on p106. the ultimate in lightness, simplicity of construction, and
low cost as well as providing the easiest means of
Collinear arrays gaining all-important height. Mounted horizontally
Prior to becoming interested in the quad the author had between two supports, a two-element wire beam can be
for many years been using collinear pairs backed by made reversible by employing two feeders as described
reflectors. Two such arrays erected at a height of some earlier or, if a single feeder is used, reversed mechani-
17ft (5-2m) above a 20° slope to the west and phased cally by means of two cords attached to the centre of
together (one of them about 60ft (18-3m) in front of the the reflector. One hangs straight down whereas the
other) gave excellent results. This might be expected in other is taken over the top and hangs down on the
view of the estimated gain of 9dB relative to free-space opposite side. By releasing the cord that hangs straight
propagation for a radiation angle of zero degrees. This down and pulling on the other one, the beam can easily
may sound like a large and ambitious array but in fact be “‘flopped over” so that the position of the element
covered an area of only one-sixth of an acre (0-07ha), and the beam direction is reversed.
such as many amateurs have at their disposal, and this Horizontal dipoles can as already indicated be con-
area could have been further reduced. strained into almost any shape within reason and,
The low height rendered this system much easier to whereas the VK2ABQ or modified VK2ABQ format is
erect than even the simplest of horizontal beams cap- ideal for a rotary beam, the inverted-V configuration is
able of reasonable performance over flat ground. The more useful for fixed beams, the spreader being fixed
arrays were invisible to neighbours; design, construc- to the top of a mast before erection or hauled up by a
tion and adjustment being made very simple by virtue rope over a pulley. The apex angle should not be less
 HORIZONTAL BEAMS 181

than 90° and if there is enough ground space, the ends example to use two dipoles at right-angles. It may even
of the central cord can be brought within reach of the be possible to erect two beams at right-angles but if
ground and used to walk the beam round into any these are close together the one not in use must be
direction. Otherwise the direction may be fixed and, open-circuited.
failing suitable trees, chimneys or other supports, one Assuming sufficient space, the desirability or other-
or more auxiliary masts employed. A further alterna- wise of a beam can probably be resolved only by
tive is to use a single support with, for 14MHz, a 24ft experiment, and in this respect indoor beams have a
(7-3m) spreader which can be rotated through 120° or number of advantages since usually they are readily
sO, assuming a suitable arrangement of guy-wires. accessible, can be adjusted in situ, and do not require
In principle two beams can be mounted at right- weatherproofing.
angles, one below the other, on a single mast to provide The chances of success are obviously much greater
switched selection of four directions, in which case it is the higher the frequency and, given a space of at least
absolutely essential to detune the one not in use. This is 7ft (2:1m) diameter and 5ft (1-5m) high, a small quad
possible by suitable termination of unused feeders but based on Fig 12.16 is the best choice for 28MHz,
may need a certain amount of skill and patience. leaving its competitors a long way behind by virtue of
Similar remarks apply to quad- or delta-loop arrays. its much larger radiation resistance and smaller dia-
There is a slight problem in tuning an inverted-V meter of turning circle. At 14MHz the possibilities are
beam before erection because the requirement for likely to be limited in most cases to the VK2ABQ or
access to the centre means that the ends have to be one of its derivatives; Fig 12.11 is a strong candidate
brought very close to the ground which lowers the unless space is sufficient to accommodate something on
resonant frequency. For checking resonance, it is best the lines of Fig 12.10. It must again be strongly
to lay it out horizontally. To check for correct coupling emphasized that some form of phased array is essential,
between the elements, they should be returned to the the W8JK being the worst possible antenna for indoor
inverted-V position at a height which leaves the ends use due to its exceptionally low radiation resistance.
just within reach, and then tuned for maximum rejec- For a given type of element in a specified position
tion of signals or minimum field strength on a bearing relative to any surrounding lossy materials, it should be
of —140°. recalled (p86) that two are always better than one,
A check with a neon lamp held near the ends will subject to the provision that, if they are lossy, both
show whether the elements have equal voltages. Alter- must be driven or the coupling between them must be
natively one can compare the currents by clipping a pair increased. This could mean, for example, that a beam
of matched flash-lamp bulbs (interchange then in situ to on the lines of Fig 12.13 does not have to be neutralized
make sure of the matching) across about 4 or Sin but, if losses are due to the proximity of a wet roof,
(10-13cm) of the element wires adjacent to the feed- they will be highly variable and it is best to neutralize
points. Meters can in principle be used in the same way for fine weather, relying in all conditions on phasing by
but are rather too easily burned out and a lot more the method of Fig 5.13.
expensive to replace! If more current flows in the If losses are high it may be that a much more efficient
reflector the ends of the elements should be moved design of element could be used if it is only necessary to
further apart, and vice versa. Three-element inverted- fit one into the given space. In this case it should be
V arrays may be preferred but are more difficult to possible to use a separate small active element (p131)
support; if it is possible to erect three elements at for improving reception, although this is a field wide
height / it is usually possible to erect two elements at open to the experimenter seeking to break new ground.
height 1-12h which is enough to neutralize any gain
advantage, although three elements should result in a Conclusions and recommendations
lower average level of interference (p129) as discussed
As in the case of the last chapter it seems appropriate to
in Chapter 9.
conclude with the author’s selection of “‘best buys’,
though in this case the choice is more difficult and there
Indoor beams are numerous variations on each theme, any one of
Before embarking on an indoor beam the section on which might be preferred in the circumstances.
indoor antennas in the last chapter (p161) should be 1. For a rotary beam the first choice has to be one of
consulted. Even if there is room for a small beam it may the relatives of the VK2ABOQO antenna since, when all
be better to stick to a single element, at least for the necessary measures are adopted, these provide
transmission, since in the case of the beam there will near-optimum performance for transmission and (aided
usually be increased losses and a greater risk of rf by some exercise of skill on the part of the operator)
voltages getting into the mains wiring (p163). A beam, very deep nulls “‘off the back’’. Their particular merit
even if it does provide useful gain, may not be rotatable lies in the fact that these advantages are obtained for a
and for all-round coverage it might then be better for minimum of cost, weight, and windage and with a
182 HF ANTENNAS FOR ALL LOCATIONS

turning-circle diameter little more than half that of a elements, otherwise performance will be sacrificed.
normal full-size three-element Yagi. A further poten- The constructional method of Fig 12.5 is recommended
tial advantage is that, at least on paper, no adjustment to save weight and windage, and it also allows the use
is required in the case of the two-element. versions, of separate wire elements for 283MHz or two-band wire
other than adjusting each element for resonance at elements (eg Fig 12.6) for 21/28MHz.
band centre; this could however be affected by interac- 3. For mounting in trees, pairs of quad elements
tions between multiband elements. with resonant feeders should be used as fixed reversible
Of the various alternatives, Fig 12.12 potentially beams. The use of open-wire feeders simplifies multi-
offers the best value in respect of cost, weight and band operation, the arguments presented in Chapter 11
windage but adjustment (which may or may not be (p151) being applicable in this case also. Similarly, for
required) is more difficult, rotation is likely to be mounting between poles or trees, “squashed” delta
curtailed by guy-wires and up to the time of writing loops are recommended; in this case the ends should be
there has been only limited experience of the antenna brought in towards a common anchoring point, a
in this form. separation of about 3ft (0-9m) probably being about
Fig 12.11 in monoband form is the type of beam right for 14MHz. This should however be checked by |
which can be recommended with most confidence to observing relative currents at the bottom end of the
those without previous experience, though one or two resonant lines or, if the parallel-connected high-
additional bands should not present much of a prob- impedance matched line in Fig 11.11(f) is used, just
lem, particularly if relays are used to open-circuit the above the feedpoints.
21MHz elements during operation on 28MHz. Cases to which these recommendations are not ap-
It is quite possible that after some further develop- plicable are to be regarded as ‘‘special’’, each needing
ment one of the compact three-element designs will to be considered in the light of earlier discussion,
emerge as a “better buy’’. As it stands the arrangement though, given a requirement for the smallest possible
of Fig 12.9 is not reversible, owing to the need for turning circle, Figs 12.16 and 12.4 between them pro-
increased spacing of the reflector for the avoidance of vide suitable guidelines.
overcoupling, and its advantages relative to the latest The author feels considerable doubt as to whether
two-element designs have yet to be fully assessed. Fig the use of more than three elements for a rotary beam
12.10 is less marginal and allows greater freedom to the can often be justified, at least if one is “starting from
experimenter, but the derivation of full benefit from scratch’. As will be evident from Chapter 5, a six-
this system requires the use of three feeders and rather element beam on a SOft (15m) boom at 50ft is no better
a lot of relays which detracts from its appeal. than three elements on a 16ft (4.9m) boom at about
At the time of writing, further experiments are in 65ft (20m) or, except in regard to the average interfer-
progress, the aim being to get rid of the vertical loading ence level, a phased pair of two-element beams at 50ft.
rods, simplify or eliminate the linear resonator and Even if considerations of size, weight and cost are
evolve a “foolproof” drill for adjustment of the parasi- ignored, those with experience of switched beam rever-
tic elements. It has been found that the rods can in fact sal may well regard this possibility as an overriding
be replaced by 11ft (3-4m) (each) of 19swg (1mm) wire advantage of two- or three-element beams.
festooned onto the cord ties with very little adverse
effect on losses or radiation resistance, though in this
case the use of a separate driven element (eg a thin- References
wire folded dipole) for 283MHz appears to be essential. [1] Radio Communication Handbook, 5th edn, Vol 2,
It is reasonably certain that a near-optimum design of RSGB, 1977, p12.75.
three-element Yagi beam can be achieved with a [2] “The disappearing inductance, a new trick and
turning-circle diameter of 24ft (7-3m) or very little some better beams’’, L. A. Moxon, Radio Com-
larger; this is recommended as a useful field for the munication April/May 1977, pp284—289 and 364-367.
experimenter, not yet as a “best buy”’ for the novice! [3] F. Caton, VK2ABQ/G3ONC, Electronics Australia
For “no compromise” multiband operation two- October 1973.
element beams must use separate elements because [4] ‘““The log-Yagi array”, P. D. Rhodes and J. R.
“adequate” spacing for 14MHz is ‘“‘too wide”’ at 23MHz. Painter, The ARRL Antenna Anthology, ARRL,
2. For those who already have a three-element Yagi 1978, p49.
or the means of supporting one, Fig 12.1 provides the [5] ‘““Technical Topics’, Radio Communication Febru-
basis for an antenna having broader bandwidths (lead- ary 1977 and June 1978.
ing to greater operational convenience) and at the [6] ‘““High performance small beams’ >, L. A. Moxon,
higher frequencies about 2dB more gain than option Ham Radio March 1979.
(1). Multibanding is possible by means of linear resona- [7] “Practical design for a top-hat loaded 14MHz mini-
tors, but if stacked elements are used these must be quad”, R. G. D. Stone, G3YDX, Radio Com-
comparable in lengths and spacing with the 14MHz munication October 1976.
 CHAPTER 13

Vertical beams

It will be clear to the reader from earlier chapters that There remains the general problem of feeding verti-
vertical antennas have major virtues as well as some cal antennas so that power is not lost due to earth
awkward vices. Some of these differences are further currents or feeder radiation. The requirements laid
emphasized in the case of beams and the task in this down in earlier chapters are modified in the case of
chapter is to show how the merits of vertical arrays may beams only to the extent that radiation resistances are
be exploited to the best advantage. lower so that any losses which are allowed to occur in
For the higher-frequency bands vertical antennas the earth connection will become proportionately
with their ‘feet on the ground”’ are easy to erect, and greater. In principle any of the single radiators de-
normally provide much better performance than hori- scribed in Chapter 11 may be used as beam elements,
zontal antennas at a low mean height. On the other though it is difficult in general with close-spaced ele-
hand, a horizontal antenna at 40—S0ft (12-15m) will ments to use ground planes consisting of three or more
usually give significantly better performance than a \/4 radials. This is because adjacent sets of wires (or at
vertical one having the same gain, but in some cases a least their respective near-fields) get entangled, though
relatively elaborate low-height vertical array will be in one system to be described two elements share a
found easier to erect and will give as good or better common “artificial earth’ which can consist of /4
performance in terms of dx. The low height has many radials.
attendant advantages since the elements are accessible
for adjustment and there is no major mechanical
engineering involved. Two-element arrays
For a given gain a vertical beam is normally much
broader in azimuth than a horizontal beam, thus mak-
Fig 13.1 shows the “simplest possible’? two-element
ing it easy to obtain omnidirectional coverage by beam array. Wire elements and counterpoises are shown as
switching, and a broad beam is considered to be in itself being perhaps the easiest and cheapest system if suit-
an advantage rather than otherwise. At any given time able supports (eg trees, chimneys) are available but
the area of interest may subtend an angle of up to 90° or most other types of element are suitable. With wide
so. If the beam is narrower than this there will certainly separation of the elements and correct phasing the gain
be less risk of interference from other stations in the is 3dB for all directions; the adjustment is by altering
same distant area but such interference can usually be the relative lengths of the two feeders and with four
_ avoided if frequencies are checked before using them, switch positions plus a reversing switch, as shown, the
whereas too narrow a beam will result in contacts being worst phase error is 22:5°, corresponding to a drop in
missed. On the other hand, the relatively narrow gain of 0-7dB. This can be reduced to 0-17dB by
beamwidth in the vertical plane reduces both short-skip doubling the number of switch positions. Each feeder
interference and noise from outer space. should be accurately matched at the antenna.
The biggest problem with vertical elements is to A 4:1 balun transformer is used as a phase-reversing
prevent them interacting electrically or mechanically device but alternatively it would be possible to switch in
with supporting structures. To avoid electrical losses an extra \/2 of feeder or merely to have twice as many
positions (same line lengths between them) on the
there are three main methods:
phasing switch. The balun is preferred since it is
(a) The use of self-supporting elements. cheaper (p51) and more compact than the extra cable,
(b) Suspending the vertical elements from catenaries particularly for the lower frequency bands, and a
which can also be used to provide capacitive load- separate reversing switch is more convenient because
ing at one or both ends. the easiest method of adjustment is to find a null and
(c) Arranging that the elements “‘billow out”’ from the then reverse the phase of one element.
supporting mast so that they are well clear of it for With reduced spacing between the elements the
most of their length. Metal masts and guys are best possible gain increases to 5dB along the line of the
avoided but may be dealt with if necessary by elements, but at the same time shrinks to only just over
methods described elsewhere (p163). OdB at right-angles to this, being 4-6dB and 1-1dB

183
184 HF ANTENNAS FOR ALL LOCATIONS

Fig 13.1. Phased pair of vertical

antennas. Each step on the
phasing switch alters the rela-
tive phase by 45°. The bifilar
4:1 balun (balun-type) transformer pro-
used as phase
reverser
vides phase reversal so that a
total of eight phasing steps of
Phase reversal 45° (relative) is available. Elec-
switch trical lengths of CD, DE, are
equal. The figure shows vertical
monopoles working against
artificial earths but a similar
arrangement can be devised for
any pair of antennas. The coun-
terpoises should be at least 1-
2m above ground [1] for negligi-
ble losses due to earth currents

To transmitter

respectively for a spacing of 0-25\. For spacings less p76. A 1:1 balun provides the antiphase voltages and
than about 0-35), phasing lines used in this way usually the elements share a common two-radial ground plane
result in unequal currents as explained on p76, and a or counterpoise. Because the radiation resistance is
lumped network of the type shown in Fig 5.13 is fairly low it is not advisable to use a spacing less than
preferable. In the case of wide spacing one result of /8 on the lowest frequency band.
paralleling the feeders will be to present a mismatch at Three radiators spaced in the region of 0-15 to 0-35)
the transmitter. If necessary a separate pi-network can may be used two at a time or grouped so that two are
be used to correct this. For simple in or out-of-phase used in parallel to form one of the elements; this allows
operation the antenna may be designed to have an direction to be switched in 60° steps and makes avail-
impedance Z,/V2 or V2 Zy so that if the feeders are able a gain fairly close to 4dB for any direction. In both
even or odd multiples respectively of 4/2 the transmitter of these cases the beam remains effectively a two-
sees an impedance equal to Z,. With close-spacing and element system.
phasing by the method of Fig. 5.13 the impedance
varies with phase angle, and use of a separate pi-
network is advisable, the feeder lengths being made Beams with three or more elements
equal to a multiple of )/2. The phasing of three or more wide-spaced elements (or
The elements need not be identical and for the separate antennas) is feasible, providing a gain equal to
lower-frequency bands various assortments of anten- the number of elements (p24) if all are fed with equal
nas, towers and mast can be pressed into service as power. The effect of unequal power division has.
described on p114. It is unlikely to be worth bothering already been analysed for two elements (p100) and
with spacings much less than )/8 and even then the gain shown to have little effect for quite large ratios, pro-
is only along the line of the array. Even with this vided the elements are of equal performance. The
restriction the possibility of useful gain even on ‘“‘top easiest way to share power between elements or groups
band”’ exists for some locations. of elements is by splitting it equally, eg power is shared
The VK2ABQ antenna, either in its basic form (Fig equally between, say, elements 1 and 2. We thus have a
12.3) or using V-shaped elements, is suitable for use in group of two elements out of a set of three, one of
the vertical position since it meets the “billowing out”’ which is left over so that with further equal division of
requirement without demanding excessive height. Two power half the total goes to 3 and half to 1 + 2; in other
could be mounted at right-angles for all-round coverage words 1 and 2 each receive only a quarter of the power.
but there is little point in this in view of better ways of This reduces the gain from 4-77dB to 4-61dB, a negligi-
using four such elements as discussed below. By suit- ble difference so that there is no urgent need to resort
able arrangements of four guys, from 90° to 120° of to the complexity of more accurate power division.
rotation is possible, this being more than adequate if If a number of scattered elements are co-ordinated in
the beam is reversible (p220). Half-wave elements this way the overall pattern is likely to vary greatly with
may also be fed at their lower ends using the modified the beam heading and large lobes in unwanted direc-
Zepp feed or a tuning unit as shown in Figs 13.2(a) tions can be expected. Large arrays can in general be
and (b). Given sufficient spacing it is possible for spaced and phased to provide either maximum gain or
arrangements of this type to be pressed into service as a minimum sidelobes, but amateur requirements are
beam at half the frequency as shown in Fig 13.2(c); L more likely to be met by closely-spaced groups of
and C provide the necessary phase shift as explained on elements and, given sufficient space, up to four such
 VERTICAL BEAMS 185

groups spaced well apart from each other. Beyond this

size the cost per decibel is escalating rapidly, particular-
ly in terms of space, labour and complexity.
A number of groupings of up to four elements have
been subjected elsewhere to a computer study [2] which
N
established a clear margin of superiority for an arrange-
ment forming the basis for Fig 13.3(a); this shows what
‘\

To transmitter is in effect a three-element array, a pair at opposite

corners of the 4/4 square being used in parallel as the
main radiator and the other two operating as a reflector
and director respectively. The recommendations re-
quire equal currents in all four elements with phase
differences of minus and plus 110° respectively from the
driven element, but there is some doubt as to whether
this specification can be met adequately with parasitic
operation of the elements as shown in Fig 13.2.
The use of four separate feeders is recommended in
any case since this allows (a) beam directions to be
switched instantaneously without using relays (b) either
remote tuning of parasitic elements or the setting-up of
correct operating conditions for driven operation by
means of phasing and matching devices located in the
— Use high LC ratio shack, including the possibility of setting up different
patterns for transmission and reception.
& To Fig 13.4 suggests two methods of achieving the
transmitter correct phases and amplitudes, one based on the

y
lumped-constant phasing line of Fig 5.13 and the other
on switched lengths of feeder cable. The objections to
Artificial earth the latter (p75) can be overcome by the use of an
3 or 4ft (1m) (b)
adequate number of switch positions, together with
variable couplings in the common tuning unit for
ys

<

adjustment of amplitudes. Phase and amplitude adjust-

ments will be interdependent in both cases, this being
an intrinsic feature of close-spaced beams.
Of the two alternatives (a) provides more flexibility
r and precision but may be difficult to adjust in practice.

‘S
,
Parasitic operation is simplest and should be acceptable
provided the currents in all four elements are equal
Yao balun within 20 per cent or so. The computer results predict a
gain of 6-4dB with back and minor lobes not exceeding
—24-4dB compared with the forward field. The beam-
width is just under 90° so that with a switched beam a
loss of nearly an average S-unit can be expected in the
worst directions, though the gain exceeds 5dB over a
total of some 240°.
To transmitter
Most of the drop in signal for the 45° directions can
(c) be avoided by switching to a different configuration,
Fig 13.3(b); in this case adjacent elements are paired
Fig 13.2. The two )/2 elements of (a) or (b) may be suspended and the array becomes almost the exact equivalent of a
from a single pole. At (a) the modified (G6CJ) Zepp feed is used
and either an open-wire feeder or coaxial feeder via a 4:1 balun two-element vertically polarized cubical quad having an
may be connected at xx. An artificial earth such as the )/4 wire element spacing of 4/4. This changes the beam heading
shown dotted may be advisable. At (b) a high LC ratio (eg 25pF at by 45° and from the computer results the gain becomes
14MHz) is desirable. The length of the artificial-earth wire is not
critical. Matching is adjusted by movement of the coupling coil. 5-1dB with 16-6dB discrimination against all back and
At (c) the same elements are used as a phased monopole array at minor lobes; though not as good as before this perform-
half the frequency. The principle is similar to Fig 5.11(a); the ance is still acceptable by most standards. Changing to
balun provides the antiphased voltages and LC the required
phase shift, the right-hand element acting as a reflector. The a different configuration for the intermediate direction
radials may be replaced by a loaded counterpoise complicates not only the switching but also the phasing
186 HF ANTENNAS FOR ALL LOCATIONS

Director (a)

a i]

Direction of y To transmitter
beam

Al

a a|>
(a)

A2 A3

(b)

To
transmitter

Fig 13.4. Simplified diagram showing two methods of power

sharing and phase adjustment for three antennas or elements. (a)
uses two resonant lines; C1 determines power ratio between 1
and 2 and C2 between (1 + 2) and 3. Taps provide phase adjust-
ment. At (b) variable-length lines are provided for phase adjust-
(b) ment and amplitudes are adjusted by varying the couplings; to
simplify the diagram switching is shown for one pole only, but
both poles must be switched as in Fig 13.1 and unused portions
of lines disconnected. Phase reversal is obtained by changing
over the leads to the coupling coils

and matching systems, though only to the extent of

requiring an eight-position switch and denuding the
junk-box of a few more components.
From arguments presented in earlier chapters one
might expect a spacing of 0:25) to be excessive. This
tends to be confirmed by figures for 0°167A\ spacing
To transmitter
which show an improvement of 0-2dB in gain and 4dB
Fig 13.3. (a) Three-element Yagi making optimum use of four in discrimination, though the author of the reference
elements. BC are fed with equal currents in the same phase, A considers the extra gain will probably be offset due to
and D being respectively advanced and retarded by 60° from the
antiphase condition. For best performance the currents should be the larger currents. This seems open to doubt (unless
approximately equal in all four elements. Artificial earths have possibly there are large earth losses) in view of the
been substituted for the extensive systems of ground radials advantages which have been so clearly established for
assumed in the reference, but the principles should be equally
applicable to all types of element, including those of Fig 13.2 close spacings in the case of horizontal elements (p71),
assuming correct amplitudes and phases. Computed gain from even small ones with low radiation resistances; the
reference [2] is 6-4dB and suppression of back radiation is 18dB or more important point to emerge however is that spac-
better for all angles. If only one direction is required the elements
may be rearranged in line as a four-element Yagi to obtain ing is non-critical. Even for a spacing of 0-333A the gain
slightly better gain and directivity. (b) Alternative two-element only drops to 5-6dB and for very close spacings the
configuration analogous to quad antenna. Gain is reduced to basic geometrical principles featured in earlier chapters
5-1dB but the 45° shift in direction means that by using both
alternatives 5dB gain can be obtained even in the worst direc- must be assumed applicable; in effect gain and directiv-
tions ity remain unchanged in principle down to zero spacing,
 VERTICAL BEAMS 187

Fig 13.5. (a) and (b) show two

ways of supporting short end- a \
loaded vertical dipole elements.
Sufficient loading can be “So,
accommodated in case (a) for a a eT) ren
vertical dimension of about pee
0-18\, and some further reduc-
tion in height is possible in case
(b). (c) and (d) show two (a) Te
methods of feed and typical \
dimensions for 14MHz are given
at (d). The Zepp feed is the same ‘
as for Fig 13.2(a) but it is worth
noting that the feed does not
have to be attached to the end
of the radiator; connected as
shown here it may provide a
better match or improved band-
width or both

Gamma match
600N line

(c) (d)
To reflector for neutralization
Modified
zepp feed of excess coupling
(if necessary)

long before which losses or bandwidth become un- adequate except that antenna wires at about knee-
acceptable. height above the lawn tend to be unpopular (even if
It is reasonably certain that an array of this type they fail to cause rf burns).
could be erected in a very small space provided it is not Theoretically, though in practice it does not seem to
too closely surrounded by buildings etc. As in the case make much difference, earth losses should be reduced
of horizontal arrays with comparable element spacings, and gain slightly increased by giving the poles a 7ft
it is evident that arrays of this type can be designed to (2-1m) fencing post (or something similar) as a founda-
operate efficiently at a number of frequencies spaced tion. Fig 13.5(c) shows a form of gamma match which
over a bandwidth of at least one octave, subject to the has been used successfully with a long Yagi array
limitations of the individual elements which can make composed of similar elements. One suggestion is that
use of one or other of the multibanding methods coaxial feedlines should be brought away at right-
described in earlier chapters. angles into the centre of the array, then bunched
It is believed that the systems just described come together and brought out midway between two of the
close to being optimum designs of vertical array and it elements, preferably keeping them horizontal in order
is likely that a number of amateur requirements can to minimize possible coupling between elements and
best be met by two, three or four such groups suitably feedlines. The usual precautions against feeder radia-
phased. Many variations on these themes are possible tion (p49) should in any case be observed.
and Fig 13.5 shows two ways of constructing an efficient Fig 13.6 shows a ‘“‘monopole”’ variant of this arrange-
low-cost array using very short end-loaded dipoles ment likely to be more convenient; with a substantial
supported catenary fashion from four light poles. For a amount of end loading the current distribution is
14MHz array, 15ft (4-6m) bamboo canes would be relatively uniform and the 2:1 advantage in terms of
188 HF ANTENNAS FOR ALL LOCATIONS

assumed to be at right-angles so that interaction

between them is minimal: we have h = 0-05, t = 0-25
and P,/P,, = 0:39 for the counterpoise and similarly
for the top load. Strictly speaking, the formula should
be corrected to take account of the sinusoidal current
distributions in the long wires, but it will be seen that
even in this very extreme case the total of the unwanted
radiation is at most roughly comparable with the
wanted radiation. The loss decreases very rapidly with
improvement in element shape and for a 25ft (7-6m)
dipole loaded by 65ft (19-8m) at each end the loss is
only 1-2 per cent or about 0-01 of a S-unit! This is still a
fairly extreme case and it seems very difficult to
envisage a practical situation in which any symmetrical
arrangement of loading wires or artificial earths would
have a significant effect on gain or directivity. On the
other hand, extra radials might result in some reduction
Fig 13.6. This shows part of a monopole version of Fig 13.5,
of earth losses, whereas any lack of symmetry (eg due
allowing even more reduction in physical height. The elements to a phase shift between two radials) could lead to
may be inverted to increase the effective height and can be fed as increased losses.
at Fig 13.5(d) A multiband system analogous to Fig 11.14 is sug-
radiation resistance of dipoles over monopoles (p104) gested in Fig 13.7, which illustrates a multiband ele-
largely disappears so that for a given length of radiator ment similar to the parallel dipoles of several earlier
there is little to choose between them. A simple designs except that physical separation is achieved by
low-cost version of this can be based on the use of a multiple top-loading wires carried by spreaders as
single feeder with plugs and sockets or clips for pairing shown in the diagram. Elements may be of the same
the elements or making them into directors or reflec- size for all bands, with capacitive stretching of the
tors as required, but to achieve the convenience of higher-frequency elements as indicated. This is a re-
switched beam headings it is necessary either to use finement which should yield some improvement in
four feeders or for example a Ledex-type switch. performance but those in search of greater simplicity
Manual operation is possible also in the case of can use A/4 elements for each band. Checks for interac-
dipoles whose lower ends are accessible; in this case the tion should be made and remedies applied if necessary,
modified Zepp feed should be used (p46) in conjunc- as explained on p110. |
tion with Fig 13.5(b) but switching at high-voltage The above systems are all by way of suggestions
points is difficult so that, for a switched system, Fig 13.6 inspired by the computer study in conjunction with
is preferred to Fig 13.5. Any accessible points of high rf
voltage must of course be well protected from
accidental contact.
In cases of loading by T-pieces, including the use of
long counterpoise wires or two-radial ground planes, it
would be reasonable to expect some radiation from
these parts of the system, resulting in loss of power in
the wanted modes. To investigate this the problem was
modelled on similar lines to the treatment of dipoles on
p9, from which the ratio of power in an unwanted mode
to that in the wanted mode was found to be approx-
nately P,/P,, = t/Ah2
where ¢is half the length, in wavelengths, of a symmet-
rical loading wire or the length of one radial in the case
of a two-radial ground plane, and h is the length of the
radiator. Using this it is easy to demonstrate that
appreciable loss is possible only in very extreme cases;
consider for example a 13ft (4m) vertical at 3-SMHz Fig 13.7. Multiband monopole element. Full use is made of the
with a 130ft (39-6m) counterpoise and a top load which available height on all bands, using series capacitors to establish
resonance at the higher frequencies. Separate series-resonant
needs to be nearly as long as the counterpoise. In order artificial grounds are provided for each frequency. Feeder may be
not to complicate the example, the two long wires are connected to one or other of the coupling coils as required
 VERTICAL BEAMS 189

practical experience of some of the types of elements driven element being supported in the centre of the ring
described, and it is believed they should be near- by cords from the four corners which also carried wire
optimum systems, a number of amateur requirements crosses for end loading. Lower-end Zepp feed was
being best met by two, three, or four of such four- intended, and such an arrangement is attractive since
element groups suitably spaced and phased together. directional switching from the shack is obtained with
A 4 X 4 system with group spacing upwards of 0-8A only one feeder and four relays, though it lacks some of
could be expected to provide some 11-12dB of gain, the sophistication of the four-feeder system. One can
thereby more than making up for the deficiencies of also have a “poor man’s version’ without relays,
vertical polarization (at least for 14MHz) as compared assuming access to the lower ends of the parasitic
with good conventional horizontal rotary beams at elements so that lengths can be altered manually.
heights of 40—-SOft (12-15m). The height required for
the vertical array is dictated largely by the need to
prevent accidental contact with the lower wires of the
system, though this also helps in other respects. Good Radiator

results on 14MHz could be expected with a height of

about 6ft (1-8m) plus, say, 1-2ft (0-3-0-6m) clearance
for the counterpoises, though a loss of some 2—-4dB
might accrue from (a) extra ground losses and (b) the
smaller array factor for the antenna plus its image.
Unipole operation with T-shaped elements and an
inductively-loaded counterpoise is assumed as in Fig
a9.0.
One practical point to be noted is the risk of overcou-
pling when the ends of elements are brought close
together as in Figs 13.5 and 13.6. With all elements
driven a modest degree of overcoupling can be taken
care of automatically, but with parasitic operation there
is an optimum coupling. This will usually not be the
same for a director as for a reflector, though a reason- Side view Detail
able compromise in the case of a three-element reversi-
ble beam should be possible. If coupling is excessive the Fig 13.8. ‘““Quick-heading”’ (QH) beam. The length of the stub D is
top loading wires can usually be reduced by at least 20 adjusted to provide reflector operation of each element in the
outer ring and length C is adjusted so that closing of the relay
per cent without ill-effect, except the need for a small contacts results in director action. By converting any one element
loading coil at the top of the vertical wire in order to or adjacent pair of elements into directors, eight beam-headings
restore resonance. Alternatively for monoband opera- are obtained at intervals of 45°. B is not critical. See Table 13.1 for
dimensions. (Antennenbuch)
tion it may be sufficient to increase the spacing some-
what, though it should not exceed 4/3. The remaining From inspection there seems no obvious reason to
method is neutralization (p85) which is very simple in expect four parasitic elements to show a marked advan-
some cases, eg Fig 13.5(b) with lower-end Zepp feed as tage over three. The case of a triangular group of three
shown at (d), since the ‘‘hot”’ ends are all brought to the elements with a fourth in the middle was included in the
same point and an antiphase voltage is available from computer study quoted earlier, and found to be little
the opposite side of the open feedline. better than three elements on their own which had a
The above principles have also been applied to some gain of 4-35dB and a front/back ration of 10-45dB.
vertical arrays based on the quad. These offer a number Although the cases are not precisely comparable it
of advantages and form the subject of the next section, seems likely that the “‘pig in the middle” arrangements
but reference should first be made to one other type of (at their best) waste one element, but some readers
array which uses a ring of switched elements. This has may find the idea sufficiently attractive to be worth
appeared in various forms such as the ‘““QH”’ (quick- pursuing.
heading) beam illustrated in Fig 13.8. By switching any
single or any adjacent pair of parasitic elements to act
as directors eight beam headings are obtained. The Table 13.1. Dimensions for QH-Beam [3]
reference [3] from which this diagram was taken quotes Distance of radiator
a gain of 6-5dB but states that ‘‘its construction is Radiator from parasitic elements
Band (MHz) length (all lengths in cm) B C D
however somewhat complicated’’.
This need not be the case and in one version designed 14 1,027 253 460 10 40 90
21 686 170 307 7 27 60
(though never quite completed) by the author the 28 512 126 230 5 20 45
parasitic elements were arranged as in Fig 13.5, the
190 HF ANTENNAS FOR ALL LOCATIONS

The half-loop vertical array

The author has long been intrigued by the idea of
integrating a high-gain beam antenna with the guy-wire
system so that the mast and the antenna support each
other, but some partial solutions of this problem using
horizontal polarization fell a long way short of the main
objectives. These included the possibility of using much
less substantial structures having little visual impact,
easy to erect, light enough to cause no damage even if
they blow down and unlikely to require planning (a)
consent. This goal has now been largely achieved by the
use of vertical arrays consisting of four half-loops
mutually at right-angles and suspended from a single
pole, Fig 13.9. As an additional feature the beam is
intended to be switchable in any direction. With all
these advantages the loss of one S-unit compared with a
horizontal beam at 40ft (12m) (p135) may be thought a
small price to pay. An alternative option would be
replication of the array (2—4 units), much as discussed
earlier in the context of Fig 13.6. If the user is prepared
to dispense with remote beam switching or is able to
devise a satisfactory switch for use at points of high rf
voltage, the arrangement shown at (b) is preferred.
There is in this case no need to support the weight of
feeders and a much more slender mast structure is
possible; this uses the same method of feed as Fig
13.2(a).
An attempt to realize similar objectives has been
described by UA3IAR [4] who uses a pair of horizon- Q
tally polarized diamond-shaped quad loops at right- (b) /‘
angles. These are supported by a single mast so that the \ >

sides act as guy wires, and looking down from above \

\

one sees an X configuration. Inevitably this suffers Balun \

cn

from the disadvantages of X-beams in general (p81) ==

= —

and, referring to Fig 5.4, the shape of the published

radiation pattern suggests a phase-angle ratio /@p of
1-8, corresponding to a gain of only 3dB. If it is
necessary to lose an S-point anyway, it seems to the
author that given the choice one might as well combine Fig 13.9. This array uses four vertical half-loops suspended from a
single pole, with the top portions of the loops forming part of the
it with the advantages of low height and, incidentally, guy-wire system. In the preferred arrangement diagonally-
vastly greater reductions of rear and minor lobes, opposite pairs are fed in phase as the driven element, the other
assuming the results of the computer study [2] to be two elements being tuned as reflector and director. The equiva-
lent quad configuration, analogous to Fig 13.3(b), may also be
applicable despite the bends in the elements. used. For 360° directional switching the four feeders may be
The basic problems with X-beams are low radiation brought into the shack or a “switchbox”’ may be mounted on the
resistance and large dips in the radiation pattern with mast. The arrangements shown in Fig 13.2(a) and (b) can also be
adapted assuming access to the lower end and the use of
zero front/back ratio for the 45° directions as explained plug-and-socket connections
in Chapter 5. Both of these are avoided by changing to
vertical polarization as in Fig 13.9(a) since maximum
current is now in the middle (or perhaps one should say emerged in the course of the earlier discussion. In the
at the side corners) where separation is maximum. case of the vertically polarized quad there is no need for
Although each quarter-loop on its own produces radia- the elements to be joined at the top corner; insulating
tion in all three planes, on considering the whole of the top ends of all four wires leaves one free to connect
each element one finds that everything cancels except the bottom ends so that either the quad type of current
the wanted vertical components. What we now have is distribution is retained, or the change made to an
virtually identical with the “‘second choice”’ (quad-like) optimum Yagi in line with the computer study. In the
configuration for the four vertical elements which case of the Yagi the parasitic elements have become
 VERTICAL BEAMS 191

are used. Multibanding is possible by the use of linear

resonators but requires further duplication or switching
of tuner units.
One suggestion, shown in Fig 13.11, uses a lightweight
21/28MHz stretched dipole, similar to that of Fig 12.6,
stacked in parallel with the 14MHz element. The
lengths of twin feeder (see caption) are not sufficient to
result in serious losses or wet-weather problems, and if
any interaction occurs between elements it should be
(a) preventable by linear trapping (p110) or the use of
relays (p116). Lower-end feed with band and direction
changing by manual operations using plugs and sockets
Resonant
line
!
is possible, as are also arrangements based on Fig
| 13.2(b) with remote switching of tuning units.
|
|
Symmetry is important for cancellation of the un-
wanted modes and requires the loops to be of approxi-

YY: Uy
mately the correct size in the case of Fig 13.2. On the
other hand, Fig 13.10, derived from the DJ4VM type of
quad (p160), is less critical in this respect but involves
A B more weight and windage so that a slightly stronger
support may be needed. Loop shapes are not critical

Alternative Birds’ (or

(b) beam worms’)
directions eye view

A B

Fig 13.10. Arrangement based on DJ4VM quad is completely

symmetrical and therefore multiband if used with open-wire. For
satisfactory performance at the higher frequencies it is essential
to use the “quad” configuration, ie the half-loops must be paired
as at (b). The length of the feedlines AB is not critical and they
may be routed via a switch box (or anchoring points) located near
the mast

inverted-Vs turned on their sides, but from _pre-

vious analysis of the inverted-V (p21) this should do
no harm, particularly as the upwards and downwards
radiation from the two parasitic elements is largely
cancelled. Both configurations have been used success-
fully with no obvious practical difference in the trans-
mitting case, though this would be quite difficult to
measure and some improvement in reception is to be
expected in the case of the Yagi.
Use of opposite pairs of loops as a monopole beam at
half the frequency as in Fig 13.2(c) is possible in this
case also, but the temptation to use two loops in
parallel for each element should be avoided as it
Fig 13.11. Illustration of three-band (14/21/28MHz) operation
reduces the mean spacing by 30 per cent. using pairs of series-tuned elements. The sketch shows one
For driven elements there are two main alternatives, corner of the array only; for 14MHz the capacitors (about 40pF
end feeding as in Fig 13.2(a) or (b) and a symmetrical for open-wire line) tune out the inductance of the line BC. ABC is
resonant (3)/2) at 28MHz where C2 has no effect. At 21MHz C2
feed as in Figs 13.9 or 13.10, each with its own set of (abour 8pF each) tunes out the inductive reactance of the ele-
advantages and drawbacks. Feeding at the lower end is ment. These elements must be paired as in Fig 13.10. The length
best mechanically and allows manual beam rotation BC should be 9ft (2:4m) so that C requires supporting about 4ft
(1:2m) out from the mast. Twin feeder (300 or 7522) may be used
using plugs and sockets etc, but is difficult for remote from the balun to the 14MHz element. The )/2 shown refers to
switching unless four separate Zepp feeds or tuner units 28MHz
192 HF ANTENNAS FOR ALL LOCATIONS

4 half-wave tuning and matching components, or an open-wire line

half-loops at
right- angles can be run all the way to the shack with matching stubs
brought into circuit as required for band changing.
Director \ A fully rotatable array can be constructed on this
basis, either using a support substantial enough not to
need guy wires or, following a suggestion by G6BZ,
hanging from the centre of a catenary suspended
between two large trees or a house and a tree.
These are just a few of the many arrangements to
which this system lends itself; only a few of these have
been investigated so far, and for transmission the superi-
ority of the three-element Yagi versus the two-element

[
4:1 balun
quad expected from the computer study has failed to
emerge clearly from the author’s experiments. There is
a margin of doubt, however, both as to the validity of
applying the computer figures to a parasitic array and
the adequacy of the experiments.
The quad configuration as shown in Fig 13.13 is
inherently simpler and with lower-end feed requires
Sor ee

Fig 13.12. Rotatable array using four vertically polarized half-

loops similar to those of Fig 13.2(a). The optimum configuration
of Figs 13.3(a) and 13.9 is applicable in this case also, lower-end
Zepp feed being more convenient for rotation. Dimensions of Fig
13.2(a) are applicable and any type of feeder may be used. For
14MHz the balun should be located 40in (102cm) from the end of
the stub. Driven elements are 33ft (10-1m). Reflector and director
lengths typically 37ft (9-4m) and 34ft (10-4m) respectively but
should be found by experiment
(a) wW——

and, in the case of Fig 13.11, fine tuning by adjust-

ment of horizontal wire lengths may be found easier
than adjustment of C1 or C2. In the case of Fig 13.2
equality of element lengths is relatively unimportant;
any short bit of wire, whether it forms part of either
element or is just an “‘odd bit hanging on’’, is in effect a
small capacitance to the ground which tunes the ele-
ment but contributes nothing, useful or otherwise, to
the radiated field. Neither does its removal or addition
upset the balance of the elements since near an end the
voltage changes only slowly along the length. Putting
this another way, one finds that detuning a dipole has Feeder (2) -

almost the same effect on both voltage and current, so

that since the two dipoles have their ends tied together
they must have the same currents, near enough, even if (b)
their lengths differ by several per cent. This allows fine
tuning to be effected by varying a short length of wire
common to both elements and either in series or ‘‘just Feeder (1)

hanging on’’.
Fig 13.12 shows a rotatable arrangement which is
cheap and simple to implement but requires the use of
Fig 13.13. Directional switching with lower-end feed. The quad
what is in effect an ordinary quad spider, though the configuration is used and as shown in the bird’s-eye view at (a)
four arms can be of lighter-than-usual construction. the elements are paired 1/4, 2/3 for the east-west direction or 1/2,
Stacked elements may be used as in the previous 3/4 for north-south directions. Two separate Zepp feeds as in Fig
13.2(a) are connected at F1 and F2 respectively, allowing either
example, with relays for beam reversal, though alterna- element pair to be driven, the other pair being remotely tuned as
tively the lengths can be changed manually at the lower a reflector. Low-capacitance high-insulation switches or relays
end. Assuming four guy wires, 90° rotation is possible are required for NS, EW selection and neutralization of excess
coupling capacitance will be required as shown at (b), see text.
(eg by means of cords routed into the shack). A Two neutralizing capacitors, though the second one is not strictly
centrally located waterproof box can be used to house necessary, have been included for symmetry
 VERTICAL BEAMS 193

Fig 13.14. Vertical Yagi array us- Reflector Radiator Director

ing short end-loaded elements. pene tit Te Oreo SON ee 0
Height is not critical, no drop-off i
in performance being noticed
even with “sag” to within 1ft
(30cm) of the ground. The reflec-
tor and director are tunable by
adjusting the length of the ver-
ticals or the span of the lower
horizontal wires. Dimensions
are rough guidelines _ only.
Alternative feeder systems
which have been used success-
fully include those of Figs 13.5(c)
and (d). Additional directors
may be used to fill whatever
space is available; the optimum Any height
spacing for such additions is
probably about 0:35) but opti-
~mum design data for end-loaded
elements are not available

only two-pole two-way switching at the antenna for The vertical extent can be further reduced by mono-
selection of NS or EW directions, the two feeders being pole plus counterpoise arrangements as in Fig 13.6.
taken to switched phasing units in the shack for beam
reversal. In the case of the quad some neutralization Log-periodic arrays
was found advisable because of excess coupling, and for One solution to the problem of multiband operation
experimental purposes the neutralizing capacitance was would be to stack three separate Yagi arrays covering
successfully contrived by draping a foot (30cm) or so of 14, 21 and 28MHz directly in front instead of alongside
well-insulated wire loosely over one of the conductors. or on top of each other. This increases the overall
So far the lower-end switching operation has only been length but using four stacks it would be possible to
tried with plugs and sockets but switching may be obtain in this way 360° coverage with an average gain of
feasible with high-grade low-capacitance relay contacts nearly 6dB. A single fairly high support would be
(ceramic insulation), assuming the tuning out and neut- needed, surrounded by four shorter ones at a distance
ralization of contact capacitances which is suggested as of about A/2 at the lowest frequency.
an interesting project for the experimenter. Of the Thus far the reader will be on familiar ground,
various Yagi options Fig 13.11 seems to be the most though he may well ask, “‘why not stack the beams
generally useful arrangement if multiband operation concentrically?”” Less space is needed, but to take
plus directional switching from the shack is required, advantage of this the outer ring of supports must be so
whereas Figs 13.9 or 13.12 provide cheap and simple much higher that it is almost certainly better to use
monoband alternatives. more space if it is available. It must next be asked
whether the arrangements just described are in fact the
Yagi arrays with short end-loaded best way of using the space; this would almost certainly
elements have been the case prior to the invention of log-
Elements of the type used in Figs 13.5 and 13.6 can be periodic arrays which are one of the topics for consid-
used to form long Yagi arrays as in Fig 13.14 or added eration in the next chapter.
to arrays such as those just described to provide Though somewhat similar in appearance the log-
increased gain in a particular direction, or more than periodic antenna is in fact a major breakthrough since it
one direction provided elements not in use are thrown provides about the same gain, and in addition can
out of resonance. Good results have been obtained on readily provide continuous coverage of the entire fre-
7MHz over the long path from the UK to Australia quency spectrum from 14-28MHz. This means that the
using up to six elements suspended from one or two 18 and 24MHz bands are automatically included with
catenaries, the lower ends being at times only a foot or no need for special dispositions and, within the indi-
so and never more than 6ft (1-8m) from the ground. vidual frequency bands, there are no bandwidth wor-
The basic element size was 25ft (7-6m) for the vertical ries. For optimum design rather more elements (at least
portion and 30ft (9-1m) for the horizontal loading 12) are needed in the log-periodic case but by further
sections, one or both of the lower arms being bent back increasing the number the frequency range may be
or the vertical portion reefed in to obtain director extended. Alternatively the gain may be increased by
action. Best results were obtained with the director as much as 2dB, but this narrows the beam by some 25
elements shortened so that the currents decreased by per cent so that more arrays are needed for omni-
about 30 per cent from each element to the one in directional coverage, an alternative option being
front, the spacing being 1/4. acceptance of inferior performance in a few directions.
194 HF ANTENNAS FOR ALL LOCATIONS

The log-periodic array described on p202 can be

erected vertically for this application, assuming a cen-
tral pole height of at least 40ft (12m) and about 28ft
(8-5m) for the outer ring. Before embarking on such a
project consideration should be given to using a set of
three two- or three-element reversible horizontal arrays
arranged at 120° intervals around the same central pole;
this should give appreciably better dx performance
besides using a lot less wire and one less outer pole!
The log-periodic array provides in some respects a
much neater solution, but for unequivocal justification
of its use in this application emphasis must again be
placed on the low-height possibilities which the above
example has failed to realize.
For keeping height to a minimum monopole arrays
offer the best prospects; the earthing problem can be
resolved by using separate two-radial ground planes for
each element, but with a total of perhaps 12 elements
for coverage of one octave the amount of wire needed
is rather large and the numerous radial wires, however
useful they may be as part of the antenna system, are
not an asset to the garden. As an alternative the use of
(a) short counterpoise earths with loading coils (Fig Fig 13.15. Suggested outline for miniature log-periodic array. The
13.1) and (b) top-loading wires forming part of the design principles outlined on p202 are applicable to the vertical
elements including spacing and top-loading, and to the feeder
catenary as in Fig 13.14 is suggested. Lengthening the systems. For the artificial earths “anything goes” provided they
array would allow the length of the loading wires to be are series-resonant, since they do not contribute directly to the
increased, and the possibility exists of achieving a radiation. However, if these are allowed to swivel or are replaced
by wires capable of being pegged to run in any direction, it
height of 10ft (3m) or less for a length of 120ft (37m) or should be possible to realize some control of the coupling
so before the drop in performance becomes serious, between elements. Some inductive top-loading may be used if
provided suitable amplitudes and phase angles are necessary
maintained.
Uncertainty arises from the lack of simple explana- was suggested that radiation free from ground reflec-
tions or formulae for log-periodic antennas which take tions might be obtained by erecting a vertical antenna
into account the mutual couplings that play a vital role just above tree-top level. This idea is easier to conceive
and will be considerably influenced by measures such as than to implement; any readers with large woodland
those just described. Nevertheless there is evidence areas on or adjacent to their property who might like to
from the literature of a great deal of design latitude, try it out would be well advised to experiment with
and experiments with Fig 13.15 as a starting point a simple monopole such as Fig 11.9(a) in the first
may lead to a workable design. In the possible event instance. This could be checked at ground level for dx .
of overcoupling this can be controlled by shortening the performance in comparison with a horizontal reference
top loads and introducing some loading inductance at antenna and then raised above the tree tops, perhaps by
the top of the vertical wires, but this is a new field not using successive extensions on the principle of a chim-
recommended to the novice. Design formulae for ney sweep’s brush and choosing a period of calm
straight elements will be found in the next chapter. weather!
Another method of erection, capable of being used
for a beam if the trial proves a success, is shown in Fig
Trees and vertical beams 19.9. It is assumed that the house is in a clearing at the
It is known that tree trunks can have a serious effect on edge of the wood or that it is at least possible to clear a
vertical elements up to distances of at least a few feet, rectangular space to fit the antenna. The block and
but up to the moment of writing it has not been tackle is attached to a point on the tree trunk which is
established whether this excludes the use of trees for accessible by ladder or climbing, and if necessary the
supporting beams such as Fig 13.9 which “‘billow out” antenna support may also be secured at a higher point
from their supports. It is also known that (a) a single by means of a strategically placed projection which can
row of trees at a distance of 50ft (15m) or so from a be used to carry a rope over a suitable branch. Guys,
vertical antenna has negligible effect; (b) attenuation of even though not needed for support, will probably be
vertically polarized waves through thick woodland needed for manoeuvring the mast into position.
occurs as discussed in Chapter 10. Arising from this it Depending on height, layout and the construction of
 VERTICAL BEAMS 195

the support it may be necessary to use more than one pick-up. This is particularly important since the most
elementary block-and-tackle. This method depends for frequent criticism of vertical antennas seems to be the
its feasibility on the layout of the tree branches but was amount of noise picked up, whereas one writer faced
used successfully to erect a miniature beam above a SOft with this problem [6] found that the answer lay in using
(15m) high tree, though there were problems with a screened loop, the method of its connection to the
open-wire feeders getting entangled in branches. This is receiver being important; in other words the problem
one case where coaxial feeder is recommended rather was nothing to do with vertical polarization as such.
than open-wire line. This article should be consulted in the event that
Despite the few extra feet of support needed, a beam anyone is still bothered with excessive noise after
on the lines of Fig 13.2, but with a base tuner on the taking the precautions indicated here, but it is impossi-
lines of 13.2(b) instead of the Zepp feed is recom- ble to say at this stage whether there is a residue of
mended in preference to any of the end-loaded wire insoluble problems under this heading.
beams which require additional supports. These will be The choice of beam element can be narrowed down
found very difficult to erect because of interference by to centre-fed or end-fed dipoles, and monopoles oper-
intervening tree branches and the problem of ensuring ating against short end-loaded counterpoises. Easy
survival under severe storm conditions is also less likely methods of erecting dipoles usually involve some form
to be resolved. One half of the array shown in Fig of shortening, and the arrangement of four right-angled
13.3(b), probably with some top loading, might also be elements hanging down from a single pole and fed at
suitable. the lower end by means of the balun-type of Zepp feed
originated by G6CJ, Figs 13.9(b) or 13.12, is recom-
mended as a “‘best buy’’, provided one can tolerate a
Discussion and recommendations pole height of at least 28ft (8-5m). Note however that
A survey of the different types of antenna used by UK the support can be extremely slender (eg using 8ft by
stations has been reported by VK30OM [7]. Out of the 2in (2-4m by 13mm) bamboo garden-canes for at least
147 stations worked over the long path 56 were using the top section) since the antenna itself constitutes a
horizontal beams and 36 some form of vertical antenna, large part of the guy-wire system. The main snag of this
but there was not a single vertical beam. This is a very arrangement lies in the desirability, in general, of
remarkable finding, given the author’s conclusion that keeping switches away from high rf voltages so that in
vertical beams are nearly as good as the average order to change the beam direction one may well find
horizontal beam at 14MHz but do not have to be put in oneself having to mess around with plugs and sockets,
the air, cost very little, are easily and quickly con- though the problem should not be insuperable. Possible
structed, have few maintenance problems and are much compromises include 90° rotation using the arrange-
less likely to lead to difficulties with neighbours or ment of Fig 13.12 in conjunction with beam reversal,
planning authorities. which requires only the use of stubs with relays for
In seeking the reasons for this one finds that vertical short-circuiting them in the centre of the parasitic
antennas have a reputation for being inefficient and elements.
noisy, there are no published designs of vertical beam As a second choice, monopole elements with short
which the author would care to recommend for 14MHz counterpoises, Fig 13.16, offer probably the simplest
or above, and most of the amateur reference books possible method of constructing a beam antenna; they
insist on extensive systems of buried wires requiring a can be used when height restrictions are even more
big investment in themselves, to say nothing of the severe, and provide a simpler basis for evolving multi-
purchase of several adjoining properties! The only band and directionally-switched arrays since these func-
alternative, according to the references, would appear tions can be provided by tuning units at ground level.
to be the use of ground planes consisting of three or An interesting aspect is the portability of the elements
four A/4 radials and suitable only for single-element complete with their ground systems; this is such that
monopole antennas. It is true there are some vertical one could for example dismantle a directionally-
dipole arrays described in the literature but here again switched three-element array and re-erect it, say, as a
one comes up against the unwarranted assumption that five-element Yagi in some fixed direction in less time
the dipoles have to be straight. This means that instead than it takes to rotate the average horizontal beam!
of being the easiest of all types of beam to support, Similarly, there is little effort involved if they have to
since it needs only a single pole with no superstructure, be taken down when not in use.
the vertical array is reduced to being in most cases a A disadvantage of monopole elements and the
non-starter. reason for demoting them to second choice is a loss of
If vertical beams are to have a fair chance it is some 2-3dB in performance, assuming the base to be at
essential to make sure that energy is not radiated from a convenient height such as 3 or 4ft (0-9-1-2m). This is
the feeder or allowed to get into the mains wiring, and because the antenna with its image resembles a single
this should take care of many possible sources of noise dipole rather than a collinear pair, thus broadening the
196 HF ANTENNAS FOR ALL LOCATIONS

pattern in the vertical plane and reducing the gain. This arrangement based on the DJ4VM type of quad, Fig
can be overcome by raising the base height to some 12ft 13.10(a).
(3-7m) or so but dipole arrays then become relatively The ‘‘best buys” in their monoband form should
more attractive, except possibly in respect of direction present little difficulty even to the inexperienced, but
switching and multiband operation. regrettably it is not possible at this stage to provide a
In the absence of direct comparisons the drop in tested and recommended “blueprint” for a multiband
average performance of vertical beams compared with electrically-rotated vertical beam—this field of de-
horizontal beams at 35—40ft (11-12m) was by no means velopment has hitherto been much neglected, and due
obvious. It emerged clearly from tests over a period of to other commitments the author has been able to do
time, being roughly 4dB for the dipole and 7dB for the little more than engage in some basic fact-finding.
low-height monopole arrays; the differences were ex- However, whereas the multiband and multidirectional
tremely variable, however. operation of dipole arrays present some interesting
challenges, the monopole approach based on “‘electrota-
Reflector
Counter poise Beam tor’’ principles, plus say the use of linear resonators for
multibanding, should be relatively straightforward.
1 \
Oe Ce
Ciel
yaeNT ariven element
Approx 2ft overlap with 8 to12in spacing Even so, to “do everything” by remote switching may
require a separate set of four relays per band plus
Fig 13.16. Two-element Yagi using elements similar to Fig 11.16. additional relays for band changing, adding up to quite
Subject to correct tuning the only critical dimension is that of the
driven element, the lower end of which should be at earth a large total! In the case of a ground-based monopole
potential to minimize coupling into the feeder. A director may be system, horizontal arrays have a bigger start at the
added; satisfactory results were obtained with the director and higher frequencies and to reduce this lead the whole of
reflector resonated at 14:8 and 13-9MHz respectively, the reflec-
tor and driven element counterpoise rods being increased in the 14MHz element or an equivalent length with
length to 10ft (3m) and swivelled towards each other so as to appropriate tuning should be used in each case. Multi-
make the reflector current equal to that in the driven element. banding by means of linear resonators employing
Linear loading can be used as in Fig 11.16 but for the driven
element the author has also used four 40in (102cm) radials with a switched capacitors in weatherproof boxes is suggested,
five-turn loading coil, 3in (76mm) diameter and 2in (15mm) long, since in this case there should be no difficulty in
overwound with a single-turn coupling coil. For directional accommodating all bands from 14-28MHz, including
switching the driven element may be surrounded by at least four
parasitic elements, and relays used to short-circuit part of the 18 and 24MHz.
loading inductances This field is strongly recommended to the experimen-
ter looking for a new challenge with good prospects.
Wire beams using short end-loaded vertical dipole or The possibilities are wide-ranging and the author has
monopole elements are recommended as a “best buy” presented no more than a selection of those which seem
for dx operation on the lower-frequency bands, a most promising.
height of 30ft (9m) or so being sufficient for a two- or Further to highlight the possibilities of vertical arrays
three-element monopole array at 3-SMHz. It should be as well as some of the problems, consider the following
noted that short end-loaded dipoles and the recom- scenario. An amateur receives a generous but un-
mended monopole designs are strictly equivalent for a wanted present of a six-element rotary beam; unwanted
given mean height reckoned to the “‘centre of gravity” because he has nowhere to put it. Being reluctant to
of the current distribution, ie the centre of a dipole or hurt the feelings of the donor by failing to put it to good
roughly A/12 from the base of a monopole. Perform- use he finds himself in a dilemma until, after reading”
ance with vertical polarization is affected by ground this chapter and Chapter 5 the solution becomes ob-
constants as discussed in Chapter 10, and the figures vious: cutting the elements in half, they can be used to
quoted above relate to the author’s location, the form four separate three-element vertical monopole
ground being nearly-flat farming land with heavy clay arrays, each having a gain of 6dB compared with 9-5dB
soil and no nearby obstructions. for the original, for which a boom length of 1A has been
It may be asked, why not use a typical horizontal assumed. Cut into four pieces, the boom provides the
beam in the vertical position? The problem here is the counterpoises. Suitably spaced and phased the four
one of electrical and mechanical interaction between Yagis provide a gain of 12dB and the dx signal should
the beam and its support as mentioned earlier. The be roughly equivalent to that from the original beam
author has however made tests with a horizontal beam supposing this to have been erected at a height of 30ft
in the early stages of erection, the mast being nearly (9m). There remains the problem of beam rotation but,
horizontal and the beam rotated into a vertical position armed with the complete works of Heath Robinson and
just clear of the ground, results being comparable with a good assortment of ropes and pulleys, there should be
those for the dipole array described above. Quads have no problem in devising a solution making use of the
also been used successfully by a number of amateurs beam rotator supplied with the antenna. An alternative
but require feeding at a side corner which tends to be method, based on the principle of running round in
inconvenient. The author has also successfully used an small circles with an element in each hand is however
 VERTICAL BEAMS 197

better for the figure and (if one is young and fit enough) [3] Antennenbuch, K. Rothamel, Telekosmos Verlag,
probably quicker! Stuttgart, 1968, p179.
[4] ‘“‘Quad with switchable polar diagram’, L. Vsevol-
zhskii, UA3IAR, Radio (USSR) No 6, 1978, pp18-
19. See also “Technical Topics”, Radio Com-
References munication October 1978 et seq.
[1] Antenna Theory and Design, Vol 2, H. P. Williams, [5] Radio Communication Handbook, Vol 2, 5th edn,
Pitman, London, 1966, p73. RSGB, p12.67.
[2] ‘“‘Simple arrays of vertical antenna elements”, The [6] “‘Beat the noise with a ‘Scoop Loop’”’, The ARRL
ARRL Antenna Anthology, ARRL, 1978, p114. Antenna Anthology, ARRL, 1978, p76. Material
(Based on article by J. L. Lawson, W2PV, QST originally presented in QST by Ben Vester, K3BC.
March 1971.) [7] ““Technical Topics”, Radio Communication
Stuttgart, p179. September 1980, p906.
CHAPTER 14

Large arrays

The arrays described in earlier chapters bear little wires in accordance with the angles given in the table;
resemblance to the large rhombic, curtain or log- seen end-on the sloping wire looks vertical and polar-
periodic arrays often used by commercial services. For ization is therefore vertical but becomes mainly hori-
the most part these are highly sophisticated systems zontal as one moves to the side, while still remaining
requiring a large amount of real estate and of relatively within the cone. For a height of 50ft (15m) and 3A of
little direct interest to amateurs; nevertheless, the wire at 14MHz, the slope of the wire is 14° which puts
amateur with a one-acre (0-4ha) field at his disposal the endwise lobe at an angle of 16° above the horizon.
may be wondering just how large a rhombic has to be in At a radiation angle of 6° there is a loss of about 3dB
order to provide a worthwhile gain compared with which exactly neutralizes the gain, the total coverage in
antennas described in earlier chapters. A number of azimuth being about 75°. For the horizontal directions
log-periodic arrays have been successfully constructed (after allowing for ground reflections) gain is likely to
by amateurs and interest in them has been stimulated be about 3dB better than this, or roughly equal to a
by the prospect of additional frequency allocations. dipole at 35ft (11m). The attractiveness of such a
Small V-beams are often used by amateurs with system derives from its simplicity and the low visual
plenty of space at their disposal but the echelon version impact, assuming a suitable tree or tall building for
of this, though much easier to accommodate, seems to supporting the antenna and a long enough garden, in
have received little attention, as does also the capaci- this case over 200ft (60m). For such a span the wire
tively stretched long wire which has interesting possibi- needs to be fairly strong and it would usually be better
lities even when confined to a medium-sized garden. if possible to use it for supporting a delta loop (at the
top) plus (at a lower point) a three-element wide-
spaced vertical (end-loaded) Yagi; between them these
The long single wire would provide coverage of four directions with gain in
This requires a very long length to provide gain com- both cases roughly equal to that of a horizontal beam at
parable even with that of a close-spaced beam and is of 25ft (8m). Spreaders could be used to turn the delta
interest mainly as a building brick for V- and rhombic
systems. The gain and directive properties of single Table 14.1. Properties of long wire radiators
wires are summarized in Table 14.1 from reference [5].
V- and rhombic antennas are formed from single Gain of main
wires disposed in such a way that main lobes of Length Angle of main lobe over half- Radiation
(d) lobe to wire wave dipole resistance
radiation coincide; thus a V-antenna might consist of (dB) ()
two 4) wires fed out of phase with an apex angle
of 2 x 26° = 52°, and would provide a gain of 1 54° (90°) 0-4 90
13 42° 1:0 100
3-3 + 3 = 6:3dB. This is comparable with that of a 2 36° (58°) 1:5 110
three- or four-element close-spaced Yagi but with the 23 33° 1:8 115
disadvantage that the direction is fixed and the angles 3 30° (46°) 2:3 120
4 26° (39°) 3:3 130
of coverage very restricted. 5 22° (35°) 4-2 140
Termination of a wire in its characteristic impedance 6 20° (31°) 5:0 147
results in a unidirectional pattern as explained on p95 8 18° (26°) 6-4 153
10 16° (23°) 7:4 160
but has little effect on gain. Methods of feeding and
terminating long single-wire antennas are shown in Fig The number of complete conical lobes (see Fig 6.4) is equal to
14.1. The resonant length for long wires is given the number of half-waves in the antenna. The main lobe is the one
nearest to the direction of the wire, and the figures in this table
approximately by / = (n — 0-05)A/2 where n = the re- give its direction and gain. When a multiple 1 antenna is
quired number of half-wavelengths. centre-fed the pattern is like that of one half, but with more gain in
Sloping wire antennas are frequently used and many the main lobe. The angles in brackets correspond to this case.
When the antenna is terminated, or self-terminating, the radiation
different arrangements are possible. The radiation pat- resistance is 30 to 50 per cent greater, and the main lobe slightly
tern should be visualized as a cone surrounding the nearer to the wire.

198
 LARGE ARRAYS 199

(a) 5002 line

a|> ot
see Ps) al»
10u“H typical for 14MHz
4:1 eye transformer eg 40turns, 1in (2-5cm) dia,
4in (10cm) wih,

50082
About 5 to 6ft (1-5 to1-8m)
752 cable
A
4

ae il ¢ ,
2 Fan of wires cut for ae
resonance (4/4) at each HS,
(c) of the required frequencies NN

a
300/600
1 ! line
«32
Ys
2a ——__—_—____}>
7

(d) bens

Multiband
75 cable -CaEa-—ak
artificial earth
3 to 6ft (0+9 to 1°8m)
Ses S indicates stabilizing cords

Fig 14.1. Methods of feeding and terminating long single-wire

antennas. Artificial earths are used at each end, the radiation
resistance being high enough to mask any losses in loading coils
such as those shown at (b) and (d). Capacitive hats similar to A
4
those used in mobile work (p210), or even smaller, should be
usable. Artificial earths are not broadband but their use makes Fig 14.2. “Stretched” long wires with resistive terminations. The
the whole of the available span available for gain purposes. The angle a depends only on the stretch factor n, and the gain is
baluns shown at (d) allow the antenna to be fed with coaxial line, proportional to length, being 3dB less for a single wire. The
all artificial earths being individually self-resonant. The arrange- figures in Table 14.2 are typical, and for a stretch-factor of 4 at
ment at (a), though the one usually recommended, is unlikely to 14MHz the wire length is 132ft (40:-2m) broken with 43pF capaci-
achieve good feeder balance tors at intervals of 5ft 6in (1-68m). For a single wire the method of
feed shown at (b) in Fig 14.1 should provide a better match and
loop into a two-element beam but mechanical and less risk of feeder radiation
visual impact problems may then arise.
Better single-wire performance may be achieved by 14MHz with perhaps another 3dB if each dipole is
capacitive stretching (p97), in which case the 200ft backed by a reflector. This arrangement is singularly
(60m) wire can be turned into the equivalent of a four- unattractive and the rhombic array, Fig 14.3(b), fitting
or five-element collinear array with 7dB gain but a into the same space is of much more interest; apart
rather narrow pattern at right-angles to the wire. It is from being a lot neater it requires only four poles
stated [1] that if such a wire is terminated in its instead of 12, substantially less antenna wire and
characteristic impedance as in Fig 14.2 there is rever- several times less antenna-plus-feeder wire. With some
sion to an endwise pattern providing about 7dB gain at manipulation of ropes, terminating resistors and feed-
15° to the wire for a “‘stretch factor” of 4. One method points it can be switched to provide a choice of four
of construction is to use 80 twin lead, forming the directions, with a good chance of being able to put a
capacitors by cutting the conductors alternatively to useful sidelobe in most of the other directions that may
leave overlapping sections. The antenna is broadband be needed.
so that dimensions are not unduly critical and harmonic
operation is possible. - Table 14.2. “Stretched” long-wire antenna data
Length n o R Gain
Rhombic antennas (d) (deg) (ohms) (for Vee array)
(dB)
The most efficient way of using a one-acre (0-4ha) plot
for a transmitting antenna would be to fill it with 1:5 2 30° 300 5
dipoles as in Fig 14.3(a). This would provide a gain 2:0 3 20° 200 6
3-0 4 15° 150 10
about equal to the number of elements, ie 14-8dB at
200 HF ANTENNAS FOR ALL LOCATIONS

1-acre (O-4ha) field

fimisii on tsae 8 chee eee Bk fire

(dB)
Gain

Leg length (A)

Fig 14.4. Theoretical gain of a terminated rhombic antenna

relative to that of a 1/2 dipole in free space (ARRL Antenna Book)

for zero wave angle [2] and it will be seen that the gain
is increased by 3dB each time the leg length is doubled;
for this wave angle the required apex angles are double
the values given in Table 14.1. It is more usual, but less
convenient for the present purpose, to refer to the
“tilt”? angle which is half the corner angle, and equal
to 90° minus the angles from the table. Leg lengths
greater than 6A are not recommended [2] because the
beam is then so narrow that performance is greatly
affected by the small changes in arrival angle (or
optimum launch angle) which are typical of dx signals.
It is common commercial practice to design for a
(b) wave angle somewhat above the horizon but in general
this is not recommended in view of the stress placed
Fig 14.3. Two ways of trying to get as much gain as possible from earlier (p17) on the use of the lowest possible angle of
a one-acre (0-4ha) field. The dipoles at (a) are assumed to be
connected to a network of feedlines arranged so that the radia- radiation for long-haul dx. If it were possible to differ-
tion adds up in phase for the directions at right-angles to the entiate between the direct wave and the ground-
dipoles. The rhombic at (b) provides less gain but is a great deal reflected wave this could alter the position somewhat
less obtrusive. Details include the definition of tilt angle ¢7
but, with typical beamwidths (free-space) of 20-40° and
With the design optimized for maximum radiation in required radiation angles of 6° or less, very little
the plane of the array (zero angle of radiation in the discrimination is possible. Thus even the rhombic with
vertical plane) the angles are as shown and the leg its relatively narrow beamwidths is no exception to the
length is 2-5A, giving a gain of 9dB over a i/2 dipole, rule that for all practical purposes the effect of the
though this could be up to 3dB greater for reception ground is independent of the type of antenna. Never- -
due to noise absorbed in the terminating resistance theless, despite the emphasis placed on this rule in
(p127). Terminated long-wire antennas such as this are earlier chapters, it is usually possible with the help of
inherently broadband and unlike dipole arrays can be the rhombic, a hilltop and knowledge of Fresnel zones
used over a band of an octave or more, though the (p137) to conceive a situation which defeats it. Thus
design is only correct at one frequency and several in Fig 14.4 a cone of rays 20° wide emanates from the
decibels may be lost at, say, 283MHz by optimizing the point P; for an antenna height of SOft (15m) and a
design for 14MHz. Front-to-back ratio depends on leg required angle of radiation of 5-7° the distance to the
length but may be optimized by adjustment of termi- geometrical reflection point is 500ft (152m) but the
nating resistance to obtain high values in all cases. The Fresnel zone extends outwards from 86ft (26m).
terminating resistance should be non-inductive and However, beyond 300ft (90m) the grounds drops away
capable of dissipating at least a third of the transmitter so that the greater part of the zone is lost; the remain-
power, the optimum value being around 800Q. der, though not illuminated by the 20° cone, would
Fig 14.4 shows the variation of gain with leg length respond fully to the 60° beam from a typical Yagi.
 LARGE ARRAYS 201

oe
Operation traps may be acceptable since the radiation
of peat _—
eee
resistance is fairly large. In this last case termination is
Equivalent ageF—.
upper edg
signal source ee practicable only at the highest frequency but a simpler
: 20° Geometric centre of Fresnel method might be to retain the arrangement of Fig 14.6,
zone (flat ground case)
displacing the reflector mechanically as required.
The echelon antenna [4] is equivalent to the V but
%
requires much less space. In Fig 14.7 the wires are
Near edge of displaced, so that the line AB joining any pair of
Fresnel zone corresponding points is at right-angles to the direction
300ft
YY which corresponds to the appropriate angle of
propagation from Table 14.1. Since the wires are fed in
opposite phase there is no propagation along this line.
For the other pair of directions there is an effective
_ Fig 14.5. Illustrating possible loss of ground reflection if vertical separation distance AC between the wires so that the
beamwidth is sufficiently narrow. To demonstrate the principle it
- is asumed that the antenna can be represented by an equivalent signal is not cancelled and if AC = 2/2 the fields are
point source at P. If the beam is replaced by a dipole or a small directly additive. The wires may be terminated as in Fig
Yagi array the reflected wave, though reduced in amplitude, 14-1 or 14:2 to obtain a unidirectional pattern and
re-appears due to illumination of the foreground
harmonic operation should be possible, though no
However, most of the zone is lost in any case and the record has been found of its use hitherto; it is true that
reflection coefficient will be fairly small. the optimum geometry is frequency dependent, but this
This example is unrealistic to the extent that in applies equally to rhombic and V-antennas. At the
practice one does not get a 20° beam from a point second harmonic the angle a is roughly halved so that
source and the radiation from one half of the rhombic YY is no longer quite at right-angles to AB, and AC
will illuminate all or part of the near zone, thus further
reducing the small advantage of the large array. The
example is very much oversimplified but answers an
obvious question. It further emphasizes the basic rule
that under normal conditions the effect of ground is
independent of the type of antenna; it seems neverthe-
Feed (1), phase advanced
less worth recording that in a situation not unlike that by 90°
just described one Australian station was able to put a
consistently strong signal into the UK during “difficult”
sunspot minimum periods, including many times when
no other VK signals were audible. Although the exam-
ple indicates a possible tendency, it falls well short of
explaining the observed performance, which in the light
of all the evidence has to be attributed to the combina-
tion of a large antenna with the hill-top location and not Feed (2)
to either alone.

Two-wire systems
From earlier mention of the V-beam it may be felt
that the gain is not too impressive, having regard to the
amount of space occupied. The V- may however be
backed by a reflector at a distance of any odd multiple
of i/4 to obtain a further gain of 3dB; it now occupies
less space than the rhombic for a given gain, although
bandwidth is restricted by the spacing requirement and
additional poles are required. Both of these problems
may be overcome by the arrangement shown in Fig
14.6, the reflector being mounted a short distance
above or below the radiator; with each of its arms Fig 14.6. Addition of reflector to V-beam without additional
shortened by A/2 the reflector is effectively moved back poles or spreaders. One V is mounted a short distance below
by 4/4, thus achieving the correct spacing [3]. Front- the other, making the legs shorter by )/2 to give an effective
mean displacement of \/4. The V with the longer legs is fed
to-back ratio may be further improved if desired by with a signal advanced in phase by 90° to obtain a unidirectional
terminations as shown in Fig 14.1(a), and for multiband pattern
202 HF ANTENNAS FOR ALL LOCATIONS

for a length of 4A and is more or less proportional to

length. In the broadside case an additional 3dB can be
obtained by the use of reflectors and in the end-fire
case more gain can be obtained by increased phase
shift, thereby to some extent invoking the supergain
principle. The 11dB figure is about 1-5dB higher than
that given by the ‘“‘gain equals n’’ rule based on )/2
spacing of elements; actually gain exceeds n by 3dB in
“N
N

. k- s—
1 1
the simple broadside case and is roughly equal to n for
- -"
Transmission res
ed
Pra ae ‘ be the end-fire case. This means that with optimum spac-
line
Y sft
a s
bas
‘
“x ings there is, perhaps not surprisingly, a beneficial
ae
Pid Ss
AS '
effect rather than otherwise from the mutual couplings.
on x

Fig 14.7. The echelon antenna; though equivalent to the V, this Log-periodic arrays
requires much less space. The two long parallel wires are spaced
a distance d and staggered by the amount S. As shown, radiation Log-periodic arrays are found in a wide variety of
takes place along the lines XX whereas the fields at a distant forms, shapes and sizes but Fig 14.8 represents those of
point along YY will cancel. If the wires are fed in phase, radiation most interest to the amateur. Typically, for any fre-
takes place along YY. The distance d and the amount S are given
by quency within a specified range there is a group of
pe 492 sina elements near enough to resonance to take part in the
~ F sin 2a radiating process, all such groups having identical
s- 492 cos a proportions so that the gain is the same throughout the
fsin 2a band. The signal travels along the array, ignoring the
where «a is the angle of maximum radiation from a single wire shorter elements (which present large capacitive react-
(Table 14.1). dand S are in feet and fin megahertz (ARAL Antenna ances) until it encounters the right group of elements.
Book)
Beyond the )/2 element the current decreases rapidly
(though not halved as one would like) is reduced with the feeder looking into a highly inductive reac-
sufficiently to ensure a much bigger field along the line tance.
XX than along YY. For such operation the choice lies For any pair of adjacent elements or spaces the ratio
between acceptance of a compromise or use of the b/a = d/c = t, this and the apex angle (a) being the basic
supporting halyards for adjustment of S and possibly d. design parameters of the system [6, 7]. The longest
If the position of one wire is adjustable by ropes over element is made equal to 492/f, ie approximately \/2 at _
pulleys and each wire is fed separately via a phasing the lowest frequency f,, and the length of. the array
unit, near-optimum gain should be realizable for any should be as long as conveniently possible, though this
frequency, any spacing between wires (if not too close), is not critical—a reasonable compromise is given by
and along either of the lines XX, YY. t = 0-9, spaces between elements being made equal to
The effect of close spacing has apparently not been one-tenth the length of the longer element. This value
investigated for either the echelon or the V-plus- of t gives the smoothest radiation pattern; lower t
reflector. means fewer active elements and lower gain, whereas
increasing t to, say, 0-95 produces more back lobes,»
requires twice as much wire, and increases gain by at
Stacked dipole arrays most 1dB. About 1-6dB extra gain is possible if the
Having rejected the idea of filling the one-acre field spacing and array length is increased (ie a reduced) by
with dipoles one might perhaps still consider, say, a row a factor of three.
of five multiband dipoles backed by reflectors along A simplified design procedure is as follows:
one edge. Though still lacking aesthetic appeal this 1. Draw to scale the longest element and a centre
would yield a gain of at least 11dB, or rather more than line representing the length available plus a small
the rhombic, though average front/back ratio would amount, eg about 50 per cent if coverage is
not be as good. restricted to one octave since the hf end of the
Broadside and end-fire arrays of vertical dipoles with “possible” array is not needed. Complete the
fields adding in phase provide almost identical gain for triangle as in Fig 14.8.
the same length but the optimum spacing is 5/8 and 2. Fill in further elements, making each one 10 per
3N8 respectively, so that a lot more elements are needed cent shorter than the previous one (ie t = 0-9)
in the end-fire case which includes long Yagi arrays. until the element length is 4/2 at the highest
The 31/8 spacing, as well as being near-optimum, is also frequency. Continue for a further three elements.
a critical value which should not be exceeded and
applies both to Yagi and driven arrays; the gain is 11dB This fixes the dimensions. Since the active elements
 LARGE ARRAYS 203

d c
Elements
active at '
Elements 18MHz a a 7
yas at 74 Fae cy
25MHz

=
=— ae)
=_—
_- =-
— -_—

So A ten A A ee ee ee b a(=44/2)
——
=—
me, ~
~
_— ~
Length not
critical -can
be justa
short
Elements °F ‘jumper-wire'
active at
rAd el: Elements ot a
active at i a
Elements
active at
14MHz

Fig 14.8. Typical !og-periodic array. The ratios b/a = d/c are identical for all adjacent pairs of elements. As shown the frequency range is
one octave but this can be extended in either direction by adding more elements according to the same basic rule. At any given
frequency within the band there are three or four elements contributing appreciably to the radiation, and performance is comparable
with that of a three-element Yagi

are only three or four in number and closely-spaced, may be increased to 45°, thereby reducing the array
additive gain is not possible, and as for other close- length from 60ft to 26ft (18-2 to 7-9m) before practical
spaced arrays (p68) the connections must be crossed difficulties become apparent. This decreases the gain
over as shown. relative to a dipole from 6-5dB to 4-8dB, and a
Calculation of matching impedance and of the num- comparable rotary array has been described [8] using 12
ber of active elements is complicated but inspection of elements, the boom length being 26-5ft (8-1m), cover-
many published designs confirms that usually three or age 13-30MHz, and total weight 116lb (52-6kg). A
four elements can be assumed active, subject to the sort comparable three- or four-element Yagi would have
of design conditions likely to be imposed. It appears more gain, be very much lighter and have lower wind
also that a good-enough match (better than 2:1 swr) loading as well as less visual impact on the neighbour-
can usually be obtained using 50Q coaxial line with a hood, so that clearly a price must be paid for the wide
4:1 balun. Alternatively 600Q line may be used with an bandwidth. On the other hand, a vertical wire beam
atu. with dimensions adapted from Fig 14.8 but using
For a vertical array with top-loaded elements the monopole elements based on Fig 13.15 could be
spacing needs to be increased in order to accommodate arranged to cover 10-30MHz, assuming L is 60ft
a useful amount of loading without the risk of overcou- (18-3m) and a pole height of 30ft (9m) without (or 24ft
pling. If it is increased by 40 per cent it should be (7-3m) with) end-loading.
possible to use top loads of at least 0-1A, thus reducing There is some discrepancy between references in
the required length of vertical monopole elements from regard to gain figures, and if consulting other sources
0-25, to about 0-174 or a maximum of 12ft (3-7m) for the reader is likely to be confused by conflicting
coverage of 14~30MHz. In practice it should be satis- definitions of a, which in references [7] and [8] is half
- factory to check one element with a gdo to ensure the apex angle, and t which in reference [6] is not b/a
equivalence with a straight element, the rest of the but (b/a)*. The former definition has been used in
array being designed by appropriate applications of the references [7] and [8], and also by this author in view of
factor t to all dimensions. its more obvious physical relevance in the amateur
Also of possible interest is the log-periodic inverted- context. Gain is shown in Fig 14.9 as a function of t and
V antenna. Assuming sufficient lateral space for string- a, using figures taken from reference [6].
ing out their ends, the centres of the elements can be
supported by a catenary running between two supports.
Similar design considerations apply and, although win- Discussion and conclusions
dage may be considerable (p253), there is no feeder Exceptional signals are often attributed to the use of a
drag to put excessive strain on the masts. large array such as a rhombic but, theoretically at least,
If end-loading is not required the angle a in Fig 13.8 the performance could usually be equalled or exceeded
204 HF ANTENNAS FOR ALL LOCATIONS

given direction. The advantages of the rhombic lie in its

relative simplicity and broad bandwidth.
Long Yagi arrays have not been considered separate-
ly in this chapter since the main essentials were covered
in Chapter 5, and arrays of short end-loaded elements
were discussed in Chapter 13. In general, the long Yagi
makes better use of space than the rhombic but has the
disadvantage that gain increases rather more slowly
with increase of length as demonstrated by Fig 5.27.
Like the rhombic the log-periodic array provides broad
(dBi)
Gain
bandwidth with a ‘“‘minimum of fuss” but at the price of
increased weight, windage, and space occupancy for a
given gain as compared with, say, a monoband beam,
so that the author’s personal choice for coverage of all
bands lies rather in the direction of switched arrays of
multiple dipoles as discussed in Chapter 7.

40 50 60 70 80
a (deg) References
Fig 14.9. Gain of log-periodic array as a function of the design [1] ‘“Loaded wire aerials”, F. C. Charman, G6CJ,
parameters t and a, relative to an isotropic source RSGB Bulletin July 1961.
[2] The ARRL Antenna Book, 13th edn, ARRL, 1974,
p175.
by the use of a small number of three-element arrays [3] Radio Engineers Handbook, F. E. Terman, 1943,
suitably spaced and phased as discussed on pp100 and p808.
105; thus rhombics with leg lengths of 2-54 and 5A can [4] The ARRL Antenna Book, 12th edn, ARRL, 1970,
be expected to equal the performance of groupings of p174.
two and four close-spaced arrays respectively, the [5] Radio Communication Handbook, Vol 2, Sth edn,
required spacings being of the order of 34/4. Much less RSGB, 1977, p12.59.
space is needed for the stacked arrays, which have the [6] Antenna Theory and Design, H. P. Williams, Pit-
major advantages that they can be pointed in any man, London, 1966, p484.
direction and deployed according to the needs of the [7] The ARRL Antenna Book, 13th edn, ARRL, 1974,
moment, eg four could be used for instantaneous p160. |
switching between four directions (eight if reversible) [8] ‘““The log-periodic dipole array”, Peter D. Rhodes,
instead of being phased to provide maximum gain in a K4EWG, OST November 1973.
 CHAPTER 15

Invisible antennas

Planning regulations or restrictive covenants may pro- Concrete roof

hibit the erection of the desired antenna system, or even
of any kind of antenna, and dwellers in high-rise build- Mobile whip (outboard)
ings have particularly difficult problems. In interpreting
_ regulations there tend to be large areas of uncertainty
which are best resolved by common sense and goodwill,
but it helps if the antenna is designed to attract as little Rotator
APARTMENT
notice as possible—few of us are so isolated from our
neighbours that we need give no thought to its appear-
ance which is often of much greater importance than
its electrical qualities. For this reason visual impact
has been featured among the basic characteristics. of
antenna systems discussed in Chapters 11-14 and we
return to these topics in Chapter 17.
Fig 15.1. Use of mobile antenna for operation from an apartment
In this chapter an attempt is made to offer a few block. Beam rotator swings the antenna out of sight when not in
solutions for those cases where the erection of a visible use, eg during the daytime
antenna is expressly forbidden or ill advised. One
answer of course is the indoor antenna which has An apartment dweller located near the top of a
already been discussed in some detail, but this does not high-rise building may be quite well placed since he has
help the dweller in a steel-framed apartment block and plenty of height with which to offset the low efficiency
is a serious hardship to anyone living in a bungalow of a small antenna such as a mobile whip. Harry
unless they are lucky enough to have a steep ground- Bourne, later ZLIOI, found himself in just such a
slope or are not interested in dx. For outdoor antennas situation while stationed in Washington, DC, USA, and
there are basically two possibilities: (a) concealment reports using a mobile whip for all five bands mounted
and (b) disguise. on the balcony railing and sloping at 45° to the horizon-
tal [1]. During the day it was inside the balcony, with a
b)
concrete floor, concrete roof and steel-framed building
A
~
‘
By
APLe
just behind it, thus forming a box with a slot in one
WAE
VO
< °)STite side, Fig 15.1. At night he used a rotator to swing the
iy
tf
i
p
antenna over the side into the clear and could then
“Oe AN
Ys
a) o
a /
| »\
work worldwide with no trouble, but even in daytime it
could be swung over the side for a few minutes without
oe a4
’,bye f

bas e, \ 8 4
Rat”
"Pe
=at ae.
=|)
attracting attention. The steel railings were used as the
No ie
me AN Ip \
ground plane except on 3-SMHz where a counterpoise
EAE
% NED ’ PET
as
wire was run under the carpet indoors.
we Sie im1 3
BS
CNIN
\
Be: A somewhat similar arrangement has been used to
hy PS
cs ®
.
g ale
pe lift an antenna off a flat roof [2], and it occurs to the
author that in some cases a dipole such as Fig 15.2
\ R ELS, a
fs 2 f Oe
SB ss se “ ate BAS
(3'8
by
id
Hy ‘ el
could be folded back against the building and raised by
y Sy =A cords or a beam rotation unit at suitable times. It would
ae
\. P

Ao S :
Aj Ar | 2ENS mt require some experiment to determine a minimum size
A s a\he LON
ae i y ? na S,ee
‘AN
for the arms and careful painting to match the building.
oe eM:Fh
ne MEAS > I ego sen5 25, g
Si hte<a
CVE Po Re [ay Bey.
ee
8
SD Be =
= =
Theoretically, at a height of 80ft (24-4m) the dx
NA)ia | A eke 2
a PA a) eOe eerts ey Sal’ Uh Be ea al
He | oz
A TA|
Sy fe performance should equal that of a three-element
* eo
monoband beam at 40ft (12:2m), and with luck the
“For outdoor antennas there are two possibilities: (a) conceal- attenuation through the building would provide a good
ment and (b) disguise” (the concealed antennas are not visible!) back-to-front ratio. Given this kind of performance,

205
206 HF ANTENNAS FOR ALL LOCATIONS

Bracket with socket portions of the down lead, as discussed on p163), or

Metal conduit
to support a thin-wire antenna such as an inverted-V or
end-fed long wire (inverted-L). It appears from one
account [3] that there is no need to use anything thinner
than 28awg (30swg or 0:3mm), and that this is just thick
enough to be satisfactory as regards possible breakage
by birds, although experience might be different in
other areas and circumstances. Even 36awg (38swg or
0-15mm) was satisfactory for performance, and invisi-
ble to the naked eye at distances greater than 15ft
(4-5m).
Fig 15.2. Retractable antenna suggested for use from an upper Further to this, there is the possibilityof disguising a
floor of an apartment block. Visual profile is low but with
sufficient height the performance could equal that of the average beam as a television antenna. In the UK it has been
suburban beam. Antenna could be centre-fed bent dipole as officially stated that tv and fm broadcast receiving
shown or an end-fed wire. For best results AB, DC should be at
least 0-08\. BC may be reduced to as little as 0-14) by additional
antennas do not need planning permission; in other
folding of the ends, eg as in Fig 11.1(j) words, they are regarded as part of the general residen-
tial scene and not affecting the general character of a
a single direction and frequency (if well chosen) can property. However, this is not much help against a
provide a lot of dx interest but, since the user has restrictive covenant and in areas where uhf television
the big advantage of, as it were, living on top of his antennas are in general use something the size of a
mast with ready access to the antenna, band changing Band 1 antenna may be regarded as out of place. Even
should not prove too difficult. so, a 23MHz beam on the lines of Fig 5.22(e), being
Even the bungalow dweller need not give up hope, in about the size of an fm receiving antenna, should be
view of the description in Chapter 13 of beams using 6ft acceptable and in Band 1 areas it should be possible to
(1-8m) elements hanging from a catenary. This pro- get by easily with the 14MHz version.
vides very much more efficient end-loading than is Planning permission is not required in the UK for an
achievable in the case of say a mobile whip, and has the antenna attached to the house, provided it does not
further advantage that it can be effectively disguised if project above the highest part of the roof or beyond the
necessary by actual use as a washing line! For all-round forwardmost part of any wall of the dwelling which
coverage by beam switching the elements can be fronts on a highway. The height restriction is unhelpful
arranged in a neat square formation based on Figs 13.5 and it will probably be better in general to take
and 13.6, which neighbours may well seek to emulate. advantage of the tv antenna concession. It may be
In the case of taller houses other possibilities arise; worth noting that the diagonal dimension of a good
for example, an end-fed wire trailed over the ridge of three-element beam for fm broadcast reception could
the roof should give good results at least in dry be about 8ft (2-4m), and a rod of this length would
weather, but if it can clear the roof by a few inches this provide enough spacing for a pair of inverted-V beam
would be a help. Alternatively the use of the outer elements for 14MHz. For a gain of 4dB the radiation
conductor of a length of coaxial cable will reduce the resistance (90° apex angle) would be 15Q per element,
characteristic impedance and hence the dielectric and from Fig 3.10 the loss resistance using 24swg
losses. Two or three elements suitably spaced can be (0-6m) wire (which might well not be noticed) would be ~
used as a fixed reversible beam along the direction of 2-9Q, a loss of 0-77dB only.
the roof ridge, remembering that if there are significant Every care must of course be taken to make sure that
losses all elements must be driven, not parasitic. any wire liable to break in a storm or due to impactby
Other possibilities include the use of guttering, light- birds cannot come into contact with electric cables or
ning conductors or flagpoles in ways which will no constitute a hazard to traffic or passers by. In some
doubt suggest themselves to readers of earlier chapters. cases it may be best to slope the wires down to the rear
The need for efficient earthing of the transmitter of the building, accepting some loss of height. Alterna-
(p163), and any other steps such as harmonic filters tive suggestions are presented in Fig 15.3. Fig 15.3(b)
which may be necessary to prevent tvi or stereo avoids the need for a spreader but the height loss may
breakthrough, must be stressed since an amateur who be serious. .
may be treading delicate ground needs, above all, Trees, where available, provide a simple answer to
good relations with his neighbours. the problem of concealment and a study of Chapters 11
If tv antennas are permitted more possibilities unfold and 12 should provide the reader with plenty of en-
themselves. A tv antenna erected as high as possible couragement, assuming that a reasonable height can be
can be used, either complete with feeder as a vertical achieved. The possibility of using a low tree to conceal
on any band (with the help where necessary of linear- a vertical antenna needs further assessment as does the
resonator type traps clipped across or coupled into possibility of using trees themselves as antennas; this is
 INVISIBLE ANTENNAS 207

Boom of tv orfm antenna antenna” being reported. Few other details are given
but the principle can be applied to other objects and it
is claimed that ‘““whenever an environment detrimental-
ly loads a regular whip, we can couple a transmitter or
receiver into the environment and make it work well as
a substitute’.

Trap acts as insulator Add tuning capacitor

<== Feed-points anchored to here if necessary
/ boom in case (a) and to
guttering in Fig 15.5. Conversion of any long metal conductor into a )/2 dipole.
Boom of ~ case (b) Traps may be satisfactory if mutually coupled as in Fig 11.15(a),
tv orfm >
za but direct connection as shown is likely to result in better
antenna
efficiency. Feeder may consist of a 502 coaxial line connected to
XX via a 4:1 balun. The traps should be as long as possible, up to
(Cc) Plan view about 1/4, and should if possible be constructed with flat strip or
large diameter tubing, though wire should be satisfactory pro-
Fig 15.3. Two-element beams using thin-wire dipoles. Wire vided the traps are not too short. Spacing, particularly if wire is
should not be thinner than 24swg (0-5mm). At (a) the boom of a used, should be as wide as possible up to at least 1ft (30cm)
tv or fm antenna is used as the spreader for an inverted-V beam,
extended if necessary to provide at least 10ft (3m) spacing at
14MHz or 7ft (2:1m) at 21MHz. The two feeders may each be 502
Requirements for short- and medium-range com-
with balun and brought down alongside the tv/fm feeder. At (b) munication on all bands are probably best met by an
the ends of the wire elements are both brought in to the pole, end-fed long thin wire with some folding over of the far
allowing a separation of about 8in (20cm), and the centres are end if necessary. Attention must be given to proper
attached to points 10-12ft (3-4m) apart on the guttering, the
lower ends being also spaced 8in and strung out to a tree or small tensioning (p150) and earthing (p43).
post retaining as much height as possible. (c) Bird's-eye view Requirements under the present heading vary widely
as do the possibilities for meeting them, and it is hoped
achieved [4] by feeding energy into the tree via a that the few ideas presented here, even where not
toroidal coil wrapped around the trunk. The system has directly applicable, may at least provide encourage-
been found “particularly useful in wet jungles’’, effi- ment in the search for solutions so that situations which
ciencies of “‘up to 22dB better than a conventional whip may at first seem hopeless are construed rather as a
Linear traps challenge. It may be impossible, at least without caus-
ing tvi, to excite the steel frame of a building, but
reference [4] reports some success even in this unlikely
context, and other solutions are not necessarily ruled
out. For example, given access to the frame it would be
possible in principle (with or without making direct
Loaded dipole (copper strip) =
connections) to use the linear resonators of Fig 11.15 to
mark off a section for use as in Fig 15.4, where they act
as insulators. The same principle can be applied (with
rather more confidence) to sections of drainpipe, gut-
tering, or plumbing as in Fig 15.5. Such methods are
more likely to succeed at the higher frequencies in view
of the rather small value of the inductances which are
involved.

Fig 15.4. Part of a metal grid such as a steel framework. For rf the References
rectangle may in principle be isolated from the rest of the
structure by linear traps. In principle the rectangle may be
[1] Material presented by H. K. Bourne in CQ.
resonant at a desired frequency and can then be excited with an [2] Wire Antennas, W. I. Orr, W6SAI, and Stuart D.
additional coupling loop or delta match. Success would however Cowan, W2LX, Radio Publications Inc, 1972,
be more likely with a separate resonant element as shown,
relying on the traps to prevent current flowing in horizontal p139.
members of the framework. Broad copper strips should be used [3] “Invisible antennas’, T. J. Gordon, W6RVQ, QST
to reduce losses in walls etc. The success (if any) of this will November 1965, p87.
depend on being able to locate the framework, and on how close
to it one can get. For operation at low heights the polarization [4] “It’satree...apole. ..aman; No! a short range
should be vertical hf antenna’’, Electronic Design 20 December 1973.
 CHAPTER 16

Mobile and portable antennas

Signals from mobile stations are usually weaker than dipole or monopole, and uniform over at least 4/16
those from fixed stations, even after allowing for outwards from a point of maximum current. If part of
possible power differences. This is usually taken for the end section is replaced by a “‘capacitive hat” it is the
granted and it might be thought absurd to question the inner part of the triangle which is retained.
acceptance of anything so obvious; nevertheless, pur- These current distributions are shown in the figures
suing the themes of equality between antenna systems and it will be clear that the field strength produced will
(Chapter 3) and the differences between horizontal and be “as if” the current J flowed through lengths h/2, 3h/4
vertical polarization (Chapter 10), the performance of a and 7h/8 respectively. From the squares of these cur-
mobile station might be expected to equal that obtained rents the radiation resistances are in the ratio 1, 2-25
with the same rig and a horizontal dipole at 30ft (9-1m). and 3-063. It will be found that at the lower frequencies
There are in fact two reasons why the mobile antenna the inductance of the antenna below the coil is negligi-
is “down” in performance and one of these does not ble compared with the required loading inductance.
apply at the higher frequencies, while the other is If this equals L in case (a) we require 2L in case (b) and
somewhat obscure. The first arises because height is in case (c), since eb = h, it is again equal to L. Since
typically limited to around 8ft (2:4m), or effectively a the value of L has little effect on the best obtainable
bit less than this taking into account partial screening coil Q, which is likely to be of the order of 300 in each
by the car body, and it is mechanically impracticable to case, the coil loss resistances R, are given respectively
rely on efficient methods of end-loading except at the by wL/300, wL/150, and wL/300. It is necessary also to
highest frequencies. One is therefore forced, for assume an earth-loss resistance (unknown) of R, which
14MHz and lower frequencies, to break the rules laid is the same in all three cases.
down in Chapter 5 in the course of the search for
efficient elements for small beams.
The second problem is that of the ground connection -e@b

which is usually composed of the capacitance to the

x2 cd/,
ground of the car body. Though difficult to analyse or
even describe precisely, there must be some loss since |
* Cc an
cana
SS d

the car body breaks the rules for counterpoise earths, 4

,
/
‘
‘
a

ie that they must be low loss and sufficiently clear of the

fh ik Z
/ Mean current Zé / Mean
ground to provide a return path for the antenna current /
/ =I /
2 /
A /
te
current
= 31/,
without forcing it to flow through “‘lossy” ground. In
i ae 7 i 4

of 7 - e
other words it should be possible to regard the counter- Ri ; at
poise as the lower half of a dipole rather than as a
ground connection, the “‘car body” earth being in fact ~.

an unknown quantity somewhere between the two

extremes. ee

The antenna is normally vertical with loading coils at

the centre or base, and a “capacitive hat” is often
added at the top of the antenna. Base loading is in
_ general much less efficient, whereas the desirability of a (a) (b) (c)
capacitive hat in terms of radiation efficiency is proved
by the complaints of “‘excessive Q”’ (or insufficient Fig 16.1. Alternative methods of loading a mobile antenna. The
height h of the radiator is assumed to be the same in each case,
bandwidth) which often result from its use! and the shape of the current distribution is shown by dotted
To get this in perspective consider the situation lines. At (c) a hypothetical increase in height ab = ae has been
illustrated in Fig 16.1. As a rough approximation one replaced by horizontal loading cd which leaves the current
distribytion below a unaffected. The length of cd is roughly equal
can regard the current distribution as triangular over a to 1:5 ab. The value of L is doubled in case (b) since the length of
distance not exceeding A/8 inwards from one end of a wire above it, and hence the capacitance, is halved

208
 MOBILE AND PORTABLE ANTENNAS 209

obtaining an extra 6dB or so of height gain from its

horizontal antenna and the mobile station can expect to
be relatively down by some 9-10dB.
Radiation resistance is proportional to f? and, for
heights of 4/8 or less, also to h?. The 8ft (2-4m)
Re
base-loaded whip has a radiation resistance at 3-SMHz
Fig 16.2. Estimation of efficiency of mobile antenna, R = radiation of only 5-5/16 = 0-34Q and the efficiency is therefore
resistance, AR, =loading coil resistance, R, = earth losses. The 0-34/15-64 or 2-2 per cent only, rising to 6-6 per cent for
antenna is assumed to be resonant
case (c).
From inspection of Fig 16.2 it is found that Fig 16.3 shows another common form of mobile
the proportion of power radiated is given by antenna, the helical whip. A 3-SMHz helix might
R/(R + R, + R,.) but at low frequencies R is only a consist of about 4/4 of 14swg (2mm) wire wound on an
fraction of an ohm and very much less than R, or R., so 8ft (2-4m) rod of lin (25mm) diameter. From Fig 3.10
that the efficiency becomes R/(R, + R.) approximate- the hf resistance is 0-89 and, assuming a sinusoidal
ly. The capacitance of an 8ft (2:-4m) whip is in the current distribution, the radiation resistance may be
region of 30pF so at 3-SMHz we have for case obtained directly from the height of the whip relative to
(a) oL = 1/wC = 1,600, R. = 5-3Q and a value of that of a A/4 monopole, ie it is equal to (4h/i)? x 36Q,
10Q is often assumed for R,. For case (b), R is giving an efficiency of 3-6 per cent compared with 4 per
multiplied by 2-25 and R, by 2, and for case (c) R is cent for the centre-loaded whip without a capacitive
multiplied by 3-06. The efficiencies (relative) are 0-06, hat. Both systems can be improved by a hat, and there
0-11 and 0-2 so that centre loading is 2-6dB better than seems little to choose between them. The helix can be
base loading, and the top hat gives a further improve- somewhat improved by tapering, ie using thicker wire
ment of 2-6dB. It may be of interest to note that with and wider spacing for the lower portion.
R, = 0 these ratios become 0-5dB and 4-35dB respec- It is important to appreciate that coiling of the wire
tively so that the improvement due to centre loading has no effect on radiation resistance, except indirectly
comes about mainly because of the earth resistance. by changing somewhat the shape of the current dis-
Without this, the rise in R, largely offsets the improve- tribution as in the above example. The reason for this
ment in R. The overall Q is given by wL/(R, + R,), ie can be appreciated from inspection of Fig 16.3(b); each
105, 78 and 105 respectively, the 2-6dB improvement in turn of the helix appears to the observer as a zigzag
case (c) compared with (b) being obtained at the price and, following the rules explained in Chapter 2, the
of a 26 per cent reduction in bandwidth. field produced in the vertical plane is proportional to
The situation is slightly different at 14MHz because the extent of these wires in the vertical plane, the
the inductance of the lower portion of the antenna is no horizontal components being cancelled. As a corollary
longer negligible compared with L and the current in to this it follows that loading coils cannot contribute in
the centre of the antenna is some 13 per cent less than themselves to the radiation resistance, though their
at the base. The reactance above the coil is 400, 800 and
4002 respectively but the lower half of the antenna nal
=
contributes about 200Q of inductive reactance in cases an

(b) and (c), so that wL = 400, 600 and 200Q respective- —

sa aiS
Ev
[en
ly. Corresponding values of R, (ohms) are 1:33,
2-0 X (0-87)? and 0-67 x (0:87)?, these being the
values referred to the feedpoint for comparison with R, One turn d d I

|
of helix
the factors in brackets serving to take account of the
sinusoidal current distribution. The radiation resistance
in case (a), from Fig 3.12, is 5-5Q and the factors for
cases (b) and (c) are approximately as before, so that
(again taking R, as 10Q) the efficiency is 33, 52 and 62
per cent respectively. The improvement from centre
loading is 2dB and the further advantage from. the ieee (b)
capacitive hat is only 0-8dB. With R, removed, Fig 16.3. Helical whip. (a) shows schematically a vertical helix and.
the top hat would reduce the coil losses by four but (b) is an enlarged view of one turn showing the horizontal and
these are small enough for the improvement in overall vertical fields produced by each half-turn as seen by a distant
observer. The horizontal components cancel, the vertical ones
performance to be negligible. add to produce the equivalent of a current / flowing through a
At 28MHz the antenna is self-resonant and losses are distance (d) equal to the pitch of the turn as shown on the right.
The losses are increased compared with those in a straight
even less. One might expect this to be reflected in conductor of the same wire diameter in the ratio 1D/d, or rather
impressive performance but, for dx communication as more unless the turns are spaced by several times the wire
compared with 14MHz, an average fixed station will be diameter
210 HF ANTENNAS FOR ALL LOCATIONS

Table 16.1. Inductance values, efficiencies, and overall bandwidth for 8ft (2-4m) centre-loaded whip
without top loading
Frequency (MHz) 1:8 3-6 7:05 10 14-2 18 21-3 25 28-5
Inductance (wH) 582 142 35-5 17 7:24 4 2:1 0
Radiation resistance ((Q) 0-2 0-8 3 6 12 AY. 21 28 36
Efficiency (Q=100) (per cent) 0-26 1-9 10 22 42 62
Efficiency (Q=300) (per cent) 0:62 3:7 16 30 50 60 66 73 78
Overall bandwidth (Q=100) (kHz) 21 48 124 490 1,420
Overall bandwidth (Q=300) (kHz) 9g 24:5 78 417 1,290

length must be included in reckoning the total height of fact that inductance is proportional to the square of the
the antenna. number of turns.
Due to differences in car bodies and mounting For multiband operation the best method is to
positions, no precise dimensions can be given and each change the coil since it would appear that any attempt
installation must be tuned in situ. The position of the at variable tuning must result in some loss, particularly
coil is not critical; moving it to the middle of the top if a capacitor is used as this ‘“‘works”’ by increasing the
half at 3-SMHz requires L and therefore R, to be current in the coil; typically, if /is doubled L is halved and
doubled, but increases R in the ratio (7/6)? and reduces losses doubled.
the efficiency by 10 per cent. There is therefore a loss in To obtain maximum benefit from use of a hat it
signal strength of 0-4dB, though there could be a slight should be as large as possible, so that its removal or
additional loss due to the self-capacitance of the coil reduction as a means of band changing is not accept-
which could be approaching self-resonance. Assuming able, except perhaps at the hf end of the range where
a stray capacitance of 1-5pF, which is probably equal to efficiency is still moderately high even without the hat.
about 20 per cent of the total tuning capacitance, this
increases the current in L by 20 per cent but reduces the
required value of L by the same amount, so that there Top loading
is a net increase of 20 per cent in the effective value of A 3ft (0-9m) horizontal bar as in Fig 16.1(c) is not the
R,. This results in a signal loss of 0-3dB. most convenient method of top loading, a “‘skeleton”
The capacitance of a short whip antenna (<A/8) can disc, Fig 16.4, being generally favoured [1]. For a solid
be reckoned as roughly 3-4pF/ft (11-2pF/m). This disc the capacitance in picofarads is roughly equal to
assumes an average diameter of 0-5in (13mm) and is a 0-9 times the diameter in inches (0-36 times the dia-
bit on the low side for longer lengths, but subject to meter in centimetres) and the skeleton disc is only
normal practical constraints differences of more than 15 slightly less. To obtain the required capacitance of
per cent are unlikely. Such figures can in any case be 13-5pF a disc diameter of 15in (38cm) with at least four
used only as rough guidelines in view of the many other spokes should be satisfactory. Alternatively, for a
variables as already indicated. cylinder having a length equal to the diameter [2] the
Table 16.1 shows approximate values of loading capacitance in picofarads is equal to twice the diameter
inductance required for the various bands on the basis in inches (0-8 times the diameter in centimetres). To
of the above figures for C and, at the higher frequen- obtain the greatest possible loading effect for a given
cies, due allowance for the inductive reactance of the length of conductor this should be curled round or bent.
lower half of the antenna. A value of 10Q is assumed over, leaving the end free as in the case of the miniature
for R,. beam elements (p176). The increased loading is due in
Table 16.1 may be scaled for other lengths of whip, part to the introduction of an inductive element but this
eg for a 12ft (3-7m) whip the 21MHz column applies at is of low loss and some overall improvement might be
14MHz, subject to maintaining unchanged the coil
reactance and the overall Q. In other words, the induct-
ance for 14MHz becomes 2-1 X 1:5 = 3-15uH and the Table 16.2. Inductance values, efficiencies and
bandwidth (Q = 100) becomes 1,420/1:5 = 947kHz. overall bandwidth for 8ft (2:4m) centre-loaded
For maximum Q the coil size should be as large as whip with 7pF top-loading capacitance (AR; =10Q)
possible but it is not critical. Weather protection tends
Frequency (MHz) 1:8 3:6 TOR. 00 14:2 18
to degrade the Q but is necessary if there is any chance Inductance (1H) 2ole OF IGS 7 2-6 0
of moisture being trapped between turns. The use of a Radiation resistance (Q) 0:27. 1:09 4-1 8-3 16-3. 27
long coil with self-supporting turns should avoid this Efficiency (Q = 200)
(per cent) 1:3 6:8 24:7 43 60 72
but it is possible only at the high frequencies; too long a Overall bandwidth (kHz) 11-5 37 141 450 934
winding degrades the Q and a diameter of 3in (7-5cm) Coil turns, 3in (7-6cm)
with a length of 6in (15cm) is a reasonable compromise. diam, 6in (15-2cm)
long 47 24 15 9
Coil turns may be obtained from Table 16.2, using the
 MOBILE AND PORTABLE ANTENNAS 211

self-resonant at 21MHz and for higher frequencies it

will be necessary to reduce the amount of top loading.

Coil construction
Drilled to fit ontop section Nylon or ptfe is recommended for coil formers and the
of whip and secured with
Aluminium grub- screws wire gauge should be chosen so that the spacing
boss 1"dia
between turns is not less than 0-6 times the wire
_——=
Ve" dia aluminium tubing diameter. Self-supporting air-spaced coils are prefer-
recessed in boss
able at the higher frequencies and can be wound with
isin (Smm) or 4in (6mm) diameter copper tubing.
Mechanical fixings can be devised using, for example,
wing nuts to allow interchange of coils plus the top
Fig 16.4. Capacitance hat suitable for vertical whip antenna
sections; however, tapped coils may be used for greater
convenience if some slight drop in efficiency is accepted
expected, though this idea does not seem to have been at the higher frequencies. One method of coil assembly
exploited up to the time of writing. [1] is shown in Fig 16.5. Loading coils should be
One effect of top loading which may be advan- protected from moisture and this can be done by means
tageous mechanically is that it allows the loading coil to of shields cut from old polythene bottles.
be mounted if necessary in a lower position. The
tendency in this direction is evident if one thinks of an
extreme case in which the amount of top loading is so Earth resistance
large that it results in uniform current in the vertical Despite the crucial role of earth resistance and the
portion of the antenna even if the coil is placed at the capacitance to ground of the car body, there are few
base, the coil inductance and therefore the resistance data available for these quantities, the 10@ resistance
then being reduced without affecting the radiation assumed in the above calculations being a value which
resistance. happens to be consistent with some measurements at
To take a more practical example, assume an 8ft 4MHz [3]. As the main source of loss it is worth looking
(2:-4m) whip with the loading coil at the top and into more closely, and from inspection of Fig 16.6 it
surrounded by a cylinder of 8in (20cm) length and looks as if R, ought to be reducible by mounting the
diameter; the current distribution is virtually uniform, antenna directly above the centre of the car roof. The rf
thus increasing the radiation resistance by four com- return circuit from the antenna is partly by capacitance
pared with Fig 16.1(a), ie to 1:36Q2 at 3-SMHz. The
capacitance is l6pF so that L = 129uH (total) or TOP SECTION
125uH for the coil. The coil resistance (Q = 300) NYLON OR PTFE
comes to 9-16 and the radiation efficiency is 1-36/ ROD CONNECTING
TOP AND BOTTOM
20-52 or 6:6 per cent, ie very slightly worse than the SECTIONS
previous example in which the coil was placed
half-way down. Next consider base loading, but since
the coil no longer has to be supported, let it be
supposed that the cylinder can be increased to 14in
(35-6cm) length and diameter; this doubles C, thereby
halving L and R, but the current is now non-uniform to AIR-SPACED
LOADING COIL
the extent that the radiation resistance factor is 2-25 (as
for Fig 16.1(b)) instead of 4. The efficiency is therefore
0-8/15-4 or 5-2 per cent, which is slightly worse, being
1-5dB down in terms of signal level compared with the Sisssgsess:

original figure of 6-8 per cent efficiency for centre plus

top loading, but still 3-74dB up compared with simple
base loading.
BOTTOM SECTION
There is of course an exact optimum position for the
loading coil, but the optimum is extremely flat especial-
ly (as we have just seen) if top loading is also used, and
there is unlikely to be much advantage in departing Fig 16.5. Loading coil assembly. An insulated rod is inserted into
the bore of the whip above and below the coil, and runs through
from the centre position for the loading coil. Table 16.2 the centre of the coil. The rod must have good insulating
provides design data for top loading in accordance with properties and mechanical strength, ptfe or nylon being the most
Figs 16.1(c) or 16.4(b). The whip will be roughly suitable materials
212 HF ANTENNAS FOR ALL LOCATIONS

directly to the car body (mainly the roof), and partly by

capacitance to ground in series with the capacitance to
ground of the car body via the resistance of the ground.
It is this second path which must account for the whole
of the unavoidable losses.
A major improvement would result if this lossy path
could be short-circuited and it would appear from
inspection of Fig 16.6(a) and (b) that this objective
might be largely achieved by mounting the antenna in
the centre of the car roof. Further to encourage the rf
current to take the shortest path back to the transmit-
ter, the base of the antenna could be surrounded by two
or three tuned radials mounted a few inches above the
La roof. The antenna could then be isolated from the car
body by parallel tuning of the capacitance between the
Ca! body and the radials as shown in Fig 16.6(c) and (d)
(b) thereby, at least in principle, removing most of R,.
However, owing to the resistance of the loading coil
this is unlikely to help much at the lower frequencies,
Re
and a better course in this case might be to series-tune
the car body with a suitable inductance so that it acts
like the lower half of a dipole. This should add to the
radiation resistance, and thus to some extent offset the
(c) Loose coupling to
minimize ground Loading coil
‘ reduction in length of the radiating element which is
capacitance likely to be part of the price paid for roof mounting.
Insulator
A possible source of earth resistance is poor electric-
Loading coils al contact between different portions of the car body
Ground plane
and, if for example the antenna is mounted on a wing or
bumper, it is advisable to check the dc resistance
Tunes out capacitance between this, the chassis and other large areas of metal
between ground plane such as the roof. Up to the time of writing Fig 16.6(c)
and car body
and (d) is an untried idea, the discussion of it being
aimed at providing further insight and forestalling any
temptation to dismiss R, (prematurely) as an insoluble
problem.
End loading may
. ‘Ca bene cessar y
(d) ie \
Matching
~

a __ Loading ‘\ Matching between the mobile antenna and the trans-

4 :
rs To transmitter coils \\
Ca f°
| \ mitter is probably best achieved by means of an L
/
/
network, Fig 16.7. For the examples given earlier the
/ total resistance R of the antenna is very roughly 20Q in
/
/ One of the radials all cases, since reduction in radiation resistance is
accompanied by increases in coil resistance and a
_~ constant figure of 1092 has been assumed for R,.
By tuning the antenna “‘slightly long’ a reactance

Ly
Fig 16.6. Reduction of earth losses. (a) analyses the problem in
terms of the relevant capacitances and resistances, the equiva-
lent circuit being shown at (b), and the simplest approach
requires that everything possible should be done to reduce C,
relative to C, by careful positioning and probably top loading. (ch
is an attempt to remove earth losses almost completely. The
loaded antenna and radials can be regarded as an isolated dipole,
the transmitter being loosely coupled and taking advantage of
the high Q of the system to minimize the disturbing effect of the Fig 16.7. L-match. X, is determined as shown in Fig 16.8; use of
feeder. The capacitance between the radials and the car body is this value ensures that a match can be achieved by tuning the
tuned out by Ly, the equivalent circuit being shown at (d) antenna slightly low
 MOBILE AND PORTABLE ANTENNAS 213
WAVELENGTHS
FORWARO °
pecisets ATTENUATION

(Ra
+j Xa)

ss eae
Sa . ae

Fig 16.8. Use of Smith chart to determine

Xp. Given that R = 0:42p points are found
:
CSS SSE HH PR SSS
on the 0-4 and 1-0 resistance circles such
that OP = OP’, all three points being in a
straight line. This gives the value of X,, and
also of X,, though this latter value is found e, <S .a>-, “ 0)
in practice by tuning for minimum swr after
connecting the right value of X, SHI
oO ose anener
y- o6esecune wae, Sw,
ESS
ser sett RRS
SSTEL wr PLRKOSN
LLKRKROS LI
6 omens LZo
LS52
eeseeeed et
SO

Making OP=OP’
we read off at P’
the admittance
(1—1+25j), equivalent
_to (50 IE ei50.
7:25a j) ohms

+X, is caused to appear in series with R, and this is greater than R,, the amount of detuning required to
equivalent to a parallel combination of resistance R,’ effect the matching will be just over half the bandwidth
and reactance X,. For matching, X, is chosen so that figures given in the tables.
R,' = Z, and the resulting value of X, is then tuned out The corresponding capacitance values are given in
by a parallel reactance —X,. This can be done very Table 16.3. It will usually be sufficient to select the
easily. with the Smith chart; referring to Fig 16.8, nearest convenient values. In those cases where the
moving half-way round the circle from any point P antenna is nearly self-resonant, eg for 24 or 283MHz, the
corresponding to (R, + jX,) brings us to a point P’, the radiation resistance plus the earth resistance should
coordinates of which are equal to 1/R,' and 1/X,. For provide a reasonable match to 50Q feeder without the
this exercise one uses the “‘normalized”’ values of R, use of a capacitor, though the antenna length or loading
and X, obtained by dividing Z, into the actual quanti- must be adjusted for resonance.
ties, and the answer is interpreted via the reverse If a base loading coil is used, matching may also be
process of multiplying by Zp, as explained in Chapter 4. carried out by tapping the inner conductor of the
For matching to 50Q the normalized R, is 0-4, and coaxial feeder onto a suitable point on the coil.
going from P to P’ must be done so that, starting from a The antenna may be tuned approximately by using a
point on the R=0-4 circle and proceeding on a gdo, the feeder being preferably short-circuited which
straight line via O, we arrive on the R = 1-0 circle in will place the frequency slightly on the low side as
such a way that OP is equal to OP’. Laying a transpa- indicated earlier. After arriving at the right size of
rent ruler across the chart, it is easy to find a position loading coil with as much capacitive loading as possible,
such that intercepts with the R = 0-4 and R = 1-0 final trimming may be carried out for minimum swr
circles are equidistant from the centre as shown. Cor- using whichever method is most convenient. If it is not
responding reactance values read from scales round the possible to obtain a satisfactory swr, this may be due to
edge are X, = 0-5 and 1/X, =—1-25. The last figure is
the most interesting one since it tells us that if Table 16.3. Values of matching capacitance
R, = 20Q a capacitive reactance of 50/1-25 (ie 40Q) assuming R, =20Q
will provide a perfect match subject to adjustment of
antenna length, size of capacitive hat or loading induct- Frequencies (MHz) 1:8 3:6 7:05 10 142 21:3
Capacitance (pF) 72l\2 4,100 soe 400. 28) ~<186
ance, whichever is most convenient. Since_X, is slightly
214 HF ANTENNAS FOR ALL LOCATIONS

a really steep ground slope even lower heights down to

10ft (3m) or so can sometimes be equally effective. On
the other hand, it has at times been found more
convenient to string the dipole horizontally between
trees. At other times the dipole has been rigged
between projections near a cliff top, for example rocks
or tufts of heather, a 10ft (3m) “‘clothes prop” being
used to push the centre of the dipole out over the edge.
On other occasions a stone attached to polythene cord
A
has been thrown over a tree branch and used to haul up
Wire 22swg ri the centre of the dipole, the ends being attached to
(O-7mm) ri — Balun
50-100ft (15—30m) lengths of thin polythene cord or
nylon fishing-line (no other insulation needed) and
strung out to whatever anchorage happened to be
‘Knot available. Even a tuft of grass serves if there is nothing
better, though the anchorage should be as high and as
me “. Polythene
“ 728cable
coaxial 2 guys to ‘.cord
ae far away as conveniently possible.
: tent pegs SS
Alternatively a mast can often be improvised from
% ee
" x xX vy bits of wood that are lying around, for example drift-
4, Tent peg or Wy wood on the sea shore. On one occasion the author
4 tuft of grass used a dipole stretched across a ravine, though reliance
Fig 16.9. Inverted-V dipole over steep ground slope as used by the on natural features can be somewhat restrictive. The
author for numerous antipodal and other dx contacts with very advantages of carrying one’s own mast around are
low power. Average time for erection is about 15min
considerable, and a 20—24ft (6-7m) mast suitable for use
values of R, or coil Q different from those assumed in in the sort of weather conditions conducive to portable
the above calculations and a different capacitance operation can be constructed from tapered 4ft (1:2m)
should be tried. lengths of bamboo with short pieces of aluminium
tubing pushed over one end to form a socket for the
next section. For use on steep ground slopes two guys
Portable antennas are sufficient and in some cases the mast can merely
Some of the antennas described in earlier chapters lend lean against a fence, tied loosely to a post. Such a mast
themselves better than others to “‘fixed-portable” and need weigh no more than a camera tripod and is easily
field-day activities but broadly similar considerations carried. |
apply and there is felt to be no need for further It takes only a few minutes to rig such an antenna and
elaboration. The situation is however rather different if results will be much more interesting than anything
discussions are limited to antennas which are ‘“‘port- likely to be achieved with the more usual vertical whip
able” in the strict sense, that is to say capable of being unless one is specializing in seashore or ‘‘at sea”
carried on foot for appreciable distances. Portable locations.
operation in this case can in itself be a unique and The reasons for this will be obvious from Chapter 10,
fascinating hobby, especially if topographical features and for general use the author considers there is no
are exploited for low-power dx as outlined in Chapter sensible alternative to a light thin-wire dipole with long
10. strings attached, which can be rolled up to fit into a
The ideal portable antenna in most cases is the pocket and erected in one of the ways just described.
inverted-V dipole (Fig 16.9), 18-22swg (0-7-1-2mm) Such an antenna, erected at heights upwards of 10ft
copper wire being suitable. Since the apex angle can (3m) above a ground slope of 30° or more, is capable of
usually be made large, so that the ends are not too close reliable worldwide voice communication with radiated
to the ground, the length formula on p00 can be used. powers of the order of 1W. The comparison between a
The feeder can be short and the power is low so that home location with 400W p.e.p., plus a three-element
lighter grades of 75Q twin-lead or 72Q coaxial cable beam, and portable operation with 1W may be esti-
(though less efficient) are quite suitable, the use of a mated roughly as follows:
balun being advisable at the bottom or top respectively.
Suitable baluns can be extremely small, lightweight, Home station
and easily constructed (p51). Power advantage 26dB
It is usually desirable for dx work to use as much Gain of beam 6dB
height and as large an apex angle as possible, for 32dB
example 20—25ft (6-1-7-6m) and a base-line several
times the length of the dipole, but when making use of From this, depending on the wave angle and antenna
 MOBILE AND PORTABLE ANTENNAS 215

Home station report (typical from best

stations) = S9 + 10dB
Portable stations report (typical) = S5,
ie 26dB down, allowing 4dB per S-unit.
Portable stations (predicted for best case, see p138)
Extra gain from distant reflection 6dB
Extra low-angle gain (typical) 10dB
Portable station report (best actual) = S9
The use of a reflector is open to consideration but in
this case the wire diameter must be increased and,
unless use can be made of trees or other natural
supports, a heavier mast or two masts must be carried.
Enlarged view
8uH of tunerfor 14MHz For exploiting the possibilities of low-angle propaga-
50 or 7282 coaxial cable tion with vertical polarization over seawater the use of
a monopole (or more accurately perhaps, asymmetrical
dipole) along the lines of Fig 16.10 is recommended.
1 turn link Seto 6it This should be erected as high and as close to the sea as
coupling (0-9 to 1°8m)
possible, though height is only of secondary
Fig 16.10. Method of re-arranging the inverted-V to provide
importance. For short- or medium-range working the
vertical polarization for experiments over sea water. The balun dipole may be erected over flat ground, but dx possibili-
should be short-circuited, the feeder being cut to a length of ties are then relatively poor unless several watts are
about 15—20ft (4-5-6m) for 14MHz and the system end-fed by the
same method as the “odd bit of wire” (p43)
available and the antenna can be raised to a height of at
least 34/4 this may not prove too difficult at 283MHz.

height it may be necessary to deduct some loss due to

the reflected wave, probably 2—3dB. References
[1] Radio Communication Handbook, Vol 2, 5th edn,
Portable station (worst case)
RSGB, 1977, pp14.10-14.12.
Gain from foreground reflection 6dB
[2] The ARRL Antenna Book, 13th edn, ARRL, 1974,
Advantage of home station 25dB
pé2.
The following comparison roughly summarizes the [3] The ARRL Antenna Book, 12th edn, ARRL, 1970,
author’s experience of Australian contacts on 14MHz: p293.
CHAPTER 17

What kind of antenna?

The reader who has persevered with earlier chapters justified provided it does not escalate much further. On
may by now have a fairly good idea as to the types of this basis one would arrive at an ‘‘economic cost”
antenna best suited to his needs. Nevertheless, the formula of Sp per milli-decibel from which it follows for
factors to be taken into account are many and varied so example that, in the case of a typical small wire beam,
that important aspects can be overlooked; the aim of the improvement in performance from increasing the
this chapter is to provide some additional guidelines. wire diameter is far outweighed by the cost of the extra
It is useful to start by asking two questions: copper! Going on from this it is easy enough to justify
the cost of simple beams even without taking into
(a) What facilities are necessary to the full enjoyment
account the reduction of interference which is probably
of our hobby, and what compromises are we
more important, but the extra costs of a full-sized beam
prepared to make? plus the tower which may be needed for it requires
(b) How much is a decibel worth to us?
drastic revision of the formula; moreover a small beam
In dealing with the first question the “I want the lot” of 50ft (15m) should be at least as good as and probably
complex should, if possible, be avoided if only because less unpopular than a large one at 40ft (12m).
the least important 20 per cent is likely to be much The value of a step in performance such as 6dB can
more expensive than the remaining 80 per cent. For this be readily assessed by waiting for interference to occur
purpose cost needs to be reckoned not only in cash but and then switching in or out a linear amplifier or
the effect that too big an antenna or the “‘one antenna attenuator, or arranging for this to be done at the other
too many”’ may have on the tolerance of one’s neigh- end, whichever is appropriate. One rarely has long to
bours! Moreover, above a certain performance level wait for the opportunity and the result is likely to be
the cost rises steeply with each additional decibel. convincing! In the absence of interference, however,
How much then is a decibel worth? In a contest it the situation is very different since propagation is usually
could be the difference between winning or losing, or good or non-existent. It is rarely marginal enough for
between, say, the 12th and 14th place; in the first case it communication to be seriously affected by a step which
is clearly important, in the second case not very! One is small compared with the difference between the
argument to be strongly rebutted is the frequent asser- weakest useful signal and the strongest likely signal,
tion that a decibel is unimportant because one cannot typically of the order of 30dB. In practice, though 20W
measure it, or hear the difference. After possibly sacri- may well result in a lengthy dx contact “Q5 all the
ficing several decibels in this way the next step is way’’, the proportion of such contacts is increased and ©
probably to go out and purchase a linear amplifier, thus the number of repetitions or serious frustrations greatly
obtaining an advantage of 6dB at a cost of, say, £300, or reduced with a power level of 200W. .
£50 for each ‘‘worthless”’ decibel. Another basis for assessment is suggested by the
It may be suggested that decibels are perhaps not all author’s experience during the early years of the
of equal importance, and the author has mdeed found BERU Contest when a ratio of 1-6 was deduced for the
that being without a 6dB linear is more of a handicap increase in number of contacts corresponding to the
when working at the 100mW level than at the 100W 6dB power difference between the Senior and Junior
level. Though based on much experience, this is little contests, or roughly an increase in the number of
more than a vague personal impression and serves contacts of 8 per cent per decibel. As a statistical
merely to confirm that within any small range, such as exercise the above examples may be open to criticism
6dB, all decibels can be regarded as equal. Com- but until a more up-to-date figure is available the
munication in the amateur bands is usually restricted by author sticks to his view that the decibels wasted, for
interference rather than noise so one might argue that example in traps, add up to a poor bargain.
little would be lost and much gained from a power limit The interest of amateur radio can often be enhanced
of 30W, universally applied. Nevertheless, as things are by specialization and this reacts favourably on antenna
we live in a competitive world and, taking a consensus specifications. For those not yet sure of what they want,
of reasonable opinion, the cost of the linear is probably a simple beam such as Fig 12.11 or 12.12 is a good

216
 WHAT KIND OF ANTENNA? 217

starting point, and for the lower frequencies inverted-V Tubing elements for 10]14MHz
‘
—
Clttree
18-30 MHz elements not shown,
see Fig17.2
dipoles will provide good coverage to ranges within and
somewhat beyond the 14MHz skip distance. At this
stage, unless an eventual mast height of at least 40ft
(12m) is contemplated, it will be useful to carry out —~VK2 ABQ type
comparative checks of vertical and horizontal polariza- array for 7MHz
(see Fig 12.10)
tion, preferably using a vertical beam such as in Fig
13.9(b). The vertical antenna for this test must not be a
ground-plane type and must be well clear of tree trunks
or wet wood for most of its length. For a taste of dx at
the lower frequencies the beam with its feeders may be
4-half loop array
excited as a vertical antenna working against a short for 3°-5MHz similar
counterpoise (p43). to Fig 13.9
Since the best choice of antenna systems depends on (not shown in full)

personal priorities as well as the means at hand there

cannot be two cases alike. An earlier idea of trying to
7
anticipate reader’s problems and come up with ready- 7
/

made answers has therefore been discarded; instead a 7’ Multiband inverted

V's for short-skip
number of scenarios have been devised for their merit |
- it dantacts on 3-5 and
as vehicles for discussion of the topics judged to be 7MHz

most important. d 1¢8 MHz tuner

The ideal antenna system (A) V2

The first example is an attempt to “provide everything” Fig 17.1. Ideal system based on 100ft (30m) mast which acts as a
inclusive of the best possible performance on all bands; 4 top-loaded vertical antenna for 1-:8MHz and supports four
it is further supposed there are no planning restrictions vertically polarized half-loops which operate as a three-element
Yagi for 3‘5MHz. The rotary beam on top of the mast uses
or financial constraints but as concessions to reality the stacked two-band arrays for coverage of all bands from 10-
following assumptions have been made: 28MHz including 10, 18 and 24MHz, elements being collinear
pairs at the higher frequencies. A modified VK2ABQ-type beam
(a) Height limited to 100ft (30m), with only one mast for 7MHz is mounted on the same boom. A low-height multiband
(or tower) of this height; though achieved infre- dipole is included for short-skip working
quently in the UK this is not unusual in some
top of the mast, since coverage of all bands from 7 to
countries.
28MHz, including 10, 18 and 24MHz and making full
(b) Some space restriction, to the extent that large
use on all bands of the available aperture, is implicit in
rhombic arrays are ruled out.
the terms of the scenario. A suggested basis for such a
(c) Maximum length for tubular beam elements and
development is given in Fig 17.2 which shows what is
booms not to exceed 45ft (13-7m) and 30ft (9-1m)
basically a conventional three-element Yagi array for
respectively.
10-1MHz. By means of the relatively simple two-band
Presented with a 100ft (30m) mast or tower the type of linear resonator these same elements can be
author’s first instinct would be to fetch a saw and cut it made to operate simultaneously as extended elements
in half, partly from ‘“‘fear of monsters” but also in the for 14MHz or for both 14 and 28MHz.
belief that a two-mast system offers important practical It may be recalled that three-band 14/21/28MHz
advantages. This will be treated in due course as case operation required a relatively complicated system of
(B) but it is instructive first to consider what might be resonators because the 21MHz resonator, though in-
done with the mast if still intact. The choices in this case ductive at 28MHz, was not quite inductive enough and
are relatively simple and lead to an arrangement such this led to narrow bandwidth and too much inter-
as that shown in Fig 17.1. Here the mast is tuned as a dependence of adjustments. The separation of 283MHz
\/4 vertical antenna for 1-8MHz (p114) and used as from the nearest lower frequency being in this case a
support for a vertical-polarized 3-SMHz array on the whole octave, there should be no problem in this
lines of Fig 13.9(b), thereby achieving a gain in excess respect but the spacing of elements is excessive and the
of 6dB. An inverted-V dipole for 3-5 and possibly bandwidth of the linear resonator may be insufficient
1-8MHz is attached at about the 30ft (9-1m) level to for complete coverage of the 283MHz band. Another
provide coverage of short and medium ranges for which tool available for use in solving the multibanding
the vertical array is unsuitable because of insufficient problem is the extended two-band dipole of Fig 12.6, so
high-angle radiation. that putting these ideas together one can imagine a
A suitable beam will need to be developed for the cord-braced structure based on the 10°1MHz tubing
218 HF ANTENNAS FOR ALL LOCATIONS

So o-a eo

as | 11 oases X ...... Beam reversal relay

a, }——-x ——f ree
Som L| 10/14MHz Sec YY.... Feed points
‘ =)
Ss} LS

Wire elements are all

2- band colinear dipoles as
S| a 1S Fig 12.6.
48 /25MHz These can be identical
except for the tuning
capacitors
at 1
fetuet
‘ '
fs
21 VK2ABQ- type 7MHz
riven element ibbigg aed /28mMH>
¥ | | elements (not shown)
; ’ are strung round the edge
S! Driven element | | {S as in Fig 12.10
[Se ot oe See eee see ee cay ee ow agen et ae ne Sonbe cae
S\ | | | iS
Drive
Y element Y Y Y Y 49 |2oMuz

| | Ler ee indicates cord

Si | | iS

Sh ASS

bahagioliceet re ta ge am
S; 10/14MHz 1S

di eer X'et es
Linear TAORE Gees teem DME ee
provide 10/14MHz ees Taps
SS
operation and beam =
reversal (both bands) ee Borers

Fig 17.2. Multiband beam for 7-28MHz as specified for Fig 17.1, three elements are used on 10/14MHz, and effectively six from 21-28MHz.
For 7MHz see Fig 17.1. For beam reversal on 10 as well as 14MHz an additional pair of relays may be used. Conventional insulators are
shown in order to indicate the ends of wires but the use of polythene or nylon cord is preferable

elements but supporting whatever wire elements may currents, using the elementary type of trap shown in Fig
be required. 7-6 or possibly to use relays as described in Chap-
Fig 17.2 uses three pairs of two-band collinear wire ter 7. Note that the beam has been made reversible in
elements covering respectively 18/24 and 21/28MHz as view of the stress laid on this feature in earlier chapters;
marked; these are arranged as two separate beams with this unfortunately complicates the 14MHz linear re-
opposite directions of fire. Further wire elements may sonators but it is possible to replace them by additional
be added to form a VK2ABQ-type two-element beam relays.
for 7MHz. An alternative would be the use of a pair of This example demonstrates the complexity which is
stacked vertical arrays, but vertical stacking of horizon- the price for insisting on the “best of everything”’ yet,
tal arrays is not recommended in view of the immense although there is nothing available “‘off the shelf’, the
increase in complexity for an advantage (calculated) of author is satisfied that such an array could be made to
only 1-4dB with optimum separation (about 40ft or work by the exercise of skill and patience.
12m). Incidentally, even this small advantage decreases If searching for a simpler system, one option is to
and quickly becomes negative as mast height is re- accept lower gain at the higher frequencies, virtually
duced. discarding four or five of the eight elements by using a
At this stage separate feeders are assumed for each log-periodic array. A design which covers 13-30MHz
array but some combining of feeders and/or remote has been described by P. D. Rhodes, K4EWG;; this
switching should be feasible. It will be necessary to trap uses 12 elements, ranging in length from 38ft (11-6m) to
the 10/14MHz elements just outside the linear resona- 12ft (3:7m), the boom length being 26-5ft (8-1m) and
tors against 18MHz currents, 18MHz elements against the gain roughly equal to that of a three-element Yagi
21MHz currents and 21MHz elements against 24MHz [1]. Using the procedure indicated on p202 this design
 WHAT KIND OF ANTENNA? 219

could be extended to include 10-1MHz by adding three It is possible as explained in Chapter 13 to use the
elements with a maximum length of 50ft (15-2m) and 7MHz arrays as monopole arrays for 3-SMHz, in which
extending the boom length to 41ft (12-5m), though it case the 3dB gain from phasing should at least equal
might be a better compromise to retain the previous any loss incurred in going from dipoles to monopoles
antenna and surround it with VK2ABQ-type arrays for and at all of the higher frequencies the theoretical gain
7 and 10MHz. As described, the log-periodic antenna should be fully realizable, so that broadly speaking
has a total weight of 116lb (52-6kg) and a wind load nothing has been lost by going from one 100ft (30m) to
area Of 10-7sq ft (0-99m?) which has to be reckoned two 70ft (21m) masts; on the other hand, use of two
(p253) as a possible horizontal force of some 300Ib antennas provides a worthwhile improvement in direc-
(136kg) acting on the top of the tower. tivity. A possibly more important bonus lies in having
These figures, together with a high “visual impact the option of being able to switch instantly between any
factor’, would probably rule out such an array for most two directions. Switched coverage of another direction
locations in the UK regardless of the height of mast, plus its reciprocal could be obtained by suspending a
but an alternative form of construction is suggested in wire beam between the two towers.
Fig 17.3. Only three relatively thin tubing elements are
used, the structure being cord braced and thin wire
employed for the remainder. Mixing different gauges of
conductor in this way tends to complicate the design
because of differences in Z, but some degree of com-
pensation can be effected by using pairs of wires in
parallel as shown. It will also be necessary to compen-
sate for the inward pull of the side cords by bending
over the ends of the elements. Though yet to be tried
and clearly not for the inexperienced, this could pro-
vide a possible means of achieving the benefits of a
log-periodic array without the increased weight and
windage of conventional designs.

The ideal antenna system (B)

The two-mast system can follow similar lines to Fig
17.1, everything being duplicated except that 1-8 and Ore Insulator
A. Be: Tubing elements insulated in centre providing mechanical
3-5MHz become 3-5 and 7MHz respectively; this means support for the additional (interleaved) wire elements
that the rotary beam can be reduced in size since it no rapt a Thin wire elements using double wires to minimize
longer has to cover 7MHz. In scaling this down from a impedance discontinuities

minimum frequency of 10-1 to 14MHz the centre Fig 17.3. Suggestion for lightweight log-periodic antenna. Struc-
elements covering 18/24 and 21/28MHz have to be ture is basically that of a cord-braced three-element array with
deleted, and a VK2ABOQ-type array for 10-1MHz re- additional elements formed from pairs of thin wires
places the 7MHz array. A small amount of gain is lost
at each end of the spectrum but the mechanical advan- For those undeterred by the extra weight and win-
tage is considerable. For 3-SMHz a top-loaded /2 wire dage of quad antennas, and who would perhaps wel-
element may be suspended between the poles and this come the idea of a simpler beam which can be adjusted
is also used as a monopole for 1-8MHz, some gain from ground level, the tree-mounted quad (p179) can
probably being obtainable on both bands for directions be readily adapted for rotary operation; though slightly
aligned with the towers by tuning these as parasitic inferior in gain and front/back ratio it has the big
elements. advantage that reversing relays are at a low height and
Provision is assumed for driving the two sets of arrays accessible for maintenance.
in phase (p184), thus giving the option of a gain boost in The use of very short lengths of 300Q flat twin line
one direction or instantaneous switching between any for bypassing the beam rotator avoids problems with
two directions. The break-even height is likely to be at the twisting of open-wire lines. For adjustment in the
least 55ft (17m) for horizontal and vertical polarization case of 70ft (21m) mast height it will be necessary to
at 7MHz, so that for a mast height of 55ft there would be provide a ladder or other access to the 24ft (7-4m) level,
a height loss for low angles of 5dB compared with the or else extend the feeders by extra half-wavelengths
previous example (A), this being offset by 3dB of gain which can later be removed if relays are used for band
from phasing plus 2dB from the additional elements. changing and directional switching.
Between 55 and 70ft (21m) the extra height gain is nil in High-angle coverage for the lower frequencies re-
the vertical case and about 2dB in the horizontal case. quires inverted-V dipoles as for system A.
220 HF ANTENNAS FOR ALL LOCATIONS

Making the best of limited resources (A) visual impact and is the easiest to support, despite
the need for an extra 4ft in pole height to obtain the
It is not the intention to engage here in further discus-
same average height.
sion of invisible antennas as such, though some com-
(c) Miniature beam, eg as in Figs 12.14 or 12.13, if it is
mon ground may be perceived between this section and
impossible to accommodate anything larger.
Chapter 15.
(d) For those willing to tackle something slightly more
The first scenario to be considered assumes a two-
ambitious, a three-element beam on the lines of Fig
storey house with a tv-type antenna pole attached to
12.10, though reversal will be complicated if 10, 18
the chimney; there is a small garden at the back, not
and 24MHz bands are required.
less than about 50ft by 25ft (15m by 7-5m), with
neighbours all around. The first step is to acquire some Assuming reversibility and a suitably placed shack,
sort of solid pedestal providing a height of about 20ft any of these arrays can be rotated from the operating
(6m); this could be anything from a strong ladder to the position by means of two polythene cord lines.
bottom half of the wooden mast shown in Fig 19.8(d) For the lower-frequency bands the vertical beam can
which is adequately guyed, or a telephone pole. The be used on 7MHz (p185), and one or both of the
essential feature here is that there must be no risk of horizontal arrays can be used with series inductance
collapse which could precipitate an antenna, which may loading (both elements plus feeders being connected in
be some 20ft higher, into a neighbouring garden; if parallel) against a counterpoise as vertical radiators
anything breaks, it has got to be the top part of the for 3-S5MHz. Better than this (provided care is taken to
erection and not the pedestal! There are many ways of avoid in-band resonances of wires not in use) would be
contriving and erecting extensions of the pedestal to the use of a horizontal wire between the two masts to
support, say, a VK2ABQ-type beam at 30—40ft (9-12m) act as end loading for a vertical wire on 3-5 and
or an inverted-V array at an equivalent height, and the 1-8MHz. The ends of the loading wire can be bent over
same support can also be used for a vertical array, Figs and brought down to near ground level if necessary to
13.2 or 13.10. A number of ideas will no doubt suggest provide extra loading. For high-angle coverage on
themselves to readers after perusal of Chapters 12 and 13. 3-5/7MHz an end-fed long wire with its end folded over
The proposal is to use each of the two poles to is probably the simplest device, although if wires can be
support a pair of the horizontally polarized arrays listed carried over the house a 7/3-5MHz inverted-V dipole
below, and in addition to provide one vertical array. supported by the tv pole would be ideal, and might
Before deciding final details, however, an inverted-V even have some useful dx potential as well. It is
dipole array and a simple vertical beam such as Fig important to make sure such wires do not upset the hf
13.2, 13.9(a) or 13.16 should be erected for compara- arrays, any interaction being readily curable by detun-
tive tests on 14MHz, as a result of which it may be ing the long wire or inverted-V. |
decided for this frequency to discard either horizontal Some of these ideas have been assembled in Fig 17.4.
or vertical polarization or to retain both as part of a
system to provide switched coverage of six directions.
The use of beams for this test, though not essential,
Ba At least 35ft (11m) Roo
reduces the risk of confusion due to local reflections
besides being more realistic, allowance for any height Inverted-V
differences being made as discussed in Chapters 10 and
13.
For the final system it may be possible to dispense
i
altogether with beam rotation but, if rotation is re-

»s
quired or if there are difficulties in obtaining a suffi-
cient apex angle for the V, then one of the rotatable
designs must be selected; note however that the recom-
4 half-loop=—/7
mended beams are all reversible. Two cords (eg Vertical
polythene, 2201b (100kg) breaking strain) are therefore array,
as , 7

all that is needed for rotation. Even if only one Fig 13.9
anchoring point is available for the cords, rotation
through at least 120° should be possible and is more
than adequate, though extremes should perhaps be
avoided in stormy weather.
Fig 17.4. High-performance ail-band system for small garden
using a pair of 30-40ft (9-12m) supports spaced at least 35ft
The choice of horizontal beams is as follows:
(11m) to accommodate a phased pair of simple arrays for 14—
28MHz. The half-loop vertical array can be used to form a pair of
(a) VK2ABQ, the most versatile all-purpose choice. monopole arrays for 7MHz or, with series tuning, as a top-loaded
(b) Inverted-V, which provides minimum windage and monopole for lower frequencies
 WHAT KIND OF ANTENNA? 221

Making the most of limited resources (B) Operation as a beam should be possible at 10MHz,
since in this case the restricted bandwidth is unimpor-
In this example no assumptions are made beyond the tant, though there could be some losses.
existence of a large climbable tree in the garden. The There remains the possibility of adding a linear
advantages of loops for antennas mounted in trees and resonator as in Fig. 11.11(e) to achieve bi-square per-
some of the problems of erection have been discussed formance, giving high gain at 28MHz over a bandwidth
in Chapters 11 and 12. of about 0-5MHz. In this case an alternative feeder
There are several options, an essential requirement system and reflector short-circuits must be provided as
being an 8ft (2-4m) spreader mounted as high as explained in Chapter 11, the feedpoint being moved 8ft
possible in the tree, and if possible at the top of a short (2-4m) from the one used for all the other bands and a
pole secured to the tree and projecting above it. The 4:1 balun employed. A better arrangement is to add,
top corners of two 17ft (5:2m) square loops such as for 23MHz only, a third bisquare element in the centre;
those shown in Fig 7.2 or 11.11(d) or (e) may either be this should be the driven element, the other two being
attached to the spreader before erection or hauled up switched to act as directors or reflectors. This simplifies
afterwards using cords running through slip knots, band changing and improves performance on 28MHz.
which perform the role of pulleys at lower cost! Unless The third element should also be usable at 24MHz and
convenient branches exist a second spreader is attached it allows the spacing of the other two elements to be
to the tree about 23ft (7m) below the top one and used increased for better efficiency on 1OMHz and a better
to support the lower corners of the loops. The side front/back ratio on 14MHz.
corners may be supported by spider arms fixed to the As described so far, provision has been made for only
tree trunk or branches. Open-wire feeders about 34ft one direction and its reciprocal. In principle a second
(10-3m) long drop down to a four-pole beam-reversing beam at right-angles can be mounted a few feet lower in
relay, so that the lower ends are connected either to a the same tree, the beam not in use being open-circuited
reflector tuning stub or through a balun into a 50Q or otherwise disabled. This has been achieved success-
feeder. fully for one band, but for multiband operation difficul-
There is a slight difference in the length of open-wire ties can be expected owing to the fact that the elements
feedline required for 14 and 21MHz which can be can be resonant in a vertical mode not affected by the
compensated by high-capacitance resonant circuits position of a feeder short. This mode can be shifted by
tuned to about 18MHz (p107), but for the driven lengthening the feeder and, given a particular interest
element, unless one is particularly keen on keeping the in one or two bands, the problem would certainly be
swr down to a low value, this is usually not necessary. soluble but once again there is a complicated situation.
For the reflector, changing from 14 to 21MHz requires Hopefully it will be possible instead to contrive a
either a change of about 18in (46cm) in the length of second antenna for coverage of other directions.
stud or an intermediate position for the short, which
may be replaced by high-capacitance circuits tuned to
18MHz. If the tuning does not coincide for the two Improving the existing station
bands a larger or smaller capacitor will be required. Many readers will have already solved their antenna
At 18MHz the overall electrical length of the system problems to the extent of possessing, say, a 35—40ft
is quite close to 2:5 as compared with 2A at 14MHz and (11-12m) mast surmounted by a three-element trapped
3h at 21MHz, and there should be very little retuning beam, this being the type of antenna which appears to
needed, though the optima are unlikely to coincide be most in demand from commercial suppliers. It
exactly. With this in mind it may be decided to accept should be possible to obtain better performance (plus
the nuisance of having to alter the position of the short coverage of the 10, 18 and 24MHz bands) by removing
whenever the band is changed. At 24MHz the electrical the traps, extending the elements slightly and adding
length is 3-5A so again there should be little change in cord ties to support additional wire elements, thereby
tuning, but considerable pattern break-up can be ex- arriving at a scaled down version of Fig 17.2. The mast
pected and there may not be much (if any) gain in the and beam can be used as a A/4 top-loaded vertical for
normal direction. At 28MHz there is virtually no 3-5MHz or as a 4/2 Vertical for 7MHz on the lines of Fig
response in the normal direction, although a single loop 7.13, but it would be worth trying a vertical beam for
is an efficient low-gain radiator for the 45° directions. 14MHz which could be used to provide (a) at least one
A single 14MHz loop of this type has been used with extra switched direction, (b) in some cases quieter
no retuning on both 28 and 7MHz, good dx signal reception due to less short-skip interference and cosmic
reports being obtained on 7MHz and dx worked with a noise, and (c) possibly, operation as a two-element
small fraction of a watt on 283MHz. For 3-5MHz the monopole array for 7MHz.
loops may be used separately or in parallel (both Compared with the “‘ideal station case (B)”’, perform-
alternatives should be tried) as \/4 radiators working ance will be down about 8dB in all, or two average
against a counterpoise or two-radial ground plane. S-units at 14MHz, due to the lower height (about 5dB)
 222 HF ANTENNAS FOR ALL LOCATIONS

and loss of the 3dB of extra gain provided by the have been sought with much better chances of
phased pair of case B. Compared with the original success by other methods.
trapped beam, up to 0-5 S-unit advantage can be (b —
Use of traps. Readers of earlier chapters are aware
expected at 14MHz and at least one S-unit at 283MHz, of the author’s views on this topic; nevertheless,
though the main benefit from the exercise has been to applied to single-element antennas the losses need
widen the choice of operating frequencies. not be excessive and the author’s preference for
multiband dipoles may not be the best choice in all
circumstances. For example, in the case of the ideal
Sloping ground antenna system (B) earlier in this chapter, a trap-
This case has been largely covered elsewhere (pp138- ped dipole slung between the two towers might,
140) but some additional options suggest themselves; despite the need for a relatively heavy-wire gauge
the example which follows is based on an actual case to support the weight, be preferred to coping with
where the ground slope was 20° and the garden com- the loose ends of the inverted-V’s.
prised largely of about half-an-acre (0-2ha) of wood- —QD —— Antennas which are “‘high risk’’ in respect of feeder
land. With fixed collinear arrays a height of only 4/4 radiation. These include the ground plane (p43),
provided near-optimum performance in the down-slope end-fed arrangements other than those fed against
direction. In this case it would have been possible to a counterpoise or using a balancing stub (p46),
achieve a pair of log-periodic arrays stacked side by and any antenna fed with coaxial cable and without
side at a height of 15—20ft (4-5—6m) to give a low-angle balun. Note however that the term “high risk”’ is
gain of nearly 10dB over the entire band from 10— intended to be taken literally and implies also a
30MHz for the down-slope direction. Space could still chance that no ill effects will follow from omission
have been found for a two- or three-element Yagi for of the balun.
7MHz, though not for extending the log-periodic arrays (d) Various forms of X beams. These were analysed in
to include 7MHz. For coverage of directions other than detail in Chapter 5, and it may be recalled that when
down slope, a two-element beam was used at 4O0ft points of high current are brought together at the
(12m) and this cleared the top of the hill by about 30ft centre of an X the adjacent sections of conductor
(9m) to provide reliable short-path (up-slope) com- contribute relatively little to the radiation and
-munication with Australia, despite a performance dif- therefore radiation resistance is decreased.
ferential of at least two S-units for the two directions. Moreover, along the lines of the X the front/back
For coverage of all bands from 14~-28MHz there are ratio is unity and there is also a strong component
many alternatives to the log-periodic antennas, and if of radiation in what would be the endwise direc-
space is limited wire beams based on Fig 12.1 could be tions of a normal antenna (p81).
interleaved with two-band beams making use of Fig (e) Antennas with large built-in losses. This includes
12.6, relays being used for switching between beams the Windom (p45), short elements with resonant
and if necessary “killing” any interaction between feeders, antennas with large loading inductances
them. (unless there is no alternative, as may be the case
for mobile operation) and the W8JK, though in
this case, assuming heavy-gauge full-size elements
Ideas which have been avoided erected well in the clear, the main drawbacks are
(a) The “ZL Special’ and sundry other driven or the narrow bandwidth and lack of front/back ratio.
parasitic two-element arrays. As the reader will (f) The quad antenna. Despite the popularity of the
have appreciated from Chapter 5, efficient opera- quad its use as a rotary beam is strongly deprecated
tion of a two-element close-spaced beam depends for reasons which can be summed up by quoting
on achieving the correct phase and amplitude remarks made to the author during the writing of
balance. This is critically dependent on the mutual this book. For much of the time he was using a
coupling without which there could be no gain, fixed (reversible) quad and many of the stations
whereas most of the phasing methods described in contacted, after commenting on the excellence of
the literature rest on the implied assumption of no this type of antenna, added that they used to have
mutual coupling. Such systems are unlikely to one themselves but it blew down. This in some
provide much gain, 3dB being quoted in reference cases has reduced them to using an inferior type of
[2] as a typical figure for the ZL Special although antenna such as a typical commercial trapped beam.
this can be expected to do reasonably well in Rotary beams, moreover, fail to exploit the real
respect of front/back ratio (p77). The probability of virtues of the quad which have been put to good
efficient operation, though dependent on a number use earlier in this chapter and in Chapters 7, 11,
of coincidences, is nevertheless high enough to and 12.
account for two cases known to the author of (g) Slopers. These have some useful features but are
outstanding performance, which could however likely to be somewhat inferior as dx antennas to the
 WHAT KIND OF ANTENNA? 223

“bent dipole” arrays such as those featured in suitable, and it should be noted that it is not necessary to
Chapter 13, p190. switch power since safety interlocks can be devised to
prevent operation while rf power is applied.
Possibly the most important area for new develop-
Will it work? ment will be that of short active antennas for reception
The antennas described in these pages range from which, operating in conjunction with transmitting
simple systems well suited to the beginner to challenges arrays including (and perhaps especially) single ele-
for the ingenious and experienced, and, whereas some ment antennas, could lead to greatly improved rejec-
systems have benefited from the experience of numer- tion of unwanted stations. This could follow from
ous users Over many years, others with perhaps equal commercial developments already under way.
prospects are still at the beginning of this process. If a
design is new and not yet widely tested, it would be
unwise to attempt it without the ability to carry out The antenna and the planners
tests such as those described in Chapter 18, plus the Regulations governing the erection of antennas vary
possibility of returning the antenna to ground level for enormously from country to country, though for large
further adjustment if necessary after testing at full erections such as those considered at the beginning of
height. this chapter they are almost certain to be inescapable.
On the other hand, it might be expected that the data On the other hand, for some lesser structures the
available in various textbooks for such well-established position could well be rather obscure as indicated for
antennas as the quad would be reliable; yet frequently the UK by the following guidelines kindly supplied to
it turns out that reflector lengths are too long and other the author by Bob Price, G4BSO:
instances of the unreliability of data are reported from
time to time. The reasons for this are largely wrapped “Basically, in town & country planning we have to consider
in mystery and those purchasing commercial antennas whether the combined structure of antenna and support is
such that its erection is a ‘building operation’. Unfortunately
are perhaps even worse off since they enjoy no immun-
‘building’ has a very wide definition as ‘any structure or
ity and the remedies are less likely to be under their erection’, but we have some judicial interpretation which
own control. This will be best appreciated by home limits this definition as otherwise every piece of wire or cord
constructors who, after years of coping successfully may be a ‘building’. The judicial interpretation appears to
with their own mistakes and misfortunes, are suddenly indicate that there must be something sufficiently substantial
confronted with a faulty pcb in a commercial transceiv- or permanent in nature to affect the physical character of the
house and garden. (For the purpose of this book it is assumed
er! It is believed that the best answers to this problem that all antennas will be associated with a dwellinghouse,
are provided by antennas which can be tuned or at least otherwise some different considerations will apply).
subjected to a fine-tuning process from the shack or By letter (1969) to the RSGB, the Minister of Housing &
some other point at ground level. This includes most of Local Government agreed that an antenna system similar to a
the vertical antennas described in these pages, the domestic tv or fm broadcast antenna does not require planning
two-feeder versions of the VK2ABQO and inverted-V, permission because it does not materially affect the external
appearance of the building. Put another way, it is expected to
resonant feeder systems such as the quads described be found on most houses and is part of the general residential
above and the miniature beam of Fig 12.14. scene. It is doubtful whether we are entitled to use this
pronouncement where uhf broadcast antennas are in use but
the amateur antenna is more like the Band 1 type. This is just
What of the future? an example of the multitude of uncertainties with which we are
faced. Nevertheless the ‘local type’ of tv antenna would always
At the time of writing the impact of the 10, 18 and be a useful support for a wire antenna attached to a tree at the
24MHz bands on operating habits and antenna de- other end. One cannot imagine any planning authority
velopment is difficult to foresee, though it is believed it seriously suggesting that a long wire tied to the chimney at.one
may hasten the demise of the trap. In this context end and a tree at the other requires planning permission. It
multiband antenna design would be revolutionized by cannot be a ‘building’ even within the very wide definition
the development at an economic price of servo- indicated above.
Another Court decision on the definition of ‘building’ is that
controlled capacitors for remote tuning of beam ele- the subject must be built of constructed things of substantial
ments. Assuming linear resonators, a range of 8-SOpF size. It is thought that a trap or balun in a long wire or wire
would allow operation on any band from 14 to 28MHz, dipole could in no way be described as ‘of substantial size’.
including rapid beam reversal subject to suitable prog- It should be explained that there are vast areas of uncertain-
ramming. The same objective should however be ty simply because antennas have not been the subject of many
achievable using fixed capacitors switched by relays, and Court decisions, presumably because no amateur feels that it
is worth spending a considerable sum in pursuing an appeal
on present form this seems more probable. Relays through the Courts. The Planning Appeal Decisions of the
“designed for the job” of rf switching are likely to be Department of the Environment are of great interest and
heavy and expensive but reference has been made afford some guidance, but they do not set binding precedents in
(Chapter 7) to small light relays which appear to be the same way as do Appeal Court Decisions on points of law.
224 HF ANTENNAS FOR ALL LOCATIONS

Whether in any given circumstance planning permission is of getting the station closed down.”’ He points out also
required is a point of law; whether when planning permission that many “houses and flats now have covenants im-
is necessary it ought to be granted on the merits of the
posed on them by their deeds or tenancies which
application is not a point of law.”
absolutely prohibit the fixing of antennas of any kind to
Amateurs who feel that they may be operating in one the structure of the building or in the garden. Where
of these “areas of uncertainty” would be well advised such covenants exist we can do little about them
to cultivate good relations with neighbours, above all beyond checking that they are legally enforceable. The
avoiding tvi and planning their antenna systems with a amateur in such circumstances will have to be satisfied
view to minimum visual impact; try to avoid the skyline with an indoor antenna or put up something which is
or erecting anything which looks very unusual. From virtually invisible.” Invisible antennas are the subject
Chapter 11 onwards this has been featured as one of the of Chapter 15 and we return to this topic with further
important characteristics of antenna systems and the planning guidelines in Chapter 19, p250. .
“ideal” arrangements depicted in Figs 17.1 and 17.2 get
no marks under this heading. References
Bob Price comments further that in his experience:
“more people run into planning problems as a result of [1] The ARRL Antenna Book, 13th edn, ARRL, 1974,
tvi than for any other reason. This is because when the p160.
Post Office gives the amateur a ‘clean bill of health’, [2] The ARRL Antenna Book 12th edn, ARRL, 1970,
the neighbours search frantically for some other means p214.
CHAPTER 18

Making the antenna work

Ideally, it should be possible for an antenna to be made their personal requirements. This has included the
up according to instructions, erected, connected to the presentation of ideas, many of which are still in the
transmitter and guaranteed to work without imposing experimental stage, and such constructional data as
any further tasks on the operator. To achieve this, may be available does not have behind it the weight of
however, a number of conditions must be satisfied, as experience enjoyed by well-established designs such as
follows: the quad. Yet, even with this latter antenna poor
results are often obtained at the first attempt due to the
(a) The antenna must be designed to present a speci-
reflector being tuned too low.
fied load impedance to the transmitter.
It may be worth also recording the author’s personal
(b) The transmitter must be pre-tuned to work into the
experience with design data for Yagi antennas from two
above impedance, either by the operator using the
sources; when IBA tv transmissions first started from
appropriate ‘““dummy load”’, or by the manufacturer
London the author, being in a fringe area, decided to
who presets all the adjustments before it leaves the
try a double Yagi with five elements in each “‘bay”’.
factory on the basis of some commonly used impe-
There was no picture and (aurally) the vision signal was
dance, such as 50Q, to which the antenna must also
weak though the sound signal (lower in frequency) was
conform.
of usable strength. It was recalled that some weeks
(c) Differences between the environment assumed in
earlier it had been necessary to shorten the directors of
designing the antenna and that in which it is in-
a vertical wire Yagi for 14MHz by more than a foot
stalled must be negligible, ie any effect on antenna
(30cm), though at the time it was thought this might
impedance must be within the tolerances allowable
have been due to neglecting the effect of insulators. On
at the transmitter; eg an swr better than 1-05 as
the strength of this, armed with the requisite ladders
specified for the author’s FT75.
and a hacksaw, 0)-5in (13mm) was removed from all the
(d) The antenna must experience no _ significant
directors; returning to the tv set a perfect picture was
changes between the design or adjustment pro-
found on the screen.
cess and the final erection; this places a serious
In contrast to this experience, a three-element
handicap on anyone whose experimental efforts
14MHz Yagi designed from the data in Fig 5.17 worked
tend, like those of the author, to find themselves
perfectly. Nevertheless the author’s reluctance to
lying in the grass and repeatedly tripped over on
guarantee such information can perhaps’ be
their way from the tune-up position to the end of
appreciated; instead the reader may expect to be told
the mast. In the case of a commercial design the
how to adjust and check antennas so as to be assured
manufacturer, the carrier and the recipient may
that they are in fact working correctly.
take every precaution, and the purchaser’s confi-
dence may or may not be justified. When on the
other hand an antenna is built from published data, Simple antennas
even though this has been obtained from two or
In a few cases, such as the A/2 horizontal dipole,
more sources cross-checked with one another,
standard length formulae can be used with reasonable
there is no more than a fair chance that it will work
confidence, though (at the very least) some form of
well enough to be incapable of improvement; this
antenna current indicator is needed for tuning the
leaves a likelihood that adjustment will be needed
transmitter. The simple indicator shown in Fig 18.1 was
or desirable and the user may be unaware of this.
used in the author’s first portable rig operating with
In some cases, for example mobile antennas as we powers in the region of 0-25—1-0W, and it can of course
have just seen, final adjustments have to be carried out be adapted for higher powers by reducing the loop sizes
in situ; moreover this book has been addressed in and the meter sensitivity.
particular to experimenters looking for something Alternatively, an swr bridge (p232) can be used as the
which is either better than the antenna designs current- rf indicator and this is strongly advised since it provides
ly available “‘off the shelf’, or more closely tailored to other important functions. Assuming the use of coaxial

225
226 HF ANTENNAS FOR ALL LOCATIONS

Antenna current cent. The swr meter should be placed at the antenna,
partly because this is the centre of operations but also
because the swr will be reduced by any losses in the
feeder and readings may also be affected by currents
induced in the outer conductor of the line, a situation
which is more difficult to avoid with the antenna near
ground.
Even a simple antenna can acquire faults which are
not immediately obvious. For example, suppose the
wire used for a portable antenna, being thin and
subjected to a lot of flexing, breaks in the course of
erection; a twisted join allows operation to proceed,
but may then be forgotten until after corrosion has set
Fig 18.1. Simple current indicator. A loop of 1in (2‘5cm) diameter in. If a balun is used it should be tested before use and
has a reactance of only 4Q2 at 14MHz and will have negligible particular care taken to ensure that it is connected
effect on matching
correctly. Most such faults can be avoided by inspection
feeder, it affords the simplest indication of correct in the case of a portable antenna (Chapter 16) but the
matching and is a valuable tool for correcting a mis- coaxial or 300Q twin feeder of a permanent installation
match should this be necessary. Regardless of the type can deteriorate without this being obvious. Breakdown
of antenna or feeder system, it allows an atu or other can occur in a balun, though this is unusual when
matching device to be adjusted so that a sufficiently working with low impedances; on the other hand,
good match is presented to the transmitter. It can also operation may be perfectly satisfactory without a
be roughly calibrated by the method described on p235 balun—until the feeder length is changed, due perhaps
as a power meter for confirming that the transmitter is to a rearrangement of the shack. Assuming reasonable
giving its rated output, whereas the simple current precautions such as careful attention to symmetry,
meter establishes merely that the transmitter is giving weatherproofing of any vulnerable components and
whatever maximum output it happens to be capable of observance of any necessary precautions against feeder
producing at that moment. In principle the simple radiation (p49), there is not much else to go wrong,
indicator can also be calibrated in terms of rf power, though an “‘invisible”’ (thin wire) antenna could stretch
but only if the antenna impedance is known and can be under severe tension, eg under storm conditions if
assumed not to vary. anchored to a swaying tree without due safeguards;
Antennas can usually be tuned satisfactorily at a and, if there is more than one antenna at a given
convenient height of 6ft (2m) or so, allowing for an location, each should be suspected of upsetting the
upward shift in frequency of 1-13 per cent when they other “until proved innocent”’ or unless adequate pre-
are raised to their full height. In the case of one- cautions have been taken (p229).
element antennas, correct operation can usually be
secured by first tuning them approximately to reso-
nance with a grid-dip oscillator (gdo); after this the
The adjustment and testing of
feeder should be connected and any variables in the two-element beam antennas
matching system adjusted simultaneously with the tun- The findings reported in Chapter 5 cast doubt on
ing of the element so as to obtain minimum swr. traditional methods for the adjustment of two-element
When making observations on a driven element with beams since the main objective now includes equaliza-
a gdo, the feeder must always be disconnected since tion of current amplitudes by adjustment of coupling,
otherwise what is being measured is the resonance, not which has to be achieved simultaneously with correct
of the antenna itself, but of a complex system consisting phasing, though in practice it may be sufficient to fix
of antenna plus feeder plus whatever unknown impe- either the tuning or the coupling and vary the other.
dance is presented to, the feeder by the transmitter or These adjustments rigorously determine the radiation
atu. In the case of a centre-fed antenna correct reso- pattern and gain; conversely, observation of the radia-
nance will be obtained with the feeder short-circuited, tion pattern can be used as the test of whether the
but with a T- or delta match the feeder must be amplitudes and phases are correct, the design objective
physically removed and this will result in a slight shift in being based on Figs 5.4 and 5.6.
the effective resonance of the system. Reference to pp72-75 will establish whether a parasi-
Loops or V-shaped elements may be tuned in a tic element should be a director or reflector, and the
horizontal position. It is important in the case of a V beam should be tuned for this mode of operation in the
that the ends should not be allowed to trail in such a first instance, whether or not driven operation is in-
way that they are much closer to the ground than its tended eventually. The parasitic element is then tuned
centre since this can lower the resonance by several per for minimum response on a signal in the selected null
 MAKING THE ANTENNA WORK 227

rene least a Likely source of

errors
just described, though provided both elements are
!
nearly correct it is possible to drive them both as
Test
antenna
described on p77. In this case the swr meter should be
placed in the feeder handling most of the power, ie the
forward element if the mutual coupling is capacitive or
oS
the rear element if the coupling is inductive as ex-
Tune-up height
plained on p73. The other feeder has relatively little
lakteecesalia}
work to do, whether it is being used for remote tuning
of a parasitic element or as part of a two-driven-
element system, and identical results should be obtain-
Fig 18.2. Use of field-strength meter. A test antenna having the able.
same polarization as the beam under test incorporates a diode At this point it may be recalled that with straight A/2
detector with a meter for indicating the rectified current. The dipole elements spaced 4/8 the mutual coupling is
meter is wired back to a convenient position adjacent to the
antenna under test. A separation of 1) should suffice for providing resistive and both feeders of a driven system handle the
an accurate indication of the null directions, but care must be same amount of power, so that the above argument
taken to avoid reflections from surrounding objects such as might appear to be invalidated. Note, however, that
another beam as shown on the right
this case has been excluded by insisting on correct
direction, eg 140° relative to the beam heading as operation as a parasitic array before the second feeder
recommended on p130. The use if possible of a field- is connected; because the coupling is resistive the
strength meter in accordance with Fig 18.2 is recom- currents cannot be equal and to make them equal
mended for this operation but requires a number of reactance must be introduced. This in itself provides
precautions as described later in this chapter. more or less the required phase shift so that, assuming
The relative currents in the elements should be the correct choice has been made of reflector or
checked as described on p237 and, if they differ by more director operation (reflector in the case of quad), one
than 5-10 per cent, steps taken to equalize them by feeder becomes “‘idle” and dissipates comparatively
varying the spacing between the ends of dipoles or, in little power whether connected to the transmitter or
the case of the miniature beams described on p176, not.
adjusting the neutralization. After any adjustment of It might be thought simpler to retain straight ele-
spacing or neutralizing, the reflector tuning should be ments with \/8 spacing, so that the coupling is resistive,
rechecked and the whole process repeated if necessary. and match the two driven elements individually. This is
The driven element should then be matched as de- the apparent basis of several published antenna de-
scribed above for a single element. signs, except that in most cases matching appears to
The reflector is likely now to be identical with the have been assumed despite the lack of any provision for
driven element; it should be made so and given its own achieving it. Individual matching (as explained in
separate feeder which can be used for remote fine Chapter 5) is in fact quite a difficult process, largely
tuning, the two feeders being interchangeable for beam because it has to take into account the very different
reversal. The advantages of this have been pointed out reactances transferred from the leading element into
in earlier chapters, driven or parasitic operation being the lagging element and vice versa. This can in principle
optional. Among other advantages, beam reversal pro- be compensated by means of matching devices at the
vides an instantaneous audible check of correct opera- antenna but it then becomes impossible to reverse the
tion, though possibilities of asymmetry can arise and beam efficiently by any means available in the shack. A
the check needs to be made for both directions of fire. compromise is possible by resonating each of the
An antenna can often work well despite detuning of the elements individually with a gdo in the absence of the
driven element, so that if a two-element beam is other, and accepting a mismatch to each separate
working well to the west with a sharp null to the east feeder of the order of 2-5 to 3-0 swr for both directions
this is no proof that there will be gain to the east if the of fire. The transmitter will however “‘see”’ a resistive
beam is reversed by an interchange of feeders. It is match in both cases, the reactances contributed by the
even possible there could still be gain to the west and a individual feeders being equal and opposite.
minimum to the east unless the two elements with their It may be that the reader has not been impressed by
feeders have perfect symmetry. the advantages claimed earlier for instantaneous beam
The reason for first tuning one of the elements as a reversal, in which case the adjustment procedure de-
reflector may seem obscure, since eventually both are scribed above can be halted after the reflector has been
required to be identical and matched to the feeder. The tuned and, if possible, the currents equalized. The
matching process must however take due account of the resulting improvement in front/back ratio will however
impedance transferred into the driven element from the be realizable over only a small part of the band, in the
other element when this has the correct current flowing absence of the remote tuning made possible by provi-
in it. The easiest method of establishing this is the one sion of a second feeder.
228 HF ANTENNAS FOR ALL LOCATIONS

10022 Fig 18.3. Measurement of feeder loss. In

B case (a) the transmitter drive is set to give
Cable under test
'
' an output level of about 5-15W, and the
dummy load on the left incorporates an
ammeter which can be as shown in Fig
18.1; alternatively an swr meter may be
(9) used as the indicator. The power or current
(b) is observed using first a short lead between
B A and B, which is then replaced by the
1

--->
Cable under test! feeder under test, the loss being observable
directly since there should be no change in
the loading of the transmitter. (b) shows a
more rigorous method using two meters
and allowing the use of any load resistance.
Disregarding losses, the placing of a deep null in the Instead of replacing the cable the two
chosen direction results in a gain which is given precise- meters may be interchanged, or a single
ly by Fig 5.4 since the directional pattern depends only meter may be used, moving it from one end
to the other of the line. Many variations of
on the relative phases and amplitudes of the currents, these methods are possible. For temporary
and it is the phase difference alone which fixes the connections there is no need to use plugs
direction of the null. The depth of nulls depends only and sockets, ordinary plastic strip mains
connectors being suitable if leads are kept
on the current ratio; it is infinite if the currents are short
equal and in typical cases the gain is not seriously
affected until the null depth drops to less than 12dB. effects of ageing or ingress of moisture and the
The direction of nulls may be translated into the feeder will have to be changed.
equivalent phase difference by means of Fig 5.6. (c) Losses or dispersion of energy in unwanted direc-
It is best to aim for over- rather than undercoupling, tions due to feeder radiation (p49). The current
since this improves the “‘swr bandwidth’. In the case of probe (p237) should be used to check for current
a two-feeder system a simple test for overcoupling can on the outer of a coaxial feeder or on the centre
be made after erection by adding resistance to the line of a twin feeder but this is not necessarily
reflector; if it is found possible in this way to increase conclusive. With the antenna at tune-up height the
the depth of nulls, this indicates overcoupling. If the power flow should be checked at both ends of the
improvement is large, eg from 10dB to 20dB, this feeder and the difference should not exceed the
suggests severe overcoupling with possibly some loss measured in the previous test. This is still not
adverse effect on gain as well as directivity. conclusive either way, but if there is excessive loss
one should look for some (usually obvious) cause
Losses, their detection and prevention of asymmetry. If the swr is much better at the
transmitter than at the antenna this is cause for
As a preliminary to the determination of losses it is suspicion. Try a deliberate mismatch at the
helpful to start by obtaining the lowest possible value of
antenna; if there is little change in swr at the
swr as described later (p242) since otherwise the
transmitter, be even more suspicious. Coaxial
losses may be aggravated or meter readings made more
feeder used for this test must be the permanent
difficult to interpret. Losses can occur in several ways,
one, not shortened, and it should be brought away
as follows:
at right-angles to the elements but, if using open-
(a) Ordinary resistance (I?R) losses. Loss resistance wire line, there may be difficulties in achieving this
and radiation resistance are both easily calculable with the antenna at low height. If the beam is
or read from charts (p183), and relative currents satisfactory with coaxial feeder, open-wire line can
can be observed at tune-up height with a current be substituted prior to erection and will usually
probe (p237), so that provided there are no faulty result in lower losses.
joins the losses in each part of the antenna can be (d) Losses, or dispersion of energy in unwanted direc-
readily estimated and added together. Any sus- tions, due to radiation from a metal boom or even
pected joint can be checked by short-circuiting. An the mast. Radiation from the boom will show up as
example of this procedure will be found on p238. poor front-to-side ratio for at least one of the
(b) Feeder loss. This can be checked using a dummy endwise directions, and also as asymmetry of the
load and an swr meter, or any form of rf current main lobe. Make sure that maximum response
meter which can be roughly calibrated. Of these coincides with pointing the antenna in the wanted
indicators, the swr meter is preferred since it direction. If the effect is present or suspected look
reveals any mismatch between the feeder and load, for some cause of asymmetry; also test with the
and allows it to be corrected. A simple measure- probe for current flowing in the boom. Radiation
ment of feeder insertion loss can then be made as from the horizontal (or vertical if the antenna is
shown in Fig 18.3. If the loss exceeds the specifica- vertical) portion of a feeder can also cause asym-
tion for the feeder this is almost certainly due to metry, and there could be vertically polarized
 MAKING THE ANTENNA WORK 229

radiation from the mast or vertical portion of be placed at a sufficient distance, the meter leads being
feeder. This will cause asymmetry if the measuring run to the antenna position as described on p227. In
antenna also has a vertical component but other- general the reflector should be tuned for minimum
wise may not reveal itself, so that a thorough check signal “off the back’, and directors for maximum
should perhaps also include use of a vertically forward gain, repeating the process as necessary. If
polarized meter. field-strength measurement is not practicable it may be
The contingency is however fairly remote if possible to arrange for a steady ground-wave signal, eg
adequate care has been taken to prevent feeder from another local station, which must be using the
radiation and the mast is either non-resonant or same polarization as the antenna under test. This condi-
fully symmetrical with respect to the antenna. On tion applies equally to test antennas used for field-
one occasion the elimination of boom current strength measurements.
resulted in a symmetrical main lobe but strong If the antenna has been purchased or carefully con-
radiation off the ends was still registered by structed in accordance with Figs 5.16, 5.17, observa-
the field-strength meter. This was traced to the tions at tune-up height should be used as a rough check
antenna being at low height and the support not only and disregarded unless badly out. If the antenna
quite straight so that (in the end-on position only) appears to be tuned low by some 1-2 per cent this is
one end of the elements was closer to the ground! normal and should correct itself, more or less, when the
Asymmetry is less likely to occur if the centre of antenna is raised to 30—SO0ft (9-15m). If the antenna is
the elements is at earth potential as in monoband unsatisfactory when tested in situ a careful record
or trapped-dipole arrays, and is reduced if the should be made of directional patterns, both at the ends
elements are insulated from the boom. Owing to and centre of each band, and rechecked after the
the residual capacitance this is not always a com- antenna has been returned to a tune-up position. This
plete cure; in such cases a symmetrical feed, eg a will enable correct allowances to be made for the height
delta or T-match (not a gamma match) is essential. difference in the course of any subsequent adjustments.
(e) Screening by other antennas or overhead wires;
refer to discussion on p235. Useful antenna
measurements can be impossible unless other
antennas in the vicinity are (or can be made) The ultimate test
end-on, open-circuited (all elements), or lowered. The ultimate criterion of performance in the case of
A quad loop open-circuited for horizontal polariza- antennas, as of most other devices, is how well they do
tion still responds to vertical polarization and vice their job. Lacking sophisticated and usually expensive
versa. One beam 100ft (30m) in front of another test equipment this is often the only test method open
can screen to the extent of one to three S-units. to the user, and applies especially to the assessment of
When working with vertical antennas any nearby antenna gain. This, as mentioned earlier and further
masts, feeders, guy wires or other vertical conduc- explained on p236, is virtually impossible to measure in
tors should be checked by gdo (p231) to ensure the present context.
that there are no resonances close to the frequency There are right and wrong methods of assessing the
of interest. If there are, and it is not possible to fit ‘Sob performance” of an antenna and a number of
insulators, the linear resonator trick (p163) may be possible pitfalls. Comparisons with other stations on
tried. the basis of dx signal reports are usually satisfactory
when the other stations are within ground-wave range,
and often in the case of much greater separations,
Tuning and checking beams with three or particularly if large numbers of reports are averaged
more elements over a considerable period. It is desirable if possible to
In this case there is no simple way of relating gain to the find as a yardstick some other station where the
directivity pattern, though a reasonable estimate conditions are maintained reasonably constant, and
should be obtainable by plotting the directivity pattern long-range chordal-hop paths such as UK—Australia or
in the horizontal plane and for Yagi arrays dividing this South Africa are much more useful than, say, trans-
by cos 8 to obtain a ‘“‘guesstimate” of the vertical atlantic paths which are subject to multimode propaga-
pattern. Noting the two 3dB beamwidths o,, a, and tion. Due to interference between, say, two and three-
expressing them in radians (ie degrees + 57), the gain hop modes, transatlantic paths also exhibit consider-
in decibels is given by able fading which is not synchronized between different
stations, even when these are quite close to each other.
Guia (1log aoe 2-1 In checking a new antenna against its predecessor a
number of comparative signal reports averaged over
periods of a week or more are desirable and, of course,
Adjustment is difficult unless a field-strength meter can the more the better. If reports are consistently ‘““down’”’
230 HF ANTENNAS FOR ALL LOCATIONS

compared with several other stations using similar current in a feeder if the other element is energized.
powers, antenna heights, antenna gains and equipment Vertically polarized dipole or monopole antennas
this is legitimate cause for concern, though it may be should not interact with horizontal antennas unless very
due to some environmental factor (Chapter 10). Tests close, but complete loops have circular symmetry and
should be made in more than one direction before can respond to any polarization.
suspecting a fault in the antenna or other equipment. If disabling or rotating one antenna has no effect on
Allowance can be made if necessary for differences in the other as judged from signal reports, preferably with
power, though this should be equalized if possible—it the help of another station acting as yardstick so that
can be reckoned as a guide (for flat sites) that dx field instantaneous comparisons are possible, each antenna
strength should be proportional to antenna height; this may be used as a yardstick for the other, and usually
can however be invalidated by environmental factors either of them can be replaced for investigation of a
as explained in Chapter 10. third. For example, one of the first two antennas could
Some descriptions of antennas include gain estimates be a reference dipole such as an inverted-V, requiring
based on signal reports, assuming for example 6dB per only a light, temporary support and the third antenna
S-point as formally defined, though in practice the might be a new beam which it is required to evaluate
average bears little resemblance to this, being closer to against its predecessor.
3 or 4dB—some commercial S-meters are indeed In using this procedure some caution is needed
known to be calibrated on a 4dB basis. The attempt to insofar as a dipole may “‘see”’ reflections that are not
standardize on 6dB per S-unit is nevertheless achieving seen by the beam, and it would certainly be preferable
some success, and this in the author’s personal opinion to use a beam for the reference antenna. However, the
is unfortunate since 6dB is a very big step, whereas author has found comparisons using dipoles are useful
differences as small as 12 to 2dB can be reliably provided all antennas are at the same height and the
perceived to the extent that most stations asked, while reference is not moved. It is however essential for all
reporting little difference, think that the first (or results to be checked against ‘common sense’’ and if
second) antenna is slightly better. Unfortunately, due there is any doubt the test set-up should be the prime
to the current escalation of the S-point there is no suspect. If height is varied comparisons between a beam
longer an international language adequate for such and a dipole will not stay the same, if only due to the
reporting, although the ‘“‘half S-point’’ is widely used changes in radiation resistance of the dipole as depicted
and seems to be understood by most operators at least in Fig 6.17, any corresponding variations in the case of
in the form “‘your signals are Q5, S5 to 6”. In the event a beam being much less.
of a reported difference of say one S-unit, one simple Comparisons are greatly facilitated if it is possible to
way to determine the number of decibels is to switch switch rapidly between antennas, and usually more or
out the linear amplifier. The effect will probably be less any standard-sized rotary switch, or a relay with
reported as a drop of two S-units and this can be short leads, adequate contact area and good insulation,
compared with the known power difference which is can be pressed into service, any mismatch being com-
likely to be in the region of 6dB. pensated by an atu (p54). At least one atu is necessary
If a second antenna is available as a yardstick this is a unless both antennas are very accurately matched; this is
useful situation provided interaction between them is because any change in antenna impedance ‘“‘seen” by
not allowed to influence the results. Usually if antennas the transmitter and not corrected by retuning will result
are spaced by at least 3/4, at roughly the same height in a difference in radiated power—and possibly ‘‘splat-
and at least one of them is end-on to the other, there ter’! It will be recalled that a 2:1 limit on swr can
should not be much problem, but if one is in front of involve as much as a four to one change in impedance,
the other a separation of several wavelengths may be and this is a very large difference in loading of the
needed. If resonant feeders are used there is no difficul- transmitter. |
ty in detuning the antenna (or element) by suitably It is often advised that tests should be made on a
placed short-circuits (ideally at voltage points on the “steady” carrier, not a voice signal, but the author
lines) and the ends of an inverted-V can usually be dissents from this provided the speech is held at a fairly
slackened off so that the wires drop down vertically. constant level by an efficient rfprocessing system. This is
Single-element antennas can in principle be detuned because fading is largely selective, ie it does not take
via their feedline, and it may be sufficient merely to place simultaneously at all frequencies in the passband,
open- or short-circuit this, but with appreciable line loss so that useful averaging tends to occur if the whole
there could be a problem. Parasitic elements of rotary passband is used. It is worth noting incidentally that
beams are not usually accessible for detuning but the S-meter readings are the result of an averaging process
beams can be rotated to an end-on position. In the case which is different from that which takes place in the
of the two-element two-feeder designs recommended in ear, and therefore different for processed and unpro-
these pages it should also be possible to detune each cessed signals. This may need to be taken into account
element by finding conditions which result in little or no when assessing signal reports relative to other stations.
 MAKING THE ANTENNA WORK 231

L1 less in diameter and can hardly be expected to couple to

any appreciable extent into, say, a 33ft (10m) length of
Cla C1b
lin (2-5cm) tubing or a 133ft (40m) long wire.
}100p 100p | Fortunately there is a simple and very effective way
out of this difficulty as shown in Fig 18.5. In the case of
a 4
~
the gdo described in reference [2] a single turn right-
angled triangular loop with an 8in (20cm) base was used
for 14-30MHz. Two turns the same size were used for
7-15MHz and four turns for 3-5-7-SMHz. Depending
on the value of tuning capacitors, the range covered
may be Slightly less on the lower bands owing to
the self-capacitance of the coil. There is no difficulty
in extending the range except that this tends to make
the scale rather cramped at the hf end.
For the gdo of Fig 18.4 a larger coil is required; two
turns with a 12in (30cm) base should be suitable for
14-30MHz but there would be no difficulty in adapting
Fig 18.4. FET source-dipper suitable for use from 1:5 to 50MHz
the design to suit any twin variable capacitor of 40-
(ARRL Radio Amateur’s Handbook) 200pF capacitance that happens to be lying around in
the junk box. The smaller values, however, provide
increased coupling and some transistors may not oscil-
Test equipment late with the larger values. Users of existing gdos have
It has been shown that adjustment and checking of found no difficulty in finding the correct size of loop,
antennas can be carried out with very little equipment. but as a guide a single turn as shown in Fig 18.5 is
One could manage with even less, except that this roughly equivalent to six turns spaced out to occupy a
would be a foolish economy in view of the advantage to gin (22mm) length on a former of gin diameter. The
be derived from the acquisition of an swr bridge and a author is indebted to G3JKF for pointing out that
gdo. These items even if purchased involve compara- sensitivity can be further improved by using the ear
tively little outlay, and the remainder of the essential rather than a meter as the indicator; this requires the
instrumentation consists of little more than an assort- addition of a modulator and audio amplifier.
ment of diode detectors and at least one sensitive It might be thought that loops as illustrated could be
meter. As we shall now see, there are few constructional easily distorted, thus upsetting the calibration, but
problems, though measurements are beset by a varied considerable mishandling is possible if frequency errors
assortment of problems and pitfalls to which the reader of 1 or 2 per cent are acceptable. In general, even if the
needs to be fully alerted. loop is squashed and then pulled out again so that it
“looks right’’, the calibration will usually be found to
have remained ‘“‘spot on’, or near enough, but the use
The grid-dip oscillator (gdo) of heavy-gauge hard-drawn copper or even brass strip
This is something of a misnomer in view of the excellent should be satisfactory if greater precision is desired.
and much more portable transistor versions [1] such as Other difficulties arise in applying the gdo to antenna
Fig 18.4. The principle is very simple; when the oscilla-
tor is tuned to the same frequency as the circuit under
3 i bare 6 to 30in Sue neem
test, some of its power is absorbed in the test circuit so : (15 to 75cm)
that the voltage across L1 is reduced and this reduces
the gate current or, in the case of a valve oscillator, the
grid current. In the case of a bipolar transistor oscillator
the emitter rf voltage may be applied through a capaci-
Stiff wire
tive potential divider to a rectifier and dc amplifier [2]. One or more turns
The resonant frequency of any tuned circuit L1, Cl bound with plastic
tape
may therefore be found by varying the tuning of the
gdo until a dip is observed on the meter, and an
antenna, being a form of tuned circuit, might be
expected to have a similar effect on the gdo. According Fig 18.5. Large-area gdo coil for antenna measurements. A single
to several of the author’s informants this does not 8in (20cm) turn was used to cover 14-30MHz with a gdo based on
happen, thereby leaving them with something of a reference [2], but two turns 12in (30cm) wide would be required
for the capacitance value shown in Fig 18.4. The coil shape is not
problem. The reason for this is not far to seek, since the critical except that it must have one long flat side, and the design
coils normally specified for use in a gdo are only lin or can be readily adapted to suit any gdo
232 HF ANTENNAS FOR ALL LOCATIONS

i
fee flow (spurious modes)
measurements as a result of (a) the large number of
different ways a wave can wrap itself round a metal
structure and (b) multiple resonances due to interaction

i
between elements, particularly in the case of multiband
beams. In checking the resonant frequency of a beam
element other elements should be removed or detuned
in some way, eg by the temporary attachment of an
extra 3 or 4ft (1m) of wire to each end. Such attach-
ments should be varied in length, checking that this has Current flow
(correct mode)
no effect on the resonance under observation which in
this case can be taken as almost certainly genuine.
Fig 18.6(a) shows two unorthodox ways in which a
wave can wrap itself around a “plumber’s delight”
beam structure; either of these spurious modes should Ca) (b)
be identifiable by coupling the gdo into the boom and
Fig 18.6. (a) Spurious resonances detected by gdo; the beam is of
they are not excited in practice provided the feed “‘plumber’s delight” construction, ie the elements are electrically
system is symmetrical. Usually the correct mode can be bonded to the boom and the boom to the mast. Note the two
identified because its frequency is known approximate- “unauthorized” ways in which a wave can wrap itself round the
structure; a gdo at A responds to all three modes but at B, with
ly and it is the only one near enough to the observed careful positioning, the unwanted modes should cancei. The
frequency to be plausible. If the elements include elements can be either horizontal or vertical. (b) shows a two-
resonators the situation can become quite complicated, band dipole for which there are two wanted modes (solid arrows)
and two slightly different unwanted modes (dotted); for clarity
though usually still capable of being resolved by com- the angle between the wires is exaggerated
mon sense. Due to capacitance across the insulation,
resonances which involve the boom may be only stray capacitance including the self-capacitance of the
shifted, not removed, by insulating the element from coil.
the boom.
Fig 18.6(b) illustrates how spurious resonances, not
relevant to practical operation but liable to show up The swr meter
with the gdo, can occur in the case of a multiband There are two types of swr meter, the reflectometer
dipole. The use of traps can lead to further problems in shown in Fig 18.7 which is highly frequency-dependent,
the interpretation of gdo measurements. In the case of and the “reflected power meter” (Fig 18.8) which is
a driven element many resonances will be observed if a independent of frequency and can be designed to read
long feeder is connected and (as pointed out earlier) forward power subject to correct termination. In Fig
this must be removed or short-circuited, whichever is 18.7 currents flowing in the inner conductor of a
most appropriate. transmission line induce waves in the adjacent lines but
Provided these limitations are understood and
appreciated the gdo is an extremely useful tool for the
adjustment of antenna systems, although (as already
explained) completion of the matching process usually
requires the use of an swr meter.
The gdo provides a simple means for the measure-
ment of inductance or capacitance, the unknown L or C
being connected across a known C or L, the resonant
frequency observed and the “unknown”’ calculated
from the usual resonance formula or appropriate chart.
For most purposes the “known C”’ can be any close-
tolerance capacitor of suitable value, and for a “known
L” the author finds it convenient to remember the
inductance formula for single-layer windings:

Lun) = Fn2d
Fig 18.7. Reflectometer; currents flowing in the centre conductor
of a transmission line induce currents which flow in opposite
where d is the diameter in inches, nm the number of turns directions in the adjacent lines. The instrument is encased in a
and the shape-factor F is 0-017 (or roughly 1/60) for a metal box which provides the outer conductor of the central line;
1:1 ratio of length to diameter. It should be noted to avoid an impedance discontinuity when the instrument is
inserted in 50Q line the central conductor may be a strip about
however that in the case of very small inductances or lin (25cm) wide mounted about jin (6mm) above the base of the
capacitances, errors arise due to lead inductance and box
 MAKING THE ANTENNA WORK 233

in the opposite direction, this being a fundamental From transmitter

I- ————_-
To antenna
property of coupled lines. Thus the forward wave
results in a wave travelling in the lower line cd towards
the terminating resistor R, where it is absorbed so that
no power is reflected back to the detector, whereas the
backward wave sends a signal in the direction from c to
d where it registers on the meter Mg. Similarly the
forward wave, not the backward wave, registers on My,
subject in both cases to suitable termination of the
auxiliary line.
The majority of swr meters currently in use are of
Fig 18.8. Reflected power meter, illustrating principles as ex-
this type and incorporate sensitivity controls, a single plained in text. Ideally two meters are used to provide simul-
meter being used in many cases with a switch to read taneous observation of forward and reflected power. L, is the
forward or backward power. A full analysis of the total inductance of the winding CD and M is the mutual induct-
ance between the windings
. Operation is quite complicated but, bearing in mind that
a current Jin any conductor induces a voltage jaM/ ina meter is V,; — V, for one position of the switch and
second conductor where M is the mutual inductance V, + V, for the other, ie zero and 2V, respectively if
between them, one might expect the sensitivity to be the bridge is balanced. For a wave travelling in the
proportional to frequency and this is about how it opposite direction the readings are reversed and the
works out. Apart from this drawback readings can be bridge should be perfectly symmetrical with respect to
badly upset if “‘antenna current” is allowed to flow in forward and backward power. The inductance of the
the transmission line, ie errors can be expected under transformer primary is small enough for any voltage
the same conditions which result in feeder radiation. To drop across it to be ignored, so that A and B are at the
check for this an extra length, say 4/8, of feeder may be same rf potential.
inserted on the antenna side of the swr bridge; if there is The potential divider can be resistive, capacitive or
no change in swr reading there is probably nothing to inductive and, although a resistive divider as shown
worry about. might be expected to absorb some power, this amounts
Many readers will already possess meters of this typically to only 1 per cent (0-04dB) in the case illus-
kind. Despite the above defects, plus a tendency to be trated. The reactance of the secondary winding must be
extremely inaccurate for swr greater than 2 or 3, they large compared with R to ensure correct phasing, the
are usually adequate for their main purpose and can be voltage induced in the secondary being given by jwmM/,
left permanently in circuit. Calibrated by methods and the secondary current by jwM//(R + jwL,) so that
described later, they can also be used as sub-standard the voltage across R is equal to IRM/L, provided R is
power meters, remembering that a given calibration small enough compared with wL,. The phase shift is
holds for one frequency only. They are inexpensive to given in radians by R/wL, and, for a “‘residual’’ swr
buy .and easy to construct; in contrast the swr bridges reading of 1:05 at 3-5SMHz, values of R = 10Q,
advertised as power meters can be quite expensive but L, = 90H would be suitable. At a frequency somewhat
have the big advantage of being independent of fre- higher than 30MHz accuracy will be affected by stray
quency. For the constructor they are equally inexpen- capacitance, and it is important for such bridges to be
sive and even easier to build since the slightly esoteric well screened against the influence of possible stray rf
process of achieving correct dimensions and proper fields.
matching for the secondary lines is replaced by a Fig 18.9 is a practical design taken from reference
straightforward bridge-balancing operation; in view of JE
this the reflectometer will not be described in further The circuit uses a current transformer in which the
detail. low resistance at the secondary is split into two equal
The principle of the power meter is illustrated by Fig parts. The centre connection is taken to the voltage
18.8. A reading of ‘zero reflected power’’ indicates sampling network so that the sum and difference
balance between the voltages V, and V, which are voltages are available at the ends of the transformer
derived as follows. V is the voltage across the line but secondary winding.
except at very low power levels it is too large for the Layout of the sampling circuit is fairly critical. The
present purpose, and a potential divider is used to input and output sockets should be a few inches apart
produce the smaller voltage V, which is required to and connected together with a short length of coaxial
equal the voltage developed across R as a result of the cable. The coaxial outer must be earthed at one end
current J; or Jp flowing through the primary of the only so that it acts as an electrostatic screen between
transformer. Depending on which way round the the primary and secondary windings of the toroidal
secondary winding is connected, V, will be in or out of transformer. The layout of the sensing circuits in a
phase with V, so that if J = 0 the voltage across the similar instrument is shown in the photograph.
234 HF ANTENNAS FOR ALL LOCATIONS

Short length of y= 12 turns of 24swg wire on

Transmitter coaxial cable Van Mullard FX1596 core Antenna

eS —
Sy
Fig 18.9. Circuit of frequency-
independent directional watt-
meter with four ranges corres-
ponding to fsd of 0-5, 5, 50, and
500W in 50Q lines. The value of
R2 (including RV1 if fitted)
should be 220. For 75Q sys-
tems R2 is 150 and the calibra-
tion is different. The outer of the
coaxial cable acts as an elec-
trostatic screen between the
centre conductor and the secon-
dary winding of the _ trans-
former; the cable length is unim-
portant (Radio Communication
Handbook)

The primary of the toroidal transformer is formed by Other components in the sampling circuits should
simply threading a ferrite ring on to the coaxial cable. have the shortest possible leads. R1 and R2 should be
Twelve turns of 24swg (0-6mm) enamelled wire equally non-inductive carbon types. For powers above about
spaced around the entire circumference of the ring 100W, R1 can consist of several 2W carbon resistors in
form the secondary winding. The ferrite material parallel. RV1 should be a miniature skeleton potentio-
should maintain a high permeability over the frequency meter to keep stray reactance to a minimum. The
range to be used; a suitable ferrite ring is the Mullard detector diodes D1 and D2 should be matched point-
FX1596. contact types with a piv rating of about 50V. OA79 and
OA91 diodes are suitable. The current transformer
resistors should be matched to five per cent.
The ratio of the sampling resistors R1 and R2 is
determined by the sensitivity of the current sensing
circuit. As the two sampling voltages must be equal in
magnitude under matched conditions, RV1 provides a
fine adjustment of the ratio.
Accurate calibration requires a transmitter and an rf
voltmeter. The wattmeter is calibrated by feeding
power through the meter into a dummy load of correct
impedance. RV1 is adjusted for minimum reflected
power indication and the power scale calibrated accord- —
ing to the rf voltage appearing across the load. The
reflected power meter is calibrated by reversing the
connections to the coaxial line.
This instrument has full-scale deflections of 0-5, 5, 50
and 500W selected by the range switches. These should
not be ganged since the reflected power will normally
be much less than the forward power.
Germanium diodes as specified are essential if an
instrument is to be used at low power levels, otherwise
silicon diodes may be substituted. If an rf ammeter is
available it may be used instead of a voltmeter, except
that its range may be rather more restricted.
The main problem area is likely to be the dummy
load since accurate loads suitable for high power levels
are quite expensive, but it is a simple matter for
View of the sensing circuits of the fetrecuhaseandent calibration purposes to make use of the antenna itself.
directional wattmeter First a low-power load must be constructed using, say,
 MAKING THE ANTENNA WORK 235

five selected 27092 1W carbon resistors in parallel, system under test. For horizontal antennas at a low
keeping the connections as short as possible; this is height above flat ground field strength tends to be
used for setting up the power meter at the 5W level. proportional to distance squared, so that with two
The antenna may then be connected via a matching unit elements spaced i/8 and adjusted to give a null in a
which is adjusted to give unity swr, and the antenna plus particular direction, a field-strength meter at a distance
matching unit can thereafter be used as a matched load of 4 as in Fig 18.2 would indicate a minimum signal
for any required power level. level 1 — (%)* of the signal from the front element
As an alternative to using a voltmeter or ammeter a alone.
simple and accurate method of calibration is possible This would be interpreted as a front/back ratio of
with the help of a lamp load and photo-electric expo- only about 16dB instead of infinity but, even though
sure meter. Having first balanced the bridge by means the null is incomplete, tuning for a = mini-
of RV1 at roughly the power level which it is required mum Or maximum in the appropriate direction should
to measure, the antenna is replaced by a load consisting be sufficient to ensure correct phasing. On the
of one or more lamps adding up to the required other hand, halving the distance could reduce the depth
_ wattage. The resistance of the lamp load varies with the of null sufficiently to make adjustment difficult, and
power level but it is not too difficult to adjust the tuning too small a separation could also result in confusion due
unit and the controls on the transmitter so that full to near-field effects (p7) or mutual coupling between
lamp brilliance and an swr reading of unity are obtained the antenna (A) of the field-strength meter and the
simultaneously. “Full brilliance” can be judged by eye, antenna being tested (B).
more or less, if a mains-energized lamp is available for It is important in these circumstances to resist any
comparison but, using the exposure meter, the lamps can temptation to adjust the coupling between antenna
be pre-calibrated in terms of exposure reading and this elements so as to obtain a null on the meter since this
makes for greater accuracy and convenience. Assuming will reduce the null depth on distant signals to 16dB.
accurate calibration of a single point it is a simple The correct criterion is twofold, namely (a) null in the
matter to calibrate the rest of the scale, given any type right direction (b) element currents equal. With more
of rf indicator known to have a linear characteristic, elements the situation is more complicated, and in
such as an oscilloscope; thus if 400W produces 4cm general the larger the beam the more separation is
deflection, 1cm will represent 400/16 or 25W. needed, though even with three elements it is possible
For very low power levels, eg less than 1W, increased in practice to make do with a separation of about 1) if
sensitivity is obtainable if instead of being taken this cannot be increased.
straight through the centre of the toroid the feedline is Next we have to consider the effect of a possible third
looped round the core to form a complete turn. antenna C, probably located on top of the only mast
For the adjustment and testing of antennas it is available. Attenuation between antennas A and B is
desirable to work at low power levels but before an likely to be much greater than that between C and
antenna is erected it is advisable to check it with full either A or B so that the signal reflected from C may
power applied, particularly if it includes components exceed the direct signal. In an extreme case a maximum
such as capacitors, relays or transformers which might instead of a minimum may occur at A when B is
be liable to break down in the event of some fault or pointing at C. Intermediate situations are more com-
miscalculation. mon, the presence of C making it impossible to obtain
results that make sense. If C is a beam with a good
front/back ratio it might be thought sufficient to point it
Measurement of radiation pattern in the opposite direction, but this if anything makes
In the case of a two-element beam the relative phase of matters worse by bringing a reflector element closer to
the element currents determines the directions of nulls B! As explained in Chapter 5 the properties of a beam
and this in turn fixes the gain and all the other details of are defined with respect to voltages at its feeder terminals
the directional pattern, except that the depth of the and have no relation to its behaviour as a reflector of
nulls is determined by the ratio of the currents as incoming signals.
explained in Chapter 5. If the currents are equal the The situation becomes worse if A and B are moved
nulls are complete, ie the signal in or received from further apart, and reflections from telephone or power
these directions is zero. The direction of a null can wires then also become more important, so that even in
therefore be used for setting the phase and (in principle the absence of C there may be an optimum separation
at least) its depth can be used as an indication of the distance. Often the problem can be resolved by ensur-
current ratio. ing that either B or C is end-on to the other, but if both
Observations of the radiation pattern require, in are end-on this may increase the coupling if there is a
principle, either a transmitter or a receiver located far considerable height difference. The best arrangement
enough away for its distance and direction to be may be to use C as the test antenna.
regarded as the same for all elements of the antenna The experimental set-up illustrated in Fig 18.2 results
236 HF ANTENNAS FOR ALL LOCATIONS

in direct observation of the effects of adjustments as

they are made, thereby simplifying and speeding up the
process. In view of the problems outlined above, “5 or6 turns
however, it is essential to check the end-product on
actual signals; this can be difficult with a rotary beam

S
due to the time required for beam rotation, but if the (a)
beam is instantaneously reversible the necessary checks
of front-to-back ratio can be quickly and easily made on
a number of stations from both directions. This can be
g
eornrnrerenescee_ Leads to meter

done aurally, either on weak signals or with use of an rf

gain control so that the agc is not operative. On the
other hand, use of signals for the initial tune-up process
is difficult unless a stable signal can be arranged in
conjunction with a portable receiver or suitable exten-
sion leads.
It is essential to provide the test receiver with an
Bifilar rf choke——
efficient antenna having the same polarization as the
antenna under test, the usual ‘“‘tin box” type of field- (b)
strength meter with a short rod antenna being com-
pletely useless for this purpose. It is arguable that tests
should be made for cross-polarization due, for exam-
ple, to feeder radiation, and it may be instructive to
recall some experiments with rather large loop ele-
ments which produced a normal radiation pattern when
tested as described above, but a mass of inconsistent
observations (sometimes amounting to beam reversal)
on actual signals. This was traced to a strong vertical 5
component which had been overlooked in the basic 0752 twin lead
design. \
Design of the field-strength measuring device needs
careful attention and it must be properly calibrated in Coy) 0) ee : |
terms of relative power or voltage. Preferably, germa-
nium diodes such as the OA79 should be used in view
of the sharp “knee” in the characteristics of silicon

[PE oon
diodes. Calibration against a power meter or oscillo-
scope is a simple matter and it is usually adequate to
regard the meter as square law below and linear above
a certain signal level. Fig 18.10. (a) Basic field-strength meter circuit. To reduce the risk
For calibration the meter should be wired back to the that a vertically polarized component in the signal may modify
shack and its readings plotted against antenna current, the meter reading, the lead length plus that of the antenna .
should not be resonant. (b) Use of rf chokes to reduce pick-up on
voltage or relative power, preferably using an oscillo- the leads. The (/4 wire provides a much smaller impedance to
scope with the rf applied directly to its plates since this ground than the rf chokes which are ineffective if the lead length
can be relied on to provide a linear indication of voltage is such that a high rf voltage can exist at the end of it. (c)
Improved but less convenient arrangement using a tuned
applied to the antenna. receiver; this usually eliminates pick-up of extraneous signals
Fig 18.10(a) shows a very simple arrangement which
has been used successfully, but rf pick-up on the meter be used, in which case the additional selectivity is
leads can be a problem and it is essential to make sure essential though the antenna itself can then be broad-
that this does not reach the detector. RF chokes are not band as in Fig 18.11. Otherwise multiband dipoles may
always sufficient and additional precautions include the be used.
use of a balun and an artificial ground as in Fig If enough space is available it is an advantage to use
18.10(b). A tuned circuit as in Fig 18.10(c) provides two meters so that relative field strengths can be
additional sensitivity as well as improved isolation of observed in both directions simultaneously, and in
the meter leads, and some selectivity against strong appropriate cases meters may be permanently wired
signals which can cause interference when operating back to the shack for continuous monitoring of per-
over long ranges and at low power levels. If more formance. It might be thought that antenna gain could
sensitivity is needed, eg to work with lower levels of be directly measured by substitution, observing first the
transmitter power, a one- or two-stage amplifier may signal strength from a dipole and then substituting the
 MAKING THE ANTENNA WORK 237

7582 coaxial cable

2 x SL610

Fig 18.11. Broadband test antenna with tuned receiver and

amplifier provides increased range and reduction of lead pick-up.
As illustrated there is an amplifier gain of 40dB and (for the
shortest length) an antenna loss of 14dB relative to a \/2 dipole;
the loss is almost independent of proximity to other objects,
making this arrangement particularly suitable for a portable
instrument used for the investigation of environmental effects
Line

RF current probe with balanced loop as recommended for rough

estimates of relative current in different parts of an antenna
system. Devices such as this are essential for the experimenter
Square loop whose resources of time or money are limited

beam, but the ground-wave signal is greatly dependent

(a)
on the reflected wave. However, the beam illuminates
less of the Fresnel zone of the test antenna, and there is
no guarantee that the reflections will be comparable
Y) even if there are no power lines or other objects to
1000pF distort the picture.

The current probe and its uses

O-50yuA meter Fig 18.12(a) shows this in its simplest form, coupled to
a wire carrying an rf current. Part of the field surround-
Antenna or line I ing the antenna wire links with the probe as shown so
under test that an induced voltage given by the usual formula,
Unwanted voltage : / ' V = j@MI, is applied to the diode detector and meter,
pick-up causes ; Yaa SX which should be calibrated as just described for the
adding currents in }
one half-loop v v
field-strength meter.
subtracting inthe In the absence of other access to the transmitter
other
Twist or output a spare plug and socket joined by a small loop
tape of, say, lin (2:-Scm) diameter may be arranged for
together
insertion in the output lead or adjacent to the swr
meter, and this will also allow connection to an oscillo-
(b) scope. The probe or other device to be calibrated is
coupled into the small loop to obtain a full-scale
reading and the power level reduced in steps. In use,
one holds the loop close to the wire or some suitable
Small O-50uA meter
distance from it, depending on the sensitivity required.
Usually a rough estimate of current is sufficient and,
Fig 18.12. (a) Simple current probe. (b) Modification to reduce having settled on a distance such as 0-Sin (13mm) as
capacitive coupling effects; note tendency for magnetic and
capacitive couplings to add in one half-loop and subtract in the judged by eye, the repeatability of readings is ade-
other. Component values are not critical quate.
238 HF ANTENNAS FOR ALL LOCATIONS

the dipole plus OC is now equal and opposite to

that of the stub OV so the impedance at O must
be resistive. If this is 60092 the probe will indicate
that the line is matched, ie the current will be the
same everywhere along the line OS.
If there is still a mismatch a minimum or
maximum will be found at a point M about 16ft
(49m) from O. The current ratio [y/Jo or its
reciprocal is equal to the swr and the impedance
at O is given by (/y/Io) Z, where Z for this
example is 6002. As a second (and final) approx-
Fig 18.13. Use of probe to determine correct position and length imation the length OV should be multiplied by
of matching stub for a dipole fed with 600Q line. In the absence of VI,/Iy, but without altering the length COV since
the stub, Q is a point of maximum current; by adjusting OV while this establishes the essential resonant condition.
maintaining the length COV constant, unity swr is obtained in the
line OS. The length COV is correct if the current in it is the same In other words, if OV has to be shortened by 1ft
each side of O then O must be moved 1ft nearer to Q.
The system is now matched, and it will be seen
Sometimes, for example when making observations
that access to the line has enabled the swr to be
on an open-wire line having a high swr, readings may
determined using only the current probe.
be affected by capacitance to ground via the user, a
Although an swr/power meter could be devised
typical symptom of this being dependence of the read- for 600Q2 line it is generally more useful to have
ing on which way round the probe is held; this effect is
such an instrument in the low-impedance line
minimized by the balanced diode circuit shown in Fig
from the transmitter to the tuning unit or balun.
18.12(b) and as a further measure the meter should be
In principle open-ended stubs can be used for
as small as possible. A long insulated handle may also
matching, in which case one would have a stub
help. Sometimes a voltage probe is used but this is not
of length (A/4 — OV) at a point nearer to the
recommended for several reasons: (a) its dependence transmitter by the amount 2 xX OV. Use of
on capacitance to ground via the user, which is unpre-
open-ended stubs should however be avoided if
dictable and variable; (b) current is often the main, and
possible since they are more difficult to use and,
for evaluation of losses the only, interest; (c) often
unless the symmetry is perfect not only in regard
there is no need to make observations except near
to the line itself but also its relation to surround-
current points, whereas when using a voltage probe one
ing objects, line unbalance will result. Such
is forced to operate in the close proximity of voltage
effects should be readily detectable by means of
points. In this case the presence of one’s body can
the probe and, provided the position of P differs
seriously upset tuning and even introduce losses, thus
between the two wires by not more than a foot or
invalidating the observations.
so, this is more or less acceptable.
Use of the probe for equalizing element currents and
. Fig 18.14 shows a 14MHz quad loop with a linear
tracing causes of asymmetry has already been de-
resonator to open-circuit the top end at 283MHz
scribed. Further examples of its use include:
and turn it into a bi-square. To tune it correctly
1. A dipole, Fig 18.13, is fed with a long 600Q line. the loop is spread out horizontally at a height of ©
To improve bandwidth or because of losses (eg if 5ft (1-Sm) and C is tuned for maximum current in
the dipole is short or the line very long) it is the resonator. Points of minimum current are
desired to match the line by connection of a stub. located by the same method as P in the previous
Walking along the line, a point P is located where example and should occur at the side corners.
the current is almost zero; this can be done most This is achieved for the midband frequency by
accurately by finding adjacent points giving small fine adjustment of C and, to determine the useful
but definite and equal deflections, the current bandwidth, frequencies are found for which the
minimum being taken as exactly half-way between null positions shift some 3 or 4ft (1m) either way.
them. With an open-circuit in place of the resonator the
This position may be marked and a point Q of tuning is some five or six times less critical and a
maximum current found by measuring a distance fairly narrow bandwidth, about 0-5MHz, is the
of 1/4, typically 16ft (4-9m) in either direction for price paid for this method of multibanding.
14MHz. Choosing a point O at about 3 or 4ft (1m) . Fig 18.15 shows those portions of the antenna
from Q a stub OV = OQ is connected. Checking which are relevant to operation on 21MHz in the
the current either side of O moving towards V and case of the three-band array using linear resona-
C, it should be equal; if not O must be moved tors which is described on p172. It is desired to
slightly but without altering OV. The reactance of know the efficiency.
 MAKING THE ANTENNA WORK 239

As C or the frequency is the power radiated. For a single element the

yess ~_,— varied the current nulls move
~~. round the side corners effective length = 2 x 0-7 (ie 1-4) times )/2, so
“N
\
i
\
with unity gain the current would be 0-7 times,
and the radiation resistance therefore twice, that
vi

naa
BS
\

1 of a A/2 dipole. Assuming the condition of max-

Probe should read imum nominal front/back ratio the fields from the
zero current here
two elements are in quadrature for the wanted
direction, ie they are at right-angles.
The gain for the two elements, assuming no

Ae
change in the radiation resistance, would be V2
ae Constant current
times in field strength or 3dB; the gain would
ag distribution however be 4-2dB for A/2 elements and in this case
there is also a small amount of collinear gain.
Tuning stub
Assuming a gain of 4-8dB (ie three times in
power) the radiation resistance is divided by 3/2,
and one thus arrives at 73 X 2 X 2/3 or about
Fig 18.14. Use of probe for adjustment of bi-square element. A 962, corresponding to a radiated power of 96W
14MHz quad loop uses a linear resonator to open-circuit the top and an efficiency of 98-5 per cent. It follows that
end at 28MHz; if C is correctly set current minima appear at the
corners, but move away in one direction or the other as the in this case the total loss inclusive of that due to
frequency is varied. They are located precisely by finding points multibanding is only 0-065dB. Proceeding in a
of just-detectable current and bisecting the distance between similar manner, loss figures of about 0-3dB and
them. The length of tuning stub for 28MHz may be any odd
multiple of 4/4 0-2dB have been obtained for 28 and 14MHz
Figures marked against each portion of the beam respectively.
are relative average currents as estimated using
the probe. The dimensions are marked in half-
RF bridges, their uses and how to
wavelengths as it is required to use resistance values manage without them
extrapolated from Fig 3.10 which has been plotted The value of swr bridges has already been stressed and
on this basis. The figures in brackets after the most users will agree with the view that they are
currents are the average diameter of conductor indispensable if one is working with coaxial lines. It is
followed by the ohms per half-wavelength, assum- true that other devices such as an admittance bridge can
ing aluminium alloy with a resistivity 2-6 times that be used to measure the quality of matching, but at
of copper so thatits rf resistance is V2-6, ie 1-6 times considerable cost in convenience and the important
that of copper. The current distribution in the advantage of continuous indication is lost. What the
driven element comprises roughly the centre por- swr bridge does not do, at least not directly, is tell what
- tions of two half-wavelengths, and it is convenient to do about the swr in the event that it needs impro-
to choose the current scale and power level so that ving. This is where an admittance or impedance bridge
‘1-0 represents an average current of 1A in the can help in the event that one has direct access to the
outer portions of the element. Hoping to arrive at antenna or a feeder of known electrical length. In this
such alow figure for the loss that small errors are not latter case a bridge measurement plus a quick reference
important, no attempt is made in this example to
establish the exactcurrent distribution or differenti- 1:0 (0-7, 0-3) N25, G1 jOs2)
ate between average and rms values. Similarly no 2-5 (0-5, 0-4)
correction has been made for the greater sensitivity
of the probe when coupled to conductors of smaller Driven element
diameter, though this can be determined and (21 MHz)
allowance made if necessary.
For the specified currents the losses add up as
follows:
Outer portions of driven Director N
element: 0-25 (0-25W) 1°29 (0-5, 0-4)

Inner portion of driven Fig 18.15. Use of probe to determine rf losses in a complex
element: 0-052 (0-11W) system. Loss resistances and radiation resistance are calculated
Path via capacitor: 0-10Q (0-625W) from the dimensions and relative currents are observed with the
probe; this allows estimation of the ratio of power lost to power
Parasitic element: 0:34Q (0-56W) radiated. Dimensions are marked in half-wavelengths and all
Hence total loss = 1-54W important currents are indicated, followed by figures in brackets
indicating the conductor diameters and, derived from these, the
To determine the efficiency we need to know rf resistances
240 HF ANTENNAS FOR ALL LOCATIONS

to the Smith chart produces the actual value of antenna L.....11 turns of 28swg wire wound on Mullard F X1598 toroid
RF input drive —*
impedance, which can then in theory be adjusted to
the correct value. |
Noise bridges [4] use a wideband noise source as the |
|
signal and a receiver as the detector. In their original i
! 470 470
form they are suitable for the measurement of reac- {
tance only; for this the feeder length must be a whole 1
|
number of quarter-wave lengths and the antenna itself Add known Unknown
admittance admittance
exactly resonant. In the event of a general-coverage
receiver being available, a noise bridge may be used to
discover the resonant frequency of an antenna plus its
feeder system, though this will be meaningless if the
combined resonance occurs at a frequency for which
parasitic elements are significantly detuned. In the
event that an in-band balance is obtained and the
electrical length of the feeder is known, the feedpoint
impedance of the antenna may be deduced. From this, — Detector
in the case of a monoband antenna, one can by use of Through-panel mounting BNC socket with coaxial cable
the Smith chart deduce the value of components for Faraday-screened loop with the outer screen grounded only at
*he pane! socket
correcting the matching, though if the antenna is
multiband these are unlikely to be compatible with Fig 18.16. Admittance bridge. C1 is set to mid-capacitance and
RV1 to full resistance, RV2 and C2 then being adjusted for
correct operation on the other bands. In any case the minimum signal in the detector; this balances the “bridge
extra components should be unnecessary if adequate strays”. Applying the unknown admittance, the bridge is then
provision has been made for the adjustment of what- re-balanced using RV1, C1, and if necessary additional external
admittances. For measurement of inductance the unknown is
ever matching system is in use, an operation which shunted by a variable capacitor which is used for obtaining
usually has to be carried out by trial and error. balance. The reactance (ie 1/admittance) of the added capacitor is
It should be noted, and needs to be emphasized, that then measured separately, being equal numerically to the re-
quired inductive reactance
the impedance measurement is not a measurement of
radiation resistance (except in the case of a simple bridge has a wide range of uses including measurement
series feed at a current point such as is commonly of capacitors, eg for use in linear resonators or delta
employed with quad antennas and single dipoles) since matches. Many bridges have been described, some
matching components other than baluns usually involve being more difficult to use than others or requiring
an “unknown” impedance transformation. special components such as differential capacitors
On the other hand, knowledge of the current which may be difficult to obtain.
loop resistance R could be valuable in the case of a Fig 18.16 shows an admittance bridge designed by
two-element beam since, in conjunction with know- G3MYT which avoids these limitations [6], and Fig
ledge of phase angle and element spacing, it provides a 18.17 illustrates one method of use. It should be
direct indication of gain taking into account any losses possible to substitute a gdo for the rf input drive and the
in the antenna. Thus the current relative to that in a detector plus amplifier illustrated in Fig 18.11 would
free-space dipole is given by V(73/R) and, provided the appear to be suitable unless a convenient receiver is
nulls are deep enough, the currents in the elements can
be assumed equal. An “equivalent dipole’ current and
hence the gain is found by completing the triangle as on Low-level output
2-tone SsB
p68, the phase angle being obtained from the null AF generator transmitter

directions by using Fig 5.6. Any feeder loss must be

allowed for separately. RF input

If the antenna is not resonant in-band this situation

can be corrected by the use of a capacitor or
External admittance
if necessary >
thane
Bridge
Admittance to
be measured

inductor as described in Chapter 4 (p63), and use of a

calibrated capacitor in this way allows a noise or other
Detector
resistance-measuring bridge [5] to be used for impe-
dance measurements. A capacitive mismatch can be
converted into an inductive one by inserting an extra Receiver
\/4 length of feeder, thus obviating the need for a
variable inductor.
Fig 18.17. Block diagram showing method of use. The rf source
In contrast to the noise bridge in its original resist- and receiver are set to the desired frequency and the bridge
ance-only measuring form, an admittance or impedance adjusted for minimum received signal
 MAKING THE ANTENNA WORK 241

available. G3MYT pointed out that the bridge can be under test and an equally loose output coupling is taken
calibrated and a feeling for its use obtained by measur- to an rf voltage probe. Loose coupling is essential to
ing various known combinations of R, C and L, the avoid loading of the tuned circuit which may degrade
measurement of inductance being achieved by treating the Q, and to make sure it is loose enough each
it as a “negative capacitance”’ in which case coupling should be varied in turn, re-checking the Q. If
there is any change the coupling is not loose enough.
i= 2:54 x 104 The probe may be calibrated in the same way as the
fC current probe described earlier and the transmitter
where C is in picofarads, L in microhenrys and F in frequency varied to obtain the “3dB down” points. The
megahertz. Calibrated and used in this way it is Q is given by the bandwidth, ie the separation between
irrelevant whether the bridge is described as ‘‘admitt- the 3dB points, divided into the frequency. For testing
ance” or “‘impedance’’, and for application to feeder an inductance, C may be a variable capacitor, prefer-
measurements it produces answers directly in the form ably with ceramic insulation. For testing a capacitor a
of a parallel, not series, combination of R and C, high-Q inductor is needed; this can usually consist of a
_conversion from one to the other being achieved if short length of open-wire line using the heaviest gauge
necessary by use of the Smith chart as explained copper wire available. Its inductance is selected to tune
below. The importance is stressed of good screening, with the capacitor to the correct frequency, some
keeping leads as short as possible, and Faraday screen- distortion of the line being possible for fine tuning.
ing of the output loop, which means in practice that it
should consist of a short length of coaxial cable with its Cc eee under test
outer screen grounded only at the socket on the panel.
In the case of an open-wire line, points of minimum 5 aan att scone BERG
Very loose couplings

current are readily identified with the probe so that, From transmitter
knowing the feeder length and having used the probe operating at low drive
level.
also to measure the swr as already described, the Smith Power must be sufficient
chart can be used (p60) to obtain impedance at the to allow the use of very
weak coupling
antenna terminals and hence, for example, discover
whether the antenna needs to be longer or shorter.
RF
If there is no bridge available a similar measurement voltmeter
may be carried out with coaxial line by inserting at least
four short lengths of cable totalling A/2 pius a simple
diode voltmeter such as the one shown in Fig 18.18. Fig 18.18. Measurement of Q. The source and the detector must
Working at a low power level in order not to damage be coupled very loosely to avoid degradation of the circuit under
test
the rectifier, the relative voltage for each of the junc-
tions and the position of the voltage minimum may be
obtained by interpolation from the two lowest readings. The voltage probe has other uses, for example
By substituting a section consisting of even shorter checking the balance of a transformer—but take care
lengths the position of the minimum may be deter- not to burn out the diode!
mined to any desired degree of precision, but care must
be taken in making the joins to expose as little as
possible of the cable. Provided the electrical length of Feeder loss measurement
the feeder is known, antenna impedance can be One method has been described on p228. An
obtained using the Smith chart as in the previous easier method makes use of an swr meter, but requires
example. reasonable accuracy at high values of swr. The feeder is
connected to the transmitter via the meter and its*far
end is short-circuited or open-circuited; despite ab-
Measurement of Q sence of the antenna the swr will remain quite low if the
It may be necessary to measure the Q of an inductor or feeder loss is large but go “‘off scale” if the feeder loss is
tuned circuit, or check for example that a capacitor negligible. For intermediate swr values Fig 4.10 can be
used in an antenna has not become lossy due to used to read off the feeder loss.
moisture. Readers constructing their own antennas will It is possible to keep a check on possible feeder
probably have avoided the use of traps, but if the deterioration in the event that power can be applied to
performance of a commercial beam is suspected, de- the system at some frequency for which it is not
terioration of traps is the most likely cause and in this designed, preferably a higher frequency where the
case also a Q measurement is required. The recom- losses will be greater. The swr should be large initially
mended method is shown in Fig 18.18; a low-power but will decrease if the feeder deteriorates, allowing
signal is coupled very loosely into the tuned circuit this to be readily detected.
 242 HF ANTENNAS FOR ALL LOCATIONS

Appendix—The Smith chart and why it is that the line has a characteristic impedance (Zj)) of
useful 100Q2, in which case the scale markings can be read as
“hundreds of ohms” and the measured impedance
Suppose using the swr meter we obtain rather a high ought to be a point such as 50 + 43] lying on the 2-5
reading, let us say for example 2-5, and are not feeling swr circle. If after checking the measurements it still
too happy with this; we therefore want to know what is fails to do so, one may need a new swr bridge, a new
wrong with the antenna. By using an impedance bridge impedance bridge or both but agreement is unlikely to
or one of the various tricks described in this chapter it be exact—it will probably be necessary to cheat slightly
should be possible to discover the impedance “‘seen”’ by by coaxing the impedance measurement onto the
the transmitter, and before erecting the antenna the nearest point on the circle.
reader will have remembered of course to measure the The next step, and this is ‘what the chart is all
electrical length of the feeder by coupling it into a gdo. about”’, is to discover the impedance at the antenna end
You didn’t? Sorry, but it is “back to square one’; given of the line by moving along the line the correct number
this information a mere glance at the Smith chart would of wavelengths as indicated by the scale round the
have revealed exactly what was wrong and what to do outside of the chart. Each time round the circle is a
about it! distance of )/2 and if, for example, the line length is
1-2 we have to make 2:4 circuits of the chart “towards
What it is
As explained in Chapter 4 (which should be consulted if the load”’. This brings us to the point A which is an
the reader is looking for insight rather than an instruc- impedance of (67 — 70j)92, and means that by connect-
tion book) the chart is simply a sheet of graph paper ing an inductance of +j70Q in series we would be left
specially designed for the plotting of reactance against with a resistive impedance of two-thirds the line impe-
resistance. As one moves along a mismatched transmis- dance, ie an swr of 1-5. This might be acceptable but
sion line both of these quantities vary, and the unique the example also tells us that a much better match can
feature of the Smith chart is the ‘““cooking”’ of the scales be obtained by changing to a feeder impedance of 75Q,
in such a way that, provided there is no attenuation in which has the further merit of being nearly a standard
the line, all the points lie on the same circle. value.
The reader should now turn to Fig 18.19 which shows There is still the possibility that we can do even
the chart, a point (marked X) with its circle, and two better by using a shunt inductance as explained below,
dotted lines which will be discussed later. The point but first the reader must be introduced to another
where the circle crosses the swr scale indicates the swr important virtue of the chart, namely the fact that with
in the line and a set of concentric circles can be drawn the scale markings arranged as they are so that “1-0”
representing various values of swr. The chart owes its comes in the centre of the chart, it can be used for any
somewhat fearsome appearance mainly to professional value of line impedance Z,. For this purpose an impe-
users who like having a lot of lines drawn in; this means dance must be divided by Z, before entering it on the
that answers can be read off to better than one place of chart, and the final answer must be multiplied by Z, to
decimals which does not usually help much but looks get it ‘back into ohms’’, as explained on p60. This is
more impressive. Before being too critical of this known as “‘normalization” and, following from the fact
version however, the reader should count his bles- that the reciprocal of one is also one, it has the further
sings—most of them look a lot worse. virtue of allowing the same scales to be used for
Copies of the chart may be obtained from conductance/susceptance as for resistance/reactance .
various suppliers, and will be found to differ slightly which is its reciprocal.
from Fig 18.19. The author has deleted scales which the Words like “normalization” have a deterrent effect
average amateur does not need and may find slightly but the reader has just been taken through this process
confusing, and the resistance scale has also been used quite painlessly (it is hoped) by leaving him free to
as the swr scale which is what it is anyway. References think that the omission of a couple of noughts from the
to susceptance and conductance have also been deleted scale figures was due to laziness on the part of the chart
since these may confuse some readers and are merely designer. Reverting to more popular values of line
reactance and resistance turned upside down, a useful impedance, one must of course be prepared to face up
trick which the chart is rather good at. This does not to such added difficulties as dividing by 50 instead of
~ mean they have to be given new names but a lot of 100.
people think it helps and you may agree with them.
Series to parallel conversion
Using the chart Thus far, the chart enables us to find the inepectaniae at
We must now return to the plight of the operator whose any point along a line of any impedance if we know
measurement of a swr of 2-5 has caused him to appeal what is happening at any other given point in the line. It
to the chart for help and who may, by now, be getting will also be recalled from Chapter 4 that imped-
impatient. To simplify the example it will be assumed ance inversion takes place along a 2/4 line so that a
 MAKING THE ANTENNA WORK 243
WAVELENGTHS.
FORWA'RO °
DECIBELS ATTENUATION
r

me
Fig 18.19. Smith chart, showing its use for
finding the cause and cure for a high value LEERY
. e,
of swr. A measured impedance (point X) is
plotted on the chart and its circle drawn in.
By moving along the transmission line for
the appropriate distance the impedance at
the antenna end is given by a point such as
-A or its reciprocal B. By moving back slight-
ly B moves round to the point B’ on the R =
1:0 circle and a perfect match is obtainable
subject to tuning out the reactance 0-95 Zp
with the appropriate parallel capacitor

normalized impedance Z at one end appears like is defined as the susceptance so that a positive reactance
1/Z at the other. So far we have in effect been X, becomes a negative susceptance (1/X,) and vice
putting Z = R, + jX, but we know also that for the versa. Notice also however that the problem was solved
equivalent parallel circuit without using the “‘susceptance’”’ concept, but if a
bridge was used for the initial measurement it may well
have been designed to measure admittance (ie conduct-
ance plus susceptance) rather than impedance, and
because this is the way one “adds”? impedances in anyway if one is going to make a lot of use of the chart
parallel. 1/R, and 1/X, are by definition the conduct- it is not a bad idea to get used to thinking “‘both ways
ance and susceptance. up”’.
Let us therefore refer back to the impedance A further look at the chart shows how by taking a
R, + jX, = (67 — 70j)@ which has just been obtained little more trouble a perfect match can be obtained. By
from the chart. In its normalized form this was moving back. a distance of only 0-03A towards the
0-67 — 0-7j and to invert it all that is necessary is to go transmitter, B moves round to B’ which is on the
half-way round the circle. This pushes point A across to R=1-0 circle; there is a susceptance of 1-05
B at the opposite side of the circle where it may be read but this can be tuned out with a parallel reactance
off as 0-73 + 0-76), the first of these terms being 1/R, of 100/1-05 = 952, leaving a perfect match in the
and the second one 1/jX, which means that X, = remainder of the feeder.
—1/0-76. This after de-normalizing turns into a The chart has a lot more tricks up its sleeve, but these
capacitive reactance of 132Q and can be removed by are the basic ones which are most likely to result in the
connecting 132 of inductive reactance across the line, swr meter contenting itself with the left-hand half of its
leaving only the conductance 0-73 which corresponds scale. However, if the reader wants to get on really
to an swr of 1/0-73 = 1-37. This is slightly better than friendly terms with the chart and learn a few more of its
the previous result. tricks he will be well advised to persevere with the more
Note in passing that detailed explanations in Chapter 4.
For the chart to acquire its proper status among the
ale Se pt test equipment in the shack it may be mounted on a
jXp X, piece of board or stiff card, with a rotating cursor
244 HF ANTENNAS FOR ALL LOCATIONS

made from a strip of perspex having scale markings [2] Radio Communication Handbook, Vol 2, 5th edn,
corresponding to values of swr from, say, 1 to 10. An RSGB, 1977, p18.16.
alternative is to inscribe a set of swr circles directly on [3] Radio Communication Handbook, Vol 2, ee edn,
the chart using coloured ink. The chart may be mod- RSGB, 1977, p18.22-23.
ified to line up with Fig 18.19 with the aid of typing [4] Radio Gwenn Handbook, Vol 2, sth edn,
correction paint and a pair of scissors, but there is no RSGB, 1977, p18.23-34.
need for this once the user is master of the situation and [5] Radio Communication Handbook, Vol 2, 5th edn,
no longer concerned by appearances. RSGB, 1977, p18.10.
[6] “A simple rf admittance bridge’ ’, G. Garside,
G3MYT/VE3, Radio Communication January
References 1978.
[1] The Radio Amateurs Handbook, 53rd edn, ARRL, [7] ‘“‘Quads versus Yagis’’, W. Overbeck, N6NB, Ham
1976, p520. Radio May 1979.
 CHAPTER 19

Antenna construction and erection

Precise details of antenna construction are felt to have with a radiation resistance of 5Q and fed with 400W will
_ little value since whatever advice is given the construc- have between 4 and 5kV between each end of each
tor will, quite rightly, want to make the best use of his element and ground compared with about 1-2kV for a
own particular skills and experience, as well as avail- \/2 dipole, and losses in, say, an adjacent piece of wet
able materials and facilities. This applies particularly in bamboo will be about 16 times greater. The wet
the case of an experimenter who, after being in the bamboo also has a considerable detuning effect and,
game a long time, will certainly have acquired a large except possibly in the case of full-sized quad antennas
pile of surplus materials; he may moreover have only at points of maximum current, an antenna wire should
just recovered from being told the latest price of copper never be attached directly to untreated bamboo. The
wire and to insist that any he uses has got to be author does not know of any treatment guaranteed to
hard-drawn would certainly provoke a relapse. Proper- be effective in the hands of those of us endowed with
ly tensioned, soft copper wire should last a long time average incompetence or impatience, though several
and, if much pull is needed and breakage likely to result coats of polyurethane varnish applied by an expert may
in serious embarrassment, the wire can be ‘“‘festooned”’ well do the trick. (The author is at the time of writing
onto polythene cord which is very strong and relatively experimenting with ordinary plastic insulating tape.)
inexpensive. To overcome this problem in the case of bamboo
The situation is of course different if new wire has to spider-arms, polythene cord may be used as in Fig 12.5
be purchased, particularly if it is intended to use a long or insulated extensions contrived for example with the
span with a coaxial feeder hanging down from its centre help of plastic tubing which, if not of sufficient quality
(p150). After many years of experimenting on a limited in itself, can be used to form sockets for ceramic or
budget the author has acquired not merely a large nylon rods.
junk-pile but much experience of improvisations, short Some experiments at 283MHz with various material
cuts and economies—false and otherwise. The story is including low-cost insulating substances have yielded
one of many failures but some successes and, while the values of leakage resistance shown in Table 19.1.
favouring in principle a professional approach to con-
structional matters, the fascination of amateur radio for
the author lies largely in what can be achieved with
limited resources and this chapter is biassed accordingly. Table 19.1. Leakage resistance of various
insulators
Insulation Insulator Resistance Loss Loss
(ohms x 10-4) (miniature (wire dipole)
The possibility of greatly reducing the weight of anten- beam) (dB) (dB)
nas, particularly beams, by judicious bracing with thin
polythene cord or nylon line (p168) has not received Polythene cord 1mm
diameter, 6in (15cm)
the attention it deserves, possibly due to the traditional length, dry 700 0-06
view that antennas (except for self-supported and cor- Ditto, wet 430 0-09
respondingly heavy elements) need insulators. In fact if Strip 6in by 1in (15cm
by 2:-5cm) cut from
the principles of good insulator design are followed plastic piping, dry 38 1:0 0-08
through to their ultimate conclusion one is likely to end Ditto, wet 13 2:5
up with a length of nylon fishing line; unless, that is, the Wire ends sandwiched
between pair of dry
antenna is designed to impose greater tension on the bricks 2-6 1-4
line than, say, a shark trying to get itself off the hook! Bit of wet string
The need for insulation varies enormously between (proverbial!),
6in (15cm) 0:3 5
different antennas. Thus an efficient miniature beam
246 HF ANTENNAS FOR ALL LOCATIONS

11/2in dia 1 Vain dia tin dia 7/8in dia resonant frequency by some 3-5 per cent in the case of
typical plastic coverings.
In the case of tubing, structural as well as electrical
considerations apply and aluminium alloy is the only
material likely to comply with both. Suggested tube
sizes for conventional full-size beams are given in
lindia 3/4in dia
reference [1] and reproduced as Fig 19.1. Suggestions:
for a cord-braced design making optimum use of stan-
dard tubing lengths are given in Fig 19.2, though this is
not yet tested and may have to be modified in the light
of experience.
1india ¥4in dia dicsindia The requirements for telescopic construction are
a unrelated to those for minimizing weight and windage
so no attempt should be made to achieve a good fit; it is
a simple matter to construct packing pieces from small
strips of aluminium sheet, or tubing sawn in half. Some
Fig 19.1. Suggested dimensions for one side of tapered Yagi
elements for a three- or four-element beam. A single length of
flexibility should be imparted to the end of the larger
tubing extending each side of centre is advisable for the centre- tube by making two saw cuts at right-angles, as long as
piece of each element. Element lengths may be obtained from possible, with a junior hacksaw. Hose clips may then be
Figs 5.16 and 5.17. The boom may consist of 16—24ft (5—-7m) of 3in
(8cm) diameter tubing (ARRL Antenna Book)
used over the slit ends to lock the tubes in position after
final adjustment or, if the diameter is too small, a clip
may be made by wrapping an aluminium strip round
Conductors and the joints between them the join and drilling it so that it can be tightened with a
Soft copper stretches easily under tension. In one nut and bolt or self-tapping screw.
experiment a pull of 26lb (11-8kg) applied to 4ft (1-2m) Before assembly all parts should be smeared with
of 19swg (1mm) wire produced a permanent extension Vaseline (petroleum jelly), any surplus wiped off, and
of 2in (Scm), corresponding figures (estimated) for the joint taped over. There is in general no need for
breaking strain and allowable working stress being 40lb driven elements to be weakened by splitting them in the
(18-1kg) and 6lb (2-7kg) respectively. These weights centre if as recommended (p152) a balanced (delta or
can be assumed proportional to cross-sectional area T-match) feed system is used, though one of the
and, as an alternative to “‘try it and see’’, the suitability suggested designs of compact beam (Fig 12.10) employs
of a particular wire for a particular application can be split parasitic elements and the centre insulator needs
calculated with the help of the windage formula on p253, to be adequately reinforced. Vertical rods for end-
noting also the example on p150. Joins in copper wiré loading when required may be attached as shown in Fig
must be soldered and given some protective coating 19.3. The more usual cross arrangement uses slightly
such as Araldite, or insulating tape which is cheaper and shorter rods but requires two of them so that the total
does not demand careful cleaning of the joint. If the weight tends to be increased to no advantage.
joint is required to take any mechanical strain it must The method shown in Fig 19.4 may be used for the
start as a strong twisted joint capable of taking the full attachment of elements to a boom or the boom to a
strain, the solder being used only to ensure electrical mast, and the figure also shows how a bearing can be
contact and not for mechanical strength. contrived for a manually rotated beam. The collar
Resonant length of a wire may be affected by insula- should be fairly long and the screws staggered to ensure
tion or stranding and a number of experimenters that the tubing is not appreciably weakened by the
(including the author) have reported a lowering of the screw holes. With this construction it is easy to add a

3/4in dia
(13mm) (19mm) ,

: ;
lf 5ft 6in 5ft 6in 5ft 6in 5ft 6in ft
(1*°68m) (3°66m)

Fig 19.2. Suggested tubing dimensions for lightweight cord-braced beam for 14MHz, making best use of available sizes. It is suggested
that thin-wire elements supported by the cords should be used for the higher frequencies, making full use of the available aperture to
obtain extra gain plus increased radiation resistance and bandwidth. Dimensions shown should allow for full range of adjustment from
reflector to director operation. The 3/16in (5mm) extensions are not likely to be required except (probably) for the reflector. The cord
bracing should be attached to the end of the 5/16in (8mm) sections. Based on 12ft (3-7m) lengths of tubing, an overlap of 6in (15cm) has
been assumed at each joint for the longest element. The boom may be made from 13ft (4m) of 1}-14in (32-38mm) tubing with telescopic
extensions to give a total length of 16—17ft (4-9-5:2m)
 ANTENNA CONSTRUCTION AND ERECTION 247

plastic piping to insulate elements from the boom. In

the case of a reversible beam, rotation requires no
more than a pair of polythene cords (eg 220lb (100kg)
breaking strain) which should be attached to the boom
and brought down to a convenient anchorage; the angle
Antenna relative to the mast should be at least 45°, otherwise
somewhat stronger cords may be advisable. About 120°
of rotation is usually enough for a two- or three-
element reversible beam but all-round rotation is usual-
ly possible with the use of additional anchorage points.
Loading rod

Antenna

Loading rod Boom

Fig 19.3. One method of joining lengths of tubing at right-angles,

eg for end loading. If one rod is of larger diameter the clamp must
be modified accordingly. Clamps are made from squares of
aluminium sheet crimped as shown. For weather protection parts
should be smeared with petroleum jelly and the joint taped over. Fig 19.5. High-quality insulator contrived by use of three
Typical dimensions are: end section of antenna 0-5in (13mm) polythene cords as “guy wires”. AA’ is one half of a linear
diameter, loading rods 0:25in (6mm) diameter for lengths up to resonator and point A must be held rigid without introducing
6ft (1-:8m) losses. This is achieved without the use of a solid insulator, the
three cords from A (shown dotted) being anchored to a point B
light mast extension, say 3—4ft (1m) of gin (16mm) od on the boom and points C and D on the mast, respectively above
and below the boom
tubing, for supporting the radial ties of a lightweight
beam. A short length of bamboo wrapped with insulat-
ing tape for weather protection has also been used If a beam rotator is used it may be advisable to use
successfully for the extension. self-tapping screws on the centre-line of the plate to
Packing pieces may be cut from short lengths of prevent slippage. With conventional heavy-gauge ele-
ments (no cord ties), a better method of mounting is to
Exhaust clamps attach the boom to a horizontal plate which is then
bolted onto a pipe flange welded to a drive shaft, as
recommended in reference [2].
A T-match or linear resonator may require the use of
Boom
parallel lengths of tubing which can be mounted with
straps as shown in Fig 19.5. A high quality of insulation
is required in the centre of a linear resonator and this
may be achieved at the same time as good mechanical
rigidity by guying the ends of the rods in position as
shown in the diagram.
‘me ¥gin (9-5mm) Junctions between dissimilar metals, copper (or
apr | thick steel plate
brass) and aluminium in particular, suffer from elec-
AY) trolytic corrosion so there is a problem in achieving a
LIS
Fixed sleeve
low-resistance join between the feeder and the ele-
7. ment. One method is to use tinned or plated soldering
=
AG,
Rotating joint
tags secured with self-tapping screws and shakeproof
washers. Copper wires should be soldered to the tags
At least
and the joint can then be assembled using Vaseline and
Tt tape as described above. Polythene cord should be
(30cm)
attached a few inches from the joint and used to take
any strain so that there is no pull on the joint. It is
Mast
-----p-----------
important also to ensure that flexing (if any) is within
allowable limits, and in some cases it may be advisable
Fig 19.4. Method of securing boom to mast. Also shown is a to use a flexible insert such as a short length of outer
simple form of rotating joint for a non-motorized system conductor from a coaxial cable.
248 HF ANTENNAS FOR ALL LOCATIONS

Open-wire line an antenna change-over relay (not designed for rf) out
of the junk-box. Lead-in arrangements at both ends
Open-wire line can be easily constructed at a small consisted of thin flex pushed through holes in the
fraction of the cost of the more popular types of feeder window sill.
and is a great deal more efficient. There is some Due however to the high efficiency of open-wire
substance in the complaint that it is less convenient to lines, this could be a fire risk under fault conditions
use, since it cannot be trailed along the ground or taped such as a short- or open-circuit at the antenna: this
onto a metal mast, and should be protected against would cause a very large voltage to appear at the
accidental contact. The author well recalls one occasion lead-in. There is however no need to resort to the
when the swr momentarily shot up to infinity with an traditional practice of drilling holes in glass window
accompaniment of sparks as an errant cow dashed past panes. A few inches of coaxial cable may be used for
the shack window with a farming neighbour in hot each lead-in wire, disregarding the outer conductor, or
pursuit. almost any bit of high-tension lead will serve. Insulators
may be 2-6in (5—15cm) in length cut from plastic
tubing; two holes may be drilled at the end of each, a U
of stiff wire being pushed through the holes and used to
About 3/4 in (19mm) wide for bind the insulator onto the line, Fig 19.6.
6in (15cm) spacing or
For long straight runs of feeder in an accessible
Spacer 3/8in (10mm) wide for position, a spacing of 6in (15cm) is recommended with
2in (5cm) spacing
very few insulators, the desirable number being a
Spacer matter for experience. The author has used runs of up
to 100ft (30m) with none at all—but it is as well to have
a few handy in case of a storm brewing, or problems
Fig 19.6. Simple and inexpensive 600Q line construction. Plastic with birds! For reliable performance in all weathers it
piping of 2-3in (5-8cm) diameter is cut into short lengths and will probably be found on average that about 8-10
then sawn lengthwise into strips, a pair of small holes being
drilled near each end of each strip. Spacing may be 2-6in spacers are needed per 100ft (30m). A spacing of 2in
(5-15cm), with insulation at intervals of some 12-15 times the (Scm) with spacers about every 2ft (60cm) is recom-
line spacing. With wide spacing losses are reduced and construc- mended where a feeder has to be run close to a mast,
tion is much quicker, but symmetry is more easily upset. Note
that the spacers are slightly curved and this helps to prevent trees or other “‘lossy”’ objects.
slippage If the number of spacers is kept low it can be
estimated with the help of Fig 3.10 that the losses
Subject to these constraints very considerable liber- should not exceed 0-2dB for 300ft (91m) of 14swg wire.
ties can be taken without adverse effect; contrary to Even in the case of 28swg (0-3mm) the loss should not
general belief, radiation from the line is usually negligi- exceed 1dB, which is much less than the typical figure of
ble, although as with coaxial line it can be provoked by 3dB for the same length of coaxial cable. Losses due to
failure to take adequate precautions. Almost any wire radiation from a balanced line are approximately equal
can be used provided it is not resistance wire and not to four times the power radiated from a doublet of
made of iron or steel. These have a high rf resistance length equal to the spacing [3], ie the power lost is given
due to poor dc conductivity combined with high in watts per “‘square ampere” (corresponding to 600W
magnetic permeability (uw), rf resistance being pro- total radiated power) by 6,316(D/\)? or 0-35W for a .
portional to Vu. In one case an observed loss of about spacing of 6in (15cm) at 14MHz. This figure is indepen-
6dB in 100yd (91m) of line was traced to surplus dent of feeder length (unless very short) and represents
“copper” wire which turned out to be not copper but a loss of considerably less than one thousandth of an
copper-plated steel. S-unit! Ae
The usual labour-intensive instructions for the con- Much larger amounts of radiation can be produced
struction of 600Q line are best ignored, and for most by out-of-balance currents, namely about 80W per
purposes can be substituted by the rule that “anything square ampere [3] or a loss of 0-5dB for an out-of-
goes”’ as the following example (already referred to in balance current roughly equal to the matched line
Chapter 4) illustrates. For several years the author used current; this is for a line height of A/4, and increases
a 150yd (137m) line between a shack in the house and with height but is almost independent of line length.
another on the antenna-farm; this was made from lots Unless beam rotation can be restricted to about 150°
of odd lengths of wire, suitably paired and ranging from or less there are problems in getting a 600Q line past
22swg (0:-7mm) to 14swg (2mm) with several sharp the beam rotator without undue twisting or risk of
bends which are sources of loss. An overall loss of only contact between the line and the motor housing. The
1-6dB was measured at 14MHz between 50Q termina- author’s method is to use a short 300Q insert at the
tions using 4:1 baluns for transformation into and out of rotator and, in addition, surround the motor with a
the line, an atu, several tv-type plugs and sockets and jacket made from lengths of plastic pipe split down the
 ANTENNA CONSTRUCTION AND ERECTION 249

insulating material. Radiation loss is virtually zero

Spacers cut from plastic tube
about O-5in (13mm) wide
provided no current flows on the outside of the cable.
With most of the older types of cable there is a
considerable increase in resistance with age as a result
of corrosion caused by chemicals used in the manufac-
ture, and with cables of any age serious deterioration
D
can result if joints are not adequately protected against
Fig 19.7. Construction of 300Q line. Spacers are circular rings cut
from plastic piping. A and B, likewise C and D, are connected in
moisture. A blackened inner conductor means that the
parallel at each end of the line : cable is useless but any sign of discolouration or
corrosion should be regarded with suspicion. Losses
middle; this ensures that the ribbon is kept well clear may be measured by methods described in Chapter 18.
of the motor housing. For protection against moisture the author has been
successful with ordinary plastic insulating tape, though
care is needed to ensure the joint is thoroughly co-
30022 line vered, leaving no cracks. Using short plastic-covered
‘On one occasion an experimental antenna with a extension leads it is possible to tape even the end
feedpoint impedance of 200Q was fed with four half- connections, but the best method is to bring these into a
wavelengths of 300Q line. Using a 4:1 balun the swr in waterproof box which also houses the balun.
a 50Q line was 1-0. Next morning it was 3-0. After a lot Coaxial cable is sometimes used for the construction
of time had been wasted suspecting the antenna, the of resonators but is quite unsuitable for high-Q
300Q line dried out and the swr returned to 1-0, but rose applications; typically at 21MHz 50Q cable has a loss of
again to 2 after running some 15ft (45m) of the line 0-09dB per quarter-wavelength which translates into a
through a bucket of water. Prior to this, very erratic loss resistance of about 102. Dividing this by two to
behaviour of an antenna had been traced (after weeks allow for the sinusoidal current distribution in a re-
of wasted effort) to the use of the transparent plastic sonant line gives a Q of 50/0-:5 or 100 only. This
type of line which developed a loss of several decibels contrasts with a QO of 800 in the case of a similar
per 100ft (30m), though only when wet or still drying calculation for 600Q line using heavy-gauge copper
out. The fault appeared only after several months use (12swg or 2:6mm).
and as a further indictment of the plastic it was noticed Small-value capacitors with substantial voltage and
that one of the wires (the other being tinned) was current ratings are often required for use in antennas,
suffering from severe corrosion. particularly for multiband applications including traps
Better types are said to exist but the author has found and many of the arrangements described in these
them unobtainable, and for those that are the only pages. It is often difficult or expensive to obtain the
satisfactory treatment is removal of the plastic strip, required values and the idea of making them up from
thus allowing the wires to be spaced out as described short lengths of feeder is attractive. For this application
above and used as 60002 line; the removal can be resistance loss can be disregarded but there remains the
carried out with a pair of cutting-out scissors fixed in a question of how much of the total loss is due to the
half-open position and run along each inside edge in dielectric.
turn, the feeder being stretched taut and held against Having obtained a calculated copper loss of only
the cutting edge at a slight angle. After a little practice about half the total, the author was at first inclined to
this can be done quite quickly, the present record reject this idea out of hand but Q measurements on
standing at about 2mph though younger readers should some capacitors of about 10pF made from coaxial line
be able to improve on this. yielded figures around 300-400 which is good enough
High-quality open-wire line of 300@ impedance may for most purposes including linear resonators, though if
be constructed using the four-wire principle, Fig 19.7. the loss in the capacitor is added to all the other losses,
This can be considered as a pair of separate 600Q lines it further strengthens the case against traps. In view of
in parallel, symmetry between them being such that the the relatively low dielectric loss it is evident that more
coupling is balanced out. importance must be attached to the copper loss, much
of this being accounted for by the braiding.
Coaxial cable
RF resistance of different types of
Resistance loss, calculated for the inner conductor on
the basis of Fig 3.11, accounts for about half the conductor
attenuation of new coaxial cable. The remainder is Reference was made above to the likelihood of con-
resistance loss in the outer conductor, additional resist- siderable losses if rf current is allowed to flow through
ance loss due to proximity effect arising from the close magnetic materials. When steel masts or towers are
spacing between conductors, and dielectric losses in the energized as low-frequency antennas the losses may be
250 HF ANTENNAS FOR ALL LOCATIONS

reduced to an acceptable level by virtue of the large structure. In cases of difficulty RSGB members should
surface area, particularly as the radiation resistance is get-in touch with the RSGB who may be able to help.
likely to be quite high, but the need to avoid using Assuming visual impact to be related to the required
magnetic wire is demonstrated by the following Q surface area, a mast constructed of metal tubing would
measurements on some A/4 open-wire stubs at 14MHz. appear to have the edge over wooden masts and these
are available from commercial sources. Home-
19swg (1mm) enamelled copper 220
constructed masts, both wood and metal, nevertheless
1/044 (1mm) tinned copper, plastic insulated 210
exist in considerable numbers and almost as many
Plastic-covered iron wire, comparable gauge Rie
different shapes and sizes [4], indicating a bias towards
Despite considerable corrosion and several soldered using whatever materials happen to be available. Inevit-
joints a Q of 420 was measured for 14swg (2mm) bare ably in the absence of professional expertise such masts
copper compared with a theoretical Q of 660. In tend to be overdesigned, other ones being short lived!
contrast bare stranded wire, of comparable diameter Of the many published designs for wooden masts,
and also corroded by use, yielded a Q of only 302. arrangements somewhat similar to Fig 19.8(a) are
featured in several sources and apart from their basic
simplicity lend themselves to a number of useful
adaptations such as telescopic or tiltover use.
Masts The mast is constructed from three standard 20ft
An inexpensive mast for a dipole, inverted-V or verti- (6m) lengths of 4in by 2in (10 by 5cm) timber bolted
cal beam such as Fig 13.9(b) can be contrived by tying a together, the lower post being a 6ft (1:8m) length of
light bamboo extension to a ladder, apex heights up to oak or other hardwood, the lower third of its length
40ft (12m) or more being achievable in this way. It is being set in concrete. Fig 19.8(b) shows it in course of
advisable to tie the foot of the ladder to a strong post erection with the help of a small winch and a second
which serves to prevent it slipping during erection and much shorter post which can be temporary or perma-
helps to support the ladder subsequently, one or two nent. The doubled rope and extra pulley provides a 2:1
sets of guys being also required. It is a reasonable mechanical advantage and halves the tension in the
assumption that planning permission will not be re- rope. This arrangement has been found particularly
quired for this, particularly if it is moved around convenient for work on experimental antennas, a rest
occasionally and falls down from time to time, thereby being provided so that when lowered the top of the
creating an impression of impermanence! It is likely in mast is at a convenient height for assembling the
any case to be “‘experimental”’ in the first instance, antenna. Alternatively, using an extra length of wood as
though some tact and forethought may be required; it a lever, the tiltover arrangement shown in Fig 19.8(c)
should not be possible for example for its fall to may be used; if desired this can be counterbalanced so
demolish a greenhouse next door. The fears of neigh- that the beam sails up at the touch of a finger.
bours that they may be afflicted with a permanent A frequent complication is entanglement of the beam
eyesore can probably be allayed by giving some thought with guys holding up the lower part of the mast and,
to construction and positioning. though this can be overcome by careful planning, the
At the other end of the scale it would, putting it author found that in his case (a) was generally more
mildly, be unwise to embark on a major investment convenient. This depends, however, on a lot of factors,
such as a 60ft (18m) tower without first making sure of including the space available and the existence of ~
the planning permission which is almost certain to be intervening tree branches etc. Fig 19.8(d) shows a
required. Somewhere in between these extremes is a telescopic arrangement which has also been used; two
“grey area’ where there are no clear legal precedents sets of cords, one each side, are desirable since there is
though, subject to maintenance of good neighbourly a tendency to jam if the pull is one-sided. Methods (b)
relations, there is a fair chance that common sense will and (d) can be combined. In the case of (c) and (d) it is
prevail. It is clearly of prime importance to aim for desirable to have ladder access to the top of the lower
minimum visual impact and in general the supporting section for insertion of a final bolt.
structure is likely to be more of a problem than the For supporting a rotary beam a short length of metal
beam itself, though lighter beams can be supported by tubing may be fixed to the top of the mast by U-bolts or
thinner masts. Poles attached to chimneys and similar clamps. In the case of a manually rotated compact or
to those used for supporting television antennas stand a miniature beam such as Figs 12.10, 12.11 or 12.13 the
reasonable chance of acceptance. height can be extended about 10ft (3m) by this means,
Planning requirements have been partly covered on assuming tubing of the order of 12in (38mm) od and gin
pp203, 206; beyond this it can only be suggested that in (3mm) wall thickness plus an additional set of guys. A
cases of doubt an approach should be made to the local 15ft (4-5m) bamboo extension carrying a spreader for ©
planning authority to ascertain whether they consider the support of a three-band inverted-V beam, Fig
that planning permission is required for the proposed 12.2(a), has also been used successfully. There are too
 ANTENNA CONSTRUCTION AND ERECTION 251

WN

| Light tubing or
|other extension
if; can be secured
} here with
| U-bolts

3 lengths of Pair of guys fixed to ground points either

4 by 2in side of base prevent mast from falling
(10 by 5cm) timber sideways during erection
O-5india -«tihp.
(3mm) gilt ~\ Pulley ie! Pulley
bolts . \ i
Fig 19.8. (a) Simple mast con- 2in(5cm) Vee ry Polypropylene rope
struction from three 20ft (6m) \\ |
lengths of 4 by 2in (10 by 5cm) a
timber and a ground post. Over- \S
laps should be at least 30in O-5india \\ '\| Winch bolted
(75cm). (b) Method of erection bolts ~ gl ; ‘| to board secured
using auxiliary mast, winch and ~The \\| ‘| to ground stakes ,
elementary block and _ tackle. (a) - (b) v oF
Washing-line pulleys are suit- ped |Os ;
able. The winch must be secure- ae G
ly anchored. (c) Tiltover modifi- eh hag Yy Yy
cation allows easy erection pro- oH Wy,
vided the beam does not get Me,
entangled with guy ropes, trees,
hedges etc. (d) Telescopic ver-
sion greatly reduces strain on
mast, winch etc during erection
and may allow the use of a lon-
ger mast extension. To reduce 3 by 2in (76by Simm) 1*5in (38mm)
tendency to jam, the cords and extension thick boards on
pulleys should be duplicated on either side act as
the other side of the mast guides for the
sliding top section

Pulleys)!
YS IQA
Rope or / ALTA
strong cord H 1 ‘ Polypropylene
\ | \cord
~\
. ;/ | \:
/ \

Th

(d)
Pine
ie
(c)

Us UW,
many unknowns for the survival rate of such light used, as in Fig 19.9 which provides a 4:1 mechanical
extensions to be predictable under severe storm condi- gain. This proved more than adequate for single-
tions and experience may dictate design changes; in line handed erection of the beam in Fig 12.11 (with rotator)
with the “‘bias” indicated earlier, the emphasis here is at a height of 30ft (9-1m), the block and tackle being
on low cost, minimizing the risk of serious damage in secured at the 20ft (6m) level. It will be seen that the
the event of failure, and easy repair. length of rope increases in proportion to the mechanic-
al advantage, but the tension and therefore the re-
quired cross-sectional area is reduced pro rata so that
Mast erection cost need not be greatly increased in order to obtain
In the case of fixed inverted-V or cubical quad antennas extra “‘lift’”. Using polypropylene rope and washing-
at apex heights up to S50ft (15m) the doubled rope line pulleys, it is however not possible to get below
method shown in Fig 19.8(b) was found adequate by about 600Ib (272kg) breaking strain because the rope
itself without the winch, though it needed an assistant gets jammed between the pulley and its housing.
to take up rope slack with one turn round a firm post as In some cases it is possible to press a tree into service
the mast was raised. as part of the mast system using two or more doubled
If a stronger pull is required without going to the ropes and pulleys as in Fig. 19.10. Possibilities include
expense of acquiring a winch, more pulleys may be using the lower part of a thick tree trunk in lieu of the
252 HF ANTENNAS FOR ALL LOCATIONS

We
Fig 19.9. Block-and-tackle using washing-line pulleys to give four
to one mechanical advantage for beam erection
lower post in Fig 19.8, with normal sets of guys for the
upper sections of the mast, but guys cannot be used if
the mast is attached to parts of the tree which sway in
the wind. Structural requirements vary over a wide
range depending on the tree and how easy it is to climb.
Hopefully the mast will merge into the tree and attract AES

relatively little attention.

A simple method of raising a mast single-handed is
illustrated in the photographs taken from reference [5]
and shows a 40ft (12m) mast in course of erection; to
quote the reference “‘the whole arrangement is perfect-
ly stable and under complete control at all times. It
took 45min (entirely unaided) to assemble and lay out
the mast (four 10ft (3m) poles) and guys, and 15min to
raise it and make the final adjustments.’’ Two guys
should be attached to anchorages on a line at right-
angles to the mast as shown in the upper photograph so 8

Erection of 40ft (12m) mast by the gin-pole method (top) and

attachment of ladder to base of mast (bottom)

that they stay just taut or with a small amount of slack ©

as the mast goes up; this prevents it falling sideways.
This method is advised also with the other methods of
Rope or ropes for
bracing top erection, but if there is only room for one guy (eg
section against because of tree branches) it is not difficult to impart a
tree
bias to the proceedings, so that the mast tries to
overbalance in one direction but is prevented by a
Ropes forsteering
during erection single guy pulling the opposite way. A wooden mast
may be protected with creosote but this makes painting
impossible in the event of deciding that a different
Heavy board bolted to
foot of mast Neil colour would provide a better match to the environ-
Wijix'Stout pegs (eg guy anchor- ment.
Upsages) to stop slipping

Fig 19.10. Method which has been used successfully for erecting a
lightweight beam above a tree top. Pulleys and ropes are fixed as
Guy wires
high as possible and if necessary at intervals on the tree trunk. It Advice against the use of rope, though still frequently
is useful to have a winch and at least one helper. Additional ropes
may be used as required, with each tree presenting a different set given, is valid only in respect of old-fashioned hemp
of problems rope (which is liable to rot, and stretches when wet)
 ANTENNA CONSTRUCTION AND ERECTION 253

and ignores both the major advances in synthetic woe ifs shel ie ee
materials and the shortcomings of the main alternative,
stranded galvanized wire which rusts badly after a few
years use. Polypropylene, to quote the earlier reference
[5], ‘‘is an excellent general-purpose rope equally suit-
able for use as guys or halyards at a price which will
give years of reliable service at little cost”. Terylene,
which is slightly better, is twice as expensive but may be
more suitable “‘in those parts of the world which enjoy
more than their fair share of sunshine’’.
Polythene, much the cheapest material, is excel- N
lent for halyards and suitable for light-duty guys but (a) | (b)
stretches badly if subjected to more than a small
fraction of the breaking strain; the author is puzzled by
references to difficulty in tying knots in it, having tied
very large numbers and found it much easier to handle
than ordinary string or rope. The slippery nature of the
surface has the advantage that knots are relatively easy
to untie (provided they are first relieved of any Fig 19.11. (a) Tension Tin the guy rope can be resolved as shown
tension), and this helps to make it an ideal material at (c) into equivalent horizontal and vertical forces, Tj, which is
equal and opposite to the wind pressure P, and the downward
for many uses around the “antenna farm’’. force T, which causes the mast to buckle as shown dotted.
Typical recommendations for guy anchorages include Additional guys as shown at (b) can be used to correct the
lengths of 2in (Scm) od galvanized pipe and 2in by 2in buckling tendency
angle iron, which should be driven into the ground for a constant depending on type of surface and aspect
at least 2ft (0-6m) at right-angles to the angle of pull, ratio, being typically 250 for round and 400 for flat
though this distance should be increased in the case of surfaces [4, 6, 7]. Some easement as regards wind
sandy soil, large masts, large beams or the use of pipe. loading will result from the bending of elements which
For securing the guys those in a hurry can use any type absorbs some of the wind pressure, though this is
of knot provided it tightens under tension, but for a difficult to take into account and should perhaps be
permanent installation readers will find it to their regarded merely as an addition to the factor of safety.
advantage to consult the above references for a more Wind loading on the mast takes two forms, axial and
professional approach. bending. The axial load arises from the top weight and
If possible, guys should be at an angle of 45° to the the vertical component of the tensions in the guys. In
mast, but the angle can be halved if there is no itself it is not usually a problem but the amount of
alternative; this however doubles the tension required bending stress is a complicated function of both types of
and also results in a strong downward pull which may loading and of the initial tension in the guys. Loading
cause the mast to buckle. To minimize this effect, guys by snow and ice can result in greatly increased windage
should be left slightly loose but not so slack that it and to take care of this it has been suggested [4] that
needs more than very slight movement of the mast to guys should have a breaking strain of six times the
tighten them. In general two sets of three guys (at 120° expected working load.
intervals) are recommended for heights up to 30—35ft Wind pressure on a number of different types of
(9-10m) and three up to 40—45ft (12-14m). antenna is listed in Table 19.2 assuming elements to be
oriented at right-angles to the wind direction.
The pressures in the table are applied to the top of
Windage the mast as shown in Fig 19.11, resulting in a tension T
Knowledge of wind pressure is helpful in deciding what in the guy which can be resolved into equivalent
guy ropes or halyards to use, as well as gauges of wire horizontal and vertical pulls, T sin 8 and T cos 0
and tubing. The effects of wind on a self-supporting respectively. The tangent of 0 is d/h, so that 8 and
beam element, for example, may be roughly simulated therefore sin 8 and cos 8 may be readily obtained from
by hanging a number of weights along it. tables or a pocket calculator. If, in the absence of the
Before purchasing a tower or mast the wind-load guy, the top of the mast is free to move then 7}; must
rating should be obtained from the manufacturer and exactly balance the wind pressure P so that T = P/sin
compared with the estimated wind load for the antenna @, that is, unless the mast buckles under the downward
using the formula: force 7, as illustrated by the dotted lines, a tendency
which may be corrected by additional guys. The
Wind pressure in lb/sq ft = V2/k
tension T will be increased by wind pressure on the top
where V is the wind velocity in miles per hour and k is section of the mast and a rough allowance for this can
254 HF ANTENNAS FOR ALL LOCATIONS

Table 19.2. Wind pressure on various antennas ground and the atmosphere above the structure to be
protected [8]. Radio antenna installations usually em-
Antenna Wind pressure at 80mph (129km/h)
(Ib/sq ft) (N/m?) body the main ingredients of a protective system, but
may tend to increase the hazard instead of reducing it
Miniature beam as Fig 12.14, 21MHz with because of failure to meet certain minimum specifica-
element diameter 1in (25mm), loading tions for such systems. The majority of lightning strikes
rods 3in (13mm) 28 1,340
Modified VK2ABQ, Fig 12.11 32 1,530 involve currents in the region of 2,000A to 100,000A,
Three-band beam, cord braced, Figs 12.1 with an absolute maximum in the region of 220,000A.
and 19.2 70 3,350
Monoband three-element Yagi based on
These are of short duration, eg a rise time of a few
Fig 19.1 (14MHz) 144 6,890 microseconds and decay time of a millisecond or less,
Typical three-element three-band though a complete lightning discharge may comprise a
trapped beam [4] 85 4,070
14MHz quad [4] 170 8,140
sequence of such strokes following the same path and
14MHz quad, arms 1in (25mm) diam lasting up to one second or more. Despite the short
(average) [6] 190 9,100 duration these currents cause intense heating if they
pass through a bad joint in a metal conductor or poor
be made [4] by calculating the windage for the top insulators such as trees or brickwork where they may
one-third and adding it to P. This becomes the top third cause sudden generation of steam; in each of these
of the top half only, assuming the centre of the mast to cases the effect may be explosive, fires may be started,
be held rigid by guy wires at this point. This calculation and there is danger to any individual near the path of
indicates the strength required for the top guy ropes, the discharge. The current can pass safely to ground
the design of the mast itself being rather more compli- only if the following conditions are satisfied:
cated, as indicated above.
1. The current path is of adequate conductivity and
Values quoted for K differ slightly, depending on the
cross-section.
source, but those for wind velocity vary widely with
2. The earth resistance is low enough.
geographical area. Figures on imported American
3. Other conductors in the vicinity of the lightning
antennas refer to an EIA specification based on wind
conductor are adequately isolated or bonded to it.
speeds of 80mph (129km/h) which is also in accord with
4. There are no people or animals in the vicinity of
standard building practice on the UK [4], and corres-
the earth termination, where large potential gra-
ponds to a pressure of 26lb/sq ft (1,250N/m7) for flat
dients exist at the ground surface during a light-
surfaces of large aspect ratio such as a mast or 16lb/sq ft
ning strike.
(770N/mZ2) in the case of cylindrical surfaces, the area in
the case of a cylindrical structure being reckoned as The most frequent cause of damage to equipment is
length multiplied by diameter. On the other hand, wind not a direct strike but voltage or current induced in
speeds in excess of 100mph (160km/h) are featured antennas and mains wiring by lightning strikes in the
from time to time in news bulletins, and antennas at vicinity, at distance up to several hundred yards. Static
50ft (15m) will experience higher wind speeds than charges due to an accentuation of normal atmospheric
buildings of average height. Putting V = 100mph stresses also come into this category. Protection against
brings the pressure up to 40lb/sq ft (1,900N/m2) for flat these effects may be obtained even with a high-
surfaces. An accepted figure for areas liable to experi- resistance earth-connection but the presence of an hf
ence winds of hurricane force is 50lb/sq ft (2,400N/m7) antenna, despite adequate protection against static ©
[6] but it should be safe to assume 25-30lb/sq ft electricity, could in some cases increase the likelihood
(1,200-1,400N/m2) for sheltered areas in the UK. of a direct strike without providing the means for
dissipating it.
Lightning danger varies enormously between differ-
Lightning protection ent areas of the world and this must obviously enter
In order to achieve the most efficient radiation from an into the assessment of what precautions should be
antenna, it should in general be erected in the clear and taken. The BSI Code of Practice [9] puts forward a
as high as possible. Its potential as a lightning hazard is “points system”’ for deciding whether a building needs
thereby increased and serious consideration should be protection and the majority of private dwelling houses
given to this problem, which unfortunately is one of in the British Isles would appear to be exempt, though
considerable complexity, so that no one set of simple many of them might be put at risk by erection of an hf
rules can be devised to suit all situations. Some under- beam or ground-plane antenna on the roof. Many
standing of the general principles involved and, in- antennas mounted on towers or tall masts would also
cidentally, the weakness of many common practices is appear to be in need of protection; on the other hand
therefore desirable. some can be designed so that they themselves act as |
The object of lightning protection systems is efficient protective systems. Short of full lightning
achieved by providing a safe conducting path between protection it is recommended [10] that television
 ANTENNA CONSTRUCTION AND ERECTION 255

antennas etc should be protected against atmospheric

electricity by earthing with a conductor of not less than
1-Smm2? cross-section, the outer conductor of a coaxial
cable being regarded as suitable. An hf transmitting
antenna should obviously receive at least this degree of
protection. Antennas and masts may be earthed for this
purpose by connection to an existing system of earthed
metal work, eg suitable water pipes, the point of
connection being as high as possible in the system. In
the case of vhf or uhf antennas parasitic elements need
not be earthed, but this may be advisable with the much
larger elements used for hf beams.
Whether or not full protection is provided, the earth
resistance should be as low as possible. For full protec-
tion the BSI Code of Practice [9] recommends a value SHORT- CIRCUITZ26
not exceeding 109 and conductor cross-sections of the INNER TO
OUTER
order of 60mm/?, though opinions differ on the latter
figure; the official standards of some countries run as
low as 28mmZ2, and values as low as 5mm? have been
Fig 19.12. Quarter-wave lightning protection. The balanced line is
described as adequate despite a small number of in- approximately (/4 in free space; the coaxial line is (1/4)
V where V
stances of damage to 60mm/ conductors [8]. An earth is the velocity factor of the cable used for making the stub
resistance as low as 10Q is often difficult to achieve but
in such cases an annual dose of a solution of rock salt following exposure to the weather must not be over-
can be very effective. Low earth resistances are some- looked. Otherwise, if the lightning current is allowed to
times obtained by laying conductors in trenches near pass to earth via the tower, explosive disintegration and
the surface, but it seems reasonable to suppose that this collapse could occur. The earth resistance must be as
might tend, for a given resistance, to create dangerous low as possible and a single 3ft (1m) earth spike as
potential gradients over a much larger area in the event often suggested, though adequate for the discharge of
of a strike. static electricity, is insufficient for lightning protection.
The bonding of other conductors to the lightning Balanced feeders cannot be directly earthed except via
protective system can raise complex problems and, if it resonant stubs as illustrated in Fig 19.12.
appears necessary, the references should be consulted. The usual change-over switch for earthing an anten-
Isolation is the alternative, and for full protection na (and, worse still, other ‘‘executive’’ arrangements
requires separations of 1ft (30cm) per ohm of earth likely to require the handling of a feeder during a
resistance, plus 1ft for every 15ft (4-5m) of structure thunderstorm, such as removing it from the rig and
height to allow for the inductive voltage drop in the plugging it into an earth connection) are not merely
down conductor; however, for a slight increase in risk useless for lightning protection but could prove highly
these distances may be halved. A rough idea of the dangerous. Even in the event that an operator remem-
zone of protection of a lightning conductor system is bers to “‘switch to earth” before the onset of a storm,
obtained by imagining a cone with a 90° apex angle “any idea that they provide safety is illusory” [8]. This
extending down from the top of the conductor. This is because the switch is not capable of carrying lightning
may be extended by the efficient bonding to the currents, nor is the gap sufficient for providing the
conductor of, say, a “plumber’s delight” hf beam required isolation which needs to be several feet as
erected over the roof. Most hf transmitting antennas already indicated. It should be remembered that in the
however lie wholly or partly outside the protected zone event of a lightning strike, even if the earth resistance is
and, because of inadequate conductor sizes, are unsuit- as low as 102, voltages as high as two million could be
able for its extension. In such cases the feeder (eg outer lying around in the shack!
of coaxial cable), metal masts and parasitic elements The BSI Code of Practice recommends where possi-
should be connected to the highest convenient point on ble a common earth electrode for lightning protection
the lightning conductor. If there is no lightning protec- and all other services, the lightning conductor being of
tion system they should be earthed outside the building. course routed outside the building, and the installation
In the case of an antenna mounted on a tower or tall must comply with regulations for the other services. It
metal mast, all components including parasitic ele- should be noted that an earth resistance of 10Q is not
ments should be adequately earthed, and unless a low enough to provide in itself the fault protection
separate lightning conductor is fitted the tower must be required by wiring regulations and, if this is the best
free from high-resistance joints (eg fully welded). Even figure available, an earth-leakage trip will be fitted.
if resistance is initially low, possibilities of corrosion Some older types are voltage-operated with a coil
256 HF ANTENNAS FOR ALL LOCATIONS

arrester [11], as shown in Fig 19.13, is a reasonable

compromise for open-wire lines.

References
[1] The ARRL Antenna Book, 13th edn, ARRL,
1974, p204.
[2] The ARRL Antenna Book, 12th edn, ARRL,
Same spacing 1970, p261.
as feeders
[3] Radio Engineers Handbook, F. E. Terman,
McGraw-Hill, 1943, pp 193-194.
[4] ‘‘Aerial masts and rotation systems”, R. Thorn-
; Adjustment slot ton, GM3PKV, and W. H. Allen, MBE, G2UJ,
To earth
Radio Communication August/September 1972.
Fig 19.13. Simple lightning arrester for open-wire line using [5] ‘“Masts and rigging for amateurs—a professional
stand-off or feed-through insulators and sections of 1/8 by 1/2in (3
by 13mm) brass or copper strip. It should be installed across the approach”, J. M. Gale, G3JMG, Radio Com-
line at the point of entry to the station, the ground lead being of munication March 1970.
heavy-gauge conductor and as short and direct as possible. The [6] The ARRL Antenna Handbook, 12th edn, ARRL,
gap should be just wide enough to prevent arcing during trans-
mission (ARRL Antenna Book) 1970; p277: |
[7] Antenna Theory and Design, H. P. Williams, Vol
connected between true earth and the “‘earth connec- 2, 2nd edn, Pitman, London, 1966.
tion” of the wiring system. If an earthed coaxial feeder [8] ‘“The protection of structures against lightning”’, J.
is plugged into a transmitter or receiver to which the F. Shipley, Journal of the IEE Part I, December
usual three-pin plug is fitted, the trip coil will be wholly 1943.
or partly short-circuited and a dangerous situation [9] The Protection of Structures Against Lightning,
involving the whole of the house wiring could be British Standard Code of Practice, CP326, BSI,
created. 1965.
The earthing arrangements shown in Fig 19.12, [10] The Reception of Sound and Television Broadcast-
though ideal for a monoband system, are inconvenient ing, British Standard Code of Practice CP327.201,
for multiband operation; simple earthing of the outer of BSI, 1960.
coaxial cable will, at worst, only slightly increase the [11] The ARRL Antenna Handbook, 13th edn, ARRL,
risk of damage to the cable and the use of a lightning 1974, p270. .
Index

Absorption, ionospheric, 15 transmission-line type, 54 Bridges, rf, 239

Active antennas, 131, 223 Bandwidth, 5, 44, 121, 204
Admittance, 57, 63, 243 antenna plus feeder, 125
Capacitive “‘hat’’, 84, 208, 211
Adjustment of antennas, 6, 223, 229 directive, 121
Capacitors for use in antennas, 249
height, effect of, 229 mobile antennas, 208, 210
Chordal-hop propagation, 5, 16
two-element beams, 226 operational, 121
Close-spaced beams, 3, 24, 67,
Aluminium tubing swr, 121, 124
radiation resistance, 28, 68, 70, 74, 80
making connections, 247 traps, 123
Coaxial line, 31, 37, 47
joining, 246 Beam arrays
losses, 249
rf resistance, 27, 239 broadside, 99, 202
radiation from, 50
Angle of radiation, 17, 101, 132 collinear, 95, 98, 161, 180
Collinear arrays, 95, 98, 161, 180
very low angle, 137, 200 compact (2-el), 172
Concealment of antennas, 205
Antenna tuning unit (atu), 54, 149, 230 compact (3-el), 170, 220
Conductance, 63, 243
Antennas end-fire, 88, 90, 99, 202
Construction, 174, 179, 245
active, 131, 23 fixed, 179, 221
cord-braced, 169, 173, 218, 219, 245,
band-switched, 110 “‘half-loop” vertical, 190, 217, 220
246
beam, see ‘“Beam arrays”’ indoor, 181
Counterpoise, 4, 43, 65, 87, 145, 155,
bi-square, 98, 152, 158 inverted-V, 168, 175, 180, 206, 220 162, 184, 187
capacitively stretched, 111, 169, 191 large dipole, 199
Coupling, mutual
choice of, 217 large long-wire, 198, 199
adjustment of, 72
delta loop, see “Delta loop” lightweight, cord-braced, 168, 169,
critical, 6, 69
dipole, see “Dipoles”’ 172, 246
excessive, 85, 189
electrical length, 148, 246 loops, resonant feed, 156, 158, 160,
neutralization of excess, 85, 189
full-wave, 94, 151 168, 175, 190
reactive, 3, 67, 70
GSRV, 157, 159 miniature Yagi, 83, 91, 113, 167, 220
resistive, 3, 69, 104
ground-plane, 43, 65, 104, 145 miniature quad, 84, 152, 178
wide-spaced elements, 98
inverted-V, 21, 132, 143, 148, 168, quads, see ““Quads”’
Current
175,180; 235 reversible, 131, 172, 179, 182, 194,
indicator, 225
invisible, 168, 205 218, 220
probe, 237
isotropic, 20 rotatable, 166, 181, 192
loaded, 9, 84, 104, 187, 194, 208, 210 rotatable, methods, 246
long-wire, see ‘““Long-wire antennas”’ rotatable, open-wire feed, 222, 246 Decibel (dB), 20, 216
loop, 96, 98, 151, 221, see also separate optimizing for tx/rx, 129, S-point relationship, 230
“Quads” and “Delta loop” 131, 166 Delta loop, 96, 152, 160, 164
Marconi, 103, 144 size reduction of, 168 radiation resistance, 153
mobile, 6, 208 small fixed, 180 two-band, raised by halyard, 153
portable, 214 stacked dipole (multiband), 199, 202 Delta match, 39, 149
quad, see ‘““Quads”’ switchable directivity, 180, 183, 185, Dipoles
receiving, 13, 126 189, 190, 192, 220 asymmetrical, 42, 156, 164
trapped, 108, 161 three-element, 79, 167, 169, 182, 246 capacitively stretched, 111, 149, 169,
vertical, see ‘Vertical antennas’ and UASIAR, 190 191
“Vertical arrays” vertical, short end-loaded dipoles, folded, 96, 147
Windom, 45, 222 193 full-wave, 95, 151
Yagi, see ““Yagi arrays” vertical, two-element, 183, 196 half-wave, 9, 148
Apartments, operation from, 205 VK2ABQ, 1, 3, 75, 100, 110, 167, half-wave, insulation effect, 246
Aperture, 13, 126, 217 168, 172, 180, 184, 217, 220 horizontal, 147
relation to gain, 127 Beam elements, 246 inverted-V, 21, 132, 143, 148, 215
Arrays, see “Beam arrays” insulation, 247 methods of feeding, 149
Array factor, 93 Beam reversal, 75, 77, 131, 182 reactance of, 122
Beam rotator, bypassing, 40 short, 148
Balun, 4, 39, 51 Beamwidth, 23, 26, 127 trapped, 108, 161
construction, 53 rhombic, 200 vertical, 104, 155, 187
limitations, 52 Bi-square loops, 98, 158, 160 Directive patterns, 24, 71
precautions in using, 53 Block and tackle, 252 relation between gain and null
principles, 51 Brewster angle, 103, 133 directions, 130

257
258 HF ANTENNAS FOR ALL LOCATIONS

Directivity, 20, 29 element shape/configurations, effect out-of-band signals, 130

‘Disappearing inductance’’, see of, 81, 97 Inverted-U dipole, 148
“Resonator, linear” losses, effect of, 84, 86 Inverted-V beams, 168, 175, 180, 206,
Doublet, 8 measurement, see ‘“‘“Measurements’’ 220
D-region, 15 misleading figures, 23, 99 Inverted-V dipole, 21, 132, 143, 148,
Driven arrays, 68, 75 phasing of two antennas, 101 215
DX performance, checking, 229 quad, 100 multiband, 149, 161, 164
rhombic, 200 radiation resistance of, 148
Earth subtractive principle, 2, 24 Invisible antennas, 148, 205
artificial, 199 superdirective, 127 use of thin wire, 206, 226
connection, 144, 255 two elements, from null directions Ionosphere, characteristics of, 14
currents, 104, 144, 212 240 Isotropic radiator, 20
losses, 104, 144, 209, 211 versus spacing, 72
Echelon antenna, 201 variation with current ratio, 71 J-match, 2, 47, 87, 187, 195
Efficiency of antennas, 26, 209 variation with phase-angle, 70, 72 j-notation, 56
estimation using current probe, 239 very close spacing, effect of, 79
End effect, 148 Gamma match, 39, 125, 187 Lightning protection, 41, 254
End-fed antennas, 43, 46, 153, 164, Gin pole, 252 Lines, transmission, 30
185, 187, 190 Grid dip oscillator (gdo), 226, 231 antennas regarded as, 38
End-fire arrays, 88, 99, 202 Ground fundamental properties, 30, 57
Environment, 5, 132 connection losses, see ““Earth’’ load resistance (termination), 31
antenna characteristics, 225 effects. Zenit 1335135. 2 losses calculation, 39
exceptional locations, 140 image, 93 losses due to mismatch, 36
Weal ss 132 penetration of wave, 133 practical types, see “Feeders”
obstructions, 141 reflections, 102, 132 reactances, 37
trees, 141 reflections, effect of beamwidth, 201 transformers, 35
Equalization of element currents, 227 reflections, effect on measurements, tuned circuits, 36
E-region, 15 231 Loading, antenna
Experimenter, opportunities for, 5 sloping, 104, 138, 144, 214, 222 centre (inductive), 9, 84, 208
Extended double-Zepp, 94 Ground-plane antenna, 43, 65, 104, 145 end (capacitive), 84, 104, 187, 194
Ground planes, compact, 45, 155, 184, linear, 85
Faults, 226 194 mobile, 208, 210
fire hazard, 32 Guy wires, 252 short elements, 84, 113, 148, 187
Feeders Log cell, 77
ageing, 249 Height Log-periodic antennas, 6, 115, 194,
balanced, 249 effect) 2; 101.132; 230 202, 219
characteristics, 37, 47, 48 effective, 13, 136 Log- Yagi array, 170
choice, 51 optimum, 17, 139 Long-wire antennas, 94
coaxial, 30, 37, 47 virtual, 15 centre fed, 97
losses, 34, 39, 62, 249 Helix, 209 comparison with others, 200, 204
loss measurement, 228, 241 echelon array, 201
open wire, see ““Open-wire line”’ Impedance, 12 loaded (‘‘stretched’’), 97
protection, moisture, 249 series-parallel conversion, 63, 242 rhombic array, 199
radiation from, 49, 226, 228 Impedance, characteristic, 30 terminated, 95
resonant, 32, 55, 106 antennas, 38, 60 V array, 198, 201
single-wire (Windom), 45 charts, 37 Loop antennas, 11, 96, 98, 107, 110, ©
termination, 31, 34 formulae for calculating, 31 151, 156, 158, 160, 164, 180, 190,
twin, 47, 249 lines in series or parallel, 37 192, 220, 222
variation of phase along, 33 Indoor antennas, 1, 161 Losses
Zepp, 4, 47, 87, 187, 195 adverse effects from other _ absorption, 38
Fields, 7 conductors, 163 calculation, 27, 228 )
measurement, 236 beams, 181 conductor, 27
strength, 13, 71 environmental aspects, 144 detection, 27, 228
Free space, 12 return paths for earth currents, 164 earth, 104, 144, 209, 211
F-region, 15 using loaded loops, 162 feeders, 34, 39, 47, 48, 62
Frequency predictions, 15, 18 Insulation, 147, 245, 247 mismatch, 34, 36
Fresnel zones, 137 elements from boom, 232 radiation from feeders, 49, 226, 228
Front/back ratio, 70, 72, 79, 142 polythene/nylon cord, 169, 173, 218, spurious modes, 21, 228
Full-wave antennas, 95, 151 219, 245, 246 surroundings, 29
Interactions between antennas, 221, traps, 109
Gain 230
additive principle, 2, 23 operating at different frequencies, Marconi antenna, 103, 144
aperture, relation to, 127 109 Masts, 250
calculation of, 23, 24, 69, 74, 98, 100, Interfering signals, rejection of, 128, Match, conjugate, 28
229 223 Matched line, 31
close-spaced beams, 70 advantages of three or more Matching
definitions, 20, 22, 29 elements, 129 any two impedances, 65 —
directive, 29 nulling methods, 130 impedance bridge, use of, 240
 INDEX 259

Matching—(contd) impedance, 31 Reflection coefficient, ground, 133

“odd bit of wire’’, 43 losses, 49, 248 Relays, for antennas switching, 110,
unsymmetrical feed, 42 radiation from, 51, 248 2347222189223
Zepp (end) feed, 47, 87, 187, 195 rotating joints, 248 Reliability of data, 225
Matching, antenna to coaxial line Re-radiation of interfering signals, 128
balun, use of, 4, 39 Parasitic arrays Resistance
centre feed, 39, 149 miniature, 175 earth connection, 144, 255
delta match, 39, 149 quad, 179, 221 mutual, 3, 69, 104
gamma match, 40, 187 quad, small, 169, 178 radiation, see ‘“Radiation resistance”
series feed, 39 three-element, 79, 167, 217 rf, conductors, 27, 239, 249
shunt feed, 40 three-element, compact, 171 Resonant length, bare wire, 148
T-match, 40 two-element, 68, 78, 168, 173 plastic covering, effect of, 246
Matching, antenna to open-wire line Phase, effects of difference, 23, 70, 73 Resonator, linear, 111, 119, 217
centre feed, 39 Phasing calculations, 117
delta match, 39, 149 lines, 75, 78, 185 construction, 247
resonant lines, use of, 39, 106 pairs of antennas, 100, 217, 220 loops using, 158, 160
stubs, use of, 41, 238 resonance method, 76 matching methods, 112
Matching, antenna to twin line, 39 Pi-network, 54, 149, 230 ‘“‘spurious”’ resonances with, 119
_ Measurements Planning permission, 206, 223, 250 suppression of unwanted resonances
antenna gain, 228, 229, 230, 236 ‘“‘Plumber’s delight” construction, 41, 111 by, 110, 163
antenna impedance, 34, 239, 241 Point source, 22, 93 traps, 161
antenna losses, 239 Polar diagram, 10, 24 Rhombic antenna, 96, 199
current ratios, 237 Polarization, 8, 12, 15
errors in, 50, 229 break-even height, 136 Screening, one antenna by another,
feeder loss, 241 test antenna, 227, 229 226, 229
radiation pattern (directivity), 235 Portable operation, 143 Series-parallel conversion, 63, 243
R, Cand L, 232, 241 antennas for, 214 Sidelobes, 94, 128
resonances of antennas, 231 Power, reflected, 33 Signal-to-noise ratio, 127
rf power, 234 Propagation, 14 Smith chart, 34, 56, 213, 242
swr, 33, 62, 232, 236, 238 Slopers, 81, 222
Mismatch loss, 34 ““OH” beam, 189 Sloping ground, 104, 138, 144, 214, 222
Mobile antennas, 6, 208 Q measurement, 241 Small beams, 83, 91, 167
Monopole, 43, 65, 104, 164 Quads, 11, 67, 96, 98, 110, 151, 156, Spacing of elements, 25, 182, 186
beam elements, 184, 195 158, 160, 168, 190, 192, 222 S-point, 20, 230
compared with dipole, 43 bi-square, used as, 106, 152 Spurious modes
counterpoise, use with, 44, 157, 164 DJ4VM type, 156, 158, 160, 168, 190 bent antennas, 21
short radials, use with, 44, 162 fixed, in trees, 151, 164, 182, 221 linear resonator, 119
Multiband operation miniature, with “‘top hats’’, 84, 152, Stacking gain, 99
advantages over monoband, 113 178 power sharing, effect of, 101
GS5RV antenna, 157 multiband, 110 Stub, matching, 36, 41, 238
linear resonators, 111 radiation resistance, 98 Supergain antennas, 2, 4, 88
miniature beams, 112 rotary, 222 Susceptance, 63, 243
‘“‘odd bit of wire’, 106 Swiss, 67 SWR (standing wave ratio), 32
passive switching, 111 vertically-polarized, 190, 192 losses, effect of, 33, 36, 62
quad, 110 measurement, 226
quad with resonant feeders, 158 Radials, 4, 44, 162 meters, 232
relays, 110 Radiation, 8 /
retuning at antenna, 110 angle, 17, 101, 132, 137, 200 Test equipment, main essentials, 231
retuning in shack, 106, 157 feeder, 44, 49, 51, 226, 228, 248 Testing, antennas, see “Measurements”
stacked arrays, 109, 110, 149, 168, unwanted modes, 21 T-match, 40, 152, 247
171 Radiation pattern Traps, 108, 160, 222
traps, 108, 160 Cartesian, 26 losses in, 109, 216
two-band “‘stretched”’ elements, 169, full-wave antenna, 94 Trees
pA measurement, 226 antenna supports, 151, 164, 182
vertical antennas, 162 vertical plane, 134 performance, effect on, 141, 194
vertical antennas, use of masts as, Radiation resistance, 28, 61, 68, 70, 74, Tubing, suggested sizes for beams, 246
114 75, 83, 102, 209, 230 Tuner for vertical antennas, 157, 185
dipoles, 28, 61 Turning circle, 2
Neutralization, excessive coupling, 85, height, effect of, 102, 230
189 loading, effect of, 28, 61 Uzkov formula, 28, 88
Noise, 19, 29, 127 loops, 98
atmospheric, 128 mobile antennas, 209 V beam, 96, 198
cosmic, 19, 127 mutual reactance, effect of, 74 with reflector, 201
mutual resistance, effect of, 68 Vertical antennas
Obstructions, effects of, 141 negative values, 74, 89 dipole, 104, 155, 187
Open-wire line Radiator, isotropic, 20 environmental aspects, 134, 140
construction, 248 Reactance, mutual, 3, 67, 70 feeding, 44, 65, 157, 165
four-wire, 37, 249 Receiving antennas, 13, 126 ground plane, 43, 65, 104, 145, 155
260 HF ANTENNAS FOR ALL LOCATIONS

Vertical antennas—(contd) 167, 168, 172, 180, 184, 217, 220 hard-drawn, 151
Marconi, 103, 144 VK2AOU/DJ2UT antenna, 111 joins, 246
monopole, 43, 65, 104, 157, 162, 184, Visual impact, 148, 166, 205, 206, 219, rf resistance, 27
195 224, 250 W8JK antenna, 23, 29, 68, 76, 83, 86,
precautions in using, 195 concealment, 205 90, 222
Vertical arrays disguise, 205
“arrays of arrays’, 184 X-beams, 81, 190, 222
compared with horizontal, 183, 196 Wave
“half-loop”’, 190, 217, 220 electromagnetic, 7
Yagi arrays, 23, 79, 90, 100, 218
large, 184 reflected, 31
construction of, 246
log-periodic, 194, 203 Wavelength, 8
long, 88, 90, 180, 202, 204
phased elements, 184 Weatherproofing, 226
vertical, short dipoles, 193
three-element, 184 Windage, 5, 150, 253
trees, effect of, 194 Windom antenna, 222
Yagi with end-loaded dipoles, 193 Wire “Zepp’’ feed, 2, 47, 87, 187, 195
VK2ABOQO antenna, 1, 3, 75, 100, 110, breaking strains, 150, 246 ZL Special, 222
 Mae hh! eA
ay “3 ok Pn n "

i on Aane a Ps
‘
5
oe UP aie Pash 1)
”

ee
Pk
ee
pS ale F Aja t

NE ¥ Abd
re
 *) ae. Np
Sit
ial
NYahari
oP nota hate
yy ¥ ty Noe : he eh nai
g q

iromy: 8 be A, -
prec at
Wk. i jae a)
AY ak. a ¥ ‘le Lees , "he b
Pa we, ff a u ~*~) ‘af wrcss
‘ 7 } LO hs oe, a ¥) i TR
: i a = ha ig ys Ba
ae \ 7 ‘
'] >| y ps Po Ps 4 ti ayn

RS Sah agian Ae
ra ; bY. ee pe ie he
Te d dd

\ * a a > Pio ‘
4 ’ is
; > '
Van i al |
‘i4 ? >t
au i
- i

. =a ’ yj i”

ss

] . i?

A é
4 A
4 f ¥ yt
:A | \ i
f “ht 4
t

i
43'S
Zz I y

¢ i
~
i

'
, ~.
a
aa)

rf :

.
'

ay

<a

Ang ’ i

a tt
urs ‘

t i
bj
‘
J j iP ’
y '
,
=% a |

> 12
r oe

" + “i " é ar,

. ee int Hagar”:
i " ; ap, an
k Hir- eo?
ihe #2
A FA ' ae) i
i
, Ti :
je ei A: i et Arita
} is a i” a f “Ca oF"
Pal fie
m ayy:
5 { en |

Ay ee
met
 =) y ie 7
fas o4 y

i if
i baat| hhba tei i
oy
ats
i Pity

}) ~—

k
h
1
4
cs
* Ree
ae
=eet
er
 Some other RSGB publications...
AMATEUR RADIO OPERATING MANUAL
Covers the essential operating techniques required for most aspects
of amateur radio including station organization, and features a
comprehensive set of operating aids.

AMATEUR RADIO TECHNIQUES

Basically an ideas and source book, this ever-popular work brings
together a large selection of novel antennas, circuits and devices as
well as many fault-finding and constructional hints.

RADIO COMMUNICATION HANDBOOK |

First published in 1938 and a favourite ever since, this large and
comprehensive guide to the theory and practice of amateur radio
takes the reader from first principles right through to such specialized
fields as radio teleprinters, slow-scan television and amateur satellite
communication.

RADIO DATA REFERENCE BOOK

Presents a wide range of useful reference material in convenient
form, and without needless repetition of basic theory.

TEST EQUIPMENT FOR THE RADIO AMATEUR |

Explains the principles of measurement techniques and gives con-
structional details of many items of modern equipment of interest,
not only to the radio amateur, but also to the electronics enthusiast.

VHF/UHF MANUAL
The standard textbook on theory, techniques and equipment for
amateur radio transmission and reception at frequencies from 30MHz
upwards, including microwaves.

RADIO SOCIETY OF GREAT BRITAIN

35 Doughty Street, London WC1N 2AE

ISBN 0 900612 57 6