YAGI ANTENNA

i E = ; in Niby Dr James L. Lawson, W2PV

THE AMERICAN RADIO RELAY LEAGUE

: . mei
y

De
coke
 iGLE
VD VELLL
ED E
K

,
“
aN
v2
,4
*
)!
Y,4 ‘a
»
sd
4

”,

SSS
2

=e
-
ae
rat

ws
1
‘ =

Published by the Sa
AMERICAN RADIO RELAY LEAGUE fS3an7/
225 Main Street eS
Newington, CT 06111

\
se
Fop
+ XY

vei Y

Wes
 aniemick COLLEcel
fy eee
. a
CHIOLD O77 Re a

: 621-BESS

Copyright © 1986 by

The American Radio Relay League, Inc.

Copyright secured under the Pan-American

Convention

This work is publication No. 72 of the Radio

Amateur’s Library, published by the League.
All rights reserved. No part of this work may
be reproduced in any form except by written
permission of the publisher. All rights of
translation are reserved.

Printed in USA

Quedan reservados todos los derechos

First Edition

$15.00 in USA and possessions

ISBN: 0-87259-041-0
 FOREWORD

This book deals with a subject of interest to nearly every radio amateur,
the Yagi-Uda antenna. All aspects of the design of high performance
antenna systems are methodically presented. Among the topics addressed
by Yagi Antenna Design are:
An efficient algorithm for calculating Yagi-Uda performance
characteristics with a computer
Verification of theoretical calculations by comparison with
measurements
Systematic examination of performance versus element lengths
and spacings for two, three, four, and six element Yagis
How to establish a good free space design and then optimize that
design for maximum front-to-back ratio
How to select antenna height and to compensate for the effect of
ground on the free space performance, and how tilting the boom
affects performance
The relative performance of loop elements versus dipole elements
Vertical stacking arrangements: effects of height, spacing,
number of Yagis, phasing, and non-parallel booms
Procedures for scaling Yagi designs to change the operating
frequency and account for a change in element diameter, and for
adjusting element lengths to correct for element tapering and
element-to-boom mounting fixture effects
Complete designs for 7, 14, 21, and 28 MHz Yagi-Uda antennas.
Although this book represents a valuable reference work for the radio
amateur, rapid advances in personal computing have already made
possible anayses which extend beyond that which was feasible at the time
the book was prepared. Thus, Dr. Lawson’s work no doubt will continue to
be expanded and refined, as he would have wanted.

David Sumner, K1ZZ

Executive Vice President

Newington, Connecticut
December 1986
 tA ee Lotiain Wiaci a

x
ry
‘ pie a\ |
yi ad etn sf
afin onabeNeiiey en etTNay
4 il
»
ae Na
ef ry ‘
TS
| :
"I é rhe we Ai
i i awe Se
‘ ‘ re. 2 :
}
at 4
. a : i
Sy Rae iy ‘
F Me " tyos assae

> aoe |anitnna ¥ ihedp ikon,ph tea. tai .

20” EER mf miiW 1) peers Wry ets 5
ne ~ orsyey PertRy. wih cate hea hasnineerits yin. =
any th Lees YD eae ther IMD
$y prec anuigs oben Lens gh Awitvarss keen dian
ibaa Se Rane aPree barre
Ga siongies vaanal! Vita laste sesctat ay nae see ul.
fs i. f \ a ica cr Bisdidiys Slneay iAP « wi. creates fat if ‘

segwh Bae ‘ego, Kap eR [Link] ine Rrra on a

Se a bon, ee wl ODF wihRR
Dalits
i Narom Lalikios Gy teed tenet
mo Moats a vee i poche a
ets 707:Hisnyahe Pa igs ;
Tio Ape AN Braet anes aM
+

bea 36 oe Wad cart| yy

ctrswerghe oi Aa Ang te Dr nike. 508 Assaienidies weit

rita Wt te f rps iteA
) pee bi hee ater ONG
coats yaoi |e neu ent ras
pe vo Ge iy
du) BA sacrs be be ate ine

| vale 3 soe hs ee
a Ti BpCee, ca Es TPS:ee ea
soist arty get Swe ehekprattenn. edvdeidlety w ctresogrcel
‘dae chet i ue
nv? @rit th) Uticae ani b anes
cori. MEA fect annow 2‘vowiad90
sola 2 ase

4 Ais
PREFACE

This book is a newly edited collection and an extension of a series of

articles that | authored and published in Ham Radio during the early part
of 1980. My objective was to develop design concepts for high-quality
Yagi-Uda antenna systems for potential use by Amateur Radio operators
in the lower-frequency communication bands where ionospheric refraction
occurs; that is, frequencies between 5 and 30 MHz. The approach was to
start with a reasonably good physical model of such an antenna system
and compute its properites. Computation methodology was developed and
adequately validated by comparison with a wide range of published
experimental results. This methodology, implemented on a modern com-
puter, allowed a quick and accurate investigation of the performance of a
great many complex antenna systems. This happy circumstance makes
possible a task that, experimentally, could never have been done or at
best would require many lifetimes to perform. As a result, many brand-new
conceptual ideas concerning these systems have emerged.
The series of articles in Ham Radio were constrained to meet the
requirements of a magazine publication format and schedule; these limita-
tions are removed in this book, so that a more coherent presentation is
possible. The reader will note that | have made alterations to certain equa-
tions and numerical quantities in several places; these changes reflect the
use of improved approximations but do not sensibly affect any of the
results. In addition | have added, in Chapter 8, a number of representative
designs with practical ways for the reader to use various element-taper
schedules.
| am indebted to many who have written helpful comments to me
regarding the original series. | also appreciate the efforts of the Ham
Radio editors, who have made this publication possible.
| hope that this book will be useful and stimulating, not only to those
who wish to build antennas, but to those who wish to improve the basic
theoretical framework and computational techniques.

JAMES L. LAWSON, W2PV

Schenectady, New York
January, 1980
 if

b-aoltng & ‘hs

roterenen as tyke reno toe
tamer uch ett cr, ab) Ola (Vee dpe tiee bres Dette t
;an tte « 10? RAgpIONGs copa, nae) co eam,
whites. aia wenk vd oes ey U2 etry a pt
| Ow TE oer aas Barn ofireeds, Aaits alrurnane'y Mare ¥ ie
a a8 3 fomc
aft ar OG Wh aePOCO Bea | ae
hey! Banat Oa MDA bo isan) taspegre, bidarsty, VEO : f:
ROT the
oy

be mm vat Fe
SAT NIpOio rte (MaiaN Eyr) Seaahrneion
eh) BP, ttith }
DRCOG MOREA Mey 4 A6m HOGI TIRED (S Delray, ye
{
OC PrN 4eon VEEN Vipuitiwrtnaen oeST cattumay 1G ma
~ OSTOMEd art. 10, adiherttagsed Rave ete)
RDS *
, TTT 8 ‘OTAGO Piss at oie? Geren seen marta"
+, 18 1 ST (NE ered rever Blas iaiiirohems Nae. hs ib
ae i TRG. ehh Lgntiree 2* CALM: nit gohan Sa
7. tae Avion ‘mel roars (ace prineiwne <i Im
ar haem ol Deny GAem way cay tts of halsinesJo e
O° > eee) oeflaberipe how feaarol “eens M8 ocsncapriat
ney eta —
at
TWAS CHO & aE O8 AOE abt nb Revenant
PUpa MRO OF emake ae tarerne a4 ce gra i pene hoped
wit efor nents watt Lingcaiie |‘ergtine <8 aerinasip Je t
] at to pip Fe ie fon o- 7
dRinaewmet to technun wi Tiga At [enon mvnrt. net
A
i
iti ieee tt) “een OT titans ili aya
eae.
<i
em it m4LTD hahah ation tvest ~anfer tienen, 2 Ww
een ef to c ote ap Seren (el p. peerage i ase
|. biolmend:
mi Ment hy sine abam atic edarty ihe
Acwht oy Yes est ND Seti Usoey ed hte pierce set €
ie tet Wve OL Aone G fy a Lint anges t aes
sie ivnS ton, posieat |

biped - —s

i <
' F eee
>
ee
ei Aa “ ea

\ i) *, af
¥y * aL 4 et ;
\
y ia ea
4 2 \% Ps - / ‘4

' 7 pan;
}
Ly
: ve 7 a
aS WaryLa ae | as
Ay
t eh 7
aa ve eo da am , ‘ ed,
: ie .. nes Se ated ket
® st
did T. (SMwate: eal tess
ven
oA 7 <', Lo
\ 4 P
cay gOS
ifs rs |
ial
= Vary) “ e een
eee PL |.)
ce a ai Ay
ie. i > ray : wya 7 a y :Phy o
q'
: |; we. oa _ Ma,
Gad,
ay, 25s a Al
rm ; atePi 7 nos
ea 7 i
4 Sedo Lee “ i A ro + us
rts ie
2 ‘ 5s r bd
iy. is eo . $a

Ty om pei ; A peredSa Ne ont

EDITORS’ PREFACE

Jim completed this revision of his legendary series of Ham Radio arti-
cles in early 1980. For a number of reasons, the publishers of Ham Radio
decided not to produce the manuscript as a book. Shortly before Jim’s
death in 1982, we volunteered to search for a means to publish Jim’s
work that would meet his exacting standards.
Obviously, this project has taken longer than we had hoped. Early on,
Dave Sumner, K1ZZ, arranged for ARRL to sponsor the production of a
camera-ready version. Bob Myers, W1XT, of Ethan Allen Printers, did
much of the early typesetting. Craig Clark, N1ACH, of Ham Radio tracked
down the original magazine artwork and generously offered the entire
collection for the project. All of the radiation patterns were redrawn by
computer to produce the ARRL polar plot format. (Small discrepancies
between the plots and the values found in the tables are caused by a
minor change in the models for self- and mutual resistance. Undiscovered
major discrepancies are blunders entirely the responsibility of the first
editor named below.) We must also acknowledge the considerable behind-
the-scenes efforts of Dick Newell, AK1A.
The very large ratio of figures and tables to text has made the logical
presentation of this material exceptionally challenging. The excellence of the
outcome is the work of the ARRL production team: Debby Strzeszkowski,
Shelly Chrisjohn, Sue Fagan and David Pingree. Sue Fagan designed the
striking cover. Mark Wilson, AA2Z, coordinated the project at the ARRL
end.
The surprising number and persistence of your enquiries regarding this
project has sustained our enthusiasm. We believe that Jim would have
been gratified by your interest and pleased by this outcome.

Bill Myers, K1GQ

Clarke Greene, K1JX
December 1986
 West (tae rity fs aL a, \ 5
ah vA WB Pe te
wha ty ry eR© ome Rare a beg
¢ , MJ 4 Abel
lf
if * ‘
ely + ry
m i a Ss
hifi
= ’ ia i alt Ay Tht ie
‘ of ak ey ae
haa ete «i a
Ne | ;
; . Lay
lie hy
-icaebidpe ¢
7 1
iemy
A
‘ i y . ' i

‘ HOA
* tat

J
t , Q

: Pein vel ye
he i iu 1 betty
’ i oD Ne} © ;
wi

if :

: P Be ae
r i S4e

P ' : q

Yi L

leu. COAL eet Tee iene |aioe! wei:0 ‘ieee ae

‘onl. oes teh Cyathea” eamnaiel te eon gig
hk sith teh, YAP Rate Bs Yh ENN wet Rar
a nly: PENS, ti hs eh da WE VRAIN. CP ey pI Maal aah
if _ TAMAR Eitnec et fone Si
io. hin bead CiSet vo crtnY Parimeat eat gg ceili:
#9 PAROLE eth Wren aKit \ WE eQnene SRT
oy en Beh pec hon te TACAE ewawht doa. trolgraw ity
topianet OOF eee to tA stunt So. Golitegpayg (io
pi iow wee ine yee DWN fei garners ATA
wd nea Oh ts Tedtng: Soltathen ort 16.0 Soniing pel WUE oR
pa OAR Tens) Saetnol toy ratley IRA et re
Aa) \e8) ay eendyl hestehondeBlin ‘DuNae “wearer wit Seip Bolg ath a ata
Loe CEN ABiinG tPA Bite AA FOF eho ait apart
Ma tiie Ant viernes: ay@tinwid bn reviegwing ff-
hve Pia palin bese age en eW Cmca Bibs
sake A Mawes ACL [Link] etcing
@alyed Gath stoaires wartfant daw osbrie exalttyciter Saat yoda
BT to on h@axe BAIT OTipRelntS yiemuithaone fete lant pitt Aosais
MONTE YC SRY NOME ROI SARA, at iy showweratom
of Opi wen reat me Reng, bivud artes: fei A gua. renee yeeth:
AP PUR aaAe te Pevaenicy vit puna RAR Sa ahaeven Qc
oj ai
- Heit)time “pehlupeesICD Socaniaieg beng vadirenin: priiiatat
vit Divo euiks daritnvatled WW imaplaucie wo benlehmm
ear i
" oi inti gnaatt watlara wieLaeltarvat oy yao
i 4 . 4 ay
4 - i, ae ” ; Ned Wee, Ox _
een 2 i
. OER aoe: 7

Mes,
is is, wt ’

cen ats
. , ah oy rd rhae rf | ipt—
DAP tA slorsyne) uated WN i = Bus yd Ca fiBaler a or phe asda’
pay Te Hone, ,‘
ryWikies

~ >}

J)

Davee
ten ae 2 A ‘\b

a ma r

hishls
- De #
De ea
ee 4) j cs
vig ay
}! Dee | ACU
aS ohh ii \ if
ui ( Va rT !
i. ad ; su 7 ‘ie

Zvi a) ee ay
Tye fy. BP .
AEN IA Pe ee tl
ABOUT THE AUTHOR

Dr. James L. Lawson, W2PV was born in southern

India of missionary parents. He was first licensed in 1934
as W9SSP while attending Kansas University. He went
on to obtain his PhD in Physics from the University of
Michigan, where he heid the call sign W8QUI. From 1940
to 1945, he worked at the MIT Radiation Laboratory,
where he made important contributions to the develop-
ment of radar. Following the war, he joined the General
Electric Research Laboratory (now the Research and
Development Center), where he managed a number of
key programs before his retirement in 1981. His technical
specialties included nuclear and thermonuclear physics,
particle accelerators, computers, and information
systems.
 rere -
408 —_—* ne?
7 : : i ne vf,
- aa i wold Ass
a = i # Pe -
# is ° "a.
y. Sy y/ 7
( r . 7 7

= iy - ane oay
a: Toes : 1
} a

; a ; 3 ? j rah
A } : , « ty
i fe fal { a « hd i
7 ee 7 lo
< : neo a
s : =v @
4 eh 5 f 7 ; “a?
i

wed
en
a s
. “a t ee ;
7 hk da
oe

; zi 4 Seat” ea 705°
a]
a ah =) the, x! f A Cee

7. gierlicoye: of Wye daw Vang power: tS tama, es

os, At Vagenectl tet pil 68 MEG uarncims ) O
fetes me gre cgi Saetedl prlistorts eat WaeeW ~] a:
. r fo iteeel edd Heel askayeet? Co as, rat
stds
| at
90 a
ns vay’ gOS POETS top abet Aad bhet ‘tt start é ¢
| penstte ‘ef fetus Tit apt te it SPOT
ay <yaisiet slot sorting marge: urns at et
hgrahed) ait onli!an “ew emt ombvolad wea he
bra Ho oat agi vlan Yoctencsia orqeaays te
Tosti & hadee cd a4 wert: {pany
deca les
Ah TEGT AM 8 iia GF AER ANP ae fF
Panamera NEM: Gt Oe AUN Lab a itoed ene a
tek Ane & eh) lead EMBERS aig

be
CONTENTS
Chapter 1 Performance Calculations
Antenna Properties 1-2
Modeling 1-3
Computational Methodology 1-5
Element Self-impedance 1-5
Mutual Impedance 1-9
Element Currents 1-9
Input Impedance and Directivity 1-11
Writing Computer Programs 1-12
Validation 1-15
NBS Yagi Experiments 1-15
Yagi Gain and Patterns 1-16
Effect of Director Length 1-22
Gain Variations 1-24
Comparison Summary 1-26

Chapter 2 Simple Yagi Antennas

Two-Element Beams 2-1
More Than Two Elements 2-7
Performance Characteristics 2-22
Element lilumination 2-24
Summary 2-31

Chapter 3 Yagi Antenna Performance Optimization

Parasite Length Variations 3-3
Parasite Placement Variations 3-11
Front-to-Back Optimization 3-19
Optimum Design 3-24
Design Example 3-24
Number of Reflectors 3-29
Missing Parasites 3-33
Summary 3-35

Chapter 4 Loop Antennas

Square Loop Model 4-1
Other Driven Loops 4-9
Multiloop Arrays 4-12
Summary 4-16

Chapter 5 The Effects of Ground

Reflections From A Plane Ground 5-1
Ground Curvature Effects 5-2
Image Models 5-3
Propagation Elevation Angles 5-4
Antenna Performance Over Ground 5-6
Best Height 5-12
Antenna Upward Tilt 5-14
Summary S15
Chapter 6 Stacking
Vertical Stacking Arrangements 6-2
Excitation 6-2
Two-Array Stack 6-3
Phase-Derived Fill 6-13
Three and Four-Array Stacks 6-15
Optimization of Stack Arrays 6-16
Orthogonal and Antiparallel Stacked Yagis 6-17
Summary 6-18
Chapter 7 Practical Design
Preferred Antenna Designs 7-1
Radius Scaling 7-3
Taper Corrections 7-5
Boom and Element Clamping Correction 7-11
Examples of Three-Element Beams 7-13
Summary 7-17

Chapter 8 Practical Amateur Yagi Antennas

Designs for 7.15 MHz 8-3
Designs for 14.2 MHz 8-4
Designs for 21.3 MHz 8-7
Designs for 28.5 MHz 8-8
 ile on
q ry

hoWe hiePie isit ies:

iy iacebe4Ropes iitvere ¥ holely#ele s bab onek thi“Lh <eiee ! pe ace
» aval adh angencdcperrdam cahelehe
nine MCAT” ie"d ufoi‘
| e. .pia ua oF oar: ce a gid
s) poe pi eireastiry
y a a spinesaus Rade gie
sh prcivenn. tirbeng widlak
:
et ier vivir hint J ni
9 TOO aR * abil seen! LD
Le } ad : at
n Hh;“Rate an fuesesnneuit
nt

e fra sie dy oe fy Cite pine. %. Arno

Pa t hi y’ i | iy '

e
en

it
vmpeni e) wr
hisnelta . eee! 2 ee
n’,
" oe

th 2h /*ivwiste Uo, wa
Lens a:s ‘ie imy Le 8
ees -

:
Grae ; taatacemisa an yy au aloe! ba’ Sarde ae n
‘syiview erge r
ohaAci cite
i eee oka ryyey Hy! irewiue
rd wid rer he, Fawegtinsd yi) wveganita le ty Bay i
ys keoapanwt’ fivs Waa wantetth insti
* iad a. fama. |i oi a Pieta ee* me ‘ ® *" rlWrawetestte 2 itebak is
ih 4, ven Pvgeiey, it mY,any
Bhasin is i) ce
pra. te cee
gud
e 3
* eerie ident|:Aba he ea
Pada Hey che Ah
ven,al hic. nts
hein * i
ais,hy Wa Male
“a ad Vi r4 | ait yar e gel
¢
|
ye pvedts, SoU sty, shentebie brgiyRIMES
J
e e On
hu

2 tet:pide) Heer oe ; ‘p
=n ents 9 at, aa
ti ra r : A in
, Va noe

He 3 ¥ tide"3
ai Th ih <¥

e ied BS, |
ke
‘hiv wet a

ie t Neuy ue ‘a iti, [Link] (tiAon wig we? 2oe By ah He

ey
itn ” aa vie: i
.

ne |ih Leda wiie

pledthe wcy pi
, in dey Rasta:
iteMalt we
it tap we honed an tay
ad flter o Wah wt feanit
0 tie eh
r wispee ereas daraait gre
2) Trese nin expe‘d
ky © vi: te,* a wane ea.
y Gy ‘t P i
en = f ledamram a i 7 a
ai Cenc a
ia weiitin tity:iFwc
e tne ott Wr raul | vas uh
5 Alay secs, Vhrhanayye
iit chin peapy

meesae of, ‘Ave aa ihy

‘ he:
wis ‘ iu
ion
at te
RENN
ine iaithe!
1 eut
a

art ete ais tbe,whe bie,

"ae :
‘ai na
veve
salaoh Pah AP orgs Tere
Abney pic
a
a Chase
oka BF Ayiae whenat
bees Come

EN
af

aa R
TY

yo! Va
‘
bi
my arva

. .4;
na

4a! hot
ae, y it t fone ¥ f
thi ean ve Re
g “of
Taps ane } h ante ne
; ay i i ea
i et w f VK
} j EN a| : ‘
pow oa i Teach:
ts
iF
herons
: ve
ks

1
aa
=
i

we Nw x
 7 it i ws a : * Tt eee

te eee ee = ee A eee
: a fh ; “— pF we Uy =
of = ¢ ewe iv: ‘seh ry i ye
ar ' '
, =# é Ce

TI iia leah 5 eee)natfi

yy pha a
wg Ariecipeetin., at sy
{ “ ae
: a} rd * ga ry e
~ e 4 Pa6

poh ditey Sepa 5: G4 sok’ ibe

~ ia
sonst
THO
—
Leys
=
Det en ;
id cal Re Bee
-
7
“Wy
;
a
Pla 5 ie
Te
a4
rh
; THise. he Foxy eats Seat Te Ne eS
SS = |
al 1 atari | Son ih ies a ie aes" 4 Bea
\ 2 Phy ty 7) Pets ree ‘6 yp lea - a a leu. oe 1!

aye f.. © : a ig
"a
t fe } . qb

4 ¢ eat Thy oonpte Bai Mer ke ‘s*Wee

Peg ienih Append Deadline Fel : “A

, is 4
7

ay a =
Niece oe ay iy ! ; ;i
ly : ’ ;
; at |RO YS. q § rr fe a
A, t Ones 2 nerd s te x, i wm. Tei: ws
‘
Tem x “7.43 - 7 on
iy aes o. hvatayet
f a : re
hese ary "y Vi
#4) = Yo 4 rigs
oe . 7 By
_) PA fog Aaya
) ; een 1 UF 2 *
sa ol or a . _
“i te : fet
£0 ’ 7 -

. ‘Thin tAy - a
° 4 -)
; y)

If i ; a i
sur - a -

4 : me :

is ne |
= q at .

ae
; = Pa" ai
etn
te o =} 7 .

¥ 2 Leg ak

Cl - > wa) :

roe a ie
=. % 7 hh’
m . _ et }: ae Saas!
Si

\ es cS = a Fai . f 43 i
c perth cht fe
. Laklad %y F iy yeas ‘
. 2 . r eee af
. a le eet ie : ei aioe
J ys s i ee Pt PR
‘ ee i a oO
m ir tls oe.
Pa : i iy Pe Ey yh e ame™
*, ~ i J Ml cette c
ate ios or ‘ts
3 =, ems Y iz
A ay a ea aK 4 ry lie ay? 1Ley

bak * hee
“4 a5 yard cn ag
: ef arn if
7 i We 7 c
we Arey “ cae aaa t aud:

i a
3

a ee
CHAPTER 1

PERFORMANCE CALCULATIONS

2 ince its invention by H. Yagi and S. Uda of Tohoku University in 1926,

the Yagi-Uda antenna—commonly referred to as the Yagi—has received
a great deal of theoretical and experimental attention.! The Yagi has
also become the most widely used Amateur Radio communications antenna,
not only because of its excellent performance characteristics over the rather
narrow frequency bands occupied by Amateurs, but also because of its
remarkable tolerance to construction variations and even construction faults;
it is an antenna that ‘‘wants to work.’’
For those interested in the theoretical basis of Yagi linear arrays, there
are several excellent reference works available including those by Uda and
Mushiake, Kraus, King, and Walkinshaw; these also contain many additional
relevant references.2,3.4.5 For those interested in the experimental side of
Yagi arrays, attention is drawn to Ehrenspeck and Poehler, Lindsay,
Greenblum, and Viezbicke.®.7.8,.9 Reference 9, National Bureau of Standards
Technical Note 688, gives results of a decade-long NBS experimental
investigation of Yagi antennas. It contains a wide range of experimental
results, which can be used to test theoretical ideas.
Despite the extensive publications in existence on Yagi arrays, it appears
that designers of high-performance Yagi antennas are faced with four facts:

1) Accurate antenna experiments are very difficult to make,

especially if the antennas are designed to be used over ground.
There are many variables. It is difficult, if not impossible, to avoid
unwanted reflections. Also, accurate instrumentation is simply
not available for many of the quantities to be measured—current
distribution in the parasitic elements, for example.
2) Conceptual design ideas are often misleading; for example,
the concept of ‘‘optimum element spacing.’’ Optimum, indeed,
but with respect to what?
3) Practical design of real antenna components in some cases
does not exist; a physically ‘‘tapered’’ element, for example, must
be significantly longer than an equivalent cylindrical element, but
by how much?
4) A good theoretical basis for design is uncertain. A high-

1This and other footnotes are given at the end of this chapter
1-2 Chapter 1

performance Yagi system is necessarily complex, with many

adjustable parameters. Because of this complexity, a conceptual
approach to superior design has not emerged. Moreover, the
physical and mathematical simulation of such a system is
potentially risky; not only must the physical model be valid, but
the mathematical computations, which necessarily involve
approximations, must be made with adequate accuracy.
The objectives of this book are to (1) consider all of these facts; (2) make
significant contributions to both theoretical and practical bases for Yagi
antenna design; (3) dispel misleading concepts; and (4) stimulate readers to
make good, accurate, and relevant experiments. The subject of design is
addressed primarily in terms of antenna arrays that are useful in the frequency
range from 5 to 30 MHz, including designs ranging in size up to the largest
practical levels. ‘‘Conventional’’ Yagi components are used, although there
is some discussion of quads and of the quad-Yagi hybrid known as the quagi.
For convenience, horizontal polarization is used unless otherwise stated.

Antenna Properties
Before beginning an investigation of antenna characteristics, it is necessary
to discuss design criteria: What antenna properties are important, and how
are these properties defined in quantitative terms? The antenna user is
concerned with several properties:
1) Antenna gain
2) Pattern (including front-to-back ratio, F/B)
3) Bandwidth
4) Feed line matching, or standing-wave ratio
5) Cost
6) Longevity (wind survival, corrosion resistance)
Of these, the first four are electrical properties; the last two depend basically
on construction engineering and are not discussed further.
Antenna gain, pattern, and bandwidth (items 1, 2, and 3) must be defined
rather carefully. The gain is clearly of paramount importance; the definition
of peak directivity is used for all situations.!° Directivity in a specified
direction is 47 times the ratio of the radiation intensity in that direction to
the total power (Py) radiated by the antenna. Peak directivity is the value
of directivity in the direction of its maximum value.
An isotropic antenna has the same directivity in all directions; its radiation
intensity in any direction is Py/4z and thus its peak directivity is 1 (0 dB).
It has become common practice to refer antenna gain to the gain of this
fictional antenna, which amounts to subtracting 0 dB from the directivity
and calling the result isotropic gain, denoted by the symbol dBi. The
(theoretical) peak directivity of a thin half-wavelength dipole is 1.64 (2.15 dB).
Antenna gains which are referenced to a dipole (denoted by the symbol dBd)
are converted to isotropic gain by adding 2.15 dB.
 Performance Calculations 1-3

It is basically not practical to measure or specify the complete radiation

pattern. The pattern characteristics of primary interest are the angular widths
(horizontal and vertical) of the main beam and the amount of back radiation.
The beam widths are rather simply related to the antenna gain (see Kraus,
ref. 3, page 25). Radiation in the rear hemisphere is usually variable and
consists of one or more lobes. Perhaps the single most meaningful measure
of rear radiation is the front-to-back ratio, or F/B. This quantity is the ratio
of peak directivity to directivity in the direction complementary to the
direction of the peak. For example, if the peak directivity of a Yagi over
ground occurs along the boom towards the directors at an elevation of 10
degrees above the horizon, then the complementary direction is 10 degrees
above the horizon towards the reflector end of the boom.
Note that F/B is only one parameter of the complete antenna radiation
pattern. It is perfectly possible, in principle, to have an antenna with a large
F/B as defined above, but which has serious backward (but not directly back)
lobes. Nevertheless this simple notion of F/B ratio is perhaps the most
important single index of pattern behavior.
The third antenna property of interest is the frequency bandwidth, but
there are at least three important bandwidths: (1) the bandwidth of the
driving-point impedance (the electrical input); (2) the bandwidth over which
the gain remains high; and (3) the bandwidth over which F/B remains high.
All three are important, so it is necessary to display all three. This is best
done by observing a quantitative frequency-swept plot of them all.
Feed line match is controlled by the matching system which transforms
the antenna driving-point impedance to the transmission line characteristic
impedance. The matching system contributes to (usually reduced) bandwidth
in terms of standing-wave ratio on the transmission line. This effect depends
on the details of the matching network and is beyond the scope of this book.
Antenna characteristics can be determined either by experimental
measurements on a physical model, or by calculations from a mathematical
representation of the physical model. Which should be used? Experimental
measurements are laborious, and it is difficult to ensure accuracy. By contrast,
computer calculations are fast and, in principle, can be made with great
accuracy. Using a modern computer, a large number of antenna con-
figurations can be investigated in a few days—a number it would take a
lifetime to explore experimentally. Moreover, because of the inherent accuracy
of calculations, subtle changes and radiative coupling effects can be explored.
Therefore, it appears that if a computational procedure is believable, it can
be used very effectively. Careful experimental tests are needed, however, to
validate the computation methods.

Modeling
A real Yagi antenna can be represented by a set of parallel, cylindrical,
conducting elements, each of which has space coordinates at its center of
1-4 Chapter 1

x, y, and z. As shown in Fig. 1.1

the Yagi is oriented so that the
elements lie parallel to the z axis
and the boom is parallel to the x
axis. In free space, the origin of
this coordinate system can be
placed anywhere, but when
modeling a Yagi over the ground
¥ ienepees x Axis it is advantageous to locate the
DR = DRIVER origin on the conducting plane.
peat ahd abate The elements themselves ap-
proximate a half-wave in length;
Fig. 1.1—Coordinate axes for the the reflector is usually somewhat
Yagi antenna model. longer than a half-wavelength,
the driven element is normally
about a half-wavelength, and the
directors are usually somewhat shorter than a half-wavelength. For con-
venience, and to make the representation independent of wavelength, all coor-
dinates, lengths, and dimensions are expressed in wavelengths at some chosen
design center frequency.
This representation is clearly a simplification of a real Yagi. A real Yagi,
for example, does not ordinarily have strictly cylindrical elements; the real
elements usually have telescoping diameters with connecting hardware clamps.
Moreover, the elements are mounted at their centers with plates or brackets
to the boom, which is usually a conductor.
The clamping and mounting hardware can be accounted for by correcting
the actual length of the element to an equivalent length. Similarly a telescoping
element will act as a cylindrical element of the same average radius and an
equivalent length different from the actual length. The way in which the actual
element dimensions can be converted to a equivalent cylindrical element is
discussed in Chapter 7, as are corrections to length caused by mounting
hardware. As a side note, the element radius taper corrections to convert
actual lengths to equivalent cylinder lengths are substantial; this is often
overlooked by builders who try to use someone else’s element length schedule
with different element tubing diameters.
The simplified Yagi model also neglects the conducting boom; this is
justified only if the real Yagi is completely symmetric around the boom.
Symmetry guarantees zero electrical potential at the center of each parasitic
element and no mutual coupling to the boom. Hence no current will be
induced in the boom by the parasitic elements. If the driven element is open
at its center and driven from a balanced source, then it will induce no current
along the boom. Thus, for the symmetric Yagi with balanced feed there are
no currents in the boom, and therefore no effect of the boom on antenna
performance.
 Performance Calculations 1-5

Unbalanced feed systems such as the gamma match are usually not
especially troublesome, because if the driving-point impedance of the element
is relatively low, as it usually is, the voltage impressed onto the boom is also
low, and boom current will be correspondingly low. Moreover, the loaded
Q of the driven element is usually high enough to ensure reasonable symmetry
of element currents; this also makes for low induced boom currents. Inci-
dentally, it is possible to construct the real Yagi with insulating element-to-
boom supports; this helps to ensure negligible boom currents. In any case,
small corrections to element length due to the proximate boom are discussed
in Chapter 7.
In the mathematical model one can arrange any number of parasitic
elements and any number of drivers. Chapter 4 shows how to use the model
to approximate a quad or a quagi. The model can easily be extended to include
a ‘‘broadband’’ drive (as in the KLM antennas), where there is a main driver
and one or more dependent drivers which are connected through a transmis-
sion line to the main driver.

Computational Methodology
With this conceptual model we are in a position to compute the perfor-
mance of the Yagi array—to find out how the performance of the array varies
with frequency near the design frequency. The first task is to compute the
complex currents which flow in all elements as a result of driver excitation;
to do this, it is necessary to determine both the self- and mutual impedances
of all elements.

Element Self-Impedance
The self-impedance of a single nearly half-wavelength element in free space
has been calculated by many authors; an excellent comparison of the vari-
ous methods is given by Kraus (ref. 3, pages 272-276). Uda and Mushiake
(ref. 2) use the method originated by Hallen (solution of the boundary-value
problem), and present an approximate equation and a table which show the
self-impedance of a cylindrical, nearly half-wavelength dipole as functions
of radius and length.!! It is apparent from the table that a half-wavelength
dipole has an impedance of about 73 + j40 ohms, so a somewhat shortened
antenna is needed to resonate, that is, to show zero reactance. The required
shortening is basically only a function of K, which is defined as the ratio
of wavelength to element radius. In Table 1.1, I extended Uda and Mushiake’s
reactance data to a wider range of values, and calculated the reactance of
a full halfwave dipole, Xo.5, and of an element 0.45) long, Xo.45. Also tabu-
lated is the change in reactance when shortening the element from element
from 0.5X to 0.45d, AX.
The reactance of the halfwave dipole can be easily approximated by a
limited power series in log K:

Xo.5 ~ 33.25 + 3.19 log K — 0.35 (log K)” (1.1)

1-6 Chapter 1

Table 1.1
Cylindrical dipole self-reactance (ohms) for different wavelength to radius
ratios (K). Xo.5 is the reactance of a dipole element 0.5 wavelength long, Xo 45
is the reactance of a dipole 0.45 wavelength long and AX is the reactance
change when shortening the element from 0.5 wavelength to 0.45 wavelength.
K Xo.5 Xo.5 Xo.5 Xo.45 AX AX
(Ref.2) (1.1) (1.2) (Ref. 2) (1.3)
1x10! 34.2 36.1 34.2 23.1 —11.1 -9.2
3x10! 36.7 37.2 36.6 6.4 — 30.3 — 29.8
1x 102 38.2 38.2 38.2 — 14.1 — 52.3 — 52.3
3x 102 39.0 39.0 39.1 — 33.6 —72.6 -—72.9
1x 108 39.6 39.7 39.6 — 55.5 —95.1 —95.4
3 x 103 40.0 40.1 40.0 — 75.7 — 115.7 — 116.0
1x 104 40.4 40.4 40.4 — 98.1 — 138.5 — 138.5
3x 104 40.6 40.5 40.6 —118.6 — 159.2 — 159.1
1x 105 40.8 40.5 40.8 — 141.1 — 181.9 — 181.6
3x 105 41.0 40.2 40.6 — 161.8 — 202.8 — 202.2
1-x 106 41.1 39.8 39.9 — 184.4 — 225.5 — 224.7

or

Xo5 ~ 24.1591 + 14.5151 log K — 5.3677 (log K)?

+ 0.9352 (log K)? ~— 0.06184 (log K)* (1.2)

Equation 1.1 is a three-term approximation, and (1.2) is a more accurate

five-term approximation. Either one is satisfactory in the chief region of in-
terest (102 < K < 104). The quality of both approximations is shown in
Table 1.1.
A plot of AX as a function of log K (Fig. 1.2) shows that the calculated
points lie remarkably close to a straight line. Thus AX can be represented
quite well by a simple empirical expression:

AX a Xo ero aet as log kno (1.3)

The quality of this approximation is also shown in Table 1.1.
From Uda and Mushiake’s table it is apparent that in the region of in-
terest (length around a half-wavelength) the reactance of any cylinder is
essentially linearly related to length; thus for any element length, /, expressed
in wavelengths the reactance is approximately:

Xj= X05 + 20 AX (0.5 — J) (1.4)

Reactance vanishes at the resonant length, /p, also expressed in wavelengths:

Xip = 0 = X95 + 20 AX (0.5 — LR) (1.5)

Then:

Ip = 0.5 + Xo5/20 AX (1.6)

 Performance Calculations 1-7

Xo.45
Xos
me:
ah ro)=

Reactance
change
AX

—200

Log K

Fig. 1.2—Differential cylindrical dipole self-reactance vs. wavelength-to-

radius ratio.

The resonant length is always

Table 1.2 shorter than a half-wavelength; it
Ratio of cylindrical dipole resonant is interesting to quantify the
length (/,) to a half-wavelength for
difference. Table 1.2 shows calcu-
different wavelength to radius ratios
(K). lations of the ratio of /p to a half-
wavelength computed from (1.6),
K 2Ip (1.3) and (1.2) as a function of K.
3 x 101 0.877 A corresponding plot is shown in
fiex 102 0.927
3 x 102 0.946 Fig. 1.3. Note that the resonant
1 x 103 0.958 element is substantially shorter
3 x 103 0.966 than a half-wavelength for thick
i) oe Wer 0.971
3 x 104 0.974 cylinders (low values of K) but
1 x 105 0.978 even for extremely thin wires (high
3 x 105 0.980 values of K) significant shortening
1 x 106 0.982
is required.
Using (1.4) and (1.5) the reac-
tance of an element of length / wavelengths is expressed as:

X] = 20IAX (t= Ie) (1.7)

where all variables are wavelength (or frequency) dependent. In investigating
frequency displays of element properties it is awkward to use these changing
variables.
1-8 Chapter 1

.98

©© &ro)
S

wavelengths
in
2x
length
resonant
i<e)i=)

88

.86
1 2 K) 4 5 6
Log K

Fig. 1.3—Ratio of cylindrical dipole resonant length to a half wavelength

vs. wavelength-to-radius ratio.

Therefore define:
f = frequency
fp = design frequency
F FS/fp = normalized frequency variable
Ap wavelength at design frequency
Kp = ratio of \p to element radius
lp = ratio of element length to Ap Then:

Kee Ky 7 (1.8)

and
i= ph & (1.9)

where Kp and /p are independent of the frequency scan variable, F. Now

(1.7) may be rewritten:

X; = ACF. — Fp) (1.10)

where
A= -20AXIp (1.11)
 Performance Calculations 1-9

and
Fr = Ik/Ip (1.12)

Equation 1.12 can be easily derived from (1.6), (1.1) or (1.2), (1.3) and (1.8).
The constant A in (1.11) is, of course, the slope of reactance with normal-
ized frequency. Recall that a simple series RLC circuit in the neighborhood
of resonance displays an input reactance of:

X = RO (F/ Fp — Fr/F) = 2RO (F — FR) (1.13)

where Q is the electrical Q factor and R is the series resistance. A compari-
son of (1.13) and (1.10) indicates that electrically the dipole element behaves
essentially like the lumped circuit and that the quantity A/2R can be equat-
ed to circuit Q. The Q of the dipole element is an important parameter basi-
cally determining bandwidth.
The self-impedance of the element is thus:

Zp = 73 + JX (1.14)
The accuracy of this expression should be good to a very few percent for
elements within a few percent of a half-wavelength long.

Mutual Impedance
Now we must consider the mutual impedance between two nearly half-
wavelength elements separated by a distance of s measured in wavelengths.
Several authors have made calculations of the real and imaginary compo-
nents of this complex quantity for the limiting case of infinitely thin, half-
wavelength elements. Exponential integral equations, plots, and tables are
shown in Kraus (ref. 3, pages 265-268), and Uda-Mushiake (ref. 2, pages
69-70). Kraus (page 266) also shows calculations by Tai for two cases of
thicker, half-wavelength dipole elements. Tai’s calculations suggest that in-
accuracies caused by using the limiting thin case are not very large; there-
fore, for convenience, I have used it for all calculations. I have extended
the table of Uda-Mushiake; Table 1.3 shows values in ohms for the real part
of the mutual impedance (Ryurt) and for the imaginary part (XyquT), as a
function of element separation. For separations greater than s = 2), a
reasonable approximation can be derived from:

19.06
Ruut = sin2as, Ss > 2X (1.15)
S

19.06
XMUT = oct cos27s, Ss > 2dr
(1.16)
Element Currents
Although some caveats are necessary, we now have the necessary tools to
1-10 Chapter 1

Table 1.3
Cylindrical dipole mutual impedance for values of element spacing(s)
between zero and two wavelengths.
s Ruut Xyut s Ruut Xwut
) (ohms) (ohms) () (ohms) (ohms)
0.00 73.1 42.5
0.05 TT 24.3 1.05 8.8 15.0
0.10 67.3 ips) 1.10 12.3 Ale
0.15 60.4 -7.1 Uae 14.5 6.7
0.20 51.4 -19.2 1.20 15.3 1.9
0.25 40.8 — 28.4 (ea) 14.6 -2.7
0.30 29.3 — 34.4 1.30 12.6 -6.7
0.35 17.5 - 37.4 1.35 9.6 -9.8
0.40 6.2 - 37.4 1.40 6.0 -—11.9
0.45 -4.0 — 34.8 1.45 2.0 -12.7
0.50 -—12.5 —29.9 1.50 -1.9 -12.3
0.55 -19.1 — 23.4 155 -5.4 - 10.8
0.60 — 23.3 — 15.9 1.60 -8.2 -8.4
0.65 — 25.2 -7.9 1.65 -— 10.0 -5.3
0.70 — 24.9 -0.3 1.70 - 10.9 -2.0
0.75 — 22.5 6.6 1.75 — 10.6 1.4
0.80 — 18.5 12.3 1.80 -9.4 4.5
0.85 -13.3 16.3 1.85 7.4 7.0
0.90 -7.5 18.6 1.90 -4.8 8.7
0.95 -1.6 19.0 1.95 -1.9 9.5
1.00 4.0 17.7 2.00 vet 9.4

calculate the parasitic element currents. Recall that the physical model of
the Yagi is a good representation only to the extent that proper corrections
can be made for element hardware variances (clamping and mounting hard-
ware and element radius taper). These corrections are discussed in Chapter
7. Also recall that the computation of self- and mutual impedances are
approximations.!2 Though they are probably good approximations, and
should give reasonably accurate results, one should not rely on them for ac-
curacy better than a few percent.
The first step is to calculate the complex currents (or magnitudes and
phases) of all parasitic elements given the currents or voltages applied to all
drivers. The method is uncomplicated, following the technique of P.S. Carter
shown in Kraus (ref. 3, page 302).!3 For simplicity, I illustrate how this is
done using one driver and three parasitic elements; extension to any number
of elements is obvious. For each element add all voltages gag equate to the
terminal voltage V,:
NZ, + 1 Zyo + 15 213, 4514 Z14 = Vi
Lf, Zy, + 2 + 1h 293 + 14 2x4 = V2
I, 23) + 1p 232+ Ig 233 + 14 234 = V3
Ty Za, + Ip Zan + Tg Zg3 + Ig Zgqg = Veg (1.17)
 Performance Calculations 1-11

Assume in this example that the first three elements are parasites, i.e.,
V, = V2 = V3 = 0 and that the fourth element is driven with the complex
voltage V4. Zpn is recognized as the complex self-impedance of the nth ele-
ment, and Z;, (which is the same as Z;;) is the mutual impedance between
the jth and kth element. Thus, all of these impedances can be calculated once
the positions (and hence separations) of the elements are specified. Since there
are four linear equations with four unknowns, J,, Jy, J3, and J4, the easiest
way to solve this array is by a matrix inversion. In matrix notation (1.17)
is represented by:
Zi=V (1.18)
where Z isa4 X 4 matrix, J and V are four element vectors, and all compo-
nents are complex numbers. The solution is
I=Z-1vV (1.19)
where Z~! is the matrix inverse of Z.
The process of complex matrix inversion is readily accomplished with a
computer, with a program usually called CLINEQ under FORTRAN IV.
Although the actual solution is usually done through a mathematical process
known as Gaussian elimination, the result is equivalent to matrix inversion.
With this technique, a computer will provide solutions quickly for very large
arrays of fifty elements or more.
If one wishes to specify the driven element current, /4, instead of voltage,
V4, rewrite the first three parasitic equations of (1.17) as:

QZ, + 22 + Zy2 = —I4 Zy4

Ty Z21 + 1h 222 + 13 Zon = —I14 Zr
fy Z3, + 1 232 + J3 Z32 = —14 23 (1.20)
This (smaller) array can be solved in the same way for Jj, J>, and J; and the
results used in the fourth equation:

I, Z4, + Ip Zqn + 13 Zan + [4 Zgg = Vig (1.21)

to solve for V4.

input Impedance and Directivity

Once the complex currents in each element are known, the Yagi proper-
ties of interest can be computed by straight forward formulas. First, the input
impedance is found from:

Zin = Vpr/IpR (1.22)

where Vpp and Jpr are the complex voltage and current for the driven
element (V4 and J4 in the example of the previous section). The power
1-12 Chapter 1

input to the antenna is then:

Pins sida TDR PRin (1.23)

where Rjn is the real part of the input impedance.

Two assumptions are required to compute the directivity: (1) all of the
input power is radiated; and (2) the amplitude of the current in each element
varies sinusoidally from maximum at the element center to zero at the tips.
The first assumption is reasonable for HF Yagis constructed with aluminum
tubing, and the second is quite accurate for element lengths near \/2. The
general expression for directivity is:

D=—~ Dini enF ing ; (1.24)

n :

where F is the normalized frequency, J; is the complex current at the center

of the ith element, and 7 is the number of elements. For the H-plane (that
is, the xy plane in Fig. 1.1):

ri = Xi cos@ + y; sind (1225)

and
g = 1 -— cosxr§ (1.26)

where x; and y; are the coordinates of the center of the ith element, /; is the
length of the ith element in wavelengths, and ¢ (the elevation angle) varies
from 0 degrees along the x axis through 90 degrees along the y axis to 180
degrees along the —x axis. Note that x and y are specified in terms of
wavelength at the design frequency. For the E-plane (the xz plane):

ri = xX; sind (1.27)

and
g; = [cos (ri;cosé) — cosrij] / sind (1.28)

with 6 (the azimuth angle) equal to 0 along the z axis, 90 degrees along the
x axis and — 90 degrees along the — x axis. Note that (1.28) contains a divi-
sion by zero when 0 = 0 degrees; in fact the pattern factor g; approaches
zero as 8 approaches zero (corresponding to the well-known null off the ends
of dipole elements). Therefore in the computer program set g; = 0 when
65=7 0;

Writing Computer Programs

Specific computer programs for calculating array properties are not hard
to write, but the details do depend on the specific computer and available
software libraries and subroutines. This section explains the information flow
 Performance Calculations 1-13

used in FORTRAN programs for computing element currents and Yagi per-
formance. I first create a labeled input file containing all the necessary
information about the Yagi. The label itself is usually an abbreviated reminder
of the particular Yagi. This input file contains (1) a statement of the number
of parasites and the number of drivers; (2) information on each parasite,
i.e., x and y coordinates, length and radius, all specified in terms of
wavelength at a given design frequency; and (3) information on each driver,
i.e., x and y coordinates, length and radius, and driving-point voltage and
phase.
The element-current program first asks for the input data file label and
the frequencies (in terms of the center design frequency) at which the com-
putations are to be made. With this starting information, the input file is
called and read. For each frequency specified, a computation is made to de-
termine the (complex) values of all terms in the Z matrix; self-impedances
are calculated from (1.14), and mutual impedances are given by an interpo-
lation routine from the data in Table 1.3 or from the power series approxi-
mations (1.15) and (1.16).!4
Once the matrix is complete and the voltage vector noted from the input
file, matrix inversion is accomplished; this generates the individual element
complex current solutions. The entire result is then written into an output
file for later use. The output file contains (1) a statement of frequency for
each computation; (2) the number of parasites and number of drivers; and
(3) for each element, its x and y coordinates, its resonant frequency FR, the
parameters X95, AX, /p, K, and A, and the magnitude and phase of its cur-
rent. The element-current program continues to compute until all initially
specified frequencies are satisfied.
This output file is an input file to other computer programs which are
designed to produce displays of interesting Yagi properties. One such pro-
gram calculates and plots the radiation patterns for the H-plane and the E-
plane. This is done by reading the element current file, and for a specified
sequence of angles (say every 1 or 2 degrees), computing the directivity from
(1.24). The radiation patterns are calculated for each frequency (read from
the new input file) and can be easily plotted by one of the plot routines. It
is convenient to also display the driving-point impedance, the peak direc-
tivity value found in the elevation angle search (usually in 1 degree inter-
vals), the angle at which it occurs, and the front-to-back ratio.
Another program of great utility is one which reads the element current
file and computes and displays a number of useful Yagi properties. These
are peak directivity and the elevation angle at which it occurs, reverse direc-
tivity at the same elevation angle, front-to-back ratio, and the driving-point
impedance. Since this program does not spend time displaying the complete
patterns, it is fast and therefore capable of rapidly running through a large
number of different situations.
Programs can be written that will perform equivalent calculations for differ-
1-14 Chapter 1

Fig. 1.4—Layout of a Yagi with a broadband feed system of the type

manufactured by KLM.

ent antenna systems. As an example, one can write a program to handle a

broadband drive system as used in the KLM antennas, but to do this requires
a little manipulation. In this type of antenna the drive system is not just a
single element but actually consists of a voltage-fed main driver in parallel
with a dependent driver through a crossed transmission line with characteristic
impedance Zp and phase delay angle ¢. With conventional transmission line
equations one can relate voltages and currents at both ends of this transmis-
sion line.
As an illustration, a five-element KLM-style beam can be simulated as
shown in Fig. 1.4. The parasites are elements 1, 2, and 3, the main or master
driver is 4, and the dependent or slave driver is 5 (connected through the
transposed transmission line). Start with the five linear equations:

TZ, + 1b Zy2 + 1g 213 + [4 Z14 + Is Z15 = O

Ty Zz, + Ip Z22 + 13 Z23 + Ig Zo4 + Is Zx5 = O
Ty Z3, + Ip Z32 + Ig Z33 + Iq Z3q + Is Z35 = O
I, Z4, + Ip Zq2 + 3 Z43 + [4 Zqq + Is Z45 = V4
I, Z5, + Ip Z52 + Ty Z53 + [4 Z5q + Is Z55 = V5 (1.29)

V4 (the main driver voltage) is given, but Vs (the dependent driver voltage)
must be determined from the transmission line equations which relate Vs,
Is, V4 and transmission line current at terminals xx. For a transposed loss-
less transmission line:

Vs = —V4secd — jZo Is tang (1.390)

 Performance Calculations 1-15

so that the final five linear equations become:

Ly Zi) + In Zi2 + 13 213 + 14 Z14 + Is Z15 = O

qh 22} iF bh 222 ot I 223 AF I 224 oP Ts 225 =) 0

rf Z3} + 6} Z32 + I; 233 ta 14 234 ae Ts 235 =n

Tye Zager Dy Zap Zas, + 1a Zagt Hs: Zas. = Ve

Ty Z5, + 12 Z52 + 13 Z53 + [4 Z5q + Is (Z55 + jZotang)
= —V4 seco (1.31)

Note that the input file for the antenna must contain information about the
number of parasites, the number of independent or master drivers, and the
number of dependent or slave drivers; the dependent or slave driver must
contain a statement as to which is its master driver, and must provide Zo
and ¢ for the transposed transmission line. The Z matrix and V vector must
be modified as shown, then the program can proceed as before.

Validation
With the tools for computing a wide spectrum of antenna characteristics
at hand, it is crucial to ask for some experimental validation. I have already
commented on the inaccuracies inherent in physical modeling (although such
inaccuracies are believed to be of little consequence), and have discussed ap-
proximations which have been used in various computations. These approx-
imations are expected to be most serious for those antenna properties which
depend critically on precise element currents, such as F/B ratio. On the other
hand, it should be possible to calculate other properties such as directivity
with reasonable accuracy (comparable to the accuracy of the currents them-
selves). The calculations are expected to produce superior results for short
antennas (few elements) and for frequency regions relatively close to the
design center frequency.

NBS Yagi Experiments

Experimental information suitable for validation is rather sparse. While
a great amount of experimental information has been published, it is difficult
to find examples of accurate measurements made under conditions where
external factors have been properly considered or eliminated.
For example, let’s examine the experimental results reported by Peter
Viezbicke in NBS Technical Note 688 (ref. 9). This publication provides a
rich range of gain and pattern measurements for a wide variety of Yagi con-
figurations. In most cases, all relevant dimensions of the Yagi design are
given, so this publication is a fine vehicle to test the validity of theoretical
computational methodology; shortly, I will attempt to make such com-
parisons.
I must first comment on the NBS experimental approach. Viezbicke states
that all tests on the (receiving) test antennas were carried out using a non-
 1-16 Chapter 1

conducting Plexiglas boom mounted three wavelengths (3\) above ground

(although readers of the NBS paper will note that the tirst figure shows 2)).
The (transmitting) generator antenna was mounted 320 meters (about 430h)
away and at a height above ground to illuminate the receiving antenna at
‘‘srazing angles.’’ I interpret this to mean that the transmitter is also at a
height over ground of 3d. The nature of the intervening ground, which is
highly relevant, is not described. A reference half-wavelength dipole is
mounted about 5) to the side of the antenna under test (also at 3\ above
ground). All tests were made using a frequency of ‘‘400 MHz”’’ but the fre-
quency precision is never mentioned, and this may be quite important, as
shall be seen.
If my interpretation of the experimental setup is correct, received field
strength will be the vectorial sum of the field from the direct ray and the
field from the ray reflected from the ground at a point midway between trans-
mitter and receiver. At grazing angles, the reflectivity of the ground is near
unity and for horizontally polarized radiation ground reflection will give a
phase change of 180 degrees. Thus, the two rays interfere and nearly cancel
each other, making the received energy nearly zero. It can be seen that the
actual received energy is extremely sensitive to the nature of ground midway
between transmitter and receiver, i.e., its true reflection coefficient, its height,
and its degree of flatness (which influences its focusing properties). Moreover,
since the reference antenna ‘‘sees’’ a different ground patch midway to the
transmitter, its inherent sensitivity can be different from that of the antenna
under test. For example, if the ground patch seen by the reference antenna
is only 1 foot higher than the ground patch seen by the test antenna, a sys-
tematic error of about 2 dB will occur. It is clear, therefore, that the NBS
experimental setup invites systematic errors.

Yagi Gain and Patterns

Let me turn now to the experimental results reported by Viezbicke. I will
start with the measured gain of a two-element (dipole and reflector) Yagi
as a function of spacing. Unfortunately, Viezbicke failed to state an essen-
tial dimension—the actual length of the reflector. On the assumption that
it was 0.482 (which was usually used for the other Yagis), I calculated the
theoretical result shown in Fig. 1.5. The theoretical gain is expressed in dBi;
a direct comparison can be made by adding 2.15 dB to the experimental NBS
findings.
The comparison in Fig. 1.5 is totally unsatisfactory; not only are the shapes
of the plots different, the peak gains are very different. At a spacing of 0.2,
Viezbicke measured 4.77 dBi, whereas I calculated 6.70 dBi. In an attempt
to resolve this conflict, I numerically summed and properly averaged the meas-
ured pattern of the same two-element Yagi (Viezbicke’s Fig. 13). I found
- the directivity, or isotropic gain, to be 6.71 dBi. Thus, the measured pattern
exhibits gains which agrees with my theoretical result. I can only conclude
 Performance Calculations 1-17

ira all |
nage
Poe
p+ NBS MEASUI eaten SI:

Pap
‘
(dBi)
GAIN

wet ele
Tea
RLaSezSa=
aR
tel
ay
|
an
ee Hse
Le
ateneed
faeect
[Botan
>
0.4
ELEMENT SPACING (A)

Fig. 1.5—Gain of a driven element and reflector for different spacings

between elements (from NBS Report 688, Fig. 1).

that some unexplained error was present for the NBS series of measurements
shown in their Fig. 1.
In the NBS paper a set of six specific Yagi designs is shown ranging in
overall boom length from 0.4 to 4.2 wavelengths. If you refer directly to
the NBS publication you will note that the first director for the 0.4 three-
element design is shown in their Fig. 9 to be 0.442 long rather than 0.424),
as shown in their Table 1. With this correction, the specifications for the
six NBS Yagi designs are listed here in Table 1.4.
I calculated the theoretical gains for these six Yagi designs, and also their
theoretical patterns for comparison with the measured patterns shown in the
NBS report. For convenience in making comparisons, the NBS patterns and
my calculated theoretical patterns are shown side by side in Figs. 1.6 through
1.12. It is apparent that for all cases there is striking similarity, not only in
the qualitative details of lobe structure, but also in most quantitative aspects.
Careful scrutiny of the experimental results show some variances between
the two halves; theoretically, of course, the two halves are totally symmetri-
cal. Agreement of this kind is gratifying and demonstrates that the experimen-
tal patterns were made with great care and also that the theoretical calculations
give valid answers. This is especially comforting since the 4.2\ Yagi design
is very long and contains many parasites—precisely the situation where the-
oretical approximations (cylindrical element resonant lengths, mutual, and
self-impedances) are most sensitive.
1-18 Chapter 1

Table 1.4
NBS optimized lengths of parasitic elements for Yagi antennas of six
different boom lengths (reflector spaced 0.2 wavelength behind the driven
element, element diameter 0.0085 wavelength).
Boom length (wavelength) 0.4 0.8 1.2 Pape 3.2 4.2
Number of elements 3 5 6 12 17 15
Reflector length (wavelength) 0.482 0.482 0.482 0.482 0.482 0.475
Director length (wavelength)
1st 0.442 0.428 0.428 0.432 0.428 0.424
2nd 0.424 0.420 0.415 0.420 0.424
3rd 0.428 0.420 0.407 0.407 0.420
4th 0.428 0.398 0.398 0.407
5th 0.390 0.394 0.403
6th 0.390 0.390 0.398
7th 0.390 0.386 0.394
8th 0.390 0.386 0.390
9th 0.398 0.386 0.390
10th 0.407 0.386 0.390
11th 0.386 0.390
12th 0.386 0.390
13th 0.386 0.390
14th 0.386
15th 0.386
Director spacing (wavelength) 0.200 0.200 0.250 0.200 0.200 0.308
Gain (dBd) 7.4 9.2 10.2 12.25 13.4 14.2
(dBi) 9.25 11.35 1235 14.40 15.55 16.35

NBS MEASUREMENTS COMPUTED

Se

FSts)
NORMALIZED
(dB)
RESPONSE

-50

180° 30° = 60s 90"—ss« 120° = IS0*_ =:180°

AZIMUTHAL ANGLE (DEGREES)

Fig. 1.6—Measured and computed E- and H-plane radiation patterns for a

two-element Yagi with a boom length of 0.2 wavelengths (from NBS
Report 688 Fig. 13).
 Performance Calculations 1-19

NBS MEASUREMENTS COMPUTED

(dB)
RESPONSE
NORMALIZED

120° 90° 60° 30°

AZIMUTHAL ANGLE (DEGREES)

Fig. 1.7—Measured and computed E- and H-plane radiation patterns for a

three-element Yagi with a boom length of 0.4 wavelength (from NBS
Report 688 Fig. 14).

NBS MEASUREMENTS COMPUTED

|
ea
1

<a
§ A ®

»i 9fe) a Le \ S

(dB)
SPONSE
Rees

NORMALIZED
RE.

-50
E PLANE ===
H PLANE ===

180° 150° 120° 90° 60° 30° o° 30° 60° 90° 120° 150° 180°

AZIMUTHAL ANGLE (DEGREES)

Fig. 1.8—Measured and computed E- and H-plane radiation patterns for a
five-element Yagi with a boom length of 0.8 wavelength (from NBS
Report 688 Fig. 15).
1-20 Chapter 1

NBS MEASUREMENTS COMPUTED

10

I 8

(dB)
RESPONSE
NORMALIZED

E PLANE ===
H PLANE ===

—60.65° Iso" 120° 90° 60° 30°

AZIMUTHAL ANGLE (DEGREES)
Fig. 1.9—Measured and computed E- and H-plane radiation patterns for a
six-element Yagi with a boom length of 1.2 wavelengths (from NBS
Report 688 Fig. 16).

NBS MEASUREMENTS COMPUTED

i}
8 aes
tS}

NORMALIZED
(dB)
RUSS
PPE

E PLANE
H PLANE == =e

180° 150° 120° 90° 60° 30° oe 30° 60° 90° 120°. 150° 180°

AZIMUTHAL ANGLE (DEGREES)

Fig. 1.10—Measured and computed E- and H-plane radiation patterns for
‘a twelve-element Yagi with a boom length of 2.2 wavelengths (from NBS
Report 688 Fig. 17).
 Performance Calculations 1-21

NBS MEASUREMENTS COMPUTED

i Wa,
MIVA ‘ae Baik
ALCON,
- ene
san ine
learn paral
g0° 150° 120° 90° 60° 30° o* 30° 60° 90° 120° 150° 180°
AZIMUTHAL ANGLE (DEGREES)
Fig. 1.11—Measured and computed E- and H-plane radiation patterns for
a seventeen-element Yagi with a boom length of 3.2 wavelengths (from
NBS Report 688 Fig. 18).

NBS MEASUREMENTS COMPUTED

(dB)

NORMALIZED
RESPONSE

E PLANE ———|
H PLANE ===

~©°\ao* 150" 120° 90° 60° 30° O° 30°60") §90° 1202515021802

AZIMUTHAL ANGLE (DEGREES)

Fig. 1.12—Measured and computed E- and H-plane radiation patterns for

a fifteen-element Yagi with a boom length of 4.2 wavelengths (from NBS
Report 688 Fig. 19).
1-22 Chapter 1

Table 1.5
Gain of seven different NBS-designed Yagi antennas determined by four
different methods.
calculated
NBS from
measure- half-power pattern computer
NBS Yagi ments beamwidth integration derived
(aBi) (aBi) (aBi) (ABi)
2 el. (0.2 wavelength) 4.8 7.5 6.7 6.7
3 el. (0.4 wavelength) 9.3 10.0 9.6 9:2
5 el. (0.8 wavelength) 11.4 11.9 11.4 10.7
6 el. (1.2 wavelength) 12.4 13.9 12.6 11.8
12 el. (2.2 wavelength) 14.4 15.3 14.3 14.0
17 el. (3.2 wavelength) 15.6 16.6 15.5 15.2
15 el. (4.2 wavelength) 16.4 17.4 16.2 1Oo7)

The gain of each of these Yagi designs can be obtained in several ways, and
it is illuminating to compare all methods. Table 1.5 shows a comparison of
three experimental methods all derived from the NBS data, and my theoretical
calculations. Column 1 shows the NBS measured gain converted to isotropic
gain (dBi). Column 2 shows the gain calculated from the measured half-power
main beam angles by the usual formula, dBi = 10 log (41253/6y6g), where
the H and E half-power angles are measured in degrees (Kraus, ref. 3,
page 25). The third column is derived entirely from the experimental NBS
patterns; in each case I calculated the directivity by appropriately summing
all 10 degree intervals. This pattern averaging, if carefully done, should yield
a reasonably reliable result, free from any systematic error (due to ground
reflections, for example). The last column is the result of my theoretical cal-
culations made on each design.
In Table 1.5, note that column 1 (NBS measurements) and column 4
(theory) are in agreement within Viezbicke’s estimated measurement accuracy
of about 0.5 dB with the exception of the value for the two-element Yagi
(discussed earlier). Column 2 is derived from measured half-power angles;
it is quite difficult to determine such angles with any great precision and,
moreover, the method is nothing more than a crude approximation at best.
Column 3 (derived from NBS experimental patterns) is in remarkable agree-
ment with theory in all cases, especially considering that the summation in
10 degree intervals is really too coarse and that values in these intervals were
‘*eyeballed’’ from the published NBS patterns.

Effect of Director Length

Fig. 7 in the NBS report shows an interesting experimental result: the
selected designs are superior in gain to simplified Yagis (all directors of equal
length) for booms longer than about one wavelength. Viezbicke did not give
the lengths of the directors used for the simplified Yagis; I made calculations
 Performance Calculations 1-23

Table 1.6
Measured and calculated gain of simplified Yagis with equal-length directors.
NBS
director measured computed
NBS Yagi length gain gain
(wavelength) (aBi) (Bi)
5 el. (0.8 wavelength) 0.4260 11.3 10.7
6 el. (1.2 wavelength) 0.4240 12.2 ialei,
12 el. (2.2 wavelength) 0.4017 13.9 13.6
17 el. (3.2 wavelength) 0.3946 14.8 14.7
15 el. (4.2 wavelength) 0.4008 15.6 15.2

using directors with lengths

that are the average of those
used in each of the five longer
Yagis. His measurements of
gain and my theoretical calcu-
Director Lengths
Different Constant
lations are shown in Tablefi
(dBi)
Gain
Theory 1.6. Again, these results are in
NBS 000 AAA :
satisfactory agreement.
Fig. 1.13 shows a graph of
my theoretical results for both
0 1 2 3 4 5 Tables 1.5 and 1.6, together
Overall lengtha f FS
with the published NBS ex-
Fig. 1.13—Gain vs. boom length for perimental points. Note that
equal-length and optimized-length the theoretical results support
directors. the idea that while the simpli-
fied (equal director length)
Yagi is just as good as a more
sophisticated design for booms shorter than one wavelength, a slight gain
improvement in the gain of long Yagis is apparently possible by using direc-
tors of different lengths.
Viezbicke’s Fig. 8, reproduced here as Fig. 1.14, shows the results of an
interesting experiment in which gain for a given Yagi design was measured
using a series of different director lengths, for each of eight element diameters.
These measurements allow an interesting test of the theory of the reactance
of cylindrical elements. In principle, a fiducial point on each gain curve, such
as the length to produce peak gain, should correspond to the same element
reactance for all eight cases. The reactance of the directors can be calculated
from (1.7). From Fig. 1.14 it is somewhat difficult to estimate accurately
the position of peak gain; therefore I instead estimated the (longer) director
length which drops the gain to 9 dBd. Table 1.7 shows the results. Note that
the computed reactance is essentially the same for all eight cases; the vari-
ances from the mean value of — 60.8 ohms are not systematic nor are they
1-24 Chapter 1

0.08 cm DIA

X72
(dB)
DIPOLE
TO
RELATIVE
GAIN
LIN. = 2.54 cm

10 Wl 12 13 14

LENGTH OF DIRECTORS (INCHES)

Fig. 1.14—-Measured gain vs. director length at 400 MHz for a Yagi 1.25
wavelengths long with three directors spaced 0.35 wavelength.

Table 1.7
Element lengths (estimated from Fig. 1.14) which produce 9 dBd gain, their
resonant lengths and their reactance. The average reactance is — 60.8 ohms
with standard deviation of 2.1 ohms.
element element
diameter length i Ip x;
(cm) (inches) (wavelength) (wavelength) (ohms)
0.08 13.34 0.45171 0.48139 — 63.6
0.16 13.22 0.44766 0.47896 — 58.9
0.32 13.00 0.43988 0.47579 — 58.3
0.64 12.58 0.42586 0.47147 — 62.2
0.95 12.31 0.41699 0.46824 - 62.3
127 12.09 0.40936 046535 — 62.0
1.59 11.91 0.40329 0.46272 — 60.8
2.54 11.43 0.38725 0.45565 — 58.0

large. The standard deviation corresponds to an element length variation of

less than 0.4 percent; such a variation could easily arise from length esti-
mates in Fig. 1.14 or even in physically constructing the elements used in
the experiment. Thus this experiment seems to give strong support for the
theory of cylinder reactance.

Gain Variations
Perhaps one of the most interesting experimental results of the NBS work
is the strange ‘‘oscillating’’ gain versus boom length characteristic shown in
 Performance Calculations 1-25

Viezbicke’s Figs. 4, 5, and 6. I

attempted to calculate a num-
ber of these cases and although
the oscillating gain phenome-
non does show up, the detailed
agreement between experiment
and theory is not impressive.
However, I found that the exact
behavior of these long and
(dBi)
GAIN
heavily (director) loaded arrays
is critically dependent on fre-
quency and/or exact director
lengths. Accordingly, I ran
DIRECTOR LENGTH =.0.382A
SPACING=[Link] calculations for several fre-
2
quencies around the central fre-
YAGI LENGTH(A) quency (normalized to unity).
The results are shown in Figs.
Fig. 1.15—Gain vs. boom length for 1.15 through 1.19, where the
small variations in normalized frequency theoretical calculated points for
(from NBS Report 688 Fig. 4).
each frequency are connected
by line segments for clarity.
Superimposed on the plots are
Viezbicke’s experimental
points.
The theoretical results gener-
ally show that the accurately
computed points do not really
lie on a smooth curve; this is the
reason connecting lines are
used. Theory does give the
oscillating gain phenomenon,
x) 4 but it also shows that the details
(dBi)
GAIN depend sharply on the exact
Q
transmitter frequency. Pretty
DIRECTOR LENGTH = 0.382A good agreement can be ob-
SPACING=0.20A
tained between experiment and
theory of it is assumed that
slightly different frequencies
3 4.5 6 (~ 1 to 3 percent) were used
YAGI LENGTH (A) from experiment to experiment.
It is perhaps significant that the
Fig. 1.16—Gain vs. boom length for most likely frequency does not
small variations in normalized frequency appear to be systematically high
(from NBS Report 688 Fig. 4). or systematically low.
1-26 Chapter 1

The theoretical results are in-

teresting from three points of
view. First, for a given frequen-
cy the calculated points do not
appear to lie on a smooth
curve; it is likely that the gain
(dBi)
GAIN

fAa ee agp.
varies somewhat with side- and
back-lobe structure. The calcu- ee
lated F/B ratio varies greatly
from point to point, and it PC aN
Me dlDIRECTOR LENGTH = 0.4110
would seem reasonable that SPACING = 0.10 A

there would be some reaction ! 2 3 4

YAGI LENGTH (A)
on forward gain. Second, for
long Yagis the results vary sig-
Fig. 1.17—Gain vs. boom length for
nificantly with the exact fre- small variations in normalized frequency
quency used. Third, detailed (from NBS Report 688 Fig. 5).
calculations show that beyond
the end of the first ‘‘oscillat-
ing’’ gain cycle, Yagi gain is not
maximum in the plane of the
Yagi; instead, the front beam
appears to develop a central
dimple causing maximum gain
to occur at an appreciable ele-
vation angle. (The plots in Figs.
1.15 through 1.19, however, (dBi)
GAIN
show only the gain at zero ele-
vation angle because this is
what was presumably done in
the NBS experiments.) _
Note that all three of these
behavioral aspects of long 1.5 3 4.5 6
YAGI LENGTH (A)
Yagis appeared from theoreti-
cal computations; they were not Fig. 1.18—Gain vs. boom length for
revealed by the NBS experi- small variations in normalized frequency
ments. This illustrates the point (from NBS Report 688 Fig. 5).
that an extensive variety of Yagi
configurations can be explored
much faster theoretically, and
with much more relative accuracy, than can be accomplished experimentally.
As a result, new concepts and understanding emerge.

Comparison Summary
Let me summarize the overall comparison of the experimental results in
 Performance Calculations 1-27

the NBS report with the theoretical results. The comparisons show total agree-
ment on Yagi gain for an astonishing range of models with the single excep-
tion of the direct measurements
on atwo-element beam. A gain
figure derived from the meas-
urement of the pattern of the
two-element beam, however,
does agree with theory; it is my
(dBi)
GAIN
belief that an error was made
by the NBS group in their
measurements.
The comparisons also show
excellent agreement for Yagi
DIRECTOR LENGTH =0.4242 patterns, again over a wide
SPACING=0.20A °
. ; : Ae , range of models. This should be
YAGI LENGTH (A) a very sensitive test of the
Fig. 1.19—Gain vs. boom length for accuracy with which mutual
small variations in normalized frequency and _ self-impedances are
(from NBS Report 688 Fig. 6). represented, as well as the
resonant frequency calculations
of cylindrical elements. This latter point is also strongly supported by the
consistent values of parasitic element reactance for elements whose diameters
vary over a range of greater than 30 to one.
Finally, the strange ‘‘oscillating’’ gain phenomenon observed by Viezbicke
can be reproduced theoretically. Agreement is only qualitative if the NBS
group used an accurate 400.0 MHz frequency in their tests; however, the
comparisons suggest that they may have actually used a nominal ‘‘400 MHz,”’
with frequencies within a range of + 3 percent. In this event, the compari-
sons indicate a potentially remarkable agreement in quantitative aspects as
well.
Since these comparisons include very long Yagi models and correspond-
ingly large numbers of parasitic elements, it seems certain that the computa-
tional methodology I have outlined should be generally trustworthy. This
should be especially true for shorter antennas (with fewer elements) with which
I am primarily concerned throughout this book. Especially noteworthy is
the fact that all computed results contain no adjustable constants or
parameters; they are all derived from basic physical principles using ade-
quately accurate mathematical approximations.
1-28 Chapter 1

NOTES:
1Yagi, H. and Uda, S., Proceedings of the Imperial Academy, February, 1926. Also
Uda, S., Journal of the Institute of Electrical Engineers of Japan, Volume 47-48,
1927-1928.
2Uda, S. and Mushiake, Y., Yagi-Uda Antenna, Research Institute of Electrical Com-
munication, Tohoku University, Sendai, Japan: Sasaki Ltd., 1954.
3Kraus, John D., Antennas, New York: McGraw-Hill, 1950.
4King, Ronald W. P., Theory of Linear Antennas, Cambridge, Massachusetts:
Harvard University Press, 1956.
5Walkinshaw, W., “‘Treatment of Short Yagi Aerials,’ Journal of the IEE (London),
Volume 93, Part 3A, Number 3, 1946.
6Ehrenspeck, H. and Poehler, H., ‘“A New Method of Obtaining Maximum Gain
from Yagi Antennas,” /RE Transactions on Antennas and Propagation, October,
1959.
7Lindsay, J., ‘“Quads and Yagis,’’ QST, May, 1968.
8Greenblum, C., “‘Notes on the Development of Yagi Antennas,’’ QS7, Part 1,
_ August, 1956; Part 2, September, 1956.
8Viezbicke, P., ““Yagi Antenna Design,’’ NBS Technical Note 688, U.S. Depart-
ment of Commerce, Washington, D.C., December, 1976.
10“‘|EEE Standard Test Procedures for Antennas, IEEE Std 149-1979,” The Insti-
tute of Electrical and Electronic Engineers, Inc., December 1979, page 94.
11Hallen, E., “Theoretical Investigations into the Transmitting and Receiving Qual-
ities of Antennae,’’ Nova Acta Uppsala (Sweden), Series IV, Volume 11, Number
4, 1938.
12Any improvement in these approximations for self- and mutual impedances will
require a great amount of theoretical work through a rigorous examination of
the boundary value problem with attention to (1) current distributions along driven
and parasitic elements (they are somewhat different in principle), and (2) com-
plete numerical solutions to both real and imaginary components of element
self and mutual impedance. It will be necessary to distinguish mutual coupling
coefficients between elements of different function; i.e., driver-to-driver, driver-
to-parasite, and parasite-to-parasite. In principle, all coefficients will depend on
each affected element length and radius.
13Carter, P., ‘‘Circuit Relations in Radiating Systems and Application to Antenna
Problems,”’ Proceedings of the IRE, 20, June, 1932.
14Mutual impedance can also be calculated directly from the exponential integral
equations—see Kraus, ref. 3, page 265.
CHAPTER 2

SIMPLE YAGI ANTENNAS

Yagi antenna can be characterized by one or more driver elements
A and a number of parasite elements, all supported on a boom. For
each element we must specify x and y coordinates, a length, and a
radius, all measured in terms of wavelength at the design frequency. In the
case of each driver we must also specify the excitation potential or current.
It is instantly apparent that with all these variables an exhaustive investi-
gation into all possible configurations is impractical. Instead I shall begin
with an initial consideration of simplistic Yagi antennas and will subsequently
discuss a variety of departures from the basic design. This class of Yagi
antennas has (1) a single driven element; (2) one reflector; (3) zero or more
directors with identical lengths and diameters; and (4) all elements uniformly
spaced along the boom. Additionally, the antenna will be in free space.
These restrictions may seem at first sight to be quite severe, but I hasten
to remark that free space performance relates to actual performance over
ground and that simplistic Yagi antennas, as defined above, can, in many
instances, provide performance levels fully as good as those from more
sophisticated designs. Furthermore, we can learn a great deal about Yagi
antenna performance from studying these simple designs and, as we shall
see, will develop useful conceptual ideas about Yagi behavior and ideas for
“‘best’’ design.
Throughout this investigation of simplistic antennas I will choose element
dimensions (radii) characteristic of ‘‘normal’’ 14 MHz construction— a =
0.000526 corresponding to 7/8 inches diameter. The results can be translated
to any other element dimension by proper scaling calculations; scaling rules
are given in Chapter 7.

Two-Element Beams
I shall begin with a two-element Yagi beam involving one parasite which
can act either as a reflector (for frequencies above its resonance) or as a direc-
tor. For such a beam there are only two fundamental variables, the physical
separation of the two elements along the boom and the physical length of
the parasite. The exact length of the driven element is of little consequence
as far as gain and pattern are concerned; it does, however, affect driving-
point impedance (especially reactance) which is considered later. Since we
are interested in a frequency-swept plot of the gain and F/B properties, the
physical length of the parasite can be fixed. As the frequency increases from
well below to well above the parasite free space resonant frequency we can
observe the properties of the beam first where the parasite behaves as a direc-
tor and then as a reflector.
2-2 Chapter 2

The computation methodology is that given explicitly in Chapter 1. In all

cases of the two-element beam the element lengths are 0.48167) for both para-
site and driver; this makes each element’s normalized free space resonant
frequency FR equal to unity. The normalized frequency, F, is varied in steps
from about 0.90 to 1.10; these steps were made sufficiently small to show
the detailed behavior of the beam. Element separation, s, measured in
wavelengths at the design frequency, is varied from 0.025) to 0.5, again
in steps sufficiently small to bring out essential behavior.
Fig. 2.1 shows the frequency-swept gain and front-to-back ratio of the two-
element Yagi antenna for several element spacings, where the gain is posi-

(dBi)
GAIN

0.100 8 0.300 12 0.600 FREQUENCY (%)

Fig. 2.1—Frequency-swept gain and front-to-back ratio of two-element
Yagi antennas with different element spacings, where the parasitic
element acts as a reflector.

FREQUENCY (%)

Curve Ss s Curve s
1 0.025 5 0.150 9 0.350
2 0.050 6 0.200 10 0.400
3 0.075 7 0.250 11 0.500
4 0.100 8 0.300 12 0.600 FREQUENCY (%)

Fig. 2.2—Frequency-swept gain and front-to-back ratio of two-element

Yagi antennas with different element spacings, where the parasitic
element acts as a director.
 Simple Yagi Antennas 2-3

tive in the direction of parasite towards driver (that is, where the parasite
acts like a reflector). Each curve represents a particular spacing as keyed in
the legend. Fig. 2.2, in the same format, shows the gain and F/B where the
parasite acts like a director, i.e., where the gain is positive in the direction
of driver towards parasite.

F = |.000

Merits
ea a REFLECTOR

Fig. 2.3—Gain of two element

Yagis at the design frequency
(F = 1) as a function of element
spacing.
(dBi)
GAIN

SPACING (A)

Examination of these performance plots, together with additional

information on computed driver input impedance, reveals a number of
interesting facets of the behavior of two-element Yagi beams. Fig. 2.3 shows
a plot of the gain at F = 1 as a function of element spacing. This is similar
to but not identical with a plot shown on page 147 of The ARRL Antenna
Book.’ It is possible that the differences, particularly at small spacings, are
due to greater precision in the new calculations; in any case, I believe the
implications of this plot are somewhat misleading.
One can easily see that the maximum gains obtainable at the ‘‘best’’
frequency(s) in Figs. 2.1 and 2.2 look somewhat different. These are shown
in Figs. 2.4 and 2.5, where for reference the curve for F = 1.0 is also shown.
Note that the obtainable gain does not depend greatly on whether the parasite
is a reflector or a director as implied in Fig. 2.3; moreover, the largest gain
is obtained at very small spacings. This is a result which is not intuitive.
Figs. 2.4 and 2.5 can be compared with the early analysis by Brown, and
there appears to be good agreement.” However, the result shown on page
146 of The ARRL Antenna Book which cites Brown’s analysis is somewhat
different. The fall off in maximum gain for low spacings shown by the ARRL

1This and other footnotes are given at the end of this chapter.
2-4 Chapter 2

10

(dBi)
GAIN (dBi)
GAIN
s

Bi
3 4
Eccecenmal

4 i
ais
Sal
|
t
SPACING (Xe) SPACING (Xo)

Fig. 2.4—Gain of two-element Fig. 2.5—Gain of two-element

Yagis vs. element spacing; Yagis vs. element spacing;
parasite a reflector. parasite a director.

reference does not appear to agree with Brown nor does it substantiate the
calculations I have made (Figs. 2.4 and 2.5).
If one examines the maximum gain shown in Figs. 2.1 and 2.2 for best
frequency and corresponding (driver) driving-point impedance, one obtains
the values shown in Table 2.1. A plot of R is shown in Fig. 2.6 which can
be compared with a similar diagram shown on page 147 of The ARRL
Antenna Book; except for low values of spacing, agreement is fairly
satisfactory.
One can see from Table 2.1 and Fig. 2.6 that element spacing affects

Table 2.1
Maximum gain and feed-point impedance of a two-element Yagi for various
element spacings.
parasite a reflector parasite a director
max max :
Ss gain at R Xx gain at R X
() (ABi) (freq) (2) (Q) Q (aBi) (freq) () (Q) Q
0.025 7.24 1.036 Ubi 11.2 925 7.41 1.026 1.1. -—10.0 1003
0.050 UC Oe) 4:299920.49 237 7.46 1.0014 4.1 -17.6 266
0.075 7-i Git 030 9.7 31.8 99 7.42 1.0000 98 -29.9 105
0.100 alee 02D ee.6 oD COO 7.30 0.990 15.3 -—34.1 63
0.150 6.96 1.015 29.6 38.6 29 6.80 0.970 27.5 -44.2 31
0.200 6.72 1.005° 44.6 34.7 «7 6.12 0.955 38.0 -49.4 21
0.250 6:49 (9-000) 6114 SiGe 1 5.38 0.940 48.2 -59.4 15
0.300 6.00 0.990 72.7 16.9 8 4.72 0.920 57.2 -805 12
0.350 5.49 0.980 80.5 -0.4 7. 4.16 0.900 64.2 -104.2 11
0.400 4.91 0.960 82.0 -29.2 7 3.69 0.900 70.7 -—103.3 9
 Simple Yagi Antennas 2-5

Fig. 2.6—Driving-point resistance

of two-element Yagis vs. element
spacing.

) J 2 3 4 &
SPACING (Xo)

driving-point resistance, and therefore Q, over a very large range. This factor,
as well as the gain curves shown in Figs. 2.1 and 2.2 set a practical limit to
the achievable gain over a desirable bandwidth, for example, perhaps four
percent in F. Moreover, the higher values of (radiation) loaded Q cause larger
circuit resistive losses and therefore lower the antenna efficiency. Thus, in
practice, really short booms are not desirable; one must choose between
efficiency and bandwidth on the one hand, and gain and F/B ratio on the
other.
Long booms, however, also appear undesirable because gain really falls
off (primarily due to reduced excitation of the parasite). Furthermore, for
booms longer than 0.3\ a new phenomenon can be seen from a detailed
computational analysis (not shown here). The front lobe of radiation begins
to ‘‘dimple’’ in the forward direction, resulting in a pattern where the gain
maximum occurs at an elevation angle other than zero with respect to the
boom direction. (The gain shown in Figs. 2.1 and 2.2, however, is just the
directivity in the direction of the boom.) This pattern feature was predicted
by Brown (ref. 2) and shown in Kraus, page 294.°
Note that the two-element Yagi gives respectable performance in gain for
a wide range of element separations. However, the F/B figures are not
especially impressive; moreover best F/B does not occur at the same frequency
as best gain. Thus, in designing a two-element Yagi beam a practical
compromise is necessary. If you wish to obtain good gain with at least a fair
F/B ratio over a bandwidth of say four percent, you can determine by
inspection of Figs. 2.1 and 2.2 and Table 2.1 that a two-element beam should
have a boom length of perhaps 0.15 wavelengths. For such a boom the gain
is essentially independent of whether the parasite is a reflector or a director;
moreover, the F/B is about equivalent for either situation.
To move the peak of a gain curve in Figs. 2.1 and 2.2 to the design
frequency, the parasite length is adjusted commensurately; to reduce design
frequency driver reactance, the driver length is adjusted. These character-
2-6 Chapter 2

,
g

Fig. 2.7—Gain, front-to-

GAIN
(dBi)
&\o 8F/B
&3 (OHMS)
(48) R,X
& \ ry
back ratio and feed-
point impedance as a
IN function of frequency
for a two-element Yagi
\ A eam
ENE
ES
with 0.15 wavelength
spacing (parasite a
=
Besa.
Sra =
ola
DRIVER LE = 0.470502
R, = 0,0005260A
reflector).

PARASITE LE= 0.493662.

R, = 0.0005260.

6 30 ISO

4 20 100

q ruts Fig. 2.8—Gain, front-to-

Bt otlko back ratio and feed-
point impedance as a
function of frequency
KC in of for a two-element Yagi
with 0.15 wavelength
spacing (parasite a
TT director).
DRIVER a ea
005260A
Fe ReA00 buaaaeh
= 0.0005260A

FREQUENCY (%)

istics of two-element beams are shown in Figs. 2.7 and 2.8 with a frequency-
swept plot of each design. Note that each of these figures shows gain, F/B,
R and X of the driver. They illustrate the kind of design compromises which
must be made. They also show ay frequency-swept behavior of the main
performance parameters.
The ‘‘best’’ design frequency is a matter of choice and is a compromise
 Simple Yagi Antennas 2-7

between gain and F/B ratio; it is adjusted by the length of the parasite. The
driving-point resistance and reactance vary significantly with frequency. Note
that you cannot generally specify resistance except at a single frequency; also
note that feed-point reactance is not a linear function of frequency. This is
caused by the combined effect of self- and mutual impedance of the elements.
One can adjust the frequency of the zero reactance point by adjusting the
length of the driver; this has been done only approximately in Figs. 2.7 and
2.8. However, note that the adjusted driver lengths are quite different than
the length for a single isolated resonant dipole in free space. These differences
are again caused by mutual reactance coupled into the driver by the parasite.

More Than Two Elements

I now turn to an analysis of Table 2.2
simplistic Yagi antennas
j having more Range of Yagi parameter variations.
than one parasite. In all cases there
; : parameter range
is only one reflector, all directors
; ; Number of elements 3,4,5,6,7
have identical lengths, and all Boom length up to 1.5)
elements are uniformly spaced along Reflector resonant frequency 0.98 to 0.93
:
the boom. I shall display ;
for clarity Diirector resonant frequency d
1.02 :
to 1.07

only the essential frequency-swept

gain and frequency-swept F/B be-
haviors; the driving-point impedances, all of which were computed, are of
secondary interest at this point. For these Yagi antennas there are a number
of parameters which should be systematically explored. Table 2.2 shows these
parameters and the range over which each has been varied. To display results
in a consistent way I have chosen for each frequency-swept plot a fixed
number of elements and a fixed overall boom length, /p, measured in
wavelengths at the design frequency. On each plot there are six numbered
curves; each number designates a particular parasite ‘‘tuning’’ combination;
these combinations are shown in Table 2.3. The lengths and free space
resonant frequencies of parasites are shown for each numbered combination.
6:

Table 2.3
List of parasitic lengths and resonant frequencies for the six numbered
curves on Figs. 2.9 to 2.11.

reflector director
curve length resonance length resonance
() 0)
1 0.49150 0.98 0.47223 1.02
2 0.49657 0.97 0.46764 1.03
3 0.50174 0.96 0.46314 1.04
4 0.50702 0.95 0.45873 1.05
5 0.51241 0.94 0.45441 1.06
6 0.51792 0.93 0.45016 1.07
2-8 Chapter 2

(dBi)
GAIN

(dBi)
GAIN

eNO ib edu
Pea aoa 72Sede)
| eae, |
Sanivaneyy
ea ee i |lei
TaN
SRHHHERRNGe eee)
aT CAA
A) * Oa
CT \\\ ee)
ai
FREQUENCY (%) FREQUENCY (%)

Fig. 2.9—Gain and front-to-back ratio for three-element Yagi beams with
different boom lengths and parasitic element lengths (See Table 2.3).
 Simple Yagi Antennas 2-9

Fe
(dBi)
GAIN IE NEL ENS |
7,
Oks SORA AE)
We
90 100 110 90 100 110
FREQUENCY (%)

aeLe
Te 0.35, BOOM

Pa SER
(dBi)
GAIN
‘al ra | a |
ie arg
A@GREERA\\\ ny
a a a PGA NGG
Faea STAs
FREQUENCY (%) ag

(dBi)
GAIN

MN
A
ia ry | |Rss
a
=H9°
BENG (%) mE: (%)

Fig. 2.9 (continued)—Gain and front-to-back ratio for three-element Yagi

beams with different boom lengths and parasitic element lengths (See
Table 2.3).
2-10 Chapter 2

: Geimikiel
AMP cssNIE)
Pe AeRc hie
P77 a
s7ULIL i LI a,
MMV MYA rales aaa
CEERSECT) ~ Baby kth
cM St
Ae em Lt tt %0 | Pei | |
FREQUENCY (%) FREQUENCY (%)

oar. oom | | | |
fs] el
Pe eA
ES eR | | Ll
= Gs Mi lI
z 44 ee a ese aiete
=
S BI PAR SUREead
fl Bo EN Ee (dB)
F/B

PA Pa ERAN a]
| tt TL
AY I
8
FA
amee wtbeaah
fo
il
blc
leelt
alee
es
soci
sisi
ick

Ey
i
Zi ce
=
Ej gi
a cooly
a
AS)

CC ee
=
S
cl|eSoon)
fa SS
(dB)
F/B

COTA
CS
90 100 fo}
FREQUENCY (%) FREQUENCY (%)

Fig. 2.9 (continued)—Gain and front-to-back ratio for three-element Yagi

beams with different boom lengths and parasitic element lengths (See
Table 2.3).
 Simple Yagi Antennas 2-11

The curves of Fig. 2.9 show the results for three-element beams as boom
lengths are varied from 0.100 to 0.700 wavelengths. It is apparent from an
inspection of these plots that the performance is superior to that of the
two-element beams; this is especially true in the F/B ratio. As one increases
boom length the maximum gain increases (unlike that for two-element beams);
the F/B increases spectacularly, then decreases again. For this class of Yagi
antennas there seems to be a best boom length; we shall see this kind of result
for all of the simplistic Yagis and the physical explanation will soon be
apparent.
Note that the chief parameter controlling the bandwidth over which gain
remains high is the (resonant) frequency separation of reflector and director;
this observation will also prove to be generally true for all simplistic Yagi
antennas. The bandwidth of the F/B performance (when the F/B is very high)
is small; this is due to the critical nature of low back radiation. Back radiation
is very low only when there is vectorial cancellation of field in the back
direction; this comes about only where element complex currents are
accidentally favorable for such cancellation. When this happens, very small
changes in those currents, e.g. shifting frequency slightly, will destroy the
favorable vectorial cancellation. This general result is inherent in all Yagi
antennas; if the F/B is exceptionally high, it will be so only over a very narrow
frequency band.
Similar results for four-element simplistic beams are shown in Fig. 2.10,
and for six-element simplistic beams in Fig. 2.11. A quick inspection of these
results shows the increasing complexity of the frequency-swept plots with
number of elements; this is caused primarily by the larger number of
resonances in the system. The gain ‘‘cutoff’’ at high frequencies is also
increasingly abrupt due no doubt to the large number of directors (which
shift over to ‘‘reflectors’’ at high frequencies). F/B curves become very
complex. Moreover, really high values of F/B (greater than 30 dB) are quite
rare; it is very unusual to find combinations where vectorial cancellation in
the rear direction is nearly complete.
The total range of results shows a number of characteristics of interest.
First, the bandwidth over which gain is high is determined primarily by the
frequency spread between the reflector and the director(s). Second, the shape
of the gain curve is generally not flat in the region of interest; indeed it may
be sloped and/or humped or dished. Usually the slope favors the higher
frequencies. Third, the shape of the gain curve is more complex where the
number of elements is large, and the shape of the F/B ratio varies
enormously—much more than the shape of the associated gain curve. It may
show more than one peak; moreover, the peak structure shifts very rapidly
with boom length. The height of the peak does not necessarily seem to vary
monotonically with the frequency separation of reflector and director(s). Very
high F/B values (greater than 30 dB) are quite rare and when present are
invariably very narrow-banded. In addition, the frequency bandwidth of the
F/B parameter is undefinable because of the extreme variation in shape.
2-12 Chapter 2

fonereeer [TT
fe a ee
Saeeapr' perso TTT TT
_. RE elas leapt ae aa
aCe Cea) COCO
‘ECE Te $C a
ESNNi Ont © CCC ATTANN
NN LT 2 TS
Tit) SRT
|“ARBSo
FREQUENCY (%) FREQUENCY (%)

(dBi)
GAIN

0.25. BOOM

& 4
Nypistedecile
hep a
ANAL EET TRE |
JASCO)
SVANGTT | | UT
Sait 7S)
| ree
i el
FREQUENCY (%) FREQUENCY (%)

Fig. 2.10—Gain and front-to-back ratio for four-element Yagi beams with
different boom lengths and parasitic element lengths (See Table 2.3).
 Simple Yagi Antennas 2-13

(dBi)
GAIN
(dB)
F/B

(dBi)
GAIN

Ladne aS
HATER
GAIN4
ARAL TL
RPE THHHe
(dBi)

eee
MV/1 0.5. BOOM

Ho
Bam (%) FREQUENCY (%)

Fig. 2.10 (continued)—Gain and front-to-back ratio for four-element Yagi

beams with different boom lengths and parasitic element lengths (See
Table 2.3).
2-14 Chapter 2

| eon |
10

eee tt 40

eal
2 ag
FH
(dBi)
GAIN
aia We
fzzeae” Gee
(dB)
F/B

Bor am
PS cama\an
FREQUENCY (%)

an
SIE
BEI
(dBi)
GAIN
Bau
E>
(dB)
F/B

a
ES |

a
(dB)
F/B

CEE
rarer, ecm | |WiLL
90 [o-e)
FREQUENCY (%) FREQUENCY (%)

Fig. 2.10 (continued)—Gain and front-to-back ratio for four-element Yagi beams
with different boom lengths and parasitic element lengths (See Table 2.3).
 Simple Yagi Antennas 2-15

(dBi)
GAIN
(dB)
F/B

ye
\\
At
(dBi)
GAIN Ny [|
abel|
WH CRER caEN eee
LeCR Ee he
po-3SarzOOM, LabelSalI Lo
90 100 10
FREQUENCY (%) FREQUENCY (%)

reece TT TT
its Fadla
ia| fc Pn Pa
REE EAE
Es tk |
(dBi)
GAIN
(dB)
F/B

F550 oT EE NAY
100
FREQUENCY (%) FREQUENCY (%)

Fig. 2.10 (continued)—Gain and front-to-back ratio for four-element Yagi beams
with different boom lengths and parasitic element lengths (See Table 2.3).
2-16 Chapter 2

ey peterTTT
lope intel ane ee
“CoE
ee
Wer
bee | “ei keeTTT
3 CLAem) JCC
Geo
FSRR MeeiIMM) “CofC
e COAL
nS, Cm CCC Wal
OMT CNC) © eae
CO) “ERISA
COTO = Coss
FREQUENCY (%) FREQUENCY (%)

Ppeov ereoodae PANS ie ee

Pi aa ES Pi bel
ee (as a Sa
tae VET CRA Peeteot
TT TTT
PPFFPI alia Biaie
(dBi)
GAIN

Bebe
TENTH SAT
ASL) A
LAAT [| daleale
NAN TY
eReOUENEY (%) Petey (%)

Fig. 2.11—Gain and front-to-back ratio for six-element Yagi beams with
different boom lengths and parasitic element lengths (See Table 2.3).
 Simple Yagi Antennas 2-17

dalle)
CLOPTRS (ossiabod[TTI
a ee rere = sar alaeeH ereLa
: lpia
roeataLa
Ret
Hien
SOA
. aust
Mita\| (%)

0.30 ».» BOOM

Ea
fe
Ei
=F
faa
feel
fil
al
(dBi)
GAIN

lid =p eee
See
eke
ae
a
lal
a ad =
§ ee
FS 3 °
8 3 ko} S
prettency (%) FREQUENCY (%)

Fig. 2.11 (continued)—Gain and front-to-back ratio for six-element Yagi

beams with different boom lengths and parasitic element lengths (See
Table 2.3).
2-18 Chapter 2

odo tab | 1 11
i UR alFi dG aml2 Lad
Creer man elie
10

TPP
Take twa
Te aru Lalivva alo
Sees UeeRe
wT TT
(dBi)
GAIN

WS LLU HY
NAS HEE

FREQUENCY (%)

ogo tabed | |[111

ee
Sse| ||
SHAE 50)»BoM. es mele
itil
| Tay PET TPES Ie
(dBi)
GAIN
-

aa Sti
B07 Meee
FREQUENCY (%) FREQUENCY (%)

Fig. 2.11 (continued)—Gain and front-to-back ratio for six-element Yagi

beams with different boom lengths and parasitic element lengths (See
Table 2.3).
 Simple Yagi Antennas 2-19

Rot aa eee
tl Ei AI A
(dBi)
GAIN Oa aS

FREQUENCY (%)

fore Lebo[TT
[iiuiee-
10

CEPT AST
He IS
(dBi)
GAIN

CAS ae
ABR ERRER IIE
(SAS Se 0
even (%) FREQUENCY (%)

Fig. 2.11 (continued)—Gain and front-to-back ratio for six-element Yagi

beams with different boom lengths and parasitic element lengths (See
Table 2.3).
2-20 Chapter 2

0.75 \» BOOMEa ee
a aki
(aBi)
GAIN Faa VS a GL
Pea GL
ak) ie
TaSt
Pin?
Bi am 10
FREQUENCY (%)

(dBi)
GAIN
~ MBE ee)
RPO Ne liek VE TG
1.00, BOOM lit
me
ee A LL
AAA
FREQUENCY (%) FREQUENCY (%)

Fig. 2.11 (continued)—Gainand front-to-back ratio for six-element Yagi

beams with different boom lengths and parasitic element lengths (See
Table 2.3).
 "SECs
Simple Yagi Antennas 2-21
LI]
TASRSe
SIS Nal
cists"
sal
al Sie|
(rh NIVS

shales]
SSE
SSoE
Ek
s

en (%) FREQUENCY (%)

CY (sel 20 TET TT,
eINeileSanaa
(!2 NIVD

lag
eek
IE
Le
0
ag)
Eee
Se
| lees
iaeel
|aS
Se
eS
eee (%) FREQUENCY (%)

Fig. 2.11 (continued)—Gain and front-to-back ratio for six-element Yagi

beams with different boom lengths and parasitic element lengths (See
Table 2.3).
2-22 Chapter 2

Performance Characteristics
If we look carefully at one of these plots, for example, Fig. 2.9 for three
elements, boom = 0.25), it becomes clear that it is quite difficult to simply
characterize ‘‘the’’ gain and ‘‘the’’ F/B ratio. The maximum calculated gain
at a single frequency is 8.9 dBi (curve 1) but a realistic gain at the center
of a practical four percent band (curve 3) is more like 8.0 dBi. Even more
difficult is the characterization of the F/B ratio. The maximum calculated
F/B (curve 5) is a whopping 38 dB, but this occurs only at a very specific
frequency (F = 1) and for the situation where maximum gain is compromised
(reduced to 7.3 dBi). How then can we characterize the results by a single
gain figure and a meaningful F/B ratio?
Since gain is perhaps the most important parameter of antenna per-
formance, and since a practical antenna must work effectively over a
reasonable band, I have elected to specify the gain at the center of a four
percent band. For each case, the band-center is adjusted to give maximum
gain performance over the entire four percent band, and, finally the specific
curve is selected which yields best overall gain performance. I define ‘‘the’’
gain of this case as the gain at band-center and ‘‘the’’ F/B ratio as the value
at the same band-center for the same selected case. Note that the actual F/B
may be significantly higher at some other frequency inside or outside the
chosen band; we shall discuss this point shortly.

Table 2.4
Band-center gain and front to back ratio vs. boom length for various multi-
element Yagi beams. Data for three, four and six elements are plotted in
Figs. 2.9, 2.10 and 2.11.
Three Four Five Six Seven
elements elements elements elements elements

length gain F/B gain F/B gain F/B gain F/B gain F/B
(\) (ABi) (dB) (aBi) (dB) (OBi) (dB) (ABi) (dB) (aBi) (dB)
0.10 6.98 12.2 (22 mien 7.48 8.4 7.49 7.1 7.49 6.7
0.20 7.96 14.1 7.90 5.5 7.88 US 7.76 5.6 7.09 4.3
0.25 8.23 KEES 8.47 8.1 SB260M 8.26 7.8 7.92. 5.1
0.30 9.07 12.3 8.80 Tht 8.64 9:5 8.60 8.0 8.35 6.6

0.40 9.73 8.2 9.42 8.8 9:37 10.8 9.15 9.3 9.29 9.6
0.50 9.42 5:6 9 Gr 8.9 O15 SO. 9.64 11.0 S630 1S
0.60 8.88 5.3 9/805 11-8 9.81 15.4 9199 eal oat O96 8 135
0.70 7.92 2.3 10:08 551535 10.41. 21.6 10.55 21.2 10.62 16.1
0.75 10.51 29.4 1OLGS 1657, NOG 20 727, 10.82 13.5
0.80 10.36 223 O37 Set oLo) 11.02 14.2 W107 5 15:5
0.90 10.53 14.4 10.95 15.0 eer IPS Ale (eee
1.00 10.33 8.4 10.87 11.4 11.16 10.4 11.26 10.5
1.10 9.70 4.7 LOST HTH 11.09 13.4 11.40 13.1
1.20 10.40 15.1 Al51 249 Id Sees
a\.20 10.74 24.0 11.79" 23.8 U.O7 seis
1.30 10.49 20.3 11-62 ro i5 12.175 4 20:5
1.40 10.51 18.9 11:87 5 513.8 12:22). 15:6
1.50 10.33 12.3 A 1cOilgaae ative 12.10 10.8
 Simple Yagi Antennas 2-23

(dBi)
GAIN

BOOM LENGTH (A)

Fig. 2.12—Yagji gain for three-, four-, five-, six-, and seven-element
beams as a function of boom length.

With this definition of band-center gain and band-center F/B, Table 2.4
has been constructed to show performance not only for three-, four-, and
six-element beams, but also for five- and seven-element beams. Fig. 2.12
shows a plot of the gain information; this graph is remarkable in four respects.
First of all, it demonstrates a practical upper limit to the gain achievable
from a given boom length. Second, it demonstrates that this gain is almost
independent of the number of elements distributed along its length as long
as there are enough. Third, the achievable practical gain shows a slight
preference for more rather than fewer elements on a boom. Finally, the
‘*‘bhoom gain’’—achievable gain from a given boom length—is not really a
smooth function of boom length. Instead, it appears to exhibit ‘‘bumps’’
or oscillations with a fraction of a decibel amplitude and spacing of about
a half-wavelength.
This concept of boom gain, independent of the number of elements is not
new; in fact, it was suggested by Ehrenspeck and Poehler in a series of
experiments using the automatic plotter built at the Air Force Cambridge
Research Center.* No claims by Ehrenspeck and Poehler were made as to
the absolute accuracy of their results, but they were able to demonstrate
essential independence of gain on number of elements over rather wide limits
for two long Yagi models (1.2 and 6.0d). If one accepts the idea of universal
boom gain, it is instructive to compare the (upper envelope) curve of Fig.
2.12 with Ehrenspeck and Poehler’s experimental points, as well as the
experimental results of Lindsay. Lindsay made a number of models of
varying boom length (but unstated element dimension schedules) and
measured directivity at a design frequency of 440 MHz. All of these results
are shown in Fig. 2.13 where the solid curve is the theoretical maximum gain
2-24 Chapter 2

vary + LINDSAY
(dBi)
GAIN
© EHRENSPECK—
POEHLER

Oo
Oo O22 O4G SO 608 1.0 12 1.4 1.6
BOOM LENGTH (A)

Fig. 2.13—Theoretical Yagi gain envelope vs. boom length compared

with experimental data of Ehrenspeck-Poehler and Lindsay.

(from Fig. 2.12) and the keyed points are from Ehrenspeck-Poehler and from
Lindsay. The Lindsay experiments provide remarkable confirmation of the
universal boom gain curve. The Ehrenspeck-Poehler measurements all appear
to lie slightly below the theoretical curve (by a fraction of a decibel). It is
not clear that the slight discrepancy in absolute value is a real disagreement;
it may be within the expected accuracy of the gain calibration technique used
on the automatic plotter. It may also be due to lack of optimization;
Ehrenspeck and Poehler used a fixed reflector reactance and it is hard to
guess how much more gain they would have found with an optimized
configuration.

Element Illumination
Fig. 2.14 taken from Table 2.4 shows a plot of the band-center F/B ratio
as a function of overall length. It is notable that there are three empirical
values of overall length which seem to produce high values of F/B
independent of the number of elements. These apparently favorable overall
lengths are 0.25, 0.75, and 1.25\—all odd multiples of a quarter-wave.
For the 0.25) position only the three-element beam shows a high value, but
this is primarily caused by the definition of band-center F/B ratio.
This remarkable phenomenon suggests that there might be a basic physical
explanation covering all cases; indeed, such a physical basis is not hard to
find. Analogous to the optical illumination of an aperture by light, one can
think of the Yagi boom length as illuminated by electrical excitation. Unlike
the case of uniform illumination of an optical aperture, the Yagi illumination
is not uniform but can be viewed as a series of discrete excitation points (the
 Simple Yagi Antennas 2-25

F/B(dB)

(0) 02 04 O06 08 1.0 1.2 14 1.6

BOOM LENGTH (A)

Fig. 2.14—Band-center front-to-back ratio for three-, four-, five-, six-, and
seven-element Yagis vs. boom length.

elements) whose average envelope is quasi-uniform. In the optical case the

wave front is ordinarily plane (phase shift across the aperture is zero), whereas
in the Yagi case the phase shift is purposely designed to cause the main
diffracted ‘‘beam’’ to lie along the boom rather than broadside to the aperture
as in the optical case.
The aperture produces a diffraction pattern (beam pattern) consisting of
a ‘‘main beam’’ and several lobes. The number of lobes is determined basically
by the size of the aperture in wavelengths, while the amplitude of the lobes
is determined by the way the aperture is illuminated (phase and amplitude).
An informative treatment of an end-fire array (aperture illuminated by
a series of radiators having equal amplitudes) is given in Kraus (ref. 3, pages
76-89). There are two interesting cases: the ordinary end-fire array in which
the angular phase change between radiators is just equal to their spatial
separation angle, and the increased directivity end-fire array first derived by
Hansen and Woodyard where the radiator phase delay is larger than spatial
separation by an angle z/n.° The latter case is a good one with which to
compare the Yagi antenna, because although the amplitude of the current(s)
in the Yagi elements are not uniform, the phases are adjusted (by element
reactance) to give highest directivity or gain. Kraus gives expected null angle
directions for these two cases (null between lobes) as:
Ordinary end-fire:
69 = 2 sin-! [+ (m)/2ns)”] (2.1)

Increased directivity end-fire:

69 = 2 sin—!1 [+((2m —1)\/4ns)”] (2.2)

In our model the boom length /p is (n — 1)s so that we can rewrite these
equations as:
Ordinary end-fire:
6) = 2sin-!1 [+(m(n — 1)d/2lgn)”] (2.3)
2-26 Chapter 2

Increased-directivity end-fire:
6) = 2 sin-![ +(2m—1)(n—- 1)d/4/gn)”] (2.4)

Note that where the number of elements, 7, is large one would expect a high
F/B ratio (that is, a null at 180 degrees) for particular values of boom length
essentially independent of the number of elements.
Let’s now examine the patterns of the cases where F/B is relatively high
for the six-element beam at three boom lengths: 0.25\, 0.75, and 1.25) at
the precise frequency where the F/B ratio is maximum (presumably where
the back radiation ‘‘null’’ occurs). It is instructive to plot the pattern not
only at this ‘‘best’’ frequency but also at frequencies just below and just above
the best frequency. For all of these cases the reflector length is fixed at
0.50702 (F = 0.95), and all director lengths are fixed at 0.45873 (F = 1.05).
Fig. 2.15 shows the H- and E-plane patterns for all of these cases. The
H-plane pattern shows ‘‘nulls’’? between lobes which move with frequency
(equivalent to changing boom length in actual wavelengths). We can compare
the angle at which these nulls occur to those of the end-fire arrays from (2.3)
and (2.4); Table 2.5 lists these comparisons.
Note that these comparisons show qualitative agreement; also note that
the computed Yagi results are in better agreement with the increased-
directivity end-fire model. The more rapid shift of null angle(s) with frequency
for the Yagis compared to either end-fire model is not to be taken too seriously
because, as we shift frequency, not only does the effective boom length in
terms of actual wavelength change, but the element reactance(s) change
significantly, i.e., the Yagi really becomes a different Yagi.
The details of the Yagi pattern depend on the particular way in which the
boom is illuminated, i.e., on the details of element positions, and element
current magnitudes and phases. The depth of the nulls depends on the degree
of vectorial cancellation of back radiation; since the vectors themselves vary
significantly with all Yagi parameters it is no wonder that complete
cancellation is accidental and ordinarily impossible. The size of the lobes is
determined primarily by the shape and phase delay of the Yagi illumination
function. For the uniformly illuminated case the reader is referred to the
uniform end-fire arrays (see Kraus, pages 79-88).
It will be noted that for non-uniformly illuminated broadside structures
(Kraus, pages 93-121) the side-lobe level is highest for ‘‘edge’’ illumination,
next highest for uniform illumination, and zero for illumination based on
element amplitudes following the coefficients of a binomial series. One can
expect the same kind of result for an end-fire array where edge illumination
(two-element beam) produces high sidelobes, uniform illumination smaller
sidelobes, and an illumination function falling at the extreme edges (reflector
and end director currents smaller than in central elements) produces still
smaller sidelobes.
Unfortunately, for a given overall boom length the directivity suffers
somewhat as illumination is adjusted for smaller sidelobes. Moreover, in the
 al
Mane
Ney
x/
(continu ed)—E- and H-
 Simple Yagi Antennas 2-29

Fig. 2.15—(continued)—E- and H-

plane pattern s fora six-elemen
Yagi with 1.25 wavelength boom.
2-30 Chapter 2

Table 2.5
Yagi null angles compared to ordinary end-fire arrays (OEF) and increased
directivity end-fire (IDEF).
m=17 null m=2 null m=3 null
boom (degrees) (degrees) (degrees)
length freq Yagi OEF IDEF Yagi OEF IDEF Yagi OEF IDEF
(N)
0.25 0.972 —
0.25 0.982 159
0.25 0.992 120
0.75 0.970 69 98 65 — — 136
0.75 0.980 66 98 64 171 — 135
0.75 0.990 (axe SI/ 64 147 — 133
1.25 0.976 Stee 49 102 meal 2 91 — —_ 135
1.25 0.986 Sie 7A 49 99 111 91 171 — 134
1:25 0.996 48 71 48 93 110 90 150 — 132

case of a simplistic Yagi the illumination function is hard to adjust; the current
amplitudes and phases are all determined by element reactances and positions
and it is necessary to simply accept the result. Table 2.6 shows the current
amplitudes and phases in each element for the case of the six-element beam,
boom = 0.75) where the driver current is set at 1.0 ampere at 0 degrees phase.
Note that while the current amplitudes are not ‘‘uniform illumination,’’ they
seem to average out surprisingly
alike. Incidentally, the current in the
Table 2.6

higher than that in the preceding flement currents forasicelement

last director is always characteristically EI P et

director; this is due to the “‘end fF = 9 9809.

effect’? (no mutualé coupling to an current phase
element ahead of it). element (amps) (degrees)
Thus we now have a consistent Reflector 0.476 154.5
picture
:
of the high F/B ratio Yagi
A
_ Driven element
Director 1
eek
0.379
oe
— 121.3
design; the essence of correct design pjrector 2 0.467 ~ 169.7
is to place the null between lobes Director3 0.258 115.3
exactly in the back direction. This will Di"ector 4 eee vee
occur for simplistic Yagi antennas
when the overall length is approximately an odd multiple of a quarter-
wavelength. The specific best design will involve optimizing the boom length
and the boom illumination function (the particular element excitation cur-
rents and phases) to yield the best F/B ratio. Such optimization can be carried
out around the best boom lengths; this is discussed in Chapter 3. For the
present it is sufficient to note that really excellent F/B ratios are possible
with these simplistic designs as long as one is willing to accept boom lengths
which are approximately odd multiples of a quarter-wavelength.
As a final note on these simplistic best designs, not only is the back radiation
 Simple Yagi Antennas 2-31

Table 2.7
Sidelobe ratios for six-element Yagis with various boom lengths.
E-plane H-plane
boom length freq F/B peak sidelobe peak sidelobe
() (aB) (AB) (AB)
0.25 0.982 25.7 18.8 nD
0.75 0.980 32.5 19.7 12.6
1.25 0.986 35.0 18.8 14.9

low but the minimum (worst case) front-to-sidelobe ratio in the entire reverse
direction (90 to 180 degrees) is also surprisingly high. Table 2.7 shows the
result. These Yagi designs seem to be generally excellent.
With the exception of the two-element beam case I have not yet commented
on the driving-point impedance of any of the Yagis shown. Remember that
one can, by adjusting the length of the driver, always null out the driving-
point reactance at a designated frequency. The remaining resistance, however,
just like that for the two-element beam, varies enormously from case to case.
It is very low for very short beams where there is very strong coupling between
elements; moreover, in such cases it varies wildly with frequency as does the
change in reactance with frequency. Thus to ensure a reasonably reliable
electrical feed system it is wise to keep element separation well above 0.05);
all such cases investigated have reasonably well behaved driving-point im-
pedances.

Summary
Let me summarize the results for simplistic Yagi antennas:
1) Two- to seven-element beams with boom lengths to 1.5 have been
systematically explored.
2) Simplistic Yagis display a gain function where bandwidth is primarily
a function of the resonant frequency separation between reflector and
directors. The bandwidth can easily be made several percent of the central
frequency.
3) The shape of the gain function is generally not flat in the region of
interest. It is also more complex for beams which use a large number of
elements.
4) Simplistic Yagis display a F/B ratio function with a shape that varies
enormously from case to case. The shape may contain more than one peak
and changes rapidly with boom length and/or frequency. It is so compli-
cated that it is not possible to characterize its bandwidth.
5) High values of the F/B ratio (more than 30 dB) are quite rare; when
they occur F/B is high only over a very narrow band of frequencies.
6) The spacing between elements should be generally greater than 0.05X
to realize a well-behaved feed.
2-32 Chapter 2

7) The maximum practical gain of the simplistic Yagi is almost entirely

determined by boom length. Maximum gain increases, but not steadily, with
boom length.
8) Best design for a high F/B ratio requires the approximate boom length
to be an odd multiple of a quarter-wavelength at the design frequency.

NOTES:
1Hall, G. L. ed., The ARRL Antenna Book, 13th Edition, Newington, Connecticut:
American Radio Relay League, 1974.
2Brown, G. H., ‘‘Directional Antennas,’’ Proceedings of the IRE, January 1937.
3Kraus, John D., Antennas, New York: McGraw-Hill, 1950.
4Ehrenspeck, H. and Poehler, H., ‘‘A New Method of Obtaining Maximum Gain
from Yagi Antennas,” /RE Transactions on Antennas and Propagation,
October 1959.
S5Lindsay, J., “Quads and Yagis,‘‘ QST, May 1968.
6Hansen, W. W. and Woodyard, J. R., ‘“‘A New Principle in Directional Antenna
Design,’’ Proceedings of the IRE, March 1938.
CHAPTER 3

YAGI ANTENNA
PERFORMANCE OPTIMIZATION

| n Chapter 2, I explored the properties of simplistic Yagi antennas; that

is, antennas of a given boom length but having uniformly spaced
elements, one of which is a reflector, one of which is a driver, and with
director parasites all of uniform length. In reality, however, one is not
restricted to the simplistic design. It is therefore interesting to examine a
number of departures from the simplistic design to see if there are ways to
further improve performance. In this chapter I explore some ideas in a
systematic way. It will soon be apparent that some of the departures from
the simplistic design produce only subtle changes in performance, while others
produce significant changes in performance. It is fortunate that the accuracy
inherent in computation can show very subtle changes; these changes,
although small, can usually be trusted. However, the absolute accuracy of
the model on which computation is based may not be better than a few
percent.
Departures from the simplistic design may be made in a number of ways
but primarily by allowing the lengths of the directors (hence their resonant
frequencies) to vary and by allowing the placement of the elements on the
boom to change. Additionally, for a given boom length and a given total
number of elements, the number of reflectors (and hence directors) can be
changed. This is a much more drastic change and produces more pronounced
performance variations. These changes are analyzed. Only free space
performance is investigated at this time.
I shall start with a ‘‘good’’ simplistic design consisting of a six-element
Yagi on a boom 0.75) long, but will first change the lengths of all parasites
to bring the center frequency of the desired 4-percent band of maximum gain
to F = 1.0. Fig. 3.1 shows the main properties of this test antenna and the
required element lengths. Note that the position of maximum F/B ratio is
somewhat lower (F = 0.988) than the gain band center (F = 1.0); Fig. 3.2
shows the pattern of this antenna at band center. Two visible nulls can be
seen. The first one occurs at 63 degrees and the second at 144 degrees. The
second null can be identified with the peak in the F/B ratio occurring at
F = 0.988; at this lower frequency the null moves out to 180 degrees. This
antenna can be operated at best gain over the band and compromise F/B
(17 dB at F = 1.0), or operated at the frequency of best F/B (F = 0.988,
F/B = 38 dB) and somewhat compromise the gain available. In either case
this Yagi design seems to be a good one and I shall use it as a test case around
which certain departures from simplistic design can take place.
3-2 Chapter 3

r
40 200

g 2
3
30 6 ‘SO

204

90 100
FREQUENCY (%)
g00m = 0.75),
free space
element length (X9) resonance
Reflector 0.50195 0.96
Driven Element 0.48167 1.00
Director 1 0.45414 1.06
Director 2 0.45414 - 1.06
Director 3 0.45414 1.06
Director 4 0.45414 1.06

Fig. 3.1 — Performance characteristics of a six-element Yagi beam with

a boom length of 0.75).

Fig. 3.2 — Radiation pattern of a six-element Yagi beam on a 0.75)

boom. The pattern nulls occur at 63 and 144 degrees; the second null is
correlated with the peak of F/B ratio at F = 0.988.
 Yagi Antenna Performance Optimization 3-3

Table 3.1 — Computed performance characteristics of the six-element Yagi

over the frequency range from F = 0.970 to F = 1.030.
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (Bi) (cB) (ohms) (ohms)
0.970 10.06 12.9 24.7 — 38.2
0.972 10.15 14.0 24.5 -— 35.5
0.974 10.23 15.3 24.2 — 32.8
0.976 10.30 16.7 23.9 - 30.0
0.978 10.37 18.3 23.6 —27.2
0.980 10.43 20.2 23.2 — 24.3
0.982 10.50 22.7 22.8 —21.4
0.984 10.55 26.0 22.5 — 18.5
0.986 10.61 31.2 22.1 — 15.5
0.988 10.66 38.0 21.7 -12.4
0.990 10.71 31.8 21.4 -9.2
0.992 10.75 26.5 21.1 -6.0
0.994 10.79 23.0 20.8 -2.7
0.996 10.82 20.6 20.5 0.6
0.998 10.84 18.6 20.3 4.1
1.000 10.86 17.0 20.2 7.5
1.002 10.86 15.7 20.2 11.1
1.004 10.86 14.5 20.2 14.8
1.006 10.85 13.5 20.4 18.5
1.008 10.82 12.6 20.7 22.3
1.010 10.78 5inbe 4 21.1 26.2
1.012 10.73 11.0 21.7 30.2
1.014 10.67 10.3 22.5 34.3
1.016 10.59 9.7 23.5 38.4
1.018 10.50 9.1 24.8 42.6
_ 1.020 10.39 8.6 26.5 46.9
1.022 10.28 8.1 28.6 Sire
1.024 10.16 7.7 31.3 pote)
1.026 10.04 7.4 34.7 59.8
1.028 9.91 7.1 39.1 63.8
1.030 9.79 6.9 44.7 67.3

Since we are interested in very subtle changes I show in Table 3.1 the
detailed performance in the region of chief interest, accurately calculated
for this antenna over the frequency range from F = 0.970 to F = 1.030.
The driving point reactance at a given frequency is somewhat arbitrary; it
can be easily shifted or offset by changing the length of the driven element.
The length of the driven element, however, remains fixed (free space resonant
frequency = 1.0) throughout this series of explorations so that reactance
changes can be properly sensed.

Parasite Length Variations

If we keep the average value of the lengths of the directors constant we
can explore ‘‘linear’’ length tapers and ‘‘parabolic’’ tapers. I will define a
3-4 Chapter 3

linear taper as a uniform linear free space resonant frequency progression

from D1 to D4 keeping the average resonant frequency constant. In other
words, the director resonant frequencies are linearly related to director
position measured from the center of the director assembly (to keep the
average value constant). Similarly, a ‘‘parabolic’’ taper is one in which the
directors’ free space resonant frequencies are proportional to the square of
the distance from the center of the director assembly, with the further
condition that the average value of director resonance is held constant. If
we define the change in resonant frequency of D1 as A, then with a total
of four directors all other director resonances will change:

linear parabolic
element taper taper

Director 1 +A +A
Director 2 + A/3 -A
Director 3 -AI3 -A
Director 4 -A +A

The degree or magnitude of taper is fixed by the size of A; moreover, A can

be chosen either as an increase or decrease in resonant frequency.
I have selected and investigated six taper schedules which are delineated
in Table 3.2; as indicated, antenna performance in the critical region of
interest is shown in Tables 3.3 to 3.8. These results are to be compared with
the simplistic design shown in Table 3.1.
Table 3.3 shows the results for a ‘‘ —2 percent’’ linear taper and it is obvious
that all performance parameters are virtually unchanged. Table 3.4 shows
the results for a ‘‘—4 percent’’ linear taper, and even in this rather extreme
case performance is almost totally unchanged in the central frequency region.
Remember that for this case the free space resonant frequency of D1 has
dropped from 1.06 to 1.02 and it is easy to see that performance deteriorates
at higher frequencies (D1 no longer behaves like a director). Nevertheless,
it is remarkable that the linear taper — even of this magnitude — has virtu-
ally no effect on the performance of the Yagi antenna.
Table 3.5 shows results for a —2 percent parabolic taper; Tables 3.6, 3.7
and 3.8 show results for parabolic tapers of —1, +1, and +2 percent, respec-
tively. In comparing these tables with the standard simplistic Yagi antenna
we again see that truly minimal changes are made to the chief performance
indices. The variations in maximum F/B ratios are not of great significance;
I shall come back to this point later.
 Yagi Antenna Performance Optimization 3-5

Table 3.2 — Schedule of director lengths for the six six-element Yagi-Uda
performance characteristics listed in Table 3.3 through Table 3.8.
taper Dir 1 Dir 2 Dir 3 Dir 4
Table type () () () (\) A
3.3 Linear 0.46341 0.45719 0.45263 0.44524 —2%
3.4 Linear 0.47306 0.46028 0.44817 0.43667 —-4%
3.5 Parabolic 0.46341 0.44524 0.44524 0.46341 — 2%
3.6 Parabolic 0.45873 0.44964 0.44964 0.45873 -1%
3.7 Parabolic 0.44964 0.45873 0.45873 0.44964 +1%
3.8 Parabolic 0.44524 0.46341 0.46341 0.44524 + 2%

Table 3.3 — Performance characteristics of a six-element Yagi beam with a

boom length of 0.75), director lengths tapered linearly at — 2 percent (director
lengths shown in Table 3.2).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (AB) (ohms) (ohms)
0.970 10.07 13.0 23.9 — 39.1
0.972 10.16 14.1 23.6 — 36.4
0.974 10.24 15:3 23.4 — 33.7
0.976 10.31 16.6 23.1 -31.0
0.978 10.38 18.2 22.8 — 28.3
0.980 10.44 20.0 22.4 — 25.5
0.982 10.49 22.3 22.1 — 22.6
0.984 10.54 25.3 2t7 - 19.8
0.986 10.59 29.9 21.4 - 16.8
0.988 10.64 40.0 21.0 - 13.9
0.990 10.67 38.2 20.7 - 10.9
0.992 10.71 29.4 20.4 -7.8
0.994 10.74 25.1 20.1 -—4.7
0.996 10.76 22.2 19.9 -1.5
0.998 10.77 20.1 19.7 het
1.000 10.78 18.4 19.5 4.9
1.002 10.78 16.9 19.5 8.3
1.004 10.77 1527 19.5 11.6
1.006 10.75 14.6 19.6 15.0
1.008 10.72 13.7 19.8 18.4
1.010 10.68 12.9 20.1 21.8
1.012 10.63 12.1 20.6 25.3
1.014 10.57 1.5 21.2 28.7
1.016 10.50 10.9 22.0 32.1
1.018 10.42 10.3 23.0 35:5
1.020 10.33 9.9 24.2 38.9
1.022 10.24 9.5 20, 42.0
1.024 10.15 9.2 27.6 45.0
1.026 10.05 8.9 29.9 47.6
1.028 9.95 8.8 32.6 49.6
1.030 9.86 8.7 35.6 50.7
3-6 Chapter 3

Table 3.4 — Performance characteristics of a six-element Yagi beam with a

boom length of 0.75), director lengths tapered linearly at — 4 percent (direc-
tor lengths shown in Table 3.2).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (aBi) (dB) _ (ohms) (ohms)
0.970 10.06 13.7 20.9 - 40.3
0.972 10.15 14.9 20.6 - 37.5
0.974 10.23 16.2 20:2 — 34.7
0.976 10.31 17.7 19.9 -31.9
0.978 10.37 19.5 19.5 -29.1
0.980 10.43 21.6 19.1 — 26.2
0.982 10.49 24.2 18.6 — 23.3
0.984 10.54 27.8 18.2 - 20.3
0.986 10.58 31.9 17.8 - 17.3
0.988 10.61 31.9 17.5 - 14.3
0.990 10.64 27.7 17.1 -11.3
0.992 10.66 24.3 16.8 -8.2
0.994 10.68 21.7 16.5 -5.1
0.996 10.68 19.6 16.2 -1.9
0.998 10.67 17.9 16.0 1.3
1.000 10.65 16.5 15.8 4.5
1.002 10.62 15.3 15.7 7
1.004 10.58 14.2 15.6 10.9
1.006 10.52 13.2 15.6 14.1
1.008 10.45 12.4 18.7 17.3
1.010 10.37 11.6 15.8 20.5
1.012 10.26 10.9 16.1 23.6
1.014 10.15 10.2 16.4 - 26.7
1.016 10.01 9.7 16.7 29.7
1.018 9.86 9.1 17.1 32.5
1.020 9.69 8.6 17.6 35.2
1.022 9.49 8.2 18.0 37.6
1.024 9.25 hak 18.2 39.8
1.026 8.96 7.3 18.2 41.8
1.028 8.56 6.7 17.7 43.7
 Yagi Antenna Performance Optimization 3-7

Table 3.5 — Performance characteristics of a six-element Yagi beam with a

boom length of 0.75, director lengths tapered parabolically at - 2 percent
(director lengths shown in Table 3.2).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (Bi) (cB) (ohms) (ohms)
0.970 10.01 16.4 19.7 -— 42.3
0.972 10.11 18.1 19.2 — 39.4
0.974 10.21 20.2 18.6 — 36.4
0.976 10.30 22.8 18.1 — 33.3
0.978 10.39 26.2 17.5 - 30.3
0.980 10.47 29.7 17.0 - 27.1
0.982 10.55 29.3 16.4 -— 23.9
0.984 10.63 25.7 15.9 — 20.7
0.986 10.71 22.4 15.4 - 17.4
0.988 10.78 19.9 15.0 — 14.0
0.990 10.84 17.9 14.6 - 10.6
0.992 10.90 16.3 14.2 -7.1
0.994 10.95 14.8 13.9 —3.6
0.996 10.99 13.6 13.7 0.0
0.998 11.01 12.5 13.5 3.7
1.000 11.03 11.5 13.4 7.4
1.002 11.03 10.7 13.3 int
1.004 11.02 9.9 13.4 15.0
1.006 10.99 9.1 13.6 18.9
1.008 10.95 8.4 13.9 22.8
1.010 10.89 7.8 14.3 26.9
1.012 10.82 7.3 14.9 31.0
1.014 10.74 6.7 15:7 35.2
1.016 10.64 6.3 16.8 39.5
1.018 10.54 5.8 18.1 44.0
1.020 10.44 5:5 20.0 48.5
1.022 10.34 5.2 22.3 53.2
1.024 10.24 4.9 25.5 57.9
1.026 10.16 4.7 30.0 62.7
1.028 10.10 4.6 36.2 67.0
1.030 10.07 4.6 45.3 69.9
3-8 Chapter 3
et

Table 3.6 — Performance characteristics of a six-element Yagi beam with a

boom length of 0.75), director lengths tapered parabolically at —1 percent
(director lengths shown in Table 3.2).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (dB) (ohms) (ohms)
0.970 10.04 14.3 22.5 - 40.5
0.972 10.13 15.7 22.1 — 37.7
0.974 10.22 17.2 21.7 — 34.9
0.976 10.30 19.0 21.3 — 32.0
0.978 10.38 21.2 20.8 - 29.1
0.980 10.45 23.9 20.3 — 26.1
0.982 10.52 27.6 19.8 — 23.0
0.984 10.59 31.6 19.4 — 20.0
0.986 10.66 30.4 18.9 - 16.8
0.988 10.72 26.3 18.5 - 13.6
0.990 10.77 22.9 18.1 - 10.3
0.992 10.82 20.4 WATS -7.0
0.994 10.87 18.4 17.4 -3.7
0.996 10.91 16.8 17.1 -0.1
0.998 10.94 15.4 16.9 3.4
1.000 10.96 14.2 16.8 7.0
1.002 10.97 13.4 16.7 10.7
1.004 10.97 12.2 16.8 14.4
1.006 10.96 11.3 16.9 18.3
1.008 10.93 10.5 17.2 22.2
1.010 10.89 9.8 17.6 26.1
1.012 10.84 9.2 18.1 30.2
1.014 10.77 8.6 18.9 34.3
1.016 10.69 8.0 19.9 38.6
1.018 10.61 7.5 21.2 42.9
1.020 10.51 ria 22.9 47.4
1.022 10.41 6.7 25.0 51.9
1.024 10.31 6.4 27.9 56.5
1.026 10.22 6.1 31.6 61.1
1.028 10.13 5.9 36.6 65.5
 Yagi Antenna Performance Optimization 3-9

Table 3.7 — Performance characteristics of a six-element Yagi beam with a

boom length of 0.75), director lengths tapered parabolically at +1 percent
(director lengths shown in Table 3.2).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (Bi) (dB) (ohms) (ohms)
0.970 10.07 11.9 26.2 — 35.8
0.972 10.15 12.9 26.1 — 33.1
0.974 10.23 13.9 26.0 — 30.4
0.976 10.29 15.1 25.9 — 27.7
0.978 10.36 16.4 25.7 — 25.0
0.980 10.41 17.9 25.5 — 22.2
0.982 10.47 1OE7, 25.2 -—19.4
0.984 10.51 21.9 25.0 - 16.5
0.986 10.56 24.9 24.7 - 13.6
0.988 10.60 29.3 24.4 - 10.6
0.990 10.64 38.4 24.1 -7.6
0.992 10.67 38.8 23.9 -4.5
0.994 10.69 29.4 23.7 -1.3
0.996 10.71 25.0 23.5 2.0
0.998 10.73 22.1 23.3 5.3
1.000 10.73 19.9 23.2 8.8
1.002 10.73 18.1 23.2 12.3
1.004 10.71 16.7 23.3 15.9
1.006 10.69 15.4 23.5 19.5
1.008 10.65 14.3 23.8 23.3
1.010 10.60 13.3 24.3 27.1
1.012 10.53 12.5 24.9 31.0
1.014 10.45 1157, 25.8 35.0
1.016 10.35 10.9 26.8 39.1
1.018 10.24 10.3 28.2 43.2
1.020 10.11 9.7 29.9 47.3
1.022 9.97 9.2 32.0 1,0
1.024 9.81 8.7 34.5 55.5
1.026 9.64 8.3 37.7 59.5
1.028 9.46 7.9 41.6 63.1
3-10 Chapter 3

Table 3.8 — Performance characteristics of a six-element Yagi beam with a

boom length of 0.75), director lengths tapered parabolically at +2 percent
(director lengths shown in Table 3.2).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (dB) (ohms) (ohms)
0.970 10.07 11.0 27:1 -— 33.5
0.972 10.15 11.9 lua — 30.8
0.974 10.22 12.9 dt A. — 28.2
0.976 10.28 13.9 27.2 — 25.5
0.978 10.33 15.1 27.2 — 22.8
0.980 10.38 16.4 27.1 — 20.1
0.982 10.43 17.8 26.9 - 17.3
0.984 10.47 19.5 26.8 -14.5
0.986 10.50 21.6 26.6 -— 11.6
0.988 10.53 24.2 26.4 -8.7
0.990 10.56 27.8 26.1 -5.7
0.992 10.58 32.7 25.9 -2.6
0.994 10.59 33.6 25.7 0.5
0.996 10.59 28.7 25.6 3.7
0.998 10.59 24.8 25.4 7.1
1.000 10.58 22.0 25.3 10.5
1.002 10.56 19.8 25.3 14.0
1.004 10.53 18.1 25.4 17.5
1.006 10.48 16.6 25.6 Zine
1.008 10.41 15.3 25.9 25.0
1.010 10.33 14.1 26.3 28.9
1.012 10.23 13.1 27.0 32.9
1.014 10.11 12.2 27.8 36.9
1.016 9.92 1.3 28.8 41.0
1.018 9.79 10.6 30.1 45.2
1.020 9.60 9.8 31.7 49.5
1.022 9.38 9.2 33.6 53.7
1.024 9.14 8.5 36.0 58.0
1.026 8.87 7.9 38.9 62.2
1.028 8.57 7.4 42.3 66.2
 Yagi Antenna Performance Optimization 3-11

Parasite Placement Variations

Analogous to the length (and corresponding resonant frequency) taper var-
iations I have also investigated element placement variations along the boom.
Again it is possible to vary the space intervals between elements linearly and
pseudo-parabolically. Table 3.9 shows element positions for six schedules
I have investigated and the individual results are shown in Tables 3.10 through
Bd ay!
Table 3.10 shows the results where elements are crowded towards D4. Note
that truly large changes in placement have been made. Similarly, Table 3.13
shows the results where elements are severely crowded towards the reflector.
Tables 3.11 and 3.12 are intermediate schedules; Tables 3.14 and 3.15 show
results where end spaces are (relatively) increased. These tables all show that
these placement variations have only a very minor effect on directivity, and
while the maximum F/B ratio is somewhat affected (generally adversely),
we shall soon see that this effect may not be very significant.
Up to this point we have looked at taper schedules which are linear and
parabolic and which also involve director length, or resonant frequency, and
element placement along the boom. It is truly remarkable that all of these
schedules produce minimal changes in antenna performance; it is therefore
plausible that combinations of these schedules will also produce minimal per-
formance variations. This leads to the conclusion that the original simplistic
design (dimensions listed in Fig. 3.1) is just about as good as any. No real
improvement on gain can be expected by any new tricky design; as far as
F/B ratio is concerned, it will soon be apparent that one can ‘‘tune up’’ the
maximum F/B ratio starting with almost any of these schedules.
A summary of raw performance of all of these cases is shown in Table
3.16, where, in addition to data already shown in the previous tables, infor-
mation on pattern (not explicitly shown here) has been added. This table
shows that all cases produce about the same gain; the very small variations
are due to the effective ‘‘illumination’’ pattern of the boom aperture. The
F/B ratio at central gain frequency (F = 1.0) varies somewhat, but a very
slight change in operating frequency would easily make them all compar-
able. The frequency position of maximum F/B ratio and the angle of the
second null at F = 1.0 are related. Lower frequencies of maximum F/B
should correspond (at central frequency) to longer effective boom illumi-
nated apertures, thus corresponding to lower null angles at F = 1.0 and some-
what higher gain. An examination of Table 3.16 shows all of these quantities
to be well correlated; it appears therefore that all results are understood and
self-consistent.
3-12 Chapter 3

Table 3.9 — Schedule of director placement on boom for the six six-element
Yagi-Uda performance characteristics listed in Table 3.10 through 3.15.
taper element position on boom (h)
Table type Refl DR D1 D2 D3 D4
3.10 Linear 0 0.200 0.3750 0.5250 0.650 0.750
3.11 Linear 0 0:175. -0.3375_. 0:4875 "0.625" 0-750
3.12 Linear 0 0.125 0.2625 0.4125 0.575 0.750
3.13 Linear 0 0.100 0.2250 0.3750 0.500 0.750
3.14 Parabolic 0 0.200 0.3167 0.4333 0.550 0.750
3.15 Parabolic 0 0.175 0.3083 0.4417 0.575 0.750

From all of this information it is reasonable to draw a general conclusion

that the simplistic Yagi design gives about as much gain as any other design
of the same boom length. Tapering element lengths or element position in-
tervals along the boom is of no apparent value. The characteristic of the direc-
tors which is important is the average length (or average free space resonant
frequency). But this conclusion has been demonstrated only for a boom length
of 0.75; we must be careful not to generalize too much. Recall that the NBS
data (see Fig. 7 of the NBS report) suggested that for booms longer than
one wavelength some improvement in gain over simplistic Yagi performance
could be obtained with particular director length schedules.' My calculations
support the NBS result; nevertheless, for boom lengths shorter than one
wavelength the simplistic design is as good as any.

1Viezbicke, P., ‘“Yagi Antenna Design,’’ NBS Technical Note 688, U.S. Depart-
ment of Commerce, Washington, D.C., December, 1976.
 Yagi Antenna Performance Optimization 3-13

Table 3.10 — Performance characteristics of a six-element Yagi beam with

a boom length of 0.75), director spacing tapered linearly, large positive
interval, directors crowded toward D4.
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (aBi) (dB) (ohms) (ohms)
0.970 9.87 11.8 43.7 — 38.6
0.972 9.91 12.5 42.7 -— 36.9
0.974 9.95 13.3 41.6 -—35.1
0.976 9.99 14.1 40.4 - 33.1
0.978 10.03 15.0 39.1 — 31.0
0.980 10.08 16.0 37.7 — 28.8
0.982 10.12 17.1 36.3 — 26.4
0.984 10.17 18.3 34.9 — 23.9
0.986 10.22 19.5 33.5 -21.2
0.988 10.27 20.7 32.0 - 18.3
0.990 10.33 21.7 30.6 — 15.3
0.992 10.38 22.2 29.3 -12.2
0.994 10.44 22.0 28.0 -9.0
0.996 10.49 21.0 26.7 -5.6
0.998 10.54 19.7 25.5 —2.1
1.000 10.59 18.3 24.4 1.6
1.002 10.63 16.9 23.4 5.3
1.004 10.66 15.6 22.5 9.2
1.006 10.68 14.4 21.7 13.2
1.008 10.69 13.3 21.1 17.2
1.010 10.68 12.3 20.5 21.4
1.012 10.65 11.3 20.1 25.7
1.014 10.60 10.4 19.9 30.0
1.016 10.53 9.6 19.8 34.4
1.018 10.43 8.8 19.9 38.9
1.020 10.30 8.1 20.2 43.5
1.022 10.14 7.4 20.7 48.2
1.024 9.96 6.8 21.4 52.9
1.026 9.76 6.2 22.5 57.8
1.028 9.53 57 23.8 62.6
3-14 Chapter 3
PRAT ISN ee ohea Be ONGaNyeat erly ANDAs Meh le Mer ASS
Table 3.11 — Performance characteristics of a six-element Yagi beam with
a boom length of 0.75), director spacing tapered linearly as shown in Table
3.9 (mild positive linear interval).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (dB) (ohms) (ohms)
0.970 10.04 12.8 33.0 -— 37.5
0.972 10.10 137i 32.5 — 35.3
0.974 10.15 14.7 31.8 - 32.9
0.976 10.20 15.8 31.1 -— 30.5
0.978 10.26 A7e1 30.4 — 28.0
0.980 10.31 18.5 29.6 — 25.4
0.982 10.36 20.2 28.8 — 22.8
0.984 10.41 22.1 27.9 — 20.0
0.986 10.46 24.2 2tal -17.1
0.988 10.51 26.3 26.3 - 14.2
0.990 10.56 27.2 25.5 —11.1
0.992 10.60 26.0 24.7 -—8.0
0.994 10.65 23.7 23.9 -—4.7
0.996 10.69 21.5 23.2 -—1.4
0.998 10.72 19.6 22.6 2.1
1.000 10.75 17.9 22.0 5.6
1.002 10.77 16.4 21.5 9.3
1.004 10.78 15.1 21.1 13.0
1.006 10.78 14.0 20.8 16.8
1.008 10.76 12.9 20.6 20.7
1.010 10.73 12.0 20.6 24.7
1.012 10.68 11.1 20.7 28.7
1.014 10.62 10.4 20.9 32.9
1.016 10.53 9.6 21.3 37.1
1.018 10.43 9.0 21.9 41.5
1.020 10.31 8.3 22.8 45.9
1.022 10.17 7.8 23.9 50.4
1.024 10.02 7.3 25.4 54.9
1.026 9.86 6.8 27.3 59.6
1.028 9.68 6.4 29.8 64.3
 Yagi Antenna Performance Optimization 3-15

Table 3.12 — Performance characteristics of a six-element Yagi beam with

a boom length of 0.75, director spacing tapered linearly as shown in Table
3.9 (mild negative linear interval).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (dB) (ohms) (ohms)
0.970 9.88 LEA AS) - 41.2
0.972 10.02 13.0 17.8 - 38.1
0.974 10.16 14.3 17.8 — 35.0
0.976 10.27 15.9 17.8 -31.9
0.978 10.38 17.7 17.8 — 28.7
0.980 10.47 19.8 17.8 — 25.6
0.982 10.55 22.6 17.8 — 22.5
0.984 10.63 26.3 17.8 -— 19.3
0.986 10.69 31.9 17.8 -— 16.1
0.988 10.75 34.6 17.8 - 12.9
0.990 10.81 28.6 17.8 -9.6
0.992 10.85 24.4 17.9 -6.3
0.994 10.89 21.5 18.0 -2.9
0.996 10.91 19.3 18.1 0.5
0.998 10.93 17.6 18.3 4.0
1.000 10.94 16.2 18.6 7.5
1.002 10.94 15.0 19.0 Ti
1.004 10.93 13.9 19.5 14.8
1.006 10.91 13.0 20.1 18.5
1.008 10.88 ea 20.9 22.3
1.010 10.84 11.4 21.9 26.2
1.012 10.79 10.8 23.2 30.1
1.014 10.73 10.2 24.7 34.1
1.016 10.66 9.7 26.6 38.1
1.018 10.59 9.2 28.9 42.2
1.020 10.51 8.8 31.9 46.1
1.022 10.42 8.5 35.6 49.9
1.024 10.34 8.2 40.2 53.3
1.026 10.25 8.0 46.0 55.8
1.028 10.17 7.9 53.2 56.9
3-16 Chapter 3

Table 3.13 — Performance characteristics of a six-element Yagi beam with

a boom length of 0.75), director spacing tapered linearly with the elements
crowded toward the reflector (large negative interval taper).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (AB) (ohms) (ohms)
0.970 9.27 8.8 12.0 - 47.1
0.972 9.57 10.2 12.1 - 43.5
0.974 9.82 11.6 12.2 - 40.0
0.976 10.04 13.2 12.3 — 36.5
0.978 10.23 15.0 12.5 — 33.0
0.980 10.38 17.0 12.7 — 29.6
0.982 10.52 19.4 12.9 — 26.1
0.984 10.63 22.2 13.2 —22.7
0.986 10.73 25.2 13.5 - 19.2
0.988 10.81 26.6 13.8 -— 15.7
0.990 10.87 24.9 14.1 - 12.3
0.992 10.93 22.3 14.5 -8.8
0.994 10.96 20.0 15.0 -5.2
0.996 10.99 18.1 15.5 -1.7
0.998 11.01 16.6 16.1 1.9
1.000 11.01 15.3 16.8 5.6
1.002 11.01 14.2 17.6 9.2
1.004 11.00 13.3 18.6 13.0
1.006 10.97 12.5 19.8 16.7
1.008 10.94 11.7 21.2 20.5
1.010 10.90 11.1 22.9 24.4
1.012 10.85 10.6 24.9 28.2
1.014 10.80 10.1 27.4 31.9
1.016 10.74 O57, 30.5 35.6
1.018 10.69 9.3 34.3 38.9
1.020 10.63 9.0 38.9 41.7
1.022 10.57 8.8 44.6 43.5
1.024 10.52 8.7 51.2 43.6
1.026 10.47 8.6 58.4 41.0
1.028 10.44 8.6 64.6 34.7
 Yagi Antenna Performance Optimization 3-17
i

Table 3.14 — Performance characteristics of a six-element Yagi beam with

a boom length of 0.75), director spacing tapered pseudo parabolically
according to the schedule of Table 3.9 (large positive interval).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (dB) (ohms) (ohms)
0.970 10.25 1525 33.7 — 34.3
0.972 10.30 16.5 33.4 -—31.8
0.974 10.35 17.6 33.0 — 29.2
0.976 10.40 18.9 32.6 — 26.5
0.978 10.44 20.4 32.2 — 23.9
0.980 10.49 22.1 31.8 -—21.0
0.982 10.53 24.1 31.3 — 18.2
0.984 10.56 26.3 30.9 - 15.3
0.986 10.60 28.2 30.5 - 12.3
0.988 10.63 28.6 30.1 -9.2
0.990 10.66 27.0 29.7 -6.0
0.992 10.69 24.8 29.4 -2.8
0.994 10.71 22.6 29.1 0.6
0.996 10.73 20.7 28.9 4.0
0.998 11.74 19.1 28.7 7.6
1.000 11.75 17.6 28.7 11.3
1.002 PLAS, 16.4 28.7 15.0
1.004 10.74 15.3 28.9 18.9
1.006 10.72 14.2 29.2 22.9
1.008 10.69 13.3 29.7 27.1
1.010 10.63 12.5 30.5 31.3
1.012 10.59 11.7 31.5 35.7
1.014 10.52 10.9 32.8 40.3
1.016 10.44 10.3 34.5 45.0
1.018 10.35 9.6 36.7 49.9
1.020 10.25 9.0 39.5 54.9
1.022 10.13 8.5 43.2 60.1
1.024 10.01 8.0 47.9 65.2
1.026 9.88 Tiss 54.0 70.3
1.028 9.75 7.1 62.1 74.8
3-18 Chapter 3

Table 3.15 — Performance characteristics of a six-element Yagi beam with

a boom length of 0.75), director spacing tapered pseudo parabolically
according to the schedule of Table 3.9 (mild positive interval).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (AB) (ohms) (ohms)
0.970 10.20 14.0 31.0 — 35.2
0.972 10.26 15.0 30.7 — 32.7
0.974 10.32 16.1 30.4 - 30.2
0.976 10.37 17.4 30.1 — 27.5
0.978 10.42 18.9 29.7 — 24.9
0.980 10.47 20.6 29.3 — 22.2
0.982 10.52 22.8 28.9 — 19.4
0.984 10.56 25.6 28.5 — 16.6
0.986 10.60 29.6 28.1 -—13.6
0.988 10.64 35.4 27.7 — 10.6
0.990 10.67 34.8 2133 -7.6
0.992 10.70 29.1 27.0 -4.4
0.994 10.73 Zoe 26.6 -1.2
0.996 10.76 22.5 26.4 2.1
0.998 10.77 20.3 26.2 5.6
1.000 10.78 18.6 26.1 9.1
1.002 10.79 174 26.1 12.7
1.004 10.78 15.9 26.2 16.4
1.006 10.77 14.8 26.4 20.2
1.008 10.74 13.8 26.8 24.0
1.010 10.70 12.9 27.4 28.0
1.012 10.66 12h 28.2 32.1
1.014 10.60 11.3 29.3 36.3
1.016 10.52 10.7 30:2 40.6
1.018 10.44 10.0 32.8 45.0
1.020 10.34 9.5 34.8 49.4
1.022 10.24 9.0 Shi 53.9
1.024 10.12 8.5 41.5 58.2
1.026 10.00 8.1 46.3 62.4
1.028 9.88 ad. S209 66.0
1.030 9.76 7.5 60.4 68.4
 Yagi Antenna Performance Optimization 3-19

Front-to-Back Optimization
It can be seen from Table 3.16 that the best ‘‘null’’ positions give quite
different values of maximum F/B. Indeed, good nulls or correspondingly
high values of F/B must be viewed as accidental vectorial cancellations in
the reverse or back direction; such good cancellations will not generally occur
with any arbitrary boom illumination. Note that the various cases shown in
Table 3.16 display maximum F/B ratios ranging from 22 to 40 dB. It is an
interesting exercise to see if there is some way to significantly enhance the
maximum F/B ratio by some variational procedure.
Let us start with the simplistic Yagi design (Fig. 3.1) and vary the position
of, say D3, along the boom. We now know that small variations in position
will not significantly affect gain, but vectorial cancellation effects in the back
direction can be expected to be significant. If we find a position for D3 which
maximizes the F/B ratio, its vectorial contribution in the back direction should
be approximately out of phase with the residue from all other elements. At
this point some other element (say D1) can be positioned for a new (still
higher) maximum F/B ratio; after this is done D3 can be readjusted again
for anew maximum F/B ratio, etc. By iterating the two adjustments it should
be possible to continuously improve F/B ratios, presumably to as high a value
as desired. With such an iteration procedure it is desirable to start with a
fairly good value of F/B so that only small variations in element position
can have significant effect.

Table 3.16 — Performance comparison of six-element Yagi beams with vary-

ing director lengths and element positions along the boom shows little gain
variation.
=a maximum angle of
from gain F/B F/B F m = 2 null
Table (ABi) (dB) (dB) (deg)
3.1 10.86 17.0 38.0 0.988 144
3.3 10.78 18.4 40.0 0.988 144
3.4 10.65 16.5 31.9 0.986 138
3.5 11.03 11.6 29.7 0.980 138
3.6 10.96 14.2 31.6 0.984 141
3.7, 10.73 19.9 38.7 0.992 150
3.8 10.58 22.0 33.6 0.994 153
3.10 10.59 18.3 22.2 0.992 150
3.11 10.75 17.9 27.2 0.990 147
3.12 10.94 16.2 34.6 0.988 141
3.13 11.01 15.3 26.6 0.988 138
3.14 10.75 17.7 28.6 0.988 144
3.15 10.78 18.6 35.4 0.988 147
 3-20 Chapter 3

I have carried out such an iteration (using D3 and D1) for the simplistic
Yagi design and have arrived at the following positions:

element position (x)

Reflector 0.000
Driven Element 0.150
Director 1 0.28967)
Director 2 0.450
Director 3 0.58945)
Director 4 0.750

This design, optimized for maximum F/B ratio (at F = 0.990) produces the
performance displayed in Table 3.17. A careful comparison of this table with
the original simplistic model shows virtually identical performance in all
_ respects except that the F/B maximum has gone up from an excellent 38 dB
to an astounding 98 dB. Even this high value is not a real limit; it is limited
only by the number of iterations which were made.
Notice that this astronomical value of F/B is of no practical significance.
It occurs at essentially a single frequency and its effective bandwidth is vanish-
ingly small. Moreover, extremely small variations in the Yagi dimensions will
upset the cancellation; in practice one could not likely construct a mechani-
cally satisfactory, fully optimized Yagi antenna. Nevertheless, the mathe-
matical iteration shows that it is possible, in principle, to obtain (at a single
frequency) an arbitrarily high F/B ratio. It is likely that there are a large
number of potential solutions involving iterations with other elements. Fur-
thermore, we now know that the variations in F/B maxima shown in
Table 3.16 result from the particular illumination chosen, and it is very likely
that minor element placement variations could make an arbitrarily high F/B
ratio design starting from any of the cases shown.
To understand this iteration procedure it is helpful to show the vectorial
contributions of each element to the forward and back waves. The (current)
contribution from a given element will be a vector whose magnitude is the
magnitude of element current and whose phase consists of two parts. The
first part is the actual (time) phase of the element current referred to some
time origin (say the driver current) and the second is the (space) phase change
due to the element position referred to some space origin (say at x = 0 along
the boom). Note that this second part changes sign in going from a forward
wave to a reverse wave. Figure 3.3 shows these (current) vectorial contribu-
tions at F = 0.988 for the original simplistic Yagi design (Fig. 3.1) to both
forward and reverse waves.
Note that for the forward wave the individual element contributions do
not all fully reinforce the forward wave; in fact, the contribution from the
reflector is even negative with respect to the final total (current) vector. This
curious result is typical of all Yagi arrays. Note that the contributions to
 Yagi Antenna Performance Optimization 3-21

Table 3.17 — Performance vs frequency characteristics of a six-element Yagi

beam with a boom length of 0.75 \, element positions optimized for maxi-
mum F/B ratio (at F = 0.990).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (aBi) (2B) (ohms) (ohms)
0.960 9.48 8.1 26.8 — 50.7
0.962 9.63 8.8 26.9 - 47.9
0.964 9.76 9.7 27.1 - 45.2
0.966 9.88 10.5 27.2 - 42.5
0.968 9.98 11.4 2ii2 — 39.8
0.970 10.07 12.4 Zine - 37.2
0.972 10:15 13.4 20 — 34.5
0.974 10.23 14.5 27.0 -—31.8
0.976 10.30 15.8 26.9 - 29.2
0.978 10.36 17.2 26.7 — 26.5
0.980 10.42 18.8 26.5 — 23.7
0.982 10.48 20.8 26.2 — 20.9
0.984 10.53 23.4 26.0 -— 18.0
0.986 10.58 26.9 Con — 15.1
0.988 10.62 33.0 25.5 -12.1
0.990 10.66 98.8 (dey 72 -9.1
0.992 10.70 33.0 25.0 -5.9
0.994 10.73 27.0 24.9 -2.7
0.996 10.76 23.5 24.7 0.6
0.998 10.79 21.0 24.7 3.9
1.000 10.80 19.1 24.7 7.4
1.002 10.81 17.5 24.8 10.9
1.004 10.81 16.2 25.1 14.6
1.006 10.80 15.0 25.5 18.3
1.008 10.78 14.0 26.0 22.1
1.010 10.75 13.1 26.8 26.0
1.012 10.71 12.3 27.8 30.0
1.014 10.65 71.5 29.2 34.0
1.016 10.59 10.9 30.8 38.1
1.018 10.51 10.3 33.0 42.3
1.020 10.42 9.8 35.8 46.4
1.022 10.32 9.3 39.3 50.4
1.024 10.22 8.9 43.7 54.1
1.026 10.11 8.6 49.4 57.1
1.028 10.00 8.3 56.5 58.9
1.030 9.88 8.1 65.1 58.4
1.032 9.78 7.9 74.6 54.2
1.034 9.68 7.9 83.0 44.4
1.036 9.59 8.0 85.9 28.9
1.038 9.51 8.2 79.0 12.2
3-22 Chapter 3

Fig. 3.3 — Current vectorial contributions at F = 0.988 for six-element

Yagi (boom = 0.75)), forward and reverse waves.

the back or reverse wave, in total, nearly cancel out, leaving only a small
residue which accounts for the 38 dB F/B ratio.
Now it is easy to see conceptually what happens in the iterative procedure
to reduce the reverse wave residual. If you look at the reverse wave vector
plot, it is easy to imagine that as D3 is moved along the boom, the D3 vector
rotates around its origin. The reverse wave residual will then be changed along
an axis at right angles to the D3 vector and can be minimized by the D3 posi-
tion. After this is done another element, say D1, can be moved along the
boom; its vector contribution is at a different angle and can therefore reduce
the residual still further. Thus, in principle, iterative motions of two elements
whose reverse wave vectors contribute at different angles can ultimately reduce
the reverse wave residual to as low a value as desired. The iterative conver-
gence will be most rapid if the two element vectors are orthogonal; neverthe-
less, it can converge adequately for many element combinations.
Of course, this conceptual picture is oversimplified; as any element is moved
 Yagi Antenna Performance Optimization 3-23

Fig. 3.4 — Current vectorial contributions for the optimized

six-element Yagi at F = 0.990.

on the boom not only does its vector rotate, but all element currents and
phases readjust somewhat. However, these readjustments are usually minor
and in practice cause little difficulty as long as you start with a reasonably
small residual as shown in Fig. 3.3. Figure 3.4 shows the vectorial contribu-
tions for the optimized Yagi beam. Note that the element contributions
are only slightly modified in the optimization procedure.
At this point I must issue a warning. Recall that the mathematical model
being used in these computations involves certain approximations. These ap-
proximations make relatively little difference in the calculations for forward
gain, but they become crucial in calculations involving vectorial cancella-
tion or closure for back radiation. Thus the explicitly calculated positions
for and magnitude of a very high F/B ratio are not to be trusted. Neverthe-
less, the general behavior is still valid. The real Yagi can still be made to
have a high F/B ratio, just as our mathematical model shows, but it may
occur at a slightly different frequency and it may require slightly different
positions for D3 and D1 in the final optimization.
 3-24 Chapter 3

Optimum Design
We now have the necessary tools with which to design truly excellent Yagi
antennas. We start first from a knowledge that the boom length should be
approximately an odd number of quarter-wavelengths for an initial simplis-
tic design; we have seen that such a boom length promotes an inherently high
F/B ratio at a frequency near the center of the best gain band; boom length
also determines ultimate gain. After the boom length is chosen a resonant
frequency schedule is chosen (see appropriate figures from Chapter 2) for
reflector and director(s). A preliminary calculation is then made to accurately
determine the frequency of maximum F/B ratio, which will not necessarily
correspond with the frequency at the center of the best gain portion.
The useful band is now to be centered around the F/B point; it is neces-
sary to ensure that there is enough gain bandwidth left for the intended pur-
pose. Remember that the overall gain bandwidth is basically controlled by
-the resonant frequency schedule of the parasites. This bandwidth should not
be larger than necessary, because gain is compromised somewhat as the band-
width increases.
Now translate the frequency (F;) of best F/B to F = 1.0 by multiplying
all parasite lengths by Fj; a new preliminary calculation, possibly iterated
once more, will insure that the best F/B ratio is exactly F = 1.0. Next, alter-
nately vary the x position of D3 and then D1 to get larger and larger values
of F/B at F = 1.0 until the value is sufficiently high.

Design Example
An example illustrates this design procedure. Let’s choose a boom length
of 0.780. From an inspection of the results of our test Yagi simplistic design
(Fig. 3.1) we can probably use the same parasite resonant frequency schedule
and still obtain an adequate ultimate gain bandwidth performance. Listed
below are the initial element positions along the boom:

position initial intermediate

element x length length
() () (\)
Reflector 0.000 0.50195 0.49343
Driven El 0.156 0.48167 0.48167
Director 1 0.312 0.45414 0.44643
Director 2 0.468 0.45414 0.44643
Director 3 0.624 0.45414 0.44643
Director 4 0.780 0.45414 0.44643

Initial performance of this Yagi model is shown in Table 3.18; the fre-
quency for maximum F/B is F = 0.984. Shortening all elements by approxi-
mately this frequency factor yields the intermediate design also shown above.
 Yagi Antenna Performance Optimization 3-25

Table 3.18 — Performance characteristics of the six-element Yagi discussed

in the text (boom length = 0.78)).
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (dB) (ohms) (ohms)
0.970 10.20 14.8 23.6 — 35.8
0.972 10.29 16.2 23.3 — 32.9
0.974 10.38 17.7 23.1 —29.9
0.976 10.46 19.6 22.9 — 26.9
0.978 10.54 21.8 22.6 — 23.9
0.980 10.61 24.6 22.4 — 20.8
0.982 10.67 28.3 22.1 -17.7
0.984 10.73 31.8 21.9 =A'G
0.986 10.79 30.2 21h -11.4
0.988 10.84 26.3 2k5 -8.1
0.990 10.88 23.2 21.4 -4.8
0.992 10.92 20.8 21.3 -1.4
0.994 10.95 18.9 21.2 2.1
0.996 10.97 17.4 Pah ee 5.6
0.998 10.98 16.0 21.3 9.2
1.000 10.98 14.9 21.5 12.8
1.002 10.98 13.9 21.8 16.5
1.004 10.96 13.0 22.2 20.3
1.006 10.92 12.1 22.8 24.2
1.008 10.88 11.4 23.5 28.1
1.010 10.83 10.8 24.4 32.1
1.012 10.76 10.2 25.6 36.1
1.014 10.68 9.6 27.0 40.2
1.016 10.59 9.1 28.8 44.3
1.018 10.49 8.7 31.0 48.5
1.020 10.39 8.3 33.7 52.5
1.022 10.28 8.0 37.2 56.5
1.024 10.17 Te / 41.4 60.1
1.026 10.06 7.4 46.6 63.2
1.028 9.95 7.3 Lo ji 65.3
1.030 9.85 hee 60.9 65.5

Performance for this intermediate design is shown in Table 3.19 Note that
since all lengths were not scaled (boom not scaled), this intermediate Yagi
is not really quite the same as our starting model; the maximum F/B ratio
has, in fact, fallen to 27 dB. However, this is of no concern; it is now time
to iteratively vary D3 and D1 positions to ‘‘tune up”’ the F/B ratio. Alterna-
tively, if our concept of optimization is correct, iterative variations of D3
3-26 Chapter 3

Table 3.19 — Performance characteristics of the intermediate design six-

element Yagi described in the text; maximum F/B at F = 1.0.
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(aBi) (dB) (ohms) (ohms)
0.970 9.08 7.6 24.3 - 40.5
0.972 9.27 8.4 24.2 — 37.5
0.974 9.45 9.2 24.0 — 34.5
0.976 9.61 10.0 23.9 - 31.6
0.978 9.76 10.9 23.8 — 28.6
0.980 9.89 11.8 23.7 — 25.7
0.982 10.02 12.8 23.5 — 22.8
0.984 10.13 13.9 23.3 - 19.9
0.986 10.24 15.1 23.2 - 16.9
0.988 10.33 16.5 23.0 - 13.9
0.990 10.42 18.0 22.8 -11.0
0.992 10.51 1Oc7, 22.6 —8.0
0.994 10.59 a VA4 22.4 -4.9
0.996 10.66 24.0 22.2 -1.8
0.998 10.73 26.2 22.0 1.3
1.000 10.79 27.1 21.8 4.5
1.002 10.85 25.9 21.7 rETE
1.004 10.89 23.8 21.6 10.9
1.006 10.94 21.7 21.5 14.3
1.008 10.97 19.8 PB Bie 17.6
1.010 11.00 18.2 21.6 7m |
1.012 11.02 16.9 21.7 24.5
1.014 11.03 1537, 21.8 28.1
1.016 11.03 14.6 22.1 31.7
1.018 11.03 13.7 22.5 35.4
1.020 11.01 12.9 23.0 39.1
1.022 10.97 12.4 23.7 42.9
1.024 10.93 11.4 24.5 46.7
1.026 10.88 10.8 25.5 50.6
1.028 10.81 10.2 26.8 54.6
1.030 10.74 9.7 28.3 58.5

and DR could also tune up the F/B ratio. I have carried out both iterations
and the resulting optimized Yagi parameters are as shown:

D3-D1 opt D3-DR opt

element length x x
(\) () "MN
Reflector 0.49343 0.000 0.000
Driven el 0.48167 0.156 0.175595
Director 1 0.44643 0.291564 0.312
Director 2 0.44643 0.468 0.468
Director 3 0.44643 0.64075 0.6328873
Director 4 0.44643 0.780 0.780
 Yagi Antenna Performance Optimization 3-27

Table 3.20 — Performance of the six-element Yagi where the positions of

directors D1 and D3 have been varied to ‘‘tune up’’ the F/B ratio.
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (dB) (ohms) (ohms)
0.970 9.06 8.0 27.2 - 43.1
0.972 9.24 8.7 ert — 40.3
0.974 9.39 9.5 27.0 - 37.5
0.976 9.53 10.4 26.9 — 34.7
0.978 9.66 11.3 26.7 -31.9
0.980 9.78 12.2 26.5 — 29.2
0.982 9.89 13.2 26.3 — 26.4
0.984 9.99 14.4 26.1 — 23.6
0.986 10.09 15.6 25.8 — 20.8
0.988 10.18 17.1 25.5 -17.9
0.990 10.26 18.7 25.3 — 15.1
0.992 10.34 20.7 25.0 -—12.2
0.994 10.42 23.3 24.7 -9.2
0.996 10.49 26.9 24.4 -6.2
0.998 10.56 32.9 24.1 -3.1
1.000 10.63 119.8 23.8 0.0
1.002 10.69 33.0 23.6 3.2
1.004 10.75 27.0 23.4 6.4
1.006 10.80 23.5 23.2 9.7
1.008 10.85 21.0 23.1 13.1
1.010 10.90 19.1 200 16.6
1.012 10.94 17.6 23.1 20.1
1.014 10.99 16.2 23.2 23.7
1.016 10.99 15.1 23.4 27.4
1.018 11.00 14.1 23.7 31.1
1.020 11.01 13.2 24.2 35.0
1.022 11.00 12.4 24.8 38.9
1.024 10.97 TWF 25.6 42.9
1.026 10.96 11.0 26.6 47.0
1.028 10.92 10.4 28.0 51.2
1.030 10.87 9.9 29.6 55.4

Performance of the D3-D1 optimized antenna is shown in Table 3.20; it

is nearly the same as that of the intermediate design (Table 3.19) except that
the F/B ratio at F = 1.0 has gone up from 27 dB to an astronomical
120 dB. Similarly, the performance for the D3-DR optimized antenna is
shown in Table 3.21. Again an astounding F/B ratio figure is achieved;
moreover, the newer optimized beam performance is essentially identical with
that of the first optimized model.
It is instructive to examine the final vector contributions to forward and
reverse waves; Fig. 3.5 shows such a plot for the D3-D1 optimized Yagi and
Fig. 3.6 a similar plot for the D3-DR optimized model. Note that they look
3-28 Chapter 3

Table 3.21 — Performance of the six-element Yagi where the positions of

the driven element (DR) and director (D3) have been optimized through
computer iteration.
F/B feedpoint feedpoint
frequency gain ratio resistance reactance
(F) (ABi) (cB) (ohms) (ohms)
0.970 9.43 9.2 31.6 — 40.7
0.972 9.55 9.9 31.5 — 38.1
0.974 9.67 10.7 31.3 — 35.5
0.976 9.77 1S 31.0 — 33.0
0.978 9.87 12.3 30.8 — 30.4
0.980 9.96 13.2 30.4 —27.8
0.982 10.04 14.2 30.1 — 25.2
0.984 10.12 15.3 29.7 =22.0
0.986 10.19 16.5 29.3 - 19.9
0.988 10.27 17.9 28.9 -—17.2
0.990 10.33 19.6 28.4 - 14.5
0.992 10.40 7 ie 28.0 -11.7
0.994 10.47 24.1 27.5 -8.8
0.996 10.53 27.6 27.0 -5.8
0.998 10.59 33.7 26.6 -2.9
1.000 10.65 150.3 26.2 0.2
1.002 10.70 33.7 25.8 3.3
1.004 10.75 2I5h 25.4 6.5
1.006 10.80 24.1 25.0 9.8
1.008 10.85 21.6 24.8 13.1
1.010 10.89 19.7 24.5 16.5
1.012 10.92 18.1 24.4 20.0
1.014 10.95 16.7 24.3 23.6
1.016 10.97 15.6 24.3 27.2
1.018 10.98 14.5 24.4 30.9
1.020 10.98 13.6 24.7 34.7
1.022 10.98 12.8 25.1 38.6
1.024 10.96 12.0 25.6 42.5
1.026 10.93 11.3 26.4 46.5
1.028 10.89 10.7 27.3 50.7
1.030 10.83 10.1 28.5 54.8

similar, differing only in minute details. Incidentally, it is noteworthy that

the reverse plots show vectorial contributions going around the clock twice,
corresponding to the second null, which we have constructed.
There is one final point worth mentioning. An examination of Tables 3.1,
3.20, and 3.21 reveals that the frequency for the best F/B ratio is not gener-
ally quite the same as the frequency center of the gain bandwidth. It is offset
by an amount that depends only on the boom length. This offset is of small
importance as long as the gain bandwidth is large enough; it is nevertheless
possible to empirically measure the offset frequency as a function of boom
 Yagi Antenna Performance Optimization 3-29

REVERSE

Tee
eee
[oe
aa
Let
GESE
SOE
aS
=AST©
winnie:
imei
EAS
(BE
Esa
Fig. 3.5 — Current vectorial contributions of the D3-D1 optimized Yagi.

length. Let us fix the frequency of best F/B ratio as F = 1.0, and designate
the frequency of (central) best gain (4 percent BW) as Fg (offset frequency
= Fo — 1.0); empirical results are shown in Fig. 3.7. Note that if the boom
length is 0.63 the offset disappears. For booms shorter than this value the
offset is negative, and for booms longer than 0.63) the offset is positive.
But it is clearly possible to design a satisfactory Yagi over a considerable
range of boom lengths without incurring an offset which is comparable to
the bandwidth itself; it is only necessary to take this offset into account in
fixing the original bandwidth over which gain must be high. From a gain
consideration alone the longer booms are best; that is why the example I
used for illustrative purposes had a boom length of 0.78.

Number of Reflectors
It is interesting to consider a major change in possible Yagi antenna design:
to explore the effect of changing the number of reflectors in a Yagi array.
Up to this point we have assumed only a single reflector with a variable
number of directors. It is tempting to consider increasing the number of reflec-
3-30 Chapter 3

Fig. 3.6 — Current vectorial contributions for the D3-DR optimized Yagi
beam.

1.02

oH

Ne
eas]
Beedle)
5 0.6 0.7

BOOM LENGTH (A,)

Fig. 3.7 — Plot illustrating frequency ratio for best central gain to best
F/B. Note that frequency offset disappears for a boom length of 0.63).
 Yagi Antenna Performance Optimization 3-31

=
fea}
Sg
=
=z
oO

ieas
RES
sR
Re
Hass el
ee
hsElga?
eeeGeatBShir
Lis
HES
RBeSE
SHES
uo

FREQUENCY (%)
Fig. 3.8 — Gain of a six-element Yagi beam as a function of frequency
with number of reflectors as a parameter - overall number of elements
held constant.

tors in the hope of a significant improvement in the average F/B ratio over
the entire bandwidth to be used. This question is now easily explored. I shall
assume that the simplistic test Yagi of Fig. 3.1 will be our standard. To keep
conditions other than the number of reflectors as constant as possible I shall
keep the total boom length constant at 0.75\, and the total number of para-
sites constant at five. We shall compare the cases where the number of reflec-
tors is zero, one (our test standard), two, and three. Fig. 3.8 shows
frequency-swept gain curves for all four cases; the curves are keyed to the
legend on the diagram. Severe resonance effects are noticed near the free
space resonances of the reflector (F = 0.96) and the directors (F = 1.06);
these resonances, however, were purposely spread far enough to allow the
4 percent band of interest a good gain figure.
The highest curve (curve 1) displays gain for the standard simplistic Yagi
(same as Fig. 3.1) and it is clearly the best performer. The zero reflector case
(curve 0) yields substantially less gain in the region of interest; it also con-
3-32 Chapter 3

tains no resonance effect at the reflector frequency, because there is no reflec-

tor. The two- and three-reflector cases (curves 2 and 3) show progressive loss
of gain over the original standard; the reason is to be found in the much
lower currents induced in the additional reflectors. Shown below are the
reflector currents when the driver is excited by one ampere at the central fre-
quency (F = 1.0):

number of magnitude of reflector current

reflectors (amps)

1 0.477
2 0.043 0.538
3 0.031 0.046 0.626

Note that the reflector next to the driver has substantial current while all
other reflectors are hardly excited at all. Thus where there are multiple reflec-
tors, the effective boom length is shortened and we therefore should expect
the gain to fall appreciably.
Fig. 3.9 shows the F/B ratio for these same four cases. Clearly the stan-
dard Yagi antenna (curve 1) is superior to the zero reflector case (curve 0).
In the two-reflector case (curve 2) the peak of maximum F/B (correspond-
ing to the second null) has moved significantly higher in frequency. We have
already learned that this occurs when the effective boom length is reduced
(in this case by the relatively ineffective first reflector). This effect is exag-

re)
2
a
a

ites
ele
[5]
slsele
ANG
| el
NJ ieee
Aes
eee
perte
|||
B/sieaed.
FREQUENCY (%)

Fig. 3.9 — F/B of a six-element Yagi beam as a function of frequency

with number of reflectors as a parameter.
 Yagi Antenna Performance Optimization 3-33

gerated in the three-reflector case (curve 3) where the effective boom is still
shorter due to the first two relatively ineffective reflectors.
Thus we now see that there is a very good reason why a Yagi should con-
tain one and only one reflector in the linear boom array; one is definitely
needed to improve the gain and F/B. More than one reflector reduces the
effective boom length and therefore gain; also, because of the relatively small
currents induced in the extra reflectors, they do very little to the basic Yagi
F/B ratio potential.

Missing Parasites
A common observation among Amateurs who have had large Yagi anten-
nas in operation over a period of time is that when a parasitic element is
broken or even entirely missing the Yagi continues to perform surprisingly
well. We may now examine quantitatively just what occurs; for comparison
I shall use the same six-element simplistic Yagi design of Fig. 3.1.
When a parasite is missing, the individual element currents all readjust
to new values; such a readjustment changes the effective boom illumination
function and therefore must cause a change in Yagi antenna performance.
Starting with the standard six-element simplistic design, I have made calcu-
lations of performance when one parasite is missing. Frequency-swept plots
of gain and F/B ratio are shown in Figs. 3.10 and 3.11; the individual curves
are keyed to the legend in the diagram.
Note that there is significant gain displayed for all of the cases. The greatest

(dBi)
GAIN

FREQUENCY (%)
Fig. 3.10 — Forward gain of a six-element Yagi showing performance
when one element is missing. F/B under similar conditions is plotted in
Fig. 3.11.
 3-34 Chapter 3

(dB)
F/B

FREQUENCY (%)
Fig. 3.11 — Front-to-back ratio of a six-element Yagi showing the effect
“of a missing element. Forward gain under similar conditions is shown in
Fig. 3.10.

loss in performance occurs when the reflector, R, is missing; this, of course,

is analogous to the previously discussed zero reflector case but now with a
shorter (residual) effective boom. The most surprising aspect of Fig. 3.10
is the small but real increase in gain occasioned by the loss of D3. This can
only be understood if the readjustment element currents constitute an effec-
tive boom illumination function slightly longer than that for the fully popu-
lated beam; in this event we would expect the frequency for maximum F/B
to be lower than that for the standard case. Fig. 3.11 shows this to be true.
For all other cases of missing parasites the frequency of maximum F/B is
increased, indicating a:shortened effective boom length and hence lowered
gain. The lowered gain is verified in Fig. 3.10.
Thus a missing parasite is not always disastrous. However, if you look
at the performance at the frequency of best F/B, the original fully popu-
lated Yagi is best.
It is now apparent that a Yagi antenna really ‘‘wants to work.’’ Even major
changes, such as a missing inner director, work surprisingly well due to
automatically readjusted element currents. It is now perfectly obvious why
the Yagi antenna is so popular: it will provide reasonable performance no
matter how it is constructed. It will provide top performance, especially in
the F/B ratio, only if carefully made in accordance with the design rules
presented in this chapter.
 Yagi Antenna Performance Optimization 3-35

Summary
In this chapter, I have explored the effects of departures from the simpli-
stic design previously given. The results show:
1) Director length taper schedules have no apparent beneficial effect on
gain or F/B for boom lengths smaller than one wavelength. The impor-
tant design parameter is the average director length — not the taper
schedule.
2) Element placement schedules on the boom also have a marginal ef-
fect on gain or F/B for boom lengths less than one wavelength.
3) The simplistic design is as good as any design for boom lengths less
than one wavelength.
4) A Yagi linear array on a given boom is best when it involves one and
only one reflector element.
5) The F/B ratio at a given design frequency can, in principle, be in-
creased without limit by an iterative design procedure.
6) Very high values of F/B will be available only over very narrow band-
widths.
7) The Yagi antenna is basically very tolerant of major faults. Even
missing parasitic elements cause surprisingly little deterioration in gain.
 BER asin ahamaladl rma Cae

pe s aye orien bie die(GS: et

(eof %
“RETR PAIN unt “rie id wists intbenciefia ¢
mw Vie aN Ae » raed, exhiey ie SO, =
mp ooaily feitiaiz re
unas Se a duibadl ee Rare *
ON EE Abe pe a ara tise FiO
wey oil at yal ni4 yer raleswa Sia
Si genet ino rae biiveles haw We v
sohbetoe: my dRul ezan

ee ’ iy ni
reo! makanaral smbot 31 Aisi oe vie it ARR
rane dein See tikes a eo
| ca “natOh ian at a ‘bralorieh
‘4 ‘ 3 aoe 46: r : rm 7 SH ceva os a
wes
(tort at] ra Pa sae ? ond avis ~*~ we le
A, Fi Wt

Cy Tins Sa alia iy aft al ua ‘saboR: ees its Ay 4) eed

Te Wnet otgeueupisementaueleer 6 iss dint Eatery

foe et lige Youygieti wi kemiles yye-wlig a ra i ais ra wes

‘aig 4} oh Bi ere Ey viy aia ya abneuinen! sana |

' \ iA
yo } i C.i% et Pty ~~Cae,

9 2 yee in. sh me
TT ar £
Vy 7 “Pia ‘ a ey

‘ st? y re , Fi —

\ cue ese ions ay) p=

2
1 -
. , ‘ Da, "4
i ie ’ riety
F 4ie
! We ] ~
; raf {
s ; :
y art ] ‘y > ue ’ SA
5 j cise iy i ' ¥
Pat ie bfaee
Me © + eee Yanga?

' ore it ieee WE) A x . Bt 1 (aa Gysi

‘ , ¥ 4 AP maa
Se . a ' F a en! « * <<
* x ‘ot le ony wae eur Y oT yaa a a ° 7
a) ¥ a SSD) A er Sa '
/ baal , LeSA ' : ’ wt)
ee a re ee"en,
a ee : "Mh al oom RP AM Se oa oi
ce A 2, M4 5% \ BW) ws! : a ;
. ‘ : St j “ eee ae ee e [
5a ean: OE my hres eS aay *SHjis &
aes a oe x
‘ '
eee, 4
yb Rvsacr a Seether, A , a
ory FY
Pee i : Ra as
EAC aa TUG 7 ae wha Se ie via‘Sh [Link]: 4 Lan aa
hs We
. 4 i

ite iY: ' omy ‘30 x? smal ge WA. nora . risen

y ea tik 16 Captaigs
a0, nt mM é iy ultpits halt,ran pa fr rri me
at FM extigy cay 3 eat diy, seathe tates iy ‘ot he
7 a 7, ¢ y ari ye at"
pore Cc A Tite Gees i; a acs aa ui‘5 oy) 7
i fy) we Fi a iS")

ta
7 rw y

+
i. oe 2 ‘ha
yi Aa
+ iw : yy

; 1K a |
tus thy OG © ee OU
CHAPTER 4

LOOP ANTENNAS
U p to this point we have considered only linear cylindrical elements
about one-half wavelength long with a small diameter compared to
wavelength; there is in common use, however, a radiation ‘‘element’’
consisting of a loop of wire about one wavelength in circumference. There
are many such loop configurations: a triangular loop (commonly known as
a delta loop), a square loop with two sides parallel to the earth (known as
the quad), and a square loop oriented so that two diagonal corners are perpen-
dicular to the earth (known as a diamond loop). Loops can also be made
that are not equilateral. Triangles can be isosceles or even with three dif-
ferent sides; four-sided loops can be rectangles, and, indeed loops can have
more than four sides or can even be round.
The one-wavelength loop can be used either as a driven radiator or a
parasite. To drive the loop it is opened at some point on its circumference
where it is excited with a suitable voltage or current. As a parasite the loop
is left closed; current will flow depending upon the induced voltages from
other loops or elements and the self-impedance of the loop. To help unders-
tand the behavior of these loops it is useful to model the loop in terms of
dipole elements.

Square Loop Model

Let’s first consider a model for a square driven loop. The model starts
with two driven half-wavelength dipoles in free space as shown in Fig. 4.1.

See te | a ee OS ee
Lf ¥,
| |
| |
l l
| |
| |
| |
| |
| l
| |
| |
| |
| l
| Xi EX |
eee)” _—«

Fig. 4.1 — Diagram of two driven dipoles with one-quarter wavelength

vertical spacing. Bending the ends of the dipoles together, as shown by
the dotted lines, forms the square loop.
 4-2 Chapter 4

The dipoles are voltage driven at their centers (at xx and yy) and in this model
are separated (vertically) by one-quarter wavelength. If the excitation is equal
at xx and yy (the same voltage and phase), the dipole currents will also be
equal and therefore, the voltages at the ends of the dipoles will be equal.
The outer eighth-wavelength sections of each dipole may now be bent as
shown by the dotted line, forming a square. Since the voltages at the dipole
ends are still equal, the bent dipoles can be connected together without
changing any currents in the square.
Since for each dipole there is a unique relationship between voltage at its
ends and current at its center, the end connection of both dipoles in the square
configuration insures that the center dipole currents will be equal even though
excitation voltages at xx and yy may be different. Indeed, we can remove
voltage excitation at yy (shorting the terminals) and still be sure that whatever
dipole currents flow from excitation at xx, they will be equal in the two
dipoles. One can look at the excitation in this case as a dipole current fed
_at xx; the dipole centered at yy is voltage fed at its ends. Note that to realize
the same loop current the sum of the voltages supplied to xx and yy must
be constant. This leads to the well-known fact that the driving-point
impedance at xx with yy shorted will be just twice the impedance at xx when
yy and xx are excited equally.
Fig. 4.2 shows the square loop excited at the center of the bottom section
and also shows the relative magnitudes and directions of the currents which
flow. Note that the horizontal sections of the square show currents in the
same direction as those of the original dipoles of Fig. 4.1; therefore these
sections will provide radiation
Se nes —e fields at long distances. Since
y’ ¥ they are shorter than the original
| half-wavelength dipoles, however,
they require more current to pro-
duce a given radiation field, i.e.,
the radiation resistance will drop.
—_ rd
Moreover, the directivity and
gain of each segment will be
somewhat smaller than that of a
| . i full half-wavelength dipole. Both
of these effects will be discussed
shortly.
The vertical portions of the
=—_ x x _—S
square loop shown in Fig. 4.2 are
SS ee, eee
quite different. Note that for
eran jeaumctiee lee each of these vertical segments
ig. 4.2 — Outline of the square loop .
showing the current distribution in the LOR and bottom porate
the horizontal and vertical members. have identical currents but in
Horizontal polarization results with opposite directions. This sym-
the feed points as shown. metry assures complete cancella-
 Loop Antennas 4-3

tion of the far (radiation) field. In other words, the vertical sections do not
radiate; they simply act as a capacitive ‘‘top hat’’ loading for the horizontal
radiating segments. Thus the radiating properties of the square loop of Fig.
4.2 will be identical to that of the truncated horizontal segments alone. These
will, of course, produce only horizontally polarized radiation.
Take another look at the
Table 4.1 — Broadside gain versus two half-wavelength dipoles of
separation for two parallel half- Fig. 4.1. The two (broadside)
wavelength dipoles. dipoles separated by one-quarter
separation gain stacking gain wavelength will produce a gain
(wavelengths) (ABi) (AB) increase over one dipole. This
0.000 2.15 separation or stacking gain can be
0.125 2.43
0.250 3.24 easily calculated by using known
0.375 4.51 self- and mutual impedances; the
0.500 5.98 result is shown in Table 4.1, and
0.625 6.94
0.750 6.76 also in Fig. 4.3 where the overall
0.875 5.83 gain is plotted against the separa-
1.000 4.93 tion. Note that the gain improve-
1.250 4.37
1.500 5.28 ment peaks when the separation is
1.750 5.84 about 5/8 wavelengths, leading to
2.000 5:10 the often-quoted (but generally
incorrect for other than single
elements) ‘‘optimum stacking separation.”’
Fig. 4.3 can be understood qualitatively by remembering that the effective
aperture cross section area for a single half-wavelength dipole is about 0.13
square wavelengths. Thus, when the separation between the dipoles is large

(d8I)
GAIN

1.0 2.0
SEPARATION (A)

Fig. 4.3 — Broadside gain versus separation for two side-by-side dipoles.
4-4 Chapter 4

enough, the aperture areas are basically independent (and thus additive),
leading to an effective gain improvement of a factor of 2 (3 dB). For smaller
separations, however, the aperture areas overlap. In this overlap region both
constructive and destructive interference can occur; for very small separa-
tions the combined aperture areas reduce to that of a single dipole, but for
some intermediate regions interference occurs making the gain improvement
more than a factor of 2. This constructive interference is due to a favorable
reduction in the driving-point resistance of the dipole, which is a direct result
of the behavior of the real part of the mutual impedance of the two dipoles.
Another qualitative way of understanding this entire phenomenon is to
view the (transmitting) dipoles as an excitation of a single vertical aperture.
Broadening this aperture by separating the dipoles is tantamount to narrowing
the H-plane pattern, which will increase the gain. When the dipole separa-
tion becomes large enough, however, the quasi-uniform illumination disap-
pears and the vertical aperture acts like two dependent (small) apertures giving
rise to a diffraction pattern in the H-plane with maxima corresponding to
a gain improvement of exactly a factor of 2. In any case, the actual calcu-
lated values of gain are shown in Table 4.1. Note that for the quarter-
wavelength separation of Fig. 4.1, the gain of the stacked dipoles is 3.24 dBi,
or just 1.09 dB above that for a single half-wavelength dipole. This gain
increase is all due to beam pattern narrowing in the H-plane; the E-plane
pattern beamwidth remains the same as that for a single half-wavelength
dipole.
The square loop of Fig. 4.2 is
very similar to the stacked
dipoles of Fig. 4.1, but there are
two significant changes. Because
of the truncated or shortened
elements, the driving-point im-
pedances of the elements are
reduced and the E-plane pattern
width is somewhat increased,
resulting in a somewhat reduced
gain. To calculate these two
results I will use the same
method outlined by Kraus (pages
139-143).! Kraus’s calculation
applies to a thin radiating Fig. 4.4 — Current relationships for
element which is not capacitance the symmetrical, thin center-fed trun-

loaded; to make the calculation cated antenna of length /. This is the

eS EE ATC ORES relationship necessary to determine
appry & driving-point impedance and H- and
element fed at the maximum E-plane patterns of the square quad
current points, the limits of loop.

'This and other footnotes appear at the end of this chapter.

 Loop Antennas 4-5

integration must be suitably altered. Fig. 4.4 shows the essential geometry.
The retarded value of the current at point z on the antenna referred to
a point at distance, S, is

I = Ip [cos(2xz/d)
eset —8/0) (4.1)

where the quantity in brackets is the form factor for the current on the
antenna. Following the Kraus development, the antenna can be viewed as
a string of infinitesimal dipoles of length dz. The far fields, dE, and dHg,
at a distance, S, from an infinitesimal dipole, dz, are

dHy = j. ISoy
sind
% (4.2)

dEg = 1207dHy (4.3)

where ¢ is the angle around the z axis. The total field from the entire
antenna is then
1/2
H,= | dH (4.4)
—1/2

From (4.1), (4.2), and (4.4),

Hy = . j——
Iosind ee”4 Eej —cos(2rz/r)
1 —jwS/c
e~J%?/<dz (4.5)
2d S
—1/2

Note that
S = r—zcos6 (4.6)
and also note that at long distances the amplitude of S is the same as
the amplitude of r, so that we may write:

— ; Apsind eeeja(t—r/c)
Hay ajo sy i(wz/c)cos0 gz
J cos(2xz/djes(z/c)cos8 (4.7)
2\r ~1/2
Let B = w/c = 2z/); then (4.7) may be rewritten:

V9)
pee BIpsinO jut —r/c) j e48zc089 cos(Bz) dz (4.8)
4ar —1/2

Since

e™[acosbx + bsinbx]
j e*cosbx =
a+ D
4-6 Chapter 4

then with a = j@cosé and b = 8B, (4.8) becomes:

; ; + 1/2
ERNE GIosiné pio(t—r/c) pjBzcos0 JBcosécosBz + BsinGz
4ar B?sin20 -I/2 (4.9)
Evaluating this expression at both limits and collecting terms:

jae pa) (4.10)

2ar
where

[I] = Ipeioltr/o) (4.11)

and

J
F(@) = —_,{cos(s--cos6)sin(6—-)
sin 2, 2
—cosdsin(3-cosd)cos(8-—)]
2 2
(4.12)
:

F(@) is often referred to as the field pattern factor. Thus the far fields
of the truncated element can be written:

Hg = jE re) (4.13)
and

Ey aj an F(@) (4.14)
Note that if there is no truncation (/ = )/2) (4.12) reduces to the well-
known expression for a half-wavelength dipole:

cos(-+-cos@)
FO) = :
sin6

Kraus has shown that the self-resistance, R,;, of such a linear element can
be computed by equating the integral of the Poynting vector over a large
sphere (total power radiated) to the driving-point (current maximum) total
power supplied. The result is (see Kraus, ref. 1, equation 5-90, page 143):

Ri, = 60 § F2(6)sinedo , (4.15)

0

For the truncated element it is now a simple matter to insert the value of
F(0) from (4.12) into (4.15) and integrate. The integration is quite easily done
numerically with a simple computer program. The result of such an integra-
 Loop Antennas 4-7

Table 4.2 — Self-resistance and gain Beg SUN LG tg

versus length for a truncated dipole _—the self-resistance of a truncated
element. dipole element of length / (in
length resistance directivity gain terms of wavelengths) is shown.
(r) (ohms) (dBi) The directivity or gain can also
0.05 2.0 1.502 17, be easily computed. Once F(6) is
0.10 7.6 1.510 1.79 : : o sie Bein mt ges
0415 163 11500 4/92 given, the directivity is simply the
0.20 270 1.537 1.87 ratio of the maximum value of
0.25 38.5 1.557 1.92 F*(6) to the average value over
0.30 49.8 1.578 1.98 ; - :
0:35 sala HieGG Bas the entire 4m solid angle. That is,
0.40 67.0 1.619 2.09 the directivity D is:
0.45 71.6 1.635 2.13
0.50 73.1 1.641 2.15 pre 4nF2(6
= 1/2)
Tv
2x | F2(6)sindd6
0 (4.16)
or from (4.15)

D = 120 pegen/2) = 12 Ging5 (4.17)

Ry Ri 2

The calculated directivity, D, and the corresponding gain, expressed in dBi,

are also listed in Table 4.2.
Note that for small truncations where / is only slightly shorter than one-
half wavelength, there is not much reduction in self-resistance or in direc-
tivity; this is to be expected because the small ends of the dipole which are
truncated carry little current, so do not contribute greatly to element perfor-
mance. For heavy truncation, however, both self-resistance and directivity
decrease significantly; the limiting case where the length goes to zero is the
well-known infinitesimal dipole whose directivity is just 1.500 (1.761 dBi).
To compute the driving-point resistance of the full square loop it is neces-
sary to know the mutual resistance between two (truncated) elements sepa-
rated by one-quarter wavelength. From quite fundamental considerations
Hurwitz? has shown a mathematical expression for the real part of the com-
plex mutual impedance between two elements, R> :
TT
Ry, = 60 | F*(6)sinéJo(BSsiné)
dé (4.18)
0

where F(@) is the pattern function and Jp is the Bessel function whose argu-
ment is the product of element separation S and @siné. If S is measured in
units of wavelength the argument is simply 27Ssin@. Note that for very small
separations Jp approaches unity; thus, for very close separations Rz; ~ Rj).
Shown in Table 4.3 are values of mutual resistance versus separation for
truncated elements. Note that the mutual resistance behaves very similarly
4-8 Chapter 4

to the values for full half- :

length dipoles (see Chapter Table 4.3 — Mutual resistance versus
Wane Pe P separation for one-quarter wavelength
1) but the magnitudes aremuch truncated elements.
smaller; a careful company separation resistance separation resistance
shows that the reduction factor —) (ohms) (—) (ohms)
varies somewhat with separation. 0.00 38.5
The properties of the square 0.05 oh pee ays
loop can now be computed. ee 32.0 0.90 -~4.2
Table 4.2 shows that the gain of 0.20 27.3 0.95 —tal
a single (truncated) quarter- 0.25 21.7 1.00 Af
wavelength element is 1.92 dBi 0.30 of Nee sy
and that Ry; = [Link] 9/40 aye aioD 65
the full loop of Fig. 4.2 there 0.45 -1.7 1.40 3.2
are two ‘‘elements;”’ if they are 0.50 -6.2 1.50 pi
both equally driven, they will 0.55 3ae if cee
produce the same far-field i 13.0 1.80 —4.8
strength and power density 0.70 -12.9 1.90 -2.5
as a single driven ‘‘element’’ 0.75 — 11.8 2.00 UE
carrying twice the current.
However, the driving-point resistance of each one of the driven elements,
when both elements are equally excited, is Rj; + R2 . Therefore, the total
input power (both elements) is:

W2 = 217(Ri1 AP R)}) 4.19)

For the single element alone (same far field) the total power is:

W, = (20°Ry | (4.20)
If the directivity of a single element is designated as D, and the directivity
of the full loop as Dp, then

D> = DyWi/W = 2D\/( + Ro\/Rj)) (4.21)

From Tables 4.2 and 4.3 values for the square loop are: R,; = 38.5 ohms;
R2, = 21.7 ohms; D2/D, becomes 1.279 (1.07 dB). Note that this stacking
gain of the two truncated elements is nearly identical to the stacking gain
of half-wavelength dipoles at a separation of one-quarter wavelength (1.09
dB, see Table 4.1); the difference is due only to the details of mutual
resistance. The driving-point resistance, R, of the total loop, of course, is
twice the value for a single element (when both are driven):
R = 2(Ry, + R21) = 120.4 ohms (4.22)

Since D, (see Table 4.2) is 1.557, Dy = 1.991 (2.99 dBi).

These properties of the square loop have been obtained rather rigorously;
the main assumption in the model is total neglect of far-field radiation from
 Loop Antennas 4-9

the vertical sections, which are assumed to act as capacitance sinks for the
current at the ends of the horizontal radiating segments. Moreover, the
assumption is made that the current distribution on the horizontal segments
is strictly sinusoidal; this is valid for very thin elements. Most loops are built
with wire, which is thin compared to a wavelength, so one can be quite
confident of the model.
It is now easy to understand qualitatively the radiation pattern of the loop.
The A-plane profile does narrow, compared with a single linear element,
because of the ‘‘illumination’”’ of a wider vertical aperture; quantitatively,
this profile narrowing is almost the same for truncated loop ‘‘elements’’ and
the equivalently separated half-wavelength dipole. The E-plane profile,
however, is not as narrow for the truncated element as for a half-wavelength
dipole; this factor accounts for the somewhat reduced directivity (see
Table 4.2). The loss in gain is about 0.23 dB for the quarter-wavelength
truncated element.

Other Driven Loops

Before I consider a multiloop system I
will briefly discuss other forms of driven
loops. First, I will determine the perfor-
mance of the square loop when it is driven
at a point other than the center of the
bottom segment. Fig. 4.5 shows the square
loop with three different feed points: the
center of the bottom leg xx, the center of
the right vertical leg yy, and at an arbitrary wi
et
point, zz, placed at a counter-clockwise ————
distance d (from xx) around the loop. It has puns we
already been shown that if excitation is Fig. 4.5 — Example of a
applied at xx (with yy and zz shorted), the square loop with the excita-
loop will produce an horizontally polarized tion applied at one of three
far field; its gain is 2.99 dBi. Similarly, if different places.
excitation is applied at yy (with xx and zz
shorted), only vertically polarized far-field radiation will occur; gain is again
just 2.99 dBi. Note that excitation xx and yy are basically independent, i.e.,
unit current flow due to excitation at xx has a null in current at yy and vice
versa.
If we now excite the loop at zz with the same unit current, Jgcoswf, it is
easy to see that current flow at xx is just Jgcos(Gd)coswt and at yy is
Ipsin(6d)coswt. These currents will produce orthogonal far fields which must
be added vectorially to obtain the total far field; the total far field has exactly
the same magnitude as that which is produced by the same unit current at
xx alone. In other words, the square loop gain and the drive-point resistance
are totally independent of the feed point; only the polarization changes (from
 4-10 Chapter 4

totally horizontal if the feed is at xx to totally vertical if the feed is at yy).

This simple theorem can be easily proved by the same type of argument for
any equilateral one-wavelength loop.
It is now interesting to consider moving the feed point zz to the lower right-
hand corner of the square loop. Since we are considering free space loop
gain for which rotation is unimportant, it is clear that this configuration pro-
duces exactly the same result as the familiar square diamond loop fed at the
bottom or top corner. Thus we now know that the quad and the
diamond loop have exactly the same gain and exactly the same drive-point
resistance. Similarly, the gain and drive-point resistance of any equilateral
(one-wavelength) loop is totally independent of the position of the drive point
on the loop.
I shall now return to the horizontally polarized square of Fig. 4.2. We
have shown that a square loop (relative to a halfwave dipole) has a somewhat
enhanced gain (+ 0.84 dB), made up of an increase (+ 1.07 dB) due to the
“vertical separation of the radiating segments (H-plane narrowed somewhat)
and a decrease (— 0.23 dB) due to the truncated radiating segments (E-plane
broadening). Let’s now explore the performance of a rectangular one-
wavelength loop; Fig. 4.6 shows some examples.
Rectangle A is a wide but low loop which is recognized as a folded dipole
loop; B is a narrower and higher loop than the square, and C is a high but
very narrow loop. For each of these loops the sum of the width and height
is constrained to be 0.5 wavelengths. We are now in a position to compute

Fig. 4.6 — Diagram of three different rectangular configurations. In each

case the feed point is located for horizontally polarized radiation.
 Loop Antennas 4-11

Table 4.4 — Directivities, gain and self-resistance for one wavelength

rectangular loops with various width-to-heighi ratios.
width height D, DJ/D, Do gain resistance
(wavelengths) (wavelengths) (ABi) (ohms)
0.05 0.45 1.502 2.073 3.114 4.93 3.8
0.10 0.40 15100 181570 212.7417. 24:38 16.7
0.15 0.35 1.522 1596 2430 3.86 40.8
0.20 0.30 1.537 1.419 2.181 3.39 76.0
0.25 0.25 1.557 1.279 1.991 2.99 120.5
0.30 0.20 P57e pie el Fe B50 48 2.67 169.8
0.35 0.15 1.599 1.094 1.750 2.43 PMT (oT
0.40 0.10 1.619 1.041 1.686 2.27 257.5
0.45 0.05 1.635 1.010 1.652 2.18 283.6
0.50 0.00 1.641 1.000 1.641 2.15 292.5

loop performance. Table 4.4 shows the results of calculations for rectangular
loops by the same methods as used for the square. D2/Dy, is the H-plane
directivity increase due to the vertical separation of radiating segments (see
(4.21)), while D, is the directivity of a single horizontal radiating segment
(see Table 4.1). Note that the limiting case of the folded dipole (width =
0.5, height = 0) shows a gain of 2.15 dBi (identical to a single halfwave dipole)
and a driving-point resistance of 292.5 ohms (just four times that of a single
halfwave dipole). Since reactance effects of the full loop are double that of
a single ‘‘element,’’ the Q of the folded dipole will be just one-half the Q
of a single halfwave dipole, leading automatically to a bandwidth twice as
large. As one reduces the loop width from this limiting case the gain increases
monotonically. However, this favorable increase in gain is automatically
accompanied by a significant reduction in driving-point resistance; since reac-
tance effects are essentially identical for all loops, the circuit Q increases com-
mensurately. Thus the potential gain obtainable from high narrow loops is
always offset by unfavorably high values of Q and corresponding narrow
bandwidths.
Let us now consider other loop shapes. For equilateral shapes, we have
already seen that gain and driving-point resistance are independent of feed
point position on the loop. Moreover, a reasonably rigorous solution has
been obtained for a square loop. Because of the independence of properties
on feed point position, which is equivalent to independence of loop rotation
with feed point fixed at the bottom, it seems reasonable to assume that gain
from all equilateral loops is approximately equal to that of a circular loop
having the same enclosed area. This in turn can be equated to an equivalent
square of the same area. Such an intuitive model, together with a model of
how E-plane directivity varies with element length (see Table 4.2) and how
H-plane directivity varies with element separation (see Table 4.1), allows a
reasonable estimate of equilateral loop gain. Table 4.5 lists these values for
several equilateral loops. The values for the square are the ones already com-
 4-12 Chapter 4

Table 4.5 — Estimated directivities, gain and self-resistance for one

wavelength equilateral loops.
equivalent
square loop
equilateral loop side length D, D/D,; Dz gain resistance
type sides (wavelengths) (aBi) (ohms)
triangle 3 0.219 1.545 1.205 1.862 2.70 104.7
square 4 0.250 Lye 1.279 1.991 2.99 120.5
pentagon 5 0.262 1.562 1.309 2.044 3.10 126.4
hexagon 6 0.269 1.565 1.324 2.071 3.16 129.3
octagon 8 0.275 1.567 1.338). 2.097 4° 3:22 132.0
circle 09 0.282 1.570 1.356 2129 3.28 135.3

puted (see Table 4.4); all others are estimated by this simple model. It is quite
easy to see that the popular triangle or delta loop is slightly lower in gain
(by 0.3 dB) than the square (quad or diamond) and that the loop with the
‘highest gain is a circle. However, note that the gain of the circular loop is
estimated as 3.28 dBi which is 1.13 dB larger than that of a halfwave dipole.
This is not quite as large an increase as the approximately 2 dB which was
quoted by Lindsay, but I believe this discrepancy is probably not outside
of the experimental accuracy range of Lindsay’s measurements combined
with the estimation accuracy range of the simple model I have used.?

Multiloop Arrays
I shall now consider an array of square loops with both driven and parasitic
loop elements. From what we have just seen, the array should behave just
about like a two-array stack of equivalent Yagi antennas, separated (vertically)
by one-quarter wavelength and having elements bent together to form the
individual square loops. However, to proceed with a computation of the
properties of the parasitic loop array, we must know all complex self- and
mutual impedances of the truncated elements. Up to this point it has been
possible to rather rigorously describe the properties of a driven loop because
only the real part of self- and mutual impedances were required to obtain
gain, driving-point resistance, and pattern information (both in the E- and
H-planes).
When parasitic elements are involved, the imaginary parts of the impedance
terms are required. Computation of these reactances is a non-trivial exer-
cise, and I am unaware of any rigorous procedure for carrying out such a
calculation for truncated elements. As far as self-impedance is concerned,
the reactive value is controlled entirely by the ‘‘tuning’’ of the loop; that
is, the relationship of wavelength to loop circumference and the effective
loop Q. The complex impedance has been calculated for a linear nearly
halfwave thin cylindrical element.4 The method involves treating the metallic
cylinder as a boundary-value problem (tangential components of electric field
are made to vanish at every point on the conductor surface), from which
 Loop Antennas 4-13

an integral equation is derived. Approximate solutions of this integral equa-

tion yield the current distribution on the cylindrical element from which the
input impedance, including the imaginary component, was derived.
I have been unable to find an equivalent rigorous boundary-value calcula-
tion for the square loop; thus we do not yet have the basis for calculating
the precise reactance for a nearly one-wavelength square loop. However, it
is possible to at least estimate the loop Q (but not its precise resonant fre-
quency) by remembering that reactance changes with frequency are due to
the effective inductance and capacitance of the antenna, that is, near-field
stored energies, whereas the resistance (radiation) has to do with the far field.
Truncating the halfwave dipole changes the geometry of the element but
hardly affects its (central) inductance or (end) capacitance. It would be
reasonable, therefore, to assume the square loop contains essentially the same
total reactive impedance changes as two halfwave dipoles. Thus, following
the argument made in Chapter 1, we may write for the loop self-impedance:

Zyy= Ry + Xi ~ Riyll + LO — FR)] (4.23)

But (empirically)

R,,Q0 = 430.8 logkK — 339 (4.24)

so that
Z; ~ 120.5 + (861.6 logkK — 678)(F — Fr) (4.25)

As in Chapter 1, F is the normalized frequency and Fp is the normalized

resonant frequency of the loop. Loop Q is readily estimated from (4.24) and
(4.22). The only remaining problem is determination of Fp. Although there
is no rigorous way of calculating Fp from basic principles, it is significant
to note that the region where the ends of the two dipoles are joined must
have electric (capacitive) fields at right angles to the conducting cylindrical
element, exactly like those of an infinitely long straight cylinder near a vol-
tage node. This observation implies that there should be a negligible ‘‘end
effect’’ at the capacitive voltage loop, and, therefore, Fp should be very close
to the frequency at which the total node circumference is just one wavelength.
When we consider the imaginary part of the mutual impedance between
loop halves or ‘‘elements,’’ another computational complication arises. At
long distances or separations, the imaginary part of the mutual reactance,
X1, must be (except for a phase shift) simply related to the real part, Rp,
and this relationship should be unaffected by the precise ‘‘tuning”’ of the
‘‘elements.’’ However, at very small separation Xz; must approximate the
value of X,;, which is fundamentally fixed by circuit Q and resonant fre-
quency and not at all by R,;. How to correctly represent X >, at all in-
termediate spacings has not, to my knowledge, been solved quantitatively.
For this reason I will model the loop array as the equivalent Yagi stacked
4-14 Chapter 4

Table 4.6 — Element lengths and positions for representative beams. Radius
of all elements = 0.00052599 wavelengths.
number of | element lengths boom element Fig. 4.7
elements (wavelengths) length spacing curve
refl. driver dir. (wavelengths) (wavelengths)
2 0.4937 0.4705 — 0.15 0.150 2
3 0.4980 0.4896 0.4690 0.25 0.125 3
6 0.4953 0.4803 0.4481 0.75 0.150 6
7 0.4936 0.4762 0.4466 1.25 0.208 7

at a spacing of one-quarter wavelength, then apply necessary corrections to

directivity and gain. Compared with stacked (halfwave element) Yagis, the
loop array will have 0.23 dB less directivity and gain (all due to E-plane pat-
tern broadening).
It is interesting to calculate the free space ‘‘stacking gain’’ of Yagis as a
function of vertical separation. I chose as representative boomlengths and
number of elements the following simplistic Yagis, which were found to have
excellent free space properties (see Chapter 2): a two-element beam on a
0.15 wavelength boom, a three-element beam on a 0.25 wavelength boom,
a six-element beam on a 0.75 wavelength boom, and a seven-element beam
on a 1.25 wavelength boom. Table 4.6 lists characteristics of these beams.
Fig. 4.7 shows the computed gains of these stacked configurations as a

(d81)
GAIN

STACKING SEPARATION, S(A)

Fig. 4.7 — Stacking gain versus separation for representative two, three,
six, and seven-element Yagis.
 Loop Antennas 4-15

function of the stacking separation. Note that the variation in gain with
separation is somewhat different for each case and also somewhat different
that the case for dipoles shown in Fig. 4.3. Two things seem to be occurring
as the separation increases. For small separations, the capture area of one
Yagi essentially overlaps that of the other Yagi; therefore, the total capture
area for both is essentially the same as for one alone. As the separation
increases, the total capture area increases and ultimately doubles if the spacing
is large enough. For the larger Yagis (where the original capture area for
one Yagi is large), it is easy to see that to realize a given separation gain,
the spacing must be relatively larger than for a small Yagi or especially a
dipole. In other words, the transition from the gain of one Yagi to the
doubling of gain (3 dB increase) for two Yagis requires /arger separation for
larger Yagis. In addition to this rather gradual gain increase due to separa-
tion of capture areas, Fig. 4.7 suggests that the constructive interference due
to mutual impedances noticed for the dipoles in Fig. 4.3 also persists in
stacked Yagis. An increase in gain is noticeable at separations of 0.6 and
1.6 wavelengths. Qualitatively, Fig. 4.7 shows a combination of these two
effects; first, the constructive impedance gain increase at particular separa-
tions and second, the capture area separation gain increase which takes place
more slowly with large Yagis.

SPAIN Sela es Tet

tes Table 4.7 — Measured and comput-
the computed gain increase of ed stacking gain versus spacing for
the seven-element beam with seven-element Yagis.
the experimental
: : ‘ values
4 e reported spacing Viezbicke computed
by Viezbicke in his Fig. 11A.° (wavelengths) (aBi) (dBi)
Table 4.7 shows a comparison of 0.38 0.80 1.3
my computed results with Viezbicke’s 0.57 1.58 2.15
; A 0.78 1.36 1.8
published experimental values. The 0.99 1.90 2 35
comparison is not totally valid 1.20 2.34 3.0
because Viezbicke’s seven-element 1.40 2.53 3.25
1.61 1.93 33
beam is not the same beam I have
used in Fig. 4.7. The differences
cannot be quantified, since Viezbicke failed to include specifications for his
seven-element beam. Nevertheless, Table 4.7 shows good qualitative agree-
ment and even fair quantitative agreement (close to Viezbicke’s stated experi-
mental accuracy of 0.5 dB).
Computations represented by the data shown in Fig. 4.7 cannot really be
carried down to very small separations with any confidence because the
mathematical model uses mutual impedances of full halfwave elements. When
reactive parasites get very close together their mutual impedance has an
imaginary component quite different from that of halfwave dipoles.
I shall now examine the stacking gain increases for a separation of one-
quarter wavelength. Table 4.8 shows these increases for the four computed
cases (dipole data from Fig. 4.3). Note that the quarter-wavelength stacking
4-16 Chapter 4

Table 4.8 — Stacking gain for two Yagis 24 for very large Yagis is not
one-quarter wavelength apart, and as large as that for dipoles; this
estimated gain of a quad over a single can easily be understood
Yagi, for various number of elements. because of the (relatively)
Yagi quad gain smaller increase in capture area.
number of | boom stacking over single ; . :
elements length gain Yagi One might expect this i ues
(—\) (dB) (0B) increase to fall monotonically
1 0.00 1.09 0.86 from the value of 1.09 dB (for
re 0.15 1.03 0.80 dipoles) as the array length is
3 0.25 1.38 1.15 .
increased, however, Table 4.8
6 0.75 0.84 0.61 ‘
7 1.25 0.65 0.42 shows the actual increase to
vary somewhat due to the
detailed way in which impedances vary.
We are now in a position to estimate the performance of a loop array.
_If instead of the dipole elements of the stacked Yagi arrays I use (bent)
elements connected in square loops, I will make a loop array in which all
conditions remain about the same except that the E-plane pattern is broadened
and the gain is correspondingly reduced due to the truncated ‘‘elements.”’
The gain performance of the loop array is therefore about 0.23 dB lower
than Yagis stacked one-quarter wavelength apart. These estimations are also
shown in Table 4.8.
It must be evident by now that an array of square loops is very much like
an equivalent Yagi. Its overall gain is expected to be somewhat higher than
the Yagi because of individual loop gain, but not by very much. The E-plane
pattern is slightly broader due to the current distribution on the truncated
‘‘elements,’’ whereas the H-plane pattern is slightly narrower due to the
‘*stacking gain’’ of the loops. Because of this similarity of Yagi and loop
arrays, it is plausible that one can intermix loops and linear elements to pro-
vide a hybrid structure of roughly equal performance. Such a hybrid is known
as a ‘‘quagi’’; if properly constructed, it should provide a pattern and gain
intermediate between a similar quad and a similar Yagi. There are obviously
an enormous variety of possible quagi configurations; there will remain for
some time a challenge to the quagi designer to determine preferred configura-
tions and best dimensions for all radiating elements.

Summary
A number of interesting conclusions have been reached regarding antenna
arrays constructed with conducting loops roughly one wavelength in cir-
cumference.
1) A single (driven) loop will provide a free space gain somewhat larger
than that of a halfwave dipole. The gain increase comes about through
a narrower H-plane pattern and a slightly broader E-plane pattern.
2) Loop gain varies significantly with shape. For rectangular loops fed
in the center of the lower horizontal segment, gain depends on the ratio
 Loop Antennas 4-17

of width to height, varying from 2.15 dBi (equal to a halfwave dipole)

to about 5 dBi (3.8 dB above a halfwave dipole) as the ratio approaches
zero.
3) For equilateral loops, gain depends only on the number of sides (see
Table 4.5). For the square loop, the free space gain calculated rather
rigorously is 2.99 dBi and driving-point resistance is 120 ohms. It is signifi-
cant that these properties of the square loop are totally independent of
the feed point on the square. As an example, the gain and drive-point
resistance for a square loop is exactly the same as that of the diamond
square.
4) The free space gain of a loop array is estimated to be somewhat higher
than that of an equivalent (single) Yagi. Calculations show this difference
depends somewhat on the particular array (see Table 4.8), but ranges from
about 1 dB for short arrays to less than 0.5 dB for long arrays.
5) Quagi configurations are expected to show performance figures be-
tween those of an equivalent Yagi and equivalent quad.
6) A rigorous theory does not yet exist for self- and mutual quad loop
reactances. Consequently, quad (and quagi) parasitic loops must be
experimentally adjusted for correct resonant frequency. It is unlikely that
such experimental adjustments can be made with the same precision and
confidence that linear element lengths can now be specified by present
theory. Therefore, the slight gain advantage of the quad over the Yagi
shown in Table 4.8 may well disappear in practice.

NOTES:
1Kraus, John D., Antennas, New York: McGraw-Hill, 1950.
2Hurwitz, H., W2HH, Private communication.
3Lindsay, J., ‘Quads and Yagis,’’ QST, May, 1968.
4Hallen, E., ‘““Theoretical Investigations into the Transmitting and Receiving
Qualities of Antennae,’’ Nova Acta Uppsala (Sweden), Series IV,
Volume 11, Number 4, pp 3-44, 1938.
5Viezbicke, P. ‘‘Yagi Antenna Design,’’ NBS Technical Note 688, U.S.
Department of Commerce, Washington, D.C., December, 1976.
 : pees
(ex uy at ‘ ;

tian bisiiinalpigeiuaneat CENA Sa

ONO t
4

: . Fy i aa Ati‘iUs cy \ irs . me
a gee Ki
[Link] ‘ Lary’ Tek
° ‘ibe reais
inpa + aNiGeit wae

Ker ihNe "i 4 44 She; Wh

te eK ’ ‘ hii,aii a?ii I Ey Ate pawat

Ren eat oni wana pont sr ylcbenealy ith imide cep th

_ Tatiay beter iapoereent aa aig a9 Herat
" aN wan‘i sehen bitiSUA RNEN Nise sae
Na Ayi voit iN [Link] sai ms4) quae: aif nth)
we) :
A i
ei AM

¢ Aa sehth, wn clea: ¥ " vit rv sik afk pei:of: ies

af UL LAS

tutes Alchall owt etn wei fioath regisaanyney ti ia

9 ih ani enV whi hi ery, eit
; r ry , phi 2 i é ni

7) ated aaa wre bh ad pt theweon a!oa bh Reg }

“om sntiyet Asnnsia Me
ty pobaciet aes anyane(OER W AR):pete) tran ‘aigeet
pdr NM an wath roe udapi deritaleSR §
VATIGR, rh} Set wasaee hy
‘a Cit i HF Var, sisal ty ‘
A i .( 4 sit ih ft

pra inh ate bemcr ALD ait ac a BY

he : CRAG Hitly wn ;bia taeae juts’aS
i te

he. ve m wanes.
ie adit# Nil‘J ae

al ra me asits. idny IEMA: reas rein ve

bps ot) Aaa ae, reite waaay lpm|

la
LAN hints TAPE SER eT pannel ee thi,IG
ae ns te OES tend hyhe n ee rey te
sayAny ti
4 wh Wilae pain

(sul itabiict Vind ea hs te sais

tee
¥ li Keip
Ht SP cewed A i aie athaatarn Cte eh ba Lave ill a he ey a iy Ayeh
. ‘iuse abTen Papo oe ak thect
Ly ’ “ inh Me
laa" aarti |

\ D owl hae baked TEMA as ERS a Ve a

pil H yh Ualied ae ea lTehy AN Hil: t's big i al wigs dy Ae %
' ye vii ’ oF : 4
Y i hi rr eae eevee eae ha Raa “teat \bay
i sitMANTA. ch SUMOAE AO

MT MAN eae ae ee aot eat {hain RRM im a

! Wer haal at) + tein 1 ight yy MER Ab Hy } Ui ai x eit = y ay ried thnyeita As
ra aN Ma :
AP AHO leh. Dak PE iReateeyFabien ena ipeavnindaneya’ co
Marne
ew tent
Paya i es Orie anand
rey ;Diba
a Ly : pe
ia tical Banh ee"
f.
Nahin
uted )
ehy
igh Mi ry nf
| 4 | ili)
EAS haaxcs rs MODUL
ly ew a : ‘de

rvet CO bie ae Sint vga?

5 Gane th
|
Hy "ibsupie®eglahe

[ fit i ayeee
PatyApu ai
BAhy

Sree fed DUE clays it

* Pt riwine iy5hie 5 Ba eat i nea
p13el Wipes
a
Rae et jae
ee
iswrit cet Notary its
nett
‘
CHAPTER 5

THE EFFECTS OF GROUND

o far the antennas I’ve discussed have been considered in free space
Sy conditions. In the real world, however, the performance of an antenna
is profoundly affected by the proximity of the ground. The ground can
be beneficial in some circumstances and detrimental in others; it is important
to understand ground effects and how to use the presence of ground
advantageously.
Unfortunately, the ground near an antenna is not easily characterized;
ordinarily, it is complex in shape with highly variable radio frequency pro-
perties. The ideal ‘‘ground”’ will consist of a flat plane of material with high
electrical conductivity, usually approximated quite well by the surface of the
ocean. Actual ground, however, has a much lower conductivity and lower
dielectric constant than salt water and is ordinarily far from flat. Values of
electrical conductivity range from about 4 mhos/meter for salt water to as
low as 10~3 mhos/meter for typical residential areas; corresponding values
of dielectric constant range from about 80 for water to 5 for typical residential
areas.
The presence of ground alters antenna performance in two ways. First,
it changes the normal free space pattern (in the illuminated hemisphere) by
adding (vectorially) a reflected pattern. The combination of direct and
reflected radiation fields produces regions of enhanced gain, but at the
expense of reduced gain in other regions. Second, it alters the antenna itself;
currents that flow in the ground surface couple back into the antenna and
change all element currents. In other words, an antenna becomes a some-
what different antenna when placed near the ground.

Reflections From A Plane Ground

Let us first consider the reflection from a plane ground surface: the reflected
wave can be characterized by a reflected amplitude and a reflected phase shift,
both relative to the original wave. For horizontally polarized radiation the
reflected amplitude remains close to unity and the phase shift is close to
180 degrees for all incoming (and therefore reflected) angles with respect to
the surface. The phase shift is due to the fact that if the conductivity of the
surface is high, Maxwell’s equations require the tangential E-field at the
surface to vanish. The limiting values only occur if the conductivity of the
plane surface is infinite; however, for all practical values of ground con-
ductivity, the reflected amplitude and phase shift remain reasonably close
to the limiting values.
By contrast, the reflection of vertically polarized radiation is much more
complicated. In this case, the limiting value of the relative reflected wave
5-2 Chapter 5

amplitude (for infinite conductivity where the electric field normal to the
surface must be continuous) is also unity, but the phase shift is zero. These
values change drastically where the ground has a finite conductivity and
dielectric constant and also vary radically with the elevation angle relative
to the surface. As the elevation angle is changed from zero to 90 degrees,
the reflected amplitude varies from unity down through a minimum (at an
angle called Brewster’s angle) and slowly back up to a value usually
significantly less than unity. At the same time, the phase shift varies from
180 degrees monotonically down to zero. Brewster’s angle is a function of
the dielectric constant of the surface; for salt water it is about 6 degrees,
while for poor ground it rises to about 24 degrees.
Thus, it is easy to see that if one models flat real ground as an infinitely
conducting plane, the result should be generally trusted for horizontal
polarization, but not for vertical polarization. It is significant that far-field
radiation for vertical polarization over real (finite conductivity) ground
Vanishes at grazing angles. Only when conductivity is truly infinite does
grazing angle reflection add to direct antenna radiation.

Ground Curvature Effects

Let me return briefly to the observation that the ground surface near an
antenna is rarely flat. Even if this real ground surface were perfectly reflecting
(infinite conductivity), the pattern of reflected radiation would be exceedingly
complicated. It is useful to imagine a small optical model of the ground
surface terrain made with a good shiny reflecting surface and illuminated
by a light source (replacing the antenna). From this it is easy to understand
that the reflected pattern will show bright spots or glints where converging
rays are focused from ‘‘dished’’ (concave) surfacés; in fact, the brightness
from such dished areas is potentially much higher than the direct light from
the light source itself. If the dished surface is ideal, one can realize enormous
brightness gains; the ideal surface is, of course, a parabolic reflector at the
correct distance from the source.
Excellent optical searchlights are made with such reflectors. A modern radio
frequency example is the large Arecibo dish, which started with a natural,
roughly parabolic ground surface and was subsequently improved by high-
conductivity, accurately figured reflecting surfaces. To get effective coherent
radiation from the entire surface required that the surface maintain an
accuracy of a fraction of a wavelength.
The shiny optical model produces a reflected pattern somewhat, but not
exactly, like the real ground. The principal difference is [Link] the very short
wavelength of light compared with the size of the surface features of the
model. The brightness gain of any well-configured dished area is theoretically
proportional to the reflecting (concave) area, but inversely proportional to
the square of wavelength. Thus, the optical glints will tend to be small, bright
spots; whereas the real radio glints, occurring at the same angles, will be fuzzy,
larger, and less bright. Note, however, that the broader radio glint will
 The Effects of Ground 5-3

approximate the average value of the optical glint over the radio glint angle.
Thus, we see that ground around an antenna can not only reflect, but it
can focus (and defocus) radiation as well. To get much of a focusing effect,
the concave surface must not only be large (compared with a wavelength)
and of the correct focal length, but it must also have a surface accuracy of
a fraction of a wavelength. This type of surface is not likely to occur naturally
at wavelengths in the 10-to-100 meter range, especially if the antenna is
situated in a region where the reflecting surface is quite complicated.
Nevertheless, for those users fortunate enough to locate their antennas on
a relative smooth, concave slope of the right curvature, some significant
focusing should take place.
I shall model the ground surface in the conventional way; that is, as a
perfectly conducting flat plane. This model should be valid for horizontally
polarized radiation at antenna sites where the actual ground is reasonably
flat out to distances where specular reflection occurs at the lowest elevation
angles of interest. For high angles, the reflection takes place nearly under
the antenna, and the ground must be flat in that area to a fraction of a
wavelength for the model to apply; this is usually the case.
As the elevation angle is reduced, the reflection point recedes from the
antenna location until, at very low angles, it is many wavelengths distant.
Under this condition the real ground does not have to be very flat to reflect
energy with amplitude and phase coherence; it can, in fact, be quite rough,
with variations in height of several wavelengths. This situation is analogous
to the well-known optical reflection observed from surfaces that are rough
in comparison with a wavelength; one can observe, at grazing angles, nearly
specular reflection. A sheet of paper has roughness variations so large
compared with the wavelength of light that, at normal incidence, no reflected
images can be seen; nevertheless, at grazing angles, one can easily observe
specular reflection effects — and even fair images. For these reasons, the
model can be expected to be valid for most horizontally polarized antenna
systems.
One more point should be mentioned. Because of the finite conductivity
of the real ground, the currents that flow are not strictly at the physical surface
of the ground but are distributed throughout the top ‘‘skin depth’’ of the
ground. This skin depth is usually quite small; as an example, at a frequency
of 14 MHz, where the free space wavelength is 21 meters, the skin depth
in salt water is less than 0.1 meters; and even for poor ground, where the
conductivity is 10-3 mhos/meter, the skin depth increases to a value of less
than one meter. Therefore, the infinite conductivity plane model of the ground
should give quite acceptable results.

Image Models
Besides adding a reflected wave to the space pattern of an antenna, the
presence of the highly conducting ground plane changes the properties of
the antenna itself. Excitation of the antenna produces a current distribution
5-4 Chapter 5

in the nearby ground plane surface which, in turn, couples mutual voltages
back into all antenna elements; these mutual voltages will obviously affect
currents in all antenna elements. To model this interaction, it is useful to
replace the ground plane conducting surface by an antenna image. This image
is located just as far below the original ground plane as the real antenna is
located above. For a horizontally polarized antenna, the image is excited
equally, but exactly out of phase, with the real antenna.
Because of the geometrical symmetry of the antenna and its image around
the original ground plane surface and the opposite excitation, it is easy to
see that, at all points on the original ground plane surface, the tangential
electric fields vanish (that is, the image field cancels the antenna field). Thus,
by means of the image model, we produce exactly the same tangential field
at the ground plane coordinates as would be produced by currents flowing
in the real conducting ground plane because of antenna excitation alone. The
antenna itself cannot distinguish whether a real ground plane conductor or
‘its oppositely excited image exists. Therefore, the ground interaction with
the antenna is identical to the image interaction. One can therefore model
all antenna properties over the real ground plane by using the antenna and
its oppositely excited image in free space.
Note, however, that there is one significant difference between the image
model in free space and the real ground plane. In the real ground plane case
only a hemisphere is actually irradiated, while in the image model a full sphere
is irradiated. Although all fields in the (common) hemisphere will be exactly
the same, it is obvious that the total radiated power of the image model will
be just twice that of the real ground plane. Therefore, even though antenna
element impedances are the same for both situations, gain calculations for
the image model must be multiplied by two to obtain gain for the real antenna
over the earth.

Propagation Elevation Angles

Before I evaluate detailed antenna properties over earth, I would like to
discuss briefly the elevation angles that should be of paramount interest. It
is well known that long-distance radio communcations take place primarily
by ionospheric F, layer reflection (or, more properly, refraction). While the
F, layer can vary in (virtual) altitude over the earth (250 to 400 km), it is
instructive to make a very simple model of this layer as a reflecting shell at
an altitude or height of 300 km. A radiated wave at any elevation angle will
bounce from this shell and return to earth at the same elevation angle and
at a great circle distance which depends only on simple geometry. For a single
hop, the maximum range on the earth is limited; thus, communications at
very long distances will involve several hops.
Fig. 5.1 plots the elevation angle versus great circle distance for different
numbers of hops (up to six). This diagram shows clearly that a given distance
can be reached with different discrete elevation angles, or that a given
 The Effects of Ground 5-5

35

30

Uv)
a)—QO
9
Qa

(DEG.)
ANGLE
ELEVATION
S}

(6) / 2 3 4 5 6 Z, 8 9 10
GREAT CIRCLE RANGE (IOOOKM)

Fig. 5.1 — Diagram of the elevation angle required for propagation over
various great circle distances and number hops.

elevation angle arrives back at the earth at discrete distances. To cover all
values of distance requires a continuous spread in elevation angle, but the
limits of this spread can be narrowed somewhat by taking advantage of dif-
ferent numbers of hops. As an example, all distances beyond 1600 km can
be accommodated by the heavy line in Fig. 5.1; that is, elevation angles
varying over a range of only 3 to 17 degrees.
It is generally desirable at the higher frequencies to use low angles to
minimize attenuation resulting from multiple hops and reflection losses and
to ensure ionospheric refraction at the highest frequency. For such frequencies
(say 14 to 28 MHz) the range of elevation angles shown in Fig. 5.1 seems
quite appropriate. At lower frequencies (7 MHz and below), however, if the
propagation path at either end is in daylight (where absorption is high), a
higher range of angles (using a greater number of hops) may give a lower
overall absorption. The reasoning derived from this simplified model of
propagation effects gives expected results not inconsistent with observations
reported in The ARRL Antenna Book on page 18.! However, real
propagation is clearly more complicated than is shown in Fig. 5.1.
Kift has shown in an elegant way that long distance propagation involves
many propagation modes.2 He has shown, in measurements made between

1This and other footnotes are given at the end of this chapter.
5-6 Chapter 5

Ascension Island and Slough, England, that measured arrival angles, when
complete path ionospheric soundings are known, correlate well with ray-
tracing expectations. (Ray tracing can identify actual propagation modes.)
His results indicate that elevation angles from 3 to 20 degrees are indeed quite
important. He also shows focusing effects of a given mode and the great
variety of results that can occur in practice. Thus, in evaluating an antenna
system over ground, it is most important to ensure good gain over all lower
angles (say 3 to 17 degrees), and for the lower frequencies (7 MHz and below)
over even higher angles (up to say 30 degrees). I shall therefore show, in all
cases to be presented, a plot of H-plane gain as a function of elevation angle.

Antenna Performance Over Ground

To illustrate typical Yagi antenna performances over ideal ground, I shall
use as representative Yagi beams a three-element Yagi on a 0.3 wavelength
boom and a six-element Yagi on a 0.75 wavelength boom; the basic designs
are given in Table 5.1. I have made calculations for each beam at a number
of different elevations over ground. The results for the three-element beam

Table 5.1 — Element lengths and positions for representative three and
six-element Yagis. All elements are cylindrical with radius = 0.00052599
wavelengths.
three-element six-element
element length position length position
() () () ()
reflector 0.49403 0.000 0.49483 0.000
driver 0.48567 0.150 0.48047 0.150
D1 0.46525 0.300 0.44771 0.300
D2 0.44771 0.450
D3 0.44771 0.600
D4 0.44771 0.750

Table 5.2 — Performance of a representative three-element beam versus

height above ground.
input impedance
Fig. height gain F/B angle resistance reactance
(wavelengths) (dBi) (dB) (degrees) (ohms) (ohms)
5.2A 0.10 9.82 5.6 58 32.6 13.1
5.2B 0.25 11.08 16.8 41 15.0 -—4.0
5.2C 0.50 13.19 24.2 Pa§ 15.4 1.4
5.2D 0.75 13.84 25.3 19 16.7 -0.9
5.2E 1.00 14.04 25.5 14 15.8 0.7
5.2F 1.25 14.10 32.9 11 16.6 -0.5
5.2G 1.50 14.17, 26:5 9 15.9 0.5
5.2H 1.75 14.19 36.1 8 16.5 -0.3
5.2 2.00 14.24 27.3 rf 16.0 0.4
5.2J 2.50 14.24 27.9 6 16.0 0.3
5.2K 3.00 14.25 28.3 a 16.1 0.3
free space 8.27. 931.2 0 16.2 0.1
 The Effects of Ground 5-7

Table 5.3 — Performance of a representative six-element beam versus height

above ground.
input impedance
Fig. height gain F/B angle resistance reactance
(wavelengths) (dBi) (AB) (degrees) (ohms) (ohms)
5.3A 0.10 11.92 8.7 46 14.6 9.6
5.3B 0.25 13.00 13.2 33 23.2 -6.8
5.3C 0.50 14.85 35.5 23 21.2 -0.6
5.3D 0.75 15°60 ee 215 18 20.8 0.7
§.3E 1.00 16.36 37.7 14 21.9 -0.4
5.3F 1:25 16.23 29.3 11 20.5 0.1
5.3G 1.50 16.59 41.7 9 21.8 -0.2
5.3H 1.75 16.44 34.0 8 20.7 -0.1
5.31 2.00 16.68 45.2 i 21.6 -0.1
5.3J 2.50 16.68 48.3 6 21.6 -0.1
5.3K 3.00 16.70." =52:3. 5 rat oo} -0.1
free space 10.71 47.0 0 21.2 -0.2

are shown in Table 5.2 and those for the six-element beam in Table 5.3. As
a reference, the free space performance for each beam is also listed in these
tables. Figures 5.2 and 5.3 show the AH-plane, or vertical, pattern of each
of these cases. It is apparent from these H-plane patterns that maximum gain
in the forward direction occurs at an elevation angle which is an inverse
function of the antenna height; this relationship is tabulated quantitatively
in Tables 5.2 and 5.3.
Figures 5.2 and 5.3 also show the antenna gain has a number of lobes;
the biggest lobe is the first one (lowest elevation angle). For each succeeding
lobe the peak gain is somewhat lower. This reduction in gain is caused by
the natural free space directivity of the antenna. The overall pattern is a series
of lobes (produced by interference of the direct and reflected waves) essentially
modulated by the inherent free space pattern of the antenna. Note that the
relative gain reduction at high angles is greater for the (more directive) six-
element beam than for the three-element beam. Moreover, a careful analysis
of the lobes shows that the maximum point on each lobe is slightly altered
by the natural beam directivity. This is shown in Tables 5.2 and 5.3 by the
slightly lower elevation angles of the main lobe for the six- versus the three-
element beam at a given height above ground.
Thus, we see that the main lobe of an antenna occurs at an angle primar-
ily determined by its height over ground, but secondarily by the natural
antenna directivity. This latter effect is most pronounced at low antenna
heights; it is also responsible for the relatively poorer gain at these heights.
One would ordinarily expect the ground reflection to double the radiated
field (or to add 6.02 dB to gain), but if it occurs at a high elevation angle
(due to a low antenna height) the original free space antenna gain is signifi-
cantly lowered (at the same high angle).
The front-to-back ratio is also shown in Tables 5.2 and 5.3. Recall that
5-8 Chapter 5

0 dB = 9.8 dBi O dB = 11.1 dBi

Fig. 5.2A — H-plane radiation Fig. 5.2B — H-plane radiation

pattern for a three-element Yagi pattern for a three-element Yagi
0.1 wavelengths above the ground. 0.25 wavelengths above the
ground.

0 dB = 13.2 dBi 0 dB = 13.8 dBi

Fig. 5.2C — H-plane radiation Fig. 5.20 — H-plane radiation

pattern for a three-element Yagi pattern for a three-element Yagi
0.5 wavelengths above the ground. 0.75 wavelengths above the
ground.

0 dB = 14.1 dBi 0 dB = 14.1 dBi

* | 90°

x
NX
Fig. 5.2E — H-plane radiation Fig. 5.2F — H-plane radiation
pattern for a three-element Yagi pattern for a three-element Yagi
1.0 wavelengths above the ground. 1.25 wavelengths above the
ground.
 The Effects of Ground 5-9

0 dB = 14.2 dBi 0 dB = 14.2 dBi

ane
Fig. 5.2G — H-plane radiation Fig. 5.2H — H-plane radiation
pattern for a three-element Yagi pattern for a three-element Yagi
1.5 wavelengths above the ground. 1.75 wavelengths above the
ground.

0 dB = 14.3 dBi 0 dB = 14.3 dBi

° 90° One Yo
408

Wi,ZaN
I Whee
Fig. 5.21 — H-plane radiation Fig. 5.2J — H-plane radiation
pattern for a three-element Yagi pattern for a three-element Yagi
2.0 wavelengths above the ground. 2.5 wavelengths above the ground.

0 dB = 14.3 dBi

Fig. 5.2K — H-plane radiation

pattern for a three-element Yagi
3.0 wavelengths above the ground.
5-10 Chapter 5

0 dB = 12.0 aBi 0 dB = 13.0 dBi

90°

Fig. 5.3A — H-plane radiation Fig. 5.3B — H-plane radiation

pattern for a six-element Yagi pattern for a six-element Yagi
0.1 wavelengths above the ground. 0.25 wavelengths above the
ground.

0 dB = 14.9 cBi 0 dB = 15.6 dBi

Fig. 5.3C — H-plane radiation Fig. 5.3D — H-plane radiation

pattern for a six-element Yagi pattern for a six-element Yagi
0.5 wavelengths above the ground. 0.75 wavelengths above the
ground.

0 dB = 16.4 dBi 0 dB = 16.3 dBi

Fig. 5.3E — H-plane radiation Fig. 5.3F — H-plane radiation

pattern for a six-element Yagi pattern for a six-element Yagi
1.0 wavelengths above the ground. 1.25 wavelengths above the
ground.
 The Effects of Ground 5-11

0 dB = 16.6 dBi 0 dB = 16.5 dBi

Co?

os
VO w
SK -ESS
Fig. 5.3G — H-plane radiation Fig. 5.3H — H-plane radiation
pattern for a six-element Yagi pattern for a six-element Yagi
1.5 wavelengths above the ground. 1.75 wavelengths above the
ground.

0 dB = 16.7 dBi 0 dB = 16.7 dBi

2 90°

Fig. 5.31 — H-plane radiation Fig. 5.3J — H-plane radiation

pattern for a six-element Yagi pattern for a six-element Yagi
2.0 wavelengths above the ground. 2.5 wavelengths above the ground.

0 dB = 16.7 dBi
2 90°

Fig. 5.3K — H-plane radiation

pattern for a six-element Yagi
3.0 wavelengths above the ground.
5-12 Chapter 5

the definition I use for F/B is the ratio of forward energy flux density at
the best elevation angle to the reverse energy flux density at the same reverse
elevation angle. Tables 5.2 and 5.3 show that this quantity fluctuates rather
widely with antenna height; the cause of these fluctuations is the altered
antenna element complex currents that result from the mutual coupling of
antenna and its image. These mutual effects are large when the antenna is
low and relatively small when the antenna is high. Note, however, that even
when the antenna is three full wavelengths above ground, enough interaction
occurs to noticeably alter the free space value.
Similarly, the antenna driving-point impedance fluctuates with antenna
height. When the antenna is very low, for example, at a height of 0.1
wavelengths, driving-point resistance and reactance are far from their free
space values. This shows dramatically that if one adjusts an antenna near
the ground for best performance, it certainly will not be the best adjustment
at final operating height.
“ These ground mutual effects, which alter the antenna element currents,
are present to some degree at all antenna heights likely to be used in practice.
This is tantamount to saying that the antenna over ground is not the same
as the antenna in free space. An antenna optimized for free space will there-
fore not generally be quite optimum over ground. Obviously, one should
really optimize the antenna over ground at the desired height.

Best Height
What is the best antenna height? Recall from Fig. 5.1 that one should strive
for a large gain over a range of angles, for example, 3 to 17 degrees. An
inspection of Figs. 5.2 and 5.3 shows that this occurs when the antenna height
over ground is about 1.5 wavelengths. For 14 MHz radiation, this height
would be about 100 feet. Practical operating experience does verify that such
an antenna height gives excellent results. Note also that at a height of three
wavelengths a deep lobe null occurs at an elevation angle of 10 degrees; this
angle is sometimes important, such as for a great circle distance of 4500 km
using two F-layer hops. Such a high antenna, even though excellent as a band
Opener at very low angles, would not be expected to be a good overall
performer. I have tried a large 14 MHz antenna at a height of 2.6 wavelengths;
from my location in New York State, the average European signals were
found to be substantially inferior to those received from an antenna at a height
of 1.5 wavelengths.
It is fortunate that an antenna at a height of 1.5 As a Over ground
is not seriously degraded from its free space performance. Table 5.3 shows
that the F/B ratio (at the design frequency) for the six-element beam is a
superb 47 dB in free space and degrades only to a (still superb) value of
42 dB when the beam is mounted at 1.5 wavelengths above ground. In both
cases, optimization procedures described in Chapter 3 can tune up the F/B
ratio with only minor effects on other performance features.
 The Effects of Ground 5-13

I have carried out such an optimization by slightly varying the boom

positions of the first and third directors; final beam specifications are shown
in Table 5.4, and the performance around the frequency of best F/B is shown
in Tables 5.5 and 5.6. Note that optimization requires very delicate boom
position adjustments. These boom positions have been adjusted sufficiently
well to give a F/B well over 100 dB at the central design frequency. It is
interesting to note that different boom positions are needed for the free space
Yagi and for the Yagi to be mounted over ground. This is because they are
really slightly different antennas because of ground interaction.
Let me stress, as I did in Chapter 3, that although the Yagis are
mathematically optimized to give a very high (but narrowband) F/B ratio,
the basic model cannot really be trusted to this level of accuracy. It should

Table 5.4 — Element lengths and positions for six-element beams optimized
by slight shifts in boom positions for D1 and D3. All element radii are
0.00052599 wavelengths.
free space 1.5 wavelengths high
element length position length position
() () () (\)
reflector 0.49483 0.000 0.49483 0.000
driver 0.48073 0.150 0.48047 0.150
D1 0.44771 0.2983232 0.44771 0.303318
D2 0.44771 0.450 0.44771 0.450
D3 0.44771 0.6015036 0.44771 0.597591
D4 0.44771 0.750 0.44771 0.750

Table 5.5 — Characteristics of the six-element Yagi optimized in free space.

normalized gain F/B angle resistance reactance
frequency (dBi) (dB) (degrees) (ohms) (ohms)
0.996 10.58 26.6 0 22.0 -6.2
0.998 10.64 32.7 0 21.6 -3.1
1.000 10.70 134.8 0 21.3 -0.1
1.002 1075 ei 92.7 0 21.0 3.1
1.004 10.80 26.7 0 20.8 6.3

Table 5.6 — Characteristics of the six-element Yagi optimized at 1.5

wavelengths above ground.
normalized = gain F/B angle resistance reactance
frequency (dBi) (dB) (degrees) (ohms) (ohms)
0.996 16.50) 426.5 9 21.9 -5.4
0.998 16.55 4.32.5 9 21.6 -2.4
1.000 16.61 119.8 9 21.4 Och
1.002 16.65 32.6 9 Pah We 3.8
1.004 16.69 26.6 9 21.0 7.0
al
5-14 Chapter 5

be quite possible in principle to carry out this type of optimization experi-

mentally on a real Yagi; the basic behavior should be similar, but the final
boom positions might be slightly different.

Antenna Upward Tilt

Note that when a Yagi is mounted over ground the lowest lobe of radiation
has a maximum at an elevation angle usually sufficiently high that the direct
wave from the antenna is somewhat reduced from its peak free space value.
It is interesting to see if any improvement could be made by purposely tipping
the antenna boom upwards to increase the direct wave. Unfortunately, tipping
the antenna upward automatically tips the image downward by the same
angle; the net result is that, while the direct wave is increased, the reflected
wave is decreased, and by a greater amount.

Table 5.7 — Characteristics of a representative six-element Yagi one

wavelength above ground with the boom tipped upward.
tip angle gain F/B angle resistance reactance
Fig. (degrees) (ABi) (dB) (degrees) (ohms) (ohms)
5.3E 0 16.36 37.7 14 21.9 -0.4
5.4A 5 16.31 33.7 14 22.0 -0.6
5.4B 10 16.12 29.9 14 22.0 —0.7
5.4C “s) 15.78 26.8 13 21.8 -0.9
5.4D 20 15.26 23.8 13 21.6 -—0.9

As an example, consider the six-element Yagi mounted one wavelength

above ground. The maximum gain of 16.36 dBi occurs at an angle of
14 degrees as shown in Table 5.3. Table 5.7 and Fig. 5.4 show the result if
the antenna is tipped upward at angles of 5, 10, 15, and 20 degrees. It is

0 dB = 16.3 dBi 0 dB = 16.1 dBi

Fig. 5.4A — H-plane radiation Fig. 5.4B — H-plane radiation

patterns for a six-element beam at patterns for a six-element beam at
a height of one wavelength with a height of one wavelength with
boom tilted upward 5 degrees. the boom tilted upward
10 degrees.
 The Effects of Ground 5-15

0 dB = 15.8 dBi 0 dB = 15.3 dBi

90°

Fig. 5.4C — H-plane radiation Fig. 5.4D — H-plane radiation

patterns for a six-element beam at patterns for a six-element beam at
a height of one wavelength with a height of one wavelength with
the boom tilted upward the boom tilted upward
15 degrees. 20 degrees.

easy to see that maximum overall gain is actually best when the antenna is
parallel to the ground plane; as one tips the antenna the peak lobe gain is
reduced slightly and the deep nulls between lobes tend to become shallower.
This is precisely the behavior expected from a consideration of the vectorial
addition of direct and reflected waves.

Summary
1) Although ground is difficult to characterize, there is reason to believe
that for horizontal polarization a good model is an ideal, infinitely
conducting plane.
2) The A-plane (vertical) pattern consists of a number of lobes caused
by the interference of direct and ground-reflected waves. The first
(lowest) lobe is the strongest; succeeding lobes are reduced somewhat
in gain by the natural free space directivity of the antenna.
3) Mutual effects between the antenna and ground cause antenna
element currents to change; these changes cause significant alterations
to the antenna properties. The most noticeable variations occur in F/B
ratio, but there are also significant variations in gain and driving-point
impedance.
4) Best overall antenna performance appears to occur if the antenna
height is about 1.5 wavelengths. This is not a critical figure, but it is
believed that three wavelengths is probably too high.
5) Adjusting an antenna near the ground for best performance
guarantees that the antenna performance will not be optimum at the
operating height. Because of the significant mutual effects with the
ground, the antenna should be optimized at its final operational height.
Generally, this optimization will not be quite the same as the optimized
free space antenna.
6) Large antennas are more handicapped at low heights than are small
5-16 Chapter 5

antennas; this is due to their higher natural free space directivity.

7) For best gain the antenna boom should, in principle be parallel to
the effective ground plane surface; however, the degree of parallelism
required is not critical. Tipping the antenna upward to improve gain
will, in actual fact, decrease maximum gain.

NOTES:
1The ARRL Antenna Book, 13th Edition, American Radio Relay League, 1974.
2Kift, F. ‘‘The propagation of High-Frequency Radio Waves to Long Distances,”’
Proceedings of IEE, 107 Part B, March 1960, page 127.
CHAPTER 6
STACKING
ft his chapter addresses the use of multiple Yagi antennas arranged into
a coherent antenna system. The number of potential arrangements
is unlimited, but certain basic configurations deserve detailed analysis
because they have attractive properties. To start, I shall limit the discussion
to systems where the individual Yagi antennas are all physically identical and
aligned for maximum radiation in the same direction. Moreover, to ensure
that each Yagi contributes to the overall main radiated wave front in a
coherent manner, I shall limit the configurations to those in which the Yagi
positions (say, for example, the reflector end of the boom) lie in a plane
perpendicular to boom direction. Usually all of the Yagis are coherently
excited by the same driver current (magnitude and phase). Using identical
Yagis positioned in such a plane helps maintain a uniform radiated pattern
over a desired frequency band. The overall system beam pattern can be
pointed in azimuth only by mechanically rotating the entire system. The
radiated beam from a mechanically fixed (system) array of laterally spaced
Yagi antennas can, in principle, be steered by changing the excitation phase
to each Yagi antenna. However, the beam quality generally deteriorates. Such
mechanically fixed, electrically steered phased arrays are not considered here.
The overall system array can be viewed as a large-area aperture illumi-
nated in a quasi-uniform way by the individual Yagi antennas. So long as
the individual Yagi antennas are not too far apart (so that illumination is
relatively uniform), the system gain should be proportional to the total
effective aperture area. The system beam pattern should also show an angular
width inversely proportional to the aperture dimension. Thus, in concept,
a horizontal array of Yagi antennas (horizontally polarized) should produce
a narrow horizontal system beam pattern; similarly, a vertical array of Yagi
antennas (horizontally polarized) should produce a narrow vertical system
beam pattern.
We must consider the system array over ground; in this case all of the effects
mentioned previously (Chapter 5) will occur. Recall that ionospheric paths
over the earth primarily favor low radiation angles (up to, say, 20 degrees);
moreover, this whole range of antenna radiation angles should be covered
to accommodate a continuous range of great circle distances, as well as dif-
ferent multi-mode ionospheric paths. We shall see that, by vertically stack-
ing two or more horizontally polarized Yagis over ground, it is possible to
improve significantly low-angle performance (over that of a single Yagi
antenna over ground) without reducing the azimuthal coverage. This
improved result comes about through a suppression of otherwise useless radia-
tion at the higher angles.
6-2 Chapter 6

Vertical Stacking Arrangements

For Amateur Radio communications, relatively wide horizontal or
azimuthal coverage is generally desirable, not only to make a given contact
less sensitive to critical beam heading but to accommodate the many occa-
sions in which the communication path is somewhat skewed due to iono-
spheric conditions. Wide azimuthal coverage is especially desirable under
contest conditions, where it is advantageous to have the beam simultaneously
illuminate the largest desired Amateur population. So, a horizontal array
of Yagi antennas doesn’t appear as desirable as a vertical stack; therefore
I shall not attempt analyses of such horizontal arrays.
Vertically stacked Yagi arrays are now in reasonably wide use; these require
a supporting mast. If the stacking separation is large (which we shall find
desirable), the large mast must be entirely rotatable and, of course, very
rugged mechanically. Such a mast, including its foundation, is a major
undertaking.
Two interesting variations of this system are not as formidable. The first
variation is a stacked Yagi antenna array offset from a fixed tower. The offset
allows simultaneous rotation of the Yagi antennas over a range in azimuth
of about 300 degrees; at either end of this range, the antennas are designed
to nest around the mast. I use this construction for a 28 MHz stacked six-
element Yagi antenna system on a Rohn 45 guyed mast. It works very well
and is cost effective.
The second variation is to use a fixed mast with the top Yagi antenna fully
rotatable and a second, lower, Yagi antenna fixed in a preferred direction.
This is a particularly interesting variation for contest operation, especially
on the lower frequencies where the mast must be very high. My 7 MHz system
is a good example. A full-sized, three-element beam is fully rotatable on top
of a 180 foot Rohn 45 guyed mast. A second full-sized, three-element beam
is fixed at 90 feet, which is aimed at Europe. Thus, in the European direc-
tion, full stacking is available; in all other directions the top beam can be
used alone. Moreover, it is easy to excite both beams and activate two
azimuthal directions simultaneously, or it is also possible to switch instantly
from one direction to another without losing the normal time to turn the
large Yagi antenna. I have found the flexibility of this system to be very
helpful in many situations.

Excitation
For all types of stacked arrays, I have found it useful to provide a switching
system that allows operation of each Yagi independently or both together.
When only high-angle radiation is desired, the lower antenna is usually best.
For lower angles of radiation, the combined stack is better. It is easy to
arrange such a switch using conventional relays and quarter-wave coaxial
transformers; a practical system is shown in Fig. 6.1 for two stacked Yagi
antennas.
 Stacking 6-3

FEED LINE
TO TRANSMITTER

CONTACTS FOR RELAYS |, 2,3 SHOWN

IN NORMAL (UNEXCITED) STATE

#1,2 #/ #2
RELAY BOTH LOWER UPPER

RELAY EXCITATION SCHEDULE

Fig. 6.1 — Both, lower, or upper (BLU) switching system. T is a quarter-

wave transformer of 70 ohm coaxial cable.

The relays may have to be compensated by small shunt capacitors if their

series inductance is too large. The relay box should be mounted on the mast
about half way between the Yagi antennas. Extension to more than two
stacked Yagi antennas is equivalently easy. However, the particular scheme
will depend on the way in which power is to be split between all Yagi antennas.
Because of these various excitation techniques, it is desirable to compute
not only the properties of a vertically stacked Yagi system, but the proper-
ties of the individually excited Yagi antennas. Two complicating problems
arise. First, not only is a single Yagi antenna over ideal ground not the same
antenna as in free space, but it is further changed by all other Yagi antennas
as well as their ground images. This is true even if all other Yagi antennas
are not driven. To some extent their elements will be parasitically excited
by the single driven Yagi antenna. This means that the computation for a
single Yagi must be carefully made to account fully for all the parasites and
images in its local field.
Second, if only the top Yagi is rotatable, the performance of the single
lower antenna alone will depend on the relative azimuthal orientation of the
two antennas. In this case it is instructive to compute three cases: parallel,
orthogonal, and antiparallel orientations.

Two-Array Stack
Let us now choose some representative horizontally polarized stacking
arrangements over flat, ideal, ground and compute their theoretical perfor-
mance. I shall present computed H-plane patterns over the range of eleva-
tion angles of interest. The E-plane pattern over ideal ground is, of course,
6-4 Chapter 6
EEE

Table 6.1 — Performance of a two-array stack of three-element Yagis for

different combinations of heights.
Fig. height both lower
6.2 lower upper gain angle F/B gain angle F/B
() (\) (Bi) (deg) (dB) (aBi) (deg) (dB)
A 0.30 0.75 14.30 21 21.4 11.41 41 19.0
Bo 0.375 O75 14.38 21 20.0 11.69 36 27.4
C 0.45 0.75 14.41 21 18.8 12.01 31 23.4
D 0.60 1.50 15554 18.2 [Link]. 824 7 210
E 0.75 1.50 16:2Foca hil 13.9 13:80 20 nume2.6
F 0.90 1.50 16.58 11 19.4 ees US eye!
G 0.90 2.25 15.41 8 23.8 13: COmnl OMG ceo
le oz yes 16.06 8 28.7 14.10 13 25.3
1.35 2.25 16.74 if. eRe} 14:02 14 25.8
J 1.00 3.00 14.97 6) 930.7. 14.09 14 23.3
Ko 50 3.00 16.11 Giamu22:9 14.18 Git e2s.c
L 2.00 3.00 16.86 Sh ENS) 14.19 736.0

Fig. height upper lower alone

6.2 lower upper gain angle F/B gain angle F/B
() () (dBi) (deg) (dB) (Bi) (deg) (dB)
A 0.30 0.75 TRON) valtehey MG iiteeysh! py. ee
By 0,3758 7-075 13:39" 189 23:7, 12.20 32 47.8
C 0.45 0.75 13:32) 9918, 20:1 12.82 29 28.4
D 0.60 1.50 13.96 Ooa23:9 13-670) acs aoLO
Eee O75 1.50 13.71 9 16.0 13.84 19 25.3
F 0.90 1.50 13.83 9 18.9 13.94 16 31.2
G 0.90 2.25 14.31 6 30.8 13948" 1G motes
iso eee 25 14.19 6 29.9 14:08) 13H 2472
| 1.35 2225 14.00 655 23 14.13 11 37.2
J 1.00 3.00 14.22 5 30.8 14.04 14 25.5
K 1.50 3.00 14.31 Ly NES: 14.17 90 26:5
L 2.00 3.00 14.20 5 34.1 14.24 Thi Ss}

zero everywhere. The plots show not only how well the overall system per-
forms at the important low angles, but also what may be sacrificed at the
higher angles, which are occasionally useful. Two basic Yagi designs are used.
They are the same three-element beam (boom = 0.3 wavelengths) and the
same six-element beam (boom = 0.75 wavelengths) shown in Table 5.1.
I shall start with two stacked, identical beams over ground. In practice,
the height of the upper beam will be fixed at the overall mast height. The
placement of the lower antenna will be made at some lower position. It is
interesting to understand the tradeoffs involved in the height of the lower
antenna. I shall choose, for illustrative purposes, four different heights for
the upper beam, and for each of these cases, three different heights for the
lower beam. All heights are expressed in wavelengths at the central design
frequency. Tables 6.1 and 6.2 show computed results for all these cases. These
tables also refer to Figs. 6.2 and 6.3, which display detailed H-plane patterns
for all cases.
 Stacking 6-5

Table 6.2 — Performance of a two-array stack of six-element Yagis for

different combinations of heights.
Fig. height both lower
6.3 lower upper gain angle F/B gain angle F/B
() (\) (Bi) (deg) (dB) (Bi) (deg) (dB)
A 0.30 0.75 1552) 62023.2 12.78 33 415.8
By 0:375 10:75 15.60 20 20.2 13:06 fel 26
C 0.45 0.75 15.61 OF t8.2 13:22 fae cre coe
D 0.60 1.50 Lf:47 A 22.5 JOs01 21 21.6
Eea0275 1.50 iiveoenn tt 14.0 TOON 19.7 91657
F 0.90 1.50 180 23.3 1S)7S0M \16F 738-8
G 0.90 2.25 18.00 8 31.2 1606)" 159" 24:7
Her tgieotes 2:25 18.42 8 29.2 16.31 12 49.0
i 2:35 2.25 18.60 Zi w19:9 16.22 Tom 22.6
J 1.00 3.00 17.34 6F 739-7 TO-GSie 14g 1o0-0
Kah .00 3.00 18.71 6. 32:9 16.61 9 39.1
L 2.00 3.00 19.24 Snes 5 16.59 Vi etal

Fig. height upper lower alone

6.3 lower upper gain angle F/B gain angle F/B
() (r) (aBi) (deg) (dB) (Bi) (deg) (dB)
A 0.30 0.75 19.56 Far, ) 2127, 13.39 30 17.3
B 0.375 0.75 S22 ieee LO 13.96 27 24.0
C 0.45 0.75 14:75" 17° 14.6 14.51 25 30.0
D 0.60 1.50 16.40 9 30.3 1S Sei 21 23.7
EB, 40.75 1.50 16.06 Oni 758 15,008) 7a) 8 pee aco
F 0.90 1.50 16.08 9 24.6 16.16 15 29.1
G_ 0.90 men 15.16 6 457'5 16.1627) 15 = 29.1
Hie daha.) mee} 16.55 6 . 31.3 16.320.12) 33:5
ie cs) 2:25 16.24 6921-7 16.387" =10, 32'3
dh de 3.00 16.68 Ser 4.2 16:36" ~ 14" 37.7
K 1.50 3.00 16.76 ele} 16.59 OF aaar
L 2.00 3.00 16.58 5 ee26'2 16.68 7 45.2

Note that each figure has several curves: one for the combined stacked
performance (solid line); one for the lower antenna alone (dotted line); one
for the upper antenna alone (dashed line); and, where applicable, what the
lower antenna only would show if no upper antenna were physically present
(broken solid line). In the second and third cases, both antennas are physically
present, but only one is driven (all nondriven elements act as parasites).
An examination of Tables 6.1 and 6.2, and especially the H-plane pat-
terns of Figs. 6.2 and 6.3, reveals a number of interesting and important
characteristics of these simple, vertically stacked systems. Table 6.1 shows
the maximum gain and corresponding elevation angle for each case of a
stacked pair of three-element beams. Also shown is the F/B ratio, which we
know varies with the exact element complex currents, which in turn are
influenced by the mutual impedances to all other elements. Table 6.2 shows
the equivalent quantitites for the stacked pair of six-element beams.
Note from these tables that the smaller values of overall antenna mast
6-6 Chapter 6

combination curve

both solid
lower dots
upper dashes
lower alone chain-dots

Fig. 6.2A — H-plane radiation pat- Fig. 6.2B — H-plane radiation pat-
tern for two three-element Yagis at tern for two three-element Yagis at
0.3 and 0.75 wavelengths above 0.375 and 0.75 wavelengths above
ground. ground.

Fig. 6.2C — H-plane radiation pat- Fig. 6.2D — H-plane radiation pat-
tern for two three-element Yagis at tern for two three-element Yagis at
0.45 and 0.75 wavelengths above 0.6 and 1.5 wavelengths above
ground. ground.

Fig. 6.2E — H-plane radiation pat- Fig. 6.2F — H-plane radiation pat-
tern for two three-element Yagis at tern for two three-element Yagis at
0.75 and 1.5 wavelengths above 0.9 and 1.5 wavelengths above
ground. ground.
 Stacking 6-7

Fig. 6.2G — H-piane radiation pat- Fig. 6.2H — H-plane radiation pat-
tern for two three-element Yagis at tern for two three-element Yagis at
0.9 and 2.25 wavelengths above 1.125 and 2.25 wavelengths above
ground. ground.

Fig. 6.21 — H-plane radiation pattern Fig. 6.2J — H-plane radiation pat-
for two three-element Yagis at 1.35 tern for two three-element Yagis at
and 2.25 wavelengths above 1.0 and 3.0 wavelengths above
ground. ground.

Fig. 6.2K — H-plane radiation pat- Fig. 6.2L — H-plane radiation pat-
tern for two three-element Yagis at tern for two three-element Yagis at
1.5 and 3.0 wavelengths above 2.0 and 3.0 wavelengths above
ground. ground.
6-8 Chapter 6

combination curve

both solid
lower dots
upper dashes
lower alone chain-dots

Fig. 6.3A — H-plane radiation pat- Fig. 6.3B — H-plane radiation pat-
tern for two six-element Yagis at 0.3 tern for two six-element Yagis at
and 0.75 wavelengths above 0.375 and 0.75 wavelengths above
ground. ground.

Fig. 6.3C — H-plane radiation pat- Fig. 6.3D — H-plane radiation pat-
tern for two six-element Yagis at tern for two six-element Yagis at 0.6
0.45 and 0.75 wavelengths above and 1.5 wavelengths above ground.
ground.

Ve
Fig. 6.3E — H-plane radiation pat- Fig. 6.3F — H-plane radiation pat-
tern for two six-element Yagis at tern for two six-element Yagis at 0.9
0.75 and 1.5 wavelengths above and 1.5 wavelengths above ground.
ground.
 Stacking 6-9

Fig. 6.3G — H-plane radiation pat- Fig. 6.3H — H-plane radiation pat-
tern for two six-element Yagis at 0.9 tern for two six-element Yagis at
and 2.25 wavelengths above 1.125 and 2.25 wavelengths above
ground. ground.

Fig. 6.31 — H-plane radiation pattern Fig. 6.3J — H-plane radiation pat-
for two six-element Yagis at 1.35 tern for two six-element Yagis at 1.0
and 2.25 wavelengths above and 3.0 wavelengths above ground.
ground.

Fig. 6.3K — H-plane radiation pat- Fig. 6.3L — H-plane radiation pat-
tern for two six-element Yagis at 1.5 tern for two six-element Yagis at 2.0
and 3.0 wavelengths above ground. and 3.0 wavelengths above ground.
6-10 Chapter 6

height do not give as much overall maximum gain as the higher antennas;
this gain deficit is more severe for the six-element beams than for the three-
element beams. This is the same general result previously obtained for single
antennas over ground (Chapter 5); it results from the same phenomenon;
that is, the natural increased free space directivity of the larger Yagi antennas
reduces the gain potential at the higher elevation angles required for the lower
antennas.
Note also from these tables that the exact placement of the lower Yagi
antenna does not markedly influence the stacked maximum gain of the system
but usually does significantly affect the angle of the lower antenna radia-
tion. Note also that the excellent free space F/B ratio can be significantly
affected by stacking; it is most strongly affected when the stack spacing is
small and where the number of (adjacent) parasites is large, for example,
especially the first three cases in Table 6.2.
To properly assess all of these stacked Yagi antenna systems, it is necessary
to look at the H-plane (elevation angle) patterns shown in Figs. 6.2 and 6.3.
It is instantly clear that excellent stacked coverage (solid line) of the crucially
important zero to 20 degree elevation angles requires a reasonably high system
(more than one wavelength) but not too high (less than 2.5 wavelengths).
Above the first main lobe of radiation the patterns are quite varied; it is
helpful to understand the basic reasons for these variations. Fig. 6.4 shows
a simplified sketch of the two Yagi antennas above ground, each one
represented on this diagram by a point. The lower antenna is at height hy
and the upper one is at height /,,; also shown are the image antennas below
ground at heights of — h, and — hy, respectively. Note that at an elevation
angle, ¢, the radiation from the lower antenna lags that from the upper
antenna by a distance (h,,—h, )sin¢. This phase lag causes the pair of
antennas to interfere both constructively and destructively. At certain values
of $, which I shall designate ¢,, destructive interference will be complete
and produce a radiation pattern null. Since the phase lag between the two
antennas above ground is identical to that between the two images below
ground, the overall radiation will also show these nulls where

dp = sin™'[(n + “)/(ay — hy)I (6.1)

where 7 can take on integer values starting with zero (0, 1, 2, ...).
Now, from Fig. 6.4, note that the radiation from the image pair (which
is excited out of phase with the real antenna pair) further lags by a distance
(A, + A, )sind. Thus nulls will also occur in the overall pattern due to ground
reflections at values of ¢ which I shali designate as bg where:

og = sin7'[m/(hy + Ay) (6.2)

where m can assume integer values (0, 1, 2, ...). As an example, consider
Fig. 6.3L where hy is three wavelengths and h, is two wavelengths. Equa-
 Stacking 6-11

HU

HL
_(HU-HL) SIN &

aan DARE LEVEL

= (HUt+HL) SIN @

(HU-HL) SIN &

Fig. 6.4 — Diagram of two-array stack over ground, showing phase lags
for each antenna and its image.

tions (6.1) and (6.2) predict that nulls should occur (within the range 0 to
60 degrees as follows:

dp = 30 degrees

dg = 0, 11.5, 23.6, 36.9, 53.1 degrees

Fig. 6.3L clearly shows these minima. Moreover, note from Fig. 6.3L that
_ the upper envelope of gain falls off substantially with increasing elevation
angle; this general result is caused by the natural free space directivity of
the individual Yagi antennas. Note that this effect is much more pronounced
for the larger six-element Yagi antennas (Fig. 6.3L) than for the smaller three-
element equivalent stack (Fig. 6.2L).
Thus, the overall H-plane pattern is the result of three effects: first, the
natural free space directivity of the individual Yagi antennas; second, the
interference effect of the two real antennas; and third, the interference effect
of the above ground system with its image counterpart. All three effects have
6-12 Chapter 6

different angular dependencies; it is therefore not surprising that the overall

resultant can be quite varied and complex.
For those readers interested in constructing a vertically stacked Yagi
antenna array, a careful scrutiny of Tables 6.1 and 6.2 and especially all of
the relevant figures is quite enlightening. It is apparent that there is no single
ideal design; nevertheless, there are a number of salient points that are worth
noting.
1) A mast height (upper antenna height) of 0.75 wavelengths is really
not high enough to get very much additional gain from stacking,
especially with large Yagi antennas.
2) The higher systems provide better low-angle performance than the
lower systems but sacrifice (Sometimes needed) high-angle performance.
They also provide less gain sacrifice due to ground images for big
antennas and through increased antenna spacing provide less spoiling
of the inherently good individual Yagi free space characteristics.
3) The important (lowest) first lobe gain is only weakly dependent on
the placement of the lower antenna. The gain alone would favor h
somewhat above hy,/2 (see for example Figs. 6.3G, 6.3H, and 6.31);
nevertheless, a lower placement (wider element spacing) will result in
smaller beam interactions.
4) Mutual coupling or interaction between Yagi antennas tends to spoil
the otherwise excellent properties of a single Yagi. This spoiling is most
pronounced for low systems where spacings are small, not only to
ground but between Yagi antennas (see for example Figs. 6.2A, 6.2B,
and 6.2C). This spoiling can be easily seen in the altered patterns of
the lower beam when the upper beam is physically present (dotted line)
and when the upper beam is absent (broken solid line). You can also
see the effect that stacking has on the F/B ratios (Tables 6.1 and 6.2)
and also (not shown) the effects on the calculated driving-point
impedances of both upper and lower Yagi antennas.
5) Interactive effects are also more serious when large Yagi antennas
are used. This general result is anticipated and is due to the larger
number of adjacent parasites; it is illustrated by comparing the dotted
and broken-solid curves of Figs. 6.2B and 6.2C with those of Figs. 6.3B
and 6.3C.
6) Any good stacked array (for example, Fig. 6.3G) will benefit by the
Both, Lower, Upper (BLU) switch arrangement where at high angles
a fill in the performance can be made (usually) using the lower antenna
only. Best higher angle fill occurs when the placement of the lower
antenna is at or preferably below hy/2. A good practical height is
hy/ 3<h, < hy/2. Note that a good fill obtained in this way slightly
compromises maximum gain; however, this compromise is really not
very serious.
7) With the BLU switch available it is interesting to compare perfor-
 Stacking 6-13

mances. In all cases, at the very lowest angles, Both and Upper give
essentially identical results, that is, the stack is just as good as the upper
antenna alone. However, the stack always accepts a broader range of
vertical angles in its first lobe (due to its lower average height) and at
its peak has more gain than either upper or lower alone. This gain
advantage is one to three dB depending on the particular stack.
Although this may not seem very impressive, experience demonstrates
that the stack can indeed provide a commanding performance advan-
tage over a single Yagi antenna and, coupled with the broader vertical
coverage of the first lobe, will be more consistent.
8) A number of excellent stacked arrays can be chosen from these
figures. As a good example note Fig. 6.3D. I have operated a stack
very much like this on 14 MHz for several years; experience shows this
to be a superb performer even without a BLU switch arrangement.
Figures 6.3E and 6.3F also look very attractive, but the closer beam
spacing results in increased variations in F/B properties and probably
would require a BLU switch for best high-angle fill. For a higher stack
note the excellent gain performances of Figs. 6.3G through 6.3].
However, for any of these cases, a fill seems desirable by the use of
a BLU switch; note that for best fill at some higher angles the upper
antenna should be used. For a very high stack Fig. 6.3J provides
exceptional stacked gain, and by the additional use of the lower antenna
for fill, it accommodates radiation angles up to nearly 30 degrees.
However, at the 30 degree angles the system performance is abysmal,
giving essentially zero response for any setting of the BLU switch.

Phase Derived Fill

I shall now turn briefly to an alternative method of obtaining higher-angle
fill, a method that promises to be operationally simple and potentially very
effective. Up to this point, I have used identical driver currents in both
magnitude and phase. Let us now consider what effect is made on stacked
H-plane patterns if the phase of the drive current in the lower antenna is
changed relative to that in the upper antenna. I shall use as a test case the
stack of Fig. 6.3D and will change the relative phase angle of the lower
antenna drive current with respect to that of the upper antenna drive current.
Computation of system performance under these conditions exhibits some
_ remarkable effects. Table 6.3 shows performance as a function of the phase
shift for several discrete relative phase angles from zero to 330 degrees, and
Fig. 6.5 shows the H-plane patterns corresponding to five values of phase
shift.
The H-plane pattern for any positive value of phase angle is nearly iden-
tical to the pattern for the same negative value; minor differences (which
are also evident in Table 6.3) are caused by the detailed way in which all
mutual coupling effects take place. It is easy to see from Fig. 6.5 that reversing
the phase (shift = 180 degrees) results in excellent system performance at
6-14 Chapter 6

Table 6.3 — Performance of the stack shown in Fig. 6.3D versus relative phase
angle between lower and upper drive currents.
phase gain angle F/B resistance reactance
(deg) (dBi) (deg) (dB) (ohms) (ohms)
QO 17.47 11 22.5 21.6 -0.2
300) 17:27 11 23.0 21.6 -0.1
60 16.48 11 23.5 21.5 0.1
$0.5 15.02 10 24.2 21.5 0.1
120 15.80 comieelse 21.4 0.2
150 16.86 28 389.23.0 21.3 0.1
180, 17.25 28 3824.9 P| 2 0.0
210 17.03 20 42i.3 Zilke -0.1
240 3816.16 28 = 31.0 21.2 -0.2
270 3=614.52 28 = 339.4 21.3 -0.2
300 16.15 11 21.3 21.4 -0.3
330 =17.11 11 22.0 21.5 -0.2

phase angle
(deg) curve

0 solid
60 dots
90 dashes
120 chain-dots
180 chain-dashes

Fig. 6.5 — H-plane radiation patterns for a two-array stack of six-element

Yagis with out-of-phase excitation.

higher angles; basically giving maxima where the original H-plane pattern
showed minima. At intermediate values of ¢ an intermediate result is obtained
which is a combination of both the in-phase pattern (zero degrees phase angle)
and the out-of-phase pattern (180 degree phase angle). Note that this higher
angle fill effectively uses the extra gain potential of both Yagi antennas; it
is therefore potentially superior to a single Yagi antenna fill and is also quite
easy to implement (by switching in to only one of the antennas a coaxial line
whose electrical length is one-half wavelength).
One can also see-clearly from Fig. 6.5 that if the phase angle is relatively
small, little degradation of system performance occurs; this fact potentially
allows the stacking of dissimilar Yagi antennas. Nevertheless the use of
dissimilar antennas raises questions about how to measure the effective phase
shift and certainly increases the complications of controlling its value over
a reasonable bandwidth of frequencies.
It is important to note that only two values of the phase angle are desired.
The in-phase case (zero degrees) is best for low-angle performance and the
out-of-phase case (180 degrees) is best for higher values of elevation angle.
 Stacking 6-15

All other values give inferior results to either one or the other of these cases.

Three and Four-Array Stacks

Let me now consider the possibility of stacking more than two Yagi
antennas. It is obvious that some additional performance improvement should
be possible provided the mast height is sufficiently high. As examples I show
computations for two different evenly spaced stacks shown in Table 6.4. The
first stack shows three six-element Yagis evenly spaced with a top height of
2.25 wavelengths, and the second shows four six-element Yagis evenly spac-
ed to a top height of three wavelengths. Figs. 6.6 and 6.7 show the H-plane

Table 6.4 — Performance of three- and four-array stacks of six-element

Yagis.
all lowest
Fig. h, ho hg hg gain angle F/B angle F/B
0) 0) () () (Bi) (deg) (dB) (deg) (dB)
6.6 0.75 1.50 2.25 19.20 G 14.9 f40) Whee)
6.7 0.75 1.50 2.25 20.24 6 16.0 19 12.4
top lowest alone
Fig. gain angle F/B angle F/B
(aBi) (deg) (dB) (deg) (dB)
6.6 16.15 5 22.3 u. 2ies
6.7 16.11 5 21.6 ike AAS

combination curve

solid
lowest dots
highest dashes

Fig. 6.6 — H-plane radiation patterns for a three-array stack of six-

element beams at 0.75, 1.5 and 2.25 wavelengths above ground.

combination curve

solid
lowest dots
highest dashes

Fig. 6.7 — H-plane radiation patterns for a four-array stack of six-

element beams at 0.75, 1.5, 2.25 and 3 wavelengths above ground.
 6-16 Chapter 6

patterns for these two systems; they should be compared with Figs. 6.3G
and 6.3J, which are basically equivalent two-Yagi stacks of comparable mast
height. It is at once apparent from Fig. 6.6 that the addition of the third
Yagi antenna gives a main lobe gain over Fig. 6.3G of 1.2 dB; all other
characteristics are quite comparable. Likewise, Fig. 6.7 shows that the two
additional Yagi antennas give a main lobe gain increase of 2.9 dB over Fig.
6.3J; again, all other characteristics are quite similar.
These examples of vertically stacked Yagi antenna arrays using more than
two antennas show that a noticeable gain increase is possible over a two-
antenna stack; moreover they open up a wide range of higher-angle fill
possibilities. As an example, Fig. 6.7 shows patterns where all four antennas
are excited; for fill at higher angles, the lowest antenna (dotted line) and the
highest antenna (dashed line) might be used alone. Note, however, that
additional fill situations are possible if two or even three of the original four
antennas are excited coherently. Moreover one can also consider feedline
phasing for even better fill. Clearly, a host of possibilities exists, but the prac-
_tical use of all potentially desirable combinations not only requires a com-
plex switching system but a great deal of trouble in determining experimentally
the right combination for the prevailing circuit conditions. Surely the addi-
tional complexity and expense of these large vertical stacks reaches a point
of practical diminishing returns. Nevertheless, how fortunate we are to be
able to predict with reasonable confidence the performance of such large
systems, without ever having to build one.

Optimization of Stack Arrays

I shall conclude this chapter on stacking by referring again to the basic
two-antenna stack shown in Fig. 6.3D. Note that the F/B performance has
deteriorated from the excellent free space performance of the individual Yagi
antennas of 47 dB to 22.7 dB. Analogous to the optimization of a single Yagi
antenna over ground, it is possible to optimize the basic Yagi design for this
stacked system. Table 6.5 shows the optimized parameters of the six-element

Table 6.5 — Element lengths and positions for two-array stacks of six-
element beams, optimized in free space and for the heights shown in
Fig. 6.3D, by shifts in boom positions for the first and third directors. All
element radii are 0.00052599 wavelengths.
free space heights = 0.6 \ and 1.5
element length position length position
() () (\) (\)
reflector 0.49483 0.000 0.49483 . 0.000
driver 0.48073 0.150 0.48123 0.150
D1 0.44771 0.29832 0.44771 0.30184
D2 0.44771 0.450 0.44771 0.450
D3 0.44771 0.60150 0.44771 0.64120
D4 0.44771 0.750 0.44771 0.750
 Stacking 6-17

Table 6.6 — Performance of the free space optimized two-array stack of six-
element Yagis described in Table 6.5.
normalized gain F/B angle resistance reactance
frequency (dBi) (dB) (deg) (ohms) (ohms)
0.996 10.58 26.6 0 22.0 -6.2
0.998 10.64 32.7 0
1.000 10.70 134.8 0 :
1.002 10:75) 1432.7 0: 21.0 3.1
1.004 10.80 26.7 0

Table 6.7 — Performance over ground of the optimized two-array stack of

six-element Yagis described in Table 6.5.
normalized gain F/B angle resistance reactance
frequency (dBi) (dB) (deg) (ohms) (ohms)
0.996 17.26 24.8 11 16.3 -6.5
0.998 17.34 30.8 11 15.8 —3.4
1.000 17.40 93.0 11 15.3 -0.2
1.002 17.47 30.6 11 14.8 3.0
1.004 17.52 24.5 11 14.4 6.2

Yagi antenna first for free space and second for the stack of Fig. 6.3D (hy
= 0.6 wavelengths, hy, = 1.5 wavelengths). Tables 6.6 and 6.7 show the
swept-frequency performance of each of these cases close to the design
frequency.
The iterative optimization was carried out by adjustments of the boom
positions of the first and third directors to obtain high F/B (more than
90 dB), and by a slight adjustment of driven-element length to minimize reac-
tance at the design frequency. Note again that, because of mutual coupling
interactions, the stacked Yagi antenna is not the same Yagi antenna as it
would be in free space, nor is it the same Yagi antenna as it would be singly
over ground (compare with Table 5.4).

Orthogonal and Antiparallel Stacked Yagis

Now that an optimized Yagi design has been found, which provides the
superlative performance shown in Table 6.7 when the two stacked antennas
are parallel to each other, we can ask about performance degradation when
the two Yagi antennas are orthogonal to each other and also when they are
antiparallel. This question is relevant when a stack is used where the lower
antenna is fixed in some direction and the upper antenna is rotatable.
Table 6.8 shows the system performance for the case where both Yagis are
supported at the center of the boom.
It is clear that, in principle, optimization can be carried out for only one
configuration, and performance will automatically deteriorate somewhat for
other geometries. The extent of deterioration will be more severe for stacks with
small antenna spacings (both with respect to ground and to each other); that
6-18 Chapter 6

Table 6.8 — Performance of a two-array stack of six-element Yagis for

different relative boom orientations.
Both (angle =11°) lower (angle = 271°)
config- gain F/B R X gain F/B R X
uration (dBi) (dB) (ohms) (ohms) (dBi) (dB) (ohms) (ohms)
parallel 17.40 93.0 153 -0.2 15:070 Sii25natS.5 -0.6
orthogonal 15.14 369 13.7 -1.0
antiparallel 15.00 27.1 13.3 -—0.6
upper (angle = 9°)
config- gain F/B R X
uration (AqBi) (dB) (ohms) (ohms)
parallel 16.34 255 15.2 -—0.8
orthogonal 16.59 208 154 —1.1
antiparallel 16.39 24.0 21.0 1.9

is, with lower overall mast height. For the stack shown in Table 6.8, it is gratify-
ing to see that the performance for all situations is really quite acceptable.

Summary
1) Vertical stacking of two Yagi antennas allows both substantial
improvement in low-angle system performance and improved flexibility.
This flexibility can be used either to obtain fill at some needed higher
angles or to illuminate other azimuthal angles (one or two Yagi antennas
rotatable).
2) Mast heights of between one and perhaps 2.5 wavelengths can
provide excellent two-Yagi stacked systems.
3) Higher masts favor low-angle radiation and also give smaller mutual
interaction effects. However, they also treat ane (occasionally useful)
higher angles unfavorably.
4) For all vertical stacks, improved performance is available if excitation
is switchable to both antennas, the lower antenna, or the upper antenna
(a ‘‘BLU”’ switch). Switching must be done in a way that preserves phase
integrity and keeps the total drive impedance matched to the supply
coaxial line. For those antenna stacks where the lower and upper beams
remain aligned (rotate together), a highly useful switch is a phase
reverser to only one of the beams.
5) Vertical stacks using three or four Yagi antennas can display even
greater performance, but the stacks must be very high and must use
for best results a more complex feedline switching arrangement.
6) Optimization (very high F/B at one frequency) can be obtained for
only one physical configuration at a time. Nevertheless, there are
practical examples where an optimized antenna design for a two-Yagi
stack will still exhibit excellent properties when only the lower antenna
or only the upper antenna is excited; moreover, these excellent properties
are retained even if the azimuthal directions of the two individual Yagi
antennas are parallel, orthogonal or even antiparallel.
CHAPTER 7

PRACTICAL DESIGN
p to this point the specifications for a Yagi antenna have been made
: |only in terms of strictly cylindrical elements. Each element is
characterized by a position along the boom, a physical length, and
a radius; all three of these quantities are expressed in wavelengths at a central
design frequency. Such specifications have led to a number of rather good
antenna designs and I shall shortly list a brief selection of such designs.
However, when a real Yagi antenna is constructed it will rarely be con-
venient to rigorously adhere to the cylindrical element design. To start, the
element diameter is usually adjusted to fit a mechanical requirement (wind
loading, etc.); moreover, the element itself is usually not a cylinder but a
series of telescoping tubes starting with a large diameter section at the boom
and ‘‘tapering’’ to a small diameter section at the outer end of the element.
In addition, the element is fastened to the boom with a boom clamp which
may be a plate or angle bracket U-bolted to both boom and element. Some
mechanical designs even put the element directly through the boom.
Thus, the path from the cylindrical design to a practical antenna will in-
volve three tasks: first scaling the original design to an equivalent new de-
sign using a different (average) element radius; second, computing the change
in element length to account for the effect of tapered element section radii;
and third, making (usually minor) corrections to allow for the effect of the
element-to-boom clamping system. Methods for carrying out each of these
three tasks will be given below, following a discussion of preferred antenna
designs.

Preferred Antenna Designs

I shall show in this section one preferred design example for each case of
two, three, four, five and six-element Yagi antennas. Recall from Chapter
2 that simplistic Yagis (element spacing uniform and all directors having a
common length) are as good as any other design up to a boom length of
one wavelength. It was shown that a good two-element beam would have
a boom length of about 0.15 wavelengths; the exact length is not critical and
is a compromise between better gain and lower efficiency and bandwidth.
Best parasite element length is a compromise between better forward gain
and lower F/B ratio. For a three-element beam it was shown that a boom
length of about 0.3 wavelengths produced a naturally high F/B and similarly
for four, five, and six-element beams a boom about 0.75 wavelengths gives
a naturally good F/B ratio. Table 7.1 shows the characteristics of these good
Yagi designs. It should be emphasized that these particular antenna designs
are not unique; for example, the boom length can be varied somewhat (see
Chapter 2). Longer booms generally give larger forward gain, but make the
 7-2 Chapter 7

Table 7.1 — Characteristics for preferred Yagi designs. All dimensions are
given in wavelengths at the design frequency, and all element radii are
0.00052599 wavelengths.
two-element three-element four element
element ’ position length position length position length
() () () (r) (x) ()
reflector 0.0 0.49366 0.0 0.49404 0.0 0.49136
driver 0.150 0.47050 0.150 0.48572 0.250 0.47895
D1 0.300 0.46525 0.500 0.46273
D2 0.750 0.46273
D3
D4
Gain (dBi) 6.9 8.3 10.6
F/B (dB) 8.0 31.2 44.0

five-element six-element
element position length position length
(\) (r) (r) (x)
~ reflector 0.0 0.49944 0.0 0.49478
driver 0.1875 0.48031 0.150 0.48038
D1 0.375 0.45187 0.300 0.44766
D2 0.5625 0.45187 0.450 0.44766
D3 0.750 0.45187 0.600 0.44766
D4 0.750 0.44766
Gain (dBi) 10.5 10.7
F/B (dB) 32.4 46.8

Table 7.2 — Element lengths and positions for six-element Yagis optimized
at different heights. All dimensions are given in wavelengths at the design
frequency, and all element radii are 0.00052599 wavelengths.
free space 7.0 0.6 and 0.5»
element position length position length position length
(\) (r) (x) () () (\)
reflector 0.0 0.49478 0.0 0.49478 0.0 0.49478
driver 0.150 0.48038 0.150 0.48000 0.150 0.48140
D1 0.29884 0.44766 0.3074 0.44766 0.30215 0.44766
wipe 0.450 0.44766 0.450 0.44766 0.450 0.44766
D3 0.60189 0.44766 0.5980 0.44766 0.64165 0.44766
D4 0.750 0.44766 0.750 0.44766 0.750 0.44766

frequency for highest F/B ratio drop somewhat below the center of the band
where gain remains high.
It has also been shown (see Chapter 3) that there exists a procedure which
allows fine tuning the F/B ratio; this optimization procedure can be done
for Yagi antennas having four or more elements. Optimization must be car-
ried out for a specific end use. Table 7.2 shows optimized six-element beams
first for free space use, next for operation at one wavelength above ground
 Practical Design 7-3

and finally for operation in a two-Yagi stack at heights of 0.6 and 1.5
wavelengths. These parameters are mathematically correct, but note from
Chapter 1 that approximations used in the model really do not justify complete
confidence in the precise values in Table 7.2. Nevertheless I suspect that prac-
tical antennas (for use over ground) constructed from this table will exhibit
superior properties to the (free space) six-element case shown in Table 7.1.

Radius Scaling
Any Yagi antenna design, such as those shown in Table 7.1, can be scaled
either to other center frequencies or to elements of different diameter at the
same center frequency. Since all design parameters are dimensions expressed
in wavelengths at a central design frequency the values shown in Table 7.2
are invariant to frequency scaling and therefore the behavior of the antenna
will be unaffected by the choice of central design frequency. However this
is true only if a// physical dimensions (including element radius) are adjusted
in proportion to the desired wavelength.
Experience shows that practical element radii expressed in wavelengths are
not constant; at low frequencies (long wavelengths) relatively thin elements
are used while at high frequencies relatively fat elements are normal. How
then can a given design be altered to an equivalent design where element radius
is changed? The clue is to make the impedance of the changed element exactly
the same as the impedance of the original element; in this way exactly the
same element currents will flow, resulting in the same detailed antenna
performance.
Since the (radiation) resistance of the element is essentially unaffected by
changes in radius, we need only make the reactance invariant to scaling
element radius. Recall from Chapter 1 that element reactance at the
normalized central design frequency, F = 1, can be written:

X = A (1-F,) (7.1)

where FR is the normalized resonant frequency and

A = 430.8log K — 339 (7.2)

Here K is the ratio of wavelength (at the design frequency) to element radius:
d/a. The resonant frequency depends on the resonant length, /p, through
Fr = Ip/I|, where / is element length. The resonant length depends only on
radius (through K) and can be written:

Ip =~ 0.5 — [33.25 + 3.19log K — 0.35(log K)*]/[861.6log K — 678]

(7.3)
If we wish to scale the element radius from its original value to a new value
we must ensure that X is unchanged. Denote the original (‘‘design’’) element
 7-4 Chapter 7

by subscript d and the new (‘‘scaled’’) element by a subscript s. We are given

lq, ag, and a,, and we wish to find the new length /; such that:

X, = Xa (7.4)
The solution is as follows. First, compute Ag and A, from (7.2), and /p, and
lp,, from (7.3). Then solve (7.4) for Fr,, in terms of known quantities using
(7.1):
Fr, = 1-Aa(l—FRy/As (7.5)
Finally, compute /, from the resonant length and resonant frequency for the
scaled element:

ls cas In/FR, (7.6)

These three steps give the length of the scaled element (with new radius) which
has the same reactance as the original element. It is quite simple and most
‘convenient to set up the entire conversion procedure represented by (7.2)
through (7.6) on a small programmable calculator.
An example will illustrate the nature of results. Consider the antenna design
for the six-element antenna of Table 7.1; this would be a reasonable design
for a 14.2 MHz antenna where one wavelength is 831.76 inches and a radius
of 0.00052599 wavelengths corresponds to an element diameter of 7/8 inches.
This would=be*a‘treasonable? eee
= =
dimension for a mechanically Table 7.3 — Element lengths and resonant
adequate element. Now sup- lengths in wavelengths, normalized
: resonant frequencies, and reactances in
pose that we would like an ohms, for three values of element radius:
equivalent antenna for 28 MHz case 1 = 0.00052599), case 2 = 0.0008),
where the radius should be case 3 = 0.0012).
increased. Table 7.3 shows the reflector
computations. The original 1 2 3
design is shown as case 1, the / 0.49478 + 0.49445 ~—s:0.49408
scaled design for an_inter- Iz 0.48140 + 0.48002 0.47847
mediate radius of 0.0008 nie cope |) oboe
wavelengths
;
(about 5/8 inches ie
driver
diameter) as case 2, and a 1 2 3
one inch diameter (0.0012 1 0.48038 ~—0.47892-—«0.47729
wavelengths radius) element as Ip (0.48140 ~=—:0.48002~—s«(0.47847
case 3. Note that the changed FR 1.00212 = 1.00230 = 1.00247
Xx -2.3 -2.3 -2.3
values for element lengths are
not wholly intuitive because ‘ ma ea P
twoxthings happen’ simultane- | 0.44766 ~—«0.44393-~«S«0.43977
ously. As the radius increases | Ip 0.48140 0.48002 0.47847
the Q drops, requiring a greater FR 1.07537 1.08130 1.08801
spread in the resonant fre- X — 80.9 - 80.9 ~ 80.9
 Practical Design 7-5

quencies of reflector and director. However, at the same time the resonant
physical length also changes.
It is important to reiterate that it is conceptually wrong to scale boom length
and element lengths (for example, to convert a VHF antenna design to HF)
without also scaling the element radius. The correct way to adjust an antenna
element when the old and new radii (in wavelengths) are different is to modify
both length and radius to give the same electrical reactance.
We now have the tools to convert a given antenna design such as shown
in Table 7.1 to a new antenna design where the element radius is changed;
the new antenna will perform exactly the same as the original antenna at
the central design frequency. However the frequency-swept behavior of the
new antenna, while qualitatively similar to the original, will show a broader
or narrower bandwidth depending on the change in element Q.

Taper Corrections
Up to this point all antenna calculations have been made for strictly
cylindrical elements and the results will apply directly to most high-frequency
(small) Yagi antennas where the general practice is to use cylindrical elements.
However, for frequencies less than about 30 MHz, mechanical considerations
usually require that the element consists of one or more telescoping sections
of tubing. At the lower frequencies (say 7 MHz and below) the Yagi antenna
becomes gigantic and it is no small mechanical engineering task to even con-
struct a good element. Small diameters favor smaller wind forces but are
insufficiently rugged for long elements. It is therefore a universal practice
that these large elements be made of several telescoping sections; the largest
diameter section is clamped to the boom and succeeding monotonically
smaller diameter sections make up the outer portions of the element.
Thus, it is important to understand how to relate the actual detailed taper
schedule of an element (that is, the diameters and lengths of all sections)
to the equivalent length of a cylindrical element having the same average di-
ameter. ‘‘Equivalent’’ means that the resonant frequency and the Q are the
same for the tapered element and the equivalent cylinder.
To start, I shall introduce the concepts of pipe inductance and pipe
capacitance. Consider a cylindrical element of total length s and radius a
as shown in Fig. 7.1. Define a position coordinate, x, whose origin is at the
center of the element, and the related angle coordinate, 0, where 6 = x x/s
in radians. Excitation of this element in the neighborhood of the resonant
frequency will produce sinusoidal current and voltage distributions:

(0) = Ipsin(27ft)cosé (7.7)

and

V(0) Vocos(27ft)sind (7.8)

The driving-point impedance of the element consists of a resistance (which
7-6 Chapter 7

is directly related to far-field energy radiation) and of course a reactance.

All reactance effects including resonant frequency and electrical O are caused
by near-field (non-radiation) energy storage. Energy storage occurs in two
ways: the magnetic flux surrounding the current distribution in (7.7) and the
electrical field produced by the voltage distribution in (7.8). Note that at
certain instantaneous times (¢ = n/2f) the current everywhere is zero and
all stored energy resides in the electric field. Similarly at certain other times
(t = n/2f +1/4f) the electric field vanishes and all of the stored energy re-
sides in the magnetic flux. As time progresses the (constant) total stored energy
transfers back and forth between magnetic and electrostatic fields. The trans-
fer frequency is of course the resonant frequency of the element. Note that
as a result of this complete non-radiative
energy transfer the peak magnetic stored
energy must exactly equal the peak elec- Rgreerrerncc
trostatic stored energy. Note also that the
resonant exchange frequency must decrease ey
as the total stored energy is increased.
6
Now consider the effect of inserting an
infinitesimal length of pipe of the same Fig. 7.1 — Coordinates of a
radius into the element of Fig. 7.1, at the single Yagi element.
center. Recall from Chapter 1 that the
Original element driving-point reactance is:

X = (430.8log K — 339)\(F — Fp) (7.9)

At the resonant frequency the reactance vanishes. Inserting an additional
infinitesimal length of pipe, As, at the center will change the resonant
frequency to F’p. At this new frequency the total reactance again vanishes.
The added reactance due to the inserted pipe must be balanced by the origi-
nal pipe reactance at the new frequency:

0 = (430.8log K — 339)(F*p — Fr) + 2nfAL (7.10)

where AL is the increased inductance due to As.
Note that the pipe inserted at the element center can produce only inductive
effects (stored magnetic flux) since the electrial potential is strictly zero. Now
F*p is clearly related to Fp by the overall length of the element:

Fp 7 FR = —As/s (7.11)

from which |

AL/As = (430.8log K — 339)/(2zfs)

or, since s = )/2,

 Practical Design 7-7

AL/As = (430.8log K — 339)/(zc) (7.12)

where c is the velocity of light. Thus the addition of the small infinitesimal
pipe section causes the element to behave just as though a pure series
inductance were added. The effective inductance per unit length, which I
designate by JND, is given by (7.12), and is easily expressed in henrys/meters:

IND = (43.08logK —33.9)(1.061 x 10-8) (7.13)

Recall now that in a simple model of a resonant circuit it is easy to relate
the magnitude of voltage on the reactive components to the magnitude of
input current:

[Vo | = | Jo | RQ (7.14)
and also recall from Chapter 1:

2RQ = (430.8log K — 339) (7.15)

Now consider extending the element in Fig. 7.1 by a pipe of length As of

the same radius at its outer end (x = s/2). Here the current is zero so that
the small pipe can act to increase the electrostatic energy only (capacitive
effect). Since in this case (7.9) is still valid, the total increase in stored ener-
gy should be just the same as it was for insertion at the center. Therefore:

AL(Ip)*/2 = AC(Vo)2/2 (7.16)

where AC is the capacitance increase due to As at the element end and AL

is the increase in inductance due to As at the element center. From (7.14)
and (7.16):

AC = AL/(ROQ)? (7.17)
Using (7.15) (7.17) and (7.12):

As/AC = (43.08log K — 33.9)(252c/10) (7.18)

or in meters/farad:

As/AC = 1/CAP = (43.08log K —33.9)(2.356 x 10°) (7.19)

where CAP is the capacitance per unit length. Note that 1/CAP is directly
related to IND, differing only in a constant multiplier.
Thus we now can think of a cylindrical section of element pipe as
contributing to element inductance (7.13) and element capacitance (7.19).
Each contribution is a function of K (and therefore radius), and each will
depend on the current or voltage on the pipe section.
Let us now see what happens if a small section of pipe length AB/2 is first
removed at a position x (or corresponding 6) and for symmetry also at —x
 7-8 Chapter 7

from the element shown in Fig. 7.1. Now replace these removed sections with
equal length sections of larger radius. Thus the overall length of the element
remains at s but cylindrical ‘‘bumps’’ occur at x and —x. As a result of these
bumps the stored energy of the system is changed and therefore the resonant
frequency is changed. The bumps’ contribution to stored energy, W*, will be:

2W* = AB[IND*(Ip2cos26) + CAP*(Vo2sin76)] (7.20)

The relationship of V at the end of the element to J at the center is essentially

unchanged from the original element (see (7.16)):

CAP Vo2 = IND Ip2 (721)

Note also from (7.13) and (7.19) that

CAP*/CAP = IND/IND* (7.22)

so that (7.20) can be rewritten:

2W* = AB[IND*(Ip2cos20 + (IND2/IND*)(Ip2sin20)]. (7.23)

- Let us now find an equivalent length, AA/2, of the original pipe which, placed
at the same positions as each of the bumps, contributes an equal stored energy.

2W = AA[IND Ip*(cos26 + sin26)] = AA IND Ip? = 2W* (7.24)

so that

AA/AB = (IND*/IND)cos26 + (IND/IND*)sin6 (7.25)

Now for a longer section (longer bump) going from 6, to 02 the equivalent
length of the original pipe size can be easily calculated. Designate the ratio
IND*/IND as m, the length of the long bump as sg, and the length of the
original pipe which gives the same stored energy as sa. Then:

S,/SR = mM cos20 + - sin26 (7.26)

The angular functions are to be averaged over the complete bump section;
the result is:

s,/S_ = (m + 1/2 + (m - SiO/2 (7.27)

where

f@) = (sin20,—sin26,)/(20> —26}) (7.28)

and @ is measured in radians.
 Practical Design 7-9

Thus we can now compute, from a given element taper schedule involving
several sections with different pipe diameters, the equivalent lengths of
sections of ‘‘standard’’ cylindrical pipe. The procedure is to choose first the
standard cylinder radius which is to provide equivalent Q. A good choice
is the pipe size at the center of each half-element. Next, for each section of
the tapered element compute the starting 0,, and ending 02. Then, for each
section compute m; it is easily derived from (7.13), i.e.:

m = (43.08logK’ — 33.9)/(43.08log K — 33.9) (7.29)

From (7.29) and (7.27) compute s,/Sp which, multiplied by the tapered
section physical length, gives the equivalent section length of the standard
pipe. Adding the lengths of all equivalent sections gives the overall length
of the standard cylindrical element which should perform essentially the same
as the chosen taper schedule.
Perhaps an example will [Prt were
ilaistratcmthessprocedure tn my ictcao ee be raetaie hae 25
Fig. 7.2 shows schematically a
half-element with five dif- SECTION 1 2 3 4 5
ferent sections whose physical DiAMETER 125 1425 0875 0.625 0500
diameter ranges from 1.25
inches at the boom to 0.5 K 1330.82 1478.68 1901.17 2661.63 3327.04

inches at the outer end.

Fig. 7.2 — Example of a radius-
Readers will recognize this
tapered half-element.
taper schedule as one in com-
mon use (by Wilson) for a
14 MHz Yagi reflector antenna element. The middle pipe section (7/8 inches
in diameter) will represent the standard pipe. At a frequency of 14.2 MHz
one wavelength is 831.76 inches and thus a = 0.00052599 and K = 1901.17.
Table 7.4 illustrates the way in which one calculates equivalent cylinder sec-
tion lengths. For each section, column 2 shows the actual physical length,
Sp, column 3 shows pipe diameter, column 4 the K “value, column 5 the value
of m computed from (7.29), column 6 values for @2, column 7 values of /(8)
computed from (7.28), and column 8 equivalent section lengths, s,, com-

Table 7.4 — Equivalent length computations for the element of Fig. 7.2.
section S, 2a K m 6, f(0) S,
(inches) (inches) (degrees) (inches)
1 36 1.250 1330.82 0.93784 15.070 0.95452 33.868
2 50 1.125 1478.68 0.95620 36.000 0.61449 48.674
3 44 0.875 1901.17 1.00000 54.419 -0.00718 44.000
4 32 0.625 2661.63 1.05864 67.814 -0.52851 31.088
5 53 0.500 3327.04 1.09752 90.000 -0.90300 48.770
215 206.400
LS
 7-10 Chapter 7

puted by (7.27) and the column 2 values. The value for 0; for each section
is simply the value of 02 for the preceeding section, with 0; = 0 for the first
section.
Note that the total length of the tapered half-element is 215 inches whereas
the total length of the equivalent cylindrical half-element is only 206.40 inches.
In other words, just due to the taper effect alone the total (full length) tapered
element must be made 17.2 inches longer than an equivalent cylinder. This
taper correction is surprisingly large; it shows clearly that length alone is a
totally inadequate specification for an HF Yagi element. The physical reason
why the tapered element must be longer than an equivalent cylinder is that
the inner (larger) sections have smaller inductance than a standard cylinder
and therefore must be made longer; similarly the outer (smaller) sections have
smaller capacitance than the standard cylinder and must also be made longer.
Therefore the taper correction will be quite small if the taper is small, but
quite significant if the taper is large.
In the derivation of taper correction calculations, I have assumed that radial
-‘*bumps’’ are treated as small perturbations on the strictly cylindrical case
and that the current and voltage distributions are sinusoidal. Note that m
values for the heavily tapered element of Fig. 7.2 differ from unity by only
a few percent; thus the calculation, even though made by a perturbation
method, should be reasonably good. Moreover the current distribution should
be reasonably sinusoidal over the tapered element. Nevertheless there may
be some small inaccuracies in the overall calculation. It is important to note,
however, that we are after a length correction of only a few percent and
therefore some inaccuracy in the computation of the (small) correction is
tolerable.
One further point merits elaboration. The procedure just outlined only
allows a computation of cylinder equivalents from a given taper schedule;
how may we compute a suitable taper schedule starting from a given cylinder?
One good procedure is to initially specify all of the taper schedules from
mechanical considerations, leaving as a variable only the length of the
outermost section. Choose a guessed or estimated length for this section and
compute the overall equivalent cylinder. It will generally miss the desired
length by a differential length As. One can now readjust the length of the
outermost section by — mAs and recalculate. One or two such iterations will
bring the tapered element equivalent cylinder length into adequate agreement
with the desired figure.
A second procedure is almost as accurate and far easier to apply. Assume
that all elements can be constructed from the same basic taper schedule;
variations are made by varying the length of only the outermost section. One
can relate the tapered half-element length for this schedule only to equivalent
cylinder half-length through:

l= pleg tq (7.30)
 Practical Design 7-11

To find values for the coefficients p and q, calculate equivalent lengths from
two different tapered half-lengths, say for a reflector and a director, using
the procedure demonstrated above. Then p and g are determined as:

p= - b)/(leq, nt legs) (7.31)

and

Cf bra (leq, ay Nleqy)/ (leq , a legs) (7.32)

Note that (7.30) allows a direct computation of taper half-length starting from
a given desired cylinder length. These empirical equations are remarkably
accurate and are much easier to use than the detailed iterative procedure.

Boom and Element Clamping Correction

I now come to the subject of the boom-to-element mechanical clamping
system and its effect on the element reactance and hence resonance. It is clear
that a wide range of clamping systems are in common use; it is virtually
impossible to make valid calculations for all varieties. Nevertheless there are
two major kinds and it is helpful to understand them.
The first clamping system is simply to put the element directly through
the (round) boom. In this construction a length of element equal to the
complete boom diameter is replaced with the boom itself. Since this
replacement occurs at a current node we must determine the effective
inductance of the replacement; once this is done it can be considered as the
first section of a tapered element from which an equivalent cylinder length
can be calculated. I have not attempted a rigorous calculation of boom
inductance; instead I refer to the measurements of Viezbicke in which his
Fig. 10 shows that element overall length due to the presence of a round boom
should be increased by about 0.7 times the diameter of the boom.! This is
tantamount to saying that the inductance of the boom section of the element
is very low compared with normal element inductance; physically this is an
expected result. The low inductance of course is caused by the boom blocking
the magnetic flux which accompanies element current flow.
The second clamping system is much more widely used since it permits
easier element maintenance and replacement. In this system either a flat metal
rectangular plate or an angle bracket is interposed between element and boom;
two U-bolts fasten the boom to the bracket and two more U-bolts fasten
the element to the bracket. The U-bolts may also use saddles or cradles which
are mechanically better and which further tend to separate boom and element.
For this clamping system we wish to know the inductive effect of the boom
itself, and more importantly the inductive effect of the bracket. I have found

'This and other footnotes are given at the end of this chapter.
7-12 Chapter 7

experimentally that for this clamping system the boom itself has remarkably
little effect. Even though the boom and element are in physical contact, the
element length should be increased by only six percent of the boom diameter;
this small correction rapidly disappears as the element is spaced away from
the boom (even by a small amount). The reason this result is so different
from the through-the-boom result is the relative ease with which magnetic
flux, which results from element current flow, can squeeze between boom
and element, especially if there is any gap between them.
The correction in length due to the mounting bracket is readily calculable.
The method is to first calculate the equivalent radius of the element plus
bracket (which produces the same inductance) and second to use this
equivalent radius as the first short section of a taper schedule. This theory
is given by Mushiake and Uda. The equivalent radius, ae, of a flat thin plate
of total width, A, is simply:

ae = A/4 (7.33)
and that for a right-angled bracket of width A and height B is given by a
rather complicated expression which depends only slowly on the ratio B/A.
For ratios between 0.3 and 1.0 a good approximation (error less than five
percent) is:

ae ~ 0.2(A + B) (7.34)

Mushiake and Uda also show how to calculate an equivalent radius for
two parallel conductors. If S,; and S> are the lengths of the peripheries of
the cross sections, a; and a are the equivalent radii of the two conductors,
and d is the mean distance between them, the equivalent radius of the com-
bination of both conductors is determined from:

Ina. = (S;? Ina; + S>2 Inay + 2S,Sylnd)/(S; + S>)* (7.35)

Equations (7.33) and (7.35) permit a calculation of the equivalent radius

of an element which is proximate to a plate; similarly (7.34) and (7.35) pro-
vide a way of calculating the equivalent radius of an element proximate to
an angle bracket. To check
this method of calctlation, |. |[Link] asgeamin | aie hs en Oe oe
: 5 Table 7.5 — Increase in resonant frequency
have determined the experi- que to a proximate plate. Element diameter is
mental detuning effect of a one inch and length produces resonance at 46
plate just touching a one MdHz.
inch diameter element reso- dimensions "change in F,,
nant at 46 MHz. Table 7.5 plate ~— length ~—_—width theory experiment
shows both theoretical and (inches) (inches) (percent) (percent)
experimental results for 1 4.5 3.625 0.304 0.325+0.1
two different plates. These = 6.0 4.0 0.530 0.521 0.1
 Practical Design 7-13

experiments were not particularly accurate because the exact resonant fre-
quency is difficult to determine; nevertheless the agreement of theory and
experiment within estimated experimental accuracy is gratifying.
Note that element length corrections due to a proximate mounting plate
or bracket can easily be as much as 10 percent of the plate length. These
corrections are not especially large in practice but should be made wherever
there is a relatively large boom-to-element clamping system.

Examples of Three-Element Beams

It may be helpful to show explicitly how to specify a good three-element
beam starting with the design data specified in Table 7.1. I shall first per-
form radius scaling, then taper schedule calculations, and finally apply
reasonable boom clamping and boom corrections. To illustrate the full range
of these computations I shall use this procedure to specify a 14.2 MHz beam,
a 21.3 MHz beam, and finally a 28.5 MHz beam.
1) Choose an average element cylinder size which is sufficiently strong.
I shall assume that the final element is made of good strong seamless alumi-
num tubing such as 6061-T6 with 0.058 inch wall thickness. For all three
bands I choose a cylinder size of 7/8 inches outside diameter, although for
28.5 MHz a somewhat smaller size is probably permissible.
2) Choose a convenient taper schedule which is easily made from stan-
dard 12 foot lengths, leaving the length of the outermost section to be adjusted
for correct overall length.
The sections of seamless
tubing (excepting the last Table 7.6 — Examples of half-element taper
schedules for three-element Yagis.
section) are slit back
about three inches at 14.2 MHz roar 3
; a x X» number oO
nH ea? a . ae section inches) (inches) (inches) 12 foot lengths
igs wie ar e 1 tao 0 24 1
common stainless steel »5 1.000 24 72 3
hose clamp fastens sec- 3 0.875 72 136 3
tions together. Tubing 4 0.750 136 176 2
: 5 0.625 176 <215 2
overlap of about eight
inches gives good joint acer
strength. For 14.2 MHz pe os - PDysersot
° : A 1 2
the second section is S®¢H#ON inches) (inches) (inches) 12 foot lengths
actually a full 12 foot sec- 1 0.875 0 72 3
tion over which is slid the 2 0.750 72 112 2
3 0.625 112 <145 ?)
shorter first section; this
procedure gives added
strength at the center and ~ 28.5 MHz
2a X, Xp number of
improves the ease Of section
(inches) (inches) (inches) 12 foot lengths
clamping with U-bolts : pare , s ;
and saddles. For21.3 and 9 0.750 72 <105 2
28.5 MHz this extra inner
7-14 Chapter 7

Table 7.7 — Radius scaling computations for the three-element

Yagi of Table 7.1.
lengths
frequency nN 2a K a reflector driver director
(MHz) (inches) (inches) () () () ()
14.2 831.76 0.875 1901.17 0.00052599 0.49404 0.48572 0.46525
21.3 554.51 0.875 1267.45 0.00078899 0.49366 0.48471 0.46278
28.5 414.42 0.875 947.25 0.00105570 0.49336 0.48389 0.46078

Table 7.8 — Taper calculations for a 14.2 MHz Yagi. Wavelength = 831.76
inches, cylinder radius = 0.00052599 wavelengths, K = 1901.17.
reflector
section 2a xX, X, K m 6, f(@) Sa
(in.) (in.) — (in.) (degrees) (in.)
1 1.125 0 24 1478.69 0.95620 10.047 0.97963 22.971
me 1.000 24 72 1663.52 0.97673 30.140 0.74839 47.167
3 0.875 72 136 1901.17 1.00000 56.930 0.04928 64.000
4 0.750 136 176 2218.03 1.02686 73.674 — 0.64161 39.334
5 0.625 176 215 2661.64 1.05864 90.000 — 0.94675 36.958
210.430

director
section 2a xX, Xp K m 6, f(0) SA
(in.) (in.) — (in.) (degrees) (in.)
1 1e25 0 24 1478.69 0.95620 10.854 0.97625 22.974
2 1.000 24 72 «1663.52 0.97673 32.563 0.70911 47.212
3 0.875 72 136 1901.17 1.00000 61.507 — 0.06800 64.000
4 0.750 136 176 2218.03 1.02686 79.598 — 0.76544 39.202
§ 0.625 176 199 2661.64 1.05864 90.000 — 0.97817 21.755
195.143

section is unnecessary. Table 7.6 shows the specifications for these tapered
half-elements where x; and x2 represent the start (inner) and end (outer)
positions of each section. Note that the tubing requirements for all three ele-
ments are shown in terms of 12 foot lengths.
3) For these three cases it is necessary to scale the original data of Table 7.1
to use the desired average cylinder size. Table 7.7 shows the new scaled
cylinder lengths for all three beams using the radius scaling equations (7.2)
through (7.6).
4) We are now ready to compute the effect of tapered radii. For the 14.2
MHz elements Table 7.8 shows the flow of calculations; x; and x7 show the
start and finish of each section. First a trial guess at the overall reflector half-
length is made; I guessed 215 inches in this case. For each section, K*, m,
J(0) and s, are calculated by the technique used to develop Table 7.4. Note
 Practical Design 7-15

Table 7.9 — Taper calculations for a 21.3 MHz Yagi. Wavelength = 554.51
inches, cylinder radius = 0.00078899 wavelengths, K = 1267.45.
4 reflector
section 2a X, Xp K m 6, f(0) SA
(in.) (in.) — (in.) (degrees) (in.)
1 0.875 0 72 1267.45 1.00000 46.286 0.61831 72.000
2 0.750 72 112 1478.69 1.02891 72.000 — 0.45812 39.494
3 0.625 112 140 1774.43 1.06309 90.000 — 0.93549 26.449
137.943

. director
section 2a x, Xp K m 6, f(@) Sa
(in.) (in.) (in) (degrees) (in.)
1 0.875 0 72 1267.45 1.00000 49.846 0.56652 72.000
2 0.750 72 112 1478.69 1.02891 77.538 — 0.58380 39.351
3 0.625 112 130 1774.43 1.06309 90.000 — 0.96876 16.966
128.317
nee

Table 7.10 — Taper calculations for a 28.5 MHz Yagi. Wavelength = 414.42
inches, cylinder radius = 0.0010557 wavelengths, K = 947.25.
reflector
section 2a : X, X, K m 6, f(0) Sa
(in.) (in.) — (in.) (degrees) (in.)
1 0.875 0 72 947.25 1.00000 62.308 0.37839 72.000
2 0.750 72 104 1105.13 1.03058 90.000 — 0.85138 31.194
103.194

; ? director
section 2a x, X, K m 6, f(0) S,
(in.) (in.) (in.) (degrees) (in.)
1 0.875 0 72° 947.25 1.00000 66.804 0.31051 72.000
2 0.750 72 97 1105.13 1.03058 90.000 — 0.89426 24.338
96.338

that the sum of all cylinder equivalents is 210.43 inches. Next is shown a
second trial guess for overall director half-length of 199 inches; the same
calculational procedure shows the sum of all cylinder equivalents is 195.143
inches. Using these results and (7.31) and (7.32) we can write:

1 = 1.04664). — 5.24465 (14.2 MHz) (7.36)

which gives the overall tapered half-length, /, for any of the elements in terms
of the equivalent cylindrical length /,g, for this particular taper schedule only.
Tables 7.9 and 7.10 show exactly the same calculational procedure for the
7-16 Chapter 7

Table 7.11 — Element half-lengths and positions for the three-element Yagis
with taper schedules given in Tables 7.8, 7.9 and 7.10.
length position
freq element tapered clamp boom _ total
(MHz) (in.) (in.) (in.) (in.) () (in.)
14.2 reflector 209.799 0.66 0.09 210.55 0.0 0.0
driver 206.178 0.66 0.09 206.93 0.15 124.8
director 197.268 0.66 0.09 198.02 0.30 249.5
21.3 reflector 138.885 0.44 0.06 139.38 0.0 0.0
driver 136.307 0.44 0.06 136.81 0.15 83.2
director 129.990 0.44 0.06 130.49 0.30 166.4
28.5 reflector 103.014 0.24 0.06 103.31 0.0 0.0
driver 101.011 0.24 0.06 101.31 0.15 62.2
director 96.122 0.24 0.06 96.42 0.30 124.3

21.3 and the 28.5 MHz beam elements: the corresponding equations for
tapered half-lengths become:

1 = 1.03885/.g — 3.30251 (21.3 MHz) o Clean

l 1.02100/.g — 1.36144 (28.5 MHz) (7.38)

Application of these equations to the equivalent lengths given in Table 7.7

yields the tapered lengths listed in Table 7.11.
5) We are now ready for the final small boom and boom clamp correc-
tions. For this purpose I assume the elements are U-bolted with saddles to
flat plates which in turn are U-bolted with saddles to the boom. Boom
diameters are assumed to be three inches for the 14 MHz Yagi and two inches
for 21 and 28 MHz. Full plate dimensions are assumed to be six inches wide
and eight inches long (14 MHz), five inches wide and six inches long (21 MHz)
and four inches wide and four inches long (28 MHz). These plates reduce
central pipe inductance and thus cause electrical shortening of the half-
element. This shortening amounts to about 0.66 inches (14 MHz), 0.44 inches
(21 MHz), and 0.25 inches (28 MHz). Thus, to compensate for the boom
clamp each half-element should be lengthened by an equivalent amount; it
should be further lengthened by the empirical correction of 1/16 times the
boom diameter.
6) With all of these corrections the overall actual physical length of each
half-element is shown in Table 7.11. None of these taper schedules is very
severe; therefore the actual element lengths are not a great. deal longer than
the equivalent cylinder lengths; nevertheless the differences are important.
Also shown in Table 7.11 is the position along the boom for each element,
expressed both in wavelengths and in inches. Although I have not
experimentally tested any of these particular three-element beams I am
 Practical Design 7-17

confident that their performance will be excellent, and moreover they all
should be easy to construct.

Summary
A practical Yagi antenna design requires the selection of a suitable
cylindrical element design, such as one of those developed in Chapters 2 and
3, followed by corrections to element lengths to accommodate practical
element construction and mounting techniques. First one must scale the
cylindrical element to an equivalent cylinder of the desired radius. Next, the
overall element length must be altered to take into account the desired element
taper schedule. Finally, some small length corrections are needed to
compensate for boom mounting plates and other mechanical hardware.

NOTES:
iViezbicke, P. ‘“Yagi Antenna Design,’’ NBS Technical Note 688, U.S. Department
of Commerce, Washington, D.C., December, 1976.
2Uda, S., and Mushiake, Y., Yagi-Uda Antenna, Research Institute of Electrical
Communication, Tohoku University, Sendai, Japan: Sasaki Ltd., 1954.
 eh) 4 AP

AN] ™ renee Ne cr EAN

y
ee oe se eb healt aeaaAli ; mate
he 4

le
ae ntgg
Tai ge onaiist
Bi, a nr Hun
wads aibaa ;Ml'i
Haka
¥
4 ye iiiseyak { dle eae Hh oapyie?
FOU ue Me Beue sb wereiat as Ais i tyne
SHAE 4s!
bine eg soreise tml angtneGt
thwet ra TRI
4
CSN OS Ji ny 2 wae, R alee ied

Ca
bn Y i What bh7i ok i pa
ae bavf tismy Pas
’ ‘aul a!

ert 1S"“3 web,agehi ague anes >) ‘pin

sigiaks Iidehoviontae CONOR stipe hemitned)nie Stein

i Dedede yond aintoer eeu TRA: Tae Foliar in ish
MER
EUANhie
\
ews
ko Data
Ti King
site hd tatesMPa
q
‘ '
Pmyee onic i hahvw Pe we yea eh paverra oa
+ ' kn yt ene Unere eae j i ay |

j es ' wx "
J an ,

ents Kuee peor

ey MyNhe
‘a1ay,
8 pow
RUSS laa Wr

MC sea gree iA ipa", + ey

hatreds 2k
Lee ponerse vid
3 iit jawit Ay ably ef
veehs yet sca ihe ay we : bs
‘ Te ta 9 i As Spied ulwbeiae
ni Nbr wih) pay Lrbnieit,,CON

hee SEE) ak AO AE OE iy 1haR ee 1a:

My ie eto e AN ae ;
‘4 yo e

sbi) shibee Gea WAPOA SPN Hg oot aes. Shey ioe

Ng ) Deanery ae A Teh Ca) Chere eye N Wael ne Aid
SPOS i ety ahstas Wa eis aie cena Rape aa a
Pm athe pour ry agi AN, eadieraeteeyUIA! toleThea Ment aan
Kh an eh vith eatery: beatle” PRR a ar AMiets eye ae Mr UN
Wa hapaghion he Aeeet i i Hes |tai batPes Vaal d
hav bee Why eiGrank AOU
IW he ne Co a vn Nev i aee %
tC MN,
ait a
ery Aa Rech MUM sri Ty) Revthealthy bia miner witti
Lie ial agtY b Gla )ybani bomen, ey Nit ie
HAL Pah Macatee | reine US a Me
“nahi diet AV BEAM tsyea soviet aha ibe Fitna te
RL ANE) Wei RE ae A ay he See, te ry
Me \ ‘Ws pin rat ra Sieh |ugg NT veel
rae 3 k et Nicer oH wh MRE coevay
Ayabiie bugles " Raat . se phic
ui
fehaan
on
byvig
wie
HN i ‘ dip ah
oat a

eh uy asia
i, ipEaginlh wav
AL greeeeiaiavsah
| ag /
CHAPTER 8

PRACTICAL AMATEUR
YAGI ANTENNAS

| n this chapter I will provide examples of practical Yagi designs that

should perform quite well in each of the Amateur bands, 7 MHz,
14 MHz, 21 MHz and 28 MHz. At the lower frequencies mechanical
considerations dominate the design and only rather simple Yagi electrical
designs are worth considering; whereas, at the higher frequencies the
mechanical obstacles are far easier to overcome and permit more
sophisticated electrical design. In previous chapters it has been shown
that a Yagi antenna is somewhat dependent on its immediate environ-
ment; this is unfortunate since it means that the ‘‘best’’ design is not best
for all applications. Nevertheless a good design for free space is actually
quite good for most applications—particularly where the antenna is not
too close to ground or other similar antennas. Therefore I shall show
only free space performance and will make no attempt to adjust for nearby
structures nor will I attempt any element positioning adjustments to
optimize the F/B ratio as described in Chapter 3.
For each Yagi design there are a number of sequential steps required
to compute and present useful information. The first step is to choose
a boom length; this choice is based on three considerations. First, there
are preferred lengths; recall that for a two-element beam the boom should
be about 0.15 wavelengths long and for all others (for best F/B) it should
be near an odd multiple of one-quarter wavelength. Second, it is nice
to construct the boom from a ‘‘standard’’ length or lengths of strong
aluminum tubing, leaving a little room at the ends for element mounting
hardware. Third, there is some gain advantage to longer rather than shorter
booms. These three considerations permit a rational selection of boom
length.
The next step is to choose the element radius, that is, the average radius
at the center of tapered half-element; the lower limit of radius is deter-
mined by mechanical strength requirements. Larger values of radius will
improve the bandwidth of the system but will certainly make the antenna
heavy and more difficult to support and turn in the wind.
Next is a choice of number of elements and boom placement. It has
been shown that evenly spaced elements on the boom give performance
basically as good as that with any other arrangement and I would generally
8-2 Chapter 8

favor such a placement. I also generally favor a choice of an even number

of elements just because at the boom center there is no element to interfere
with mounting the antenna to the mast or rotor system. However a three-
element beam is a special case; its performance for boom lengths approx-
imating one-quarter wavelength is as good as four elements on the same
boom and has the advantages of excellent accessibility to the driver
matching system (from the fixed mast) and smaller wind load than the
equivalent four-element beam. For this special case the disadvantage of
driver element interference with the mounting system can be overcome
by moving the driver along the boom a short distance; ultimate beam
performance is essentially unaffected.
Having chosen a boom length and number of elements I start with the
closest preferred Yagi design, such as one of those shown in Table 7.1;
this design is then scaled to the new selected radius, to obtain preliminary
element lengths. Using the chosen boom positions and the chosen radius
and the preliminary values of length, the performance is computed. It '
will be found that the frequency of maximum F/B will generally not be
at the design frequency (F = 1), but at Fp. Now readjust all elements
by a factor Fp and recompute; it will be found that the frequency of max-
imum F/B will now be close to the design frequency. Perhaps a second
iteration will be needed to make the agreement precise.
Now it will be found that the frequency of zero driving-point reactance
is generally not F = 1 but at F,; readjust the driver element length only
by a factor F, and recompute. Again a second iteration may be needed
to make the driving-point reactance zero at F = 1. This is the final design
and computed performance can now be recorded and displayed in several
ways. I shall, for each case, show a table listing, for a series of frequencies,
the free space gain, F/B ratio, and driving-point resistance and reactance.
The gain and F/B ratio will also be shown on a plot for easy visualiza-
tion. A second diagram will show both E- and H-plane radiation patterns.
For each case it is necessary to design the actual tapered element. Choose
convenient lengths of telescoping tubing and devise a taper schedule which
is mechanically adequate and where the remaining variable is only the
length of the outermost section. Of course length corrections should be
made for both mounting hardware and boom size. Perhaps the best way
to make these corrections is to compute the effective diameter of the
mounting plate and boom combination and use this as the initial short
element section of the taper schedule for a half-element. This initial short
section has an effective length equal to the physical half-length of the
mounting plate minus the small boom correction (6 percent of the boom
radius). Thus by the same technique as that shown in Chapter 7, the entire
(corrected) element taper schedule can produce an equation relating the
 Practical Amateur Yagi Antennas 8-3

physical length of the fully corrected tapered half-element, /, to the

equivalent cylindrical half-length, /¢g.
The design and performance tables, performance plots and radiation
patterns for each of the Yagis described here are grouped together at the
end of this chapter. For convenience, the tables and figures for the various
antennas designs are summarized here:

Tables Figures
page taper design performance performance _ radiation
antenna no. table table table plot patterns
7 MHz 8-10 8.1
2-el 8-12 8.2 8.3 8.1 8.2
3-el 8.4 8.5 8.3 8.4
14 MHz 8.6
3-el 8-14 8.7 8.8 8.5 8.6
4el 8-16 8.9 8.10 8.7 8.8
4el 8-18 8.11 8.12 8.9 8.10
5-el 8-20 8.13 8.14 8.11 8.12
6-el 8-22 8.15 8.16 8.13 8.14
21 MHz 8.17
3-el 8-24 8.18 8.19 8.15 8.16
6-el 8-26 8.20 8.21 8.17 8.18
28 MHz 8.22
3-el 8-28 8.23 8.24 8.19 8.20
6-el 8-30 8.25 8.26 8.21 8.22
8-el 8-32 8.27 8.28 8.23 8.24

Designs For 7.15 MHz

I choose a taper schedule starting each half-element with a two inch
diameter tube 20 feet long. At its outer end I use a one inch diameter
tube 12 feet long which is supported from a pair of machined reducers
in the two inch tube. The overlap region is made one foot long to give
adequate strength. The outer end of the one inch diameter tube supports
an inserted 7/8 inch diameter telescoping tube long enough to complete
the half-element.
I choose a boom diameter of three inches which results in a virtually
negligible shortening of the initial (mounting plate) section of 0.09 inches.
I next choose a central element mounting plate 6 by 24 by 1/2 inches;
such a plate allows room for two U-bolts and saddles on each half-element
and two larger U-bolts and saddles to clamp to the boom. The (half)
mounting plate and two inch clamped element becomes the first section
in the element taper schedule. The effective radius is calculated as shown
in Chapter 7, and the result is 1.567 inches. The complete taper schedule
8-4 Chapter 8

is shown in Table 8.1, and by the methods shown in Chapter 7 produces:

1 = 1.06793l_.q— 11.68745 (8.1)

Two-element Beam
For a two-element beam a convenient choice for a boom is a three inch
diameter tube 20 feet long; it is close to the desired 0.15 wavelength. Some
room is needed at each end for the mounting plates of the end element;
I shall assume four inches. I choose the element diameter as two inches
diameter which experience shows is mechanically strong enough and does
not sag excessively. Using these chosen values and adjusting element
lengths as described in Chapter 7 I arrive at the final design shown in
Table 8.2. Performance over the 7 MHz band is shown in Table 8.3 and
plots of gain and F/B ratio are shown in Fig. 8.1. Fig. 8.2 shows the
angular performance of the beam in free space. Note that the 3 dB total
beamwidth is 119 degrees for the H-plane and 74 degrees for the E-plane.
By using (8.1) and the design data in Table 8.2 we can now specify
the real two-element beam in terms of boom positions and the overall
tapered element length; the result is included in Table 8.2. Note that the
actual lengths of the tapered elements are considerably longer than the
equivalent cylindrical lengths.

Three-element Beam
For this case I choose a boom three inches in diameter but 40 feet long,
again leaving four inches at each end for element mounting plates. I choose
the same element specifications as above, including the taper schedule.
However I shall move the driven element 24 inches off the boom center
to clear interference with mounting to rotor or mast. Using these values
the final adjusted (cylindrical elements) design is shown in Table 8.4 and
performance over the 7 MHz band is shown in Table 8.5. Plots of gain
and F/B ratio are shown in Fig. 8.3 and the angular performance of the
beam is shown in Fig. 8.4. The 3 dB total beamwidth is 100 degrees for the
H-plane and 68 degrees for the E-plane.
I choose the same element taper schedule and mounting plate infor-
mation as selected for the two-element design; therefore (8.1) is valid and
leads directly to the final beam specification shown in Table 8.4. Note
that the element lengths are somewhat different than those for the two-
element beam.

Designs For 14.2 MHz

I choose a basic cylindrical element 7/8 inches in diameter which
 Practical Amateur Yagi Antennas 8-5

experience shows is large enough for practical 14 MHz elements. Table 8.6
shows two different taper schedules which can be used; the first is a heavily
tapered half-element in common use, and the second is a lightly tapered
half-element that can be easily made from standard telescoping 6061-T6
aluminim seamless tubing with 0.058 inch wall thickness. In both cases
the taper schedules include an initial short section consisting of the
mounting plate clamped by U-bolts and saddles to the initial element
tubing and whose equivalent radius is computed as shown in Chapter 7.
For both schedules I have assumed an aluminum mounting plate whose
fuil dimensions are eight inches long, six inches wide, and 3/8 inches thick.
The effective length of this initial section for the half-element is therefore
four inches minus the small correction for boom diameter; in Table 8.6
I have assumed a boom which is three inches in diameter. The boom cor-
rection is so small that these half-element schedules should apply quite
well to any boom size from two inches to four inches.
These two schedules through the methods explained in Chapter 7 pro-
duce useful equations relating the cylindrical and tapered half-lengths.
For the heavy taper:

1 = 1.06938/.q
— 5.09785 (8.2)

and for the light taper:

1 = 1.04664/.,
— 4.57166 (8.3)

Three-element Beam
A good choice for a boom is a tube 20 feet long with about four inches
at each end used for element mounting hardware. To help clear mechanical
interference between the driven element and the rotor or mast mounting
bracket, I will move the driver 12 inches off center. With these boom
positions and the initial element lengths shown in Table 7.1, the procedure
explained in Chapter 7 produces the final design shown in Table 8.7. Per-
formance over the 14 MHz band is shown in Table 8.8 and plots of gain
and F/B ratio are shown in Fig. 8.5. Fig. 8.6 shows the angular perfor-
mance of the beam in free space. The 3 dB total beamwidth is 100 degrees
for the H-plane and 68 degrees for the E-plane.
Using the two taper schedules of Table 8.6 and the resulting equations
(8.2) and (8.3) with the basic beam data of Table 8.7 we can now give
complete specifications for the three-element beam; these are included
in Table 8.7. Note that the tapered element lengths are quite different
for the two different taper schedules.
 8-6 Chapter 8

Four-element Short-boom Beam

Exactly the same procedure can be used to design a four-element beam
on the same 20 foot boom. Since there is no element mechanical
interference at the center of the boom I shall space elements equally. The
final design for such a beam is shown in Table 8.9 and performance over
the 14 MHz band is shown in Table 8.10. Plots of gain and F/B ratio
are shown in Fig. 8.7 and angular performance of the beam is shown
in Fig. 8.8. The 3 dB total beamwidth is 88 degrees for the H-plane and
64 degrees for the E-plane. Note from Table 8.10 that this four-element
beam does not show a very high F/B ratio; the reason is demonstrated
in Fig. 8.8 where it can be seen that there is a small back lobe between
minima at angles of + 150 degrees. This back lobe arises because the boom
is a bit too long; it can be reduced by shortening the boom but then the
peak gain would be reduced as well. The beam is actually a pretty good
beam generating rather low back radiation over a substantial range in
“angle.

Four-element Long Beam

One could use four elements on a somewhat longer boom which results
in larger gain, but at a very significant loss in F/B ratio. However, we
know that if the boom is lengthened to about 3/4 wavelengths the F/B
ratio improves again. It is interesting to see if a beam with such a long
boom with only four elements is a good performer.
I shall choose a boom of a convenient total length, 54 feet, and diameter
three inches, and again allow four inches at each end for element mounting
hardware. With elements equally spaced along the boom the final design
for this long four-element beam is shown in Table 8.11. The beam per-
formance over the 14 MHz band is shown in Table 8.12 and plots of gain
and F/B ratio are shown in Fig. 8.9. Fig. 8.10 shows the angular perfor-
mance of the beam in free space. The 3 dB total beamwidth is 63 degrees
for the H-plane and 53 degrees for the E-plane. Observe that the perfor-
mance is indeed excellent, and as we shall shortly see, is essentially the
same as a five or a six-element implementation on the same boom.

Five-element Beam
For this case I choose the same boom as for the long-boom four-element
design, but will equally space five elements along the boom; the design
is shown in Table 8.13. The beam performance over the 14 MHz band
is shown in Table 8.14 and plots of gain and F/B ratio are shown in
Fig. 8.11. Figure 8.12 shows the free space angular performance; the 3 dB
total beamwidth is 66 degrees for the H-plane and 54 degrees for the
E-plane.
 Practical Amateur Yagi Antennas 8-7

Six-element Beam
Using exactly the same technique as for the five-element beam the com-
plete design for an equally-spaced six-element beam on the same 54 foot
boom is shown in Table 8.15, and the performance of the beam over the
14 MHz band is shown in Table 8.16. Plots of gain and F/B performance
are shown in Fig. 8.13 and of angular performance are shown in Fig. 8.15.
Full 3 dB beamwidths are 63 degrees (H-plane) and 53 degrees (Z-plane).
Note that there is not a great deal of difference between the perfor-
mances of the four, five and six-element beams when constructed on the
same 54-foot boom. Variations in peak F/B probably are not very signifi-
cant; one can in principle improve the already excellent figures by
optimizing the exact placement of elements along the boom. The gain
from the four-element beam, compared to the others, is a bit more sen-
sitive to frequency variations, and—not indicated on these figures — is
also more likely to be affected adversely by nearby scatterers. Both the
five and six-element versions are fine performers either alone or in stacked
systems; my own preference is the six-element version since it can be
mounted at its center with no mechanical interference from any element.

Designs For 21.3 MHz

I choose a basic cylindrical element 7/8 inches in diameter with the
taper schedule shown in Table 8.17. Two initial (half-length) sections are
shown; the first caused by the central mounting plate which I have chosen
to be six inches long, five inches wide, and 3/8 inches thick. There is
a small boom correction; for a three inch boom it is 0.09 inches. The
second initial section is a short piece of one inch diameter telescoping
tubing for reinforcing purposes only. This taper schedule results in the
equation:

1 = 1.03875.
— 2.59375 (8.4)

Three-element Beam
In exactly the same way as the 14 MHz beams were specified, a three-
element 21 MHz beam can be derived. The final design is shown in
Table 8.18 and its performance over the 21 MHz band shown in
Table 8.19. Fig. 8.15 shows plots of the gain and F/B performance over
the band and Fig. 8.16 shows the free space angular performance. Full
3 dB beamwidths are 96 degrees (H-plane) and 67 degrees (E-plane).

Six-element Beam
Following the same procedure a six-element 21 MHz beam can be
specified. Table 8.20 shows the design and Table 8.21 the performance
8-8 Chapter 8

over the 21 MHz band. Fig. 8.17 shows plots of gain and F/B over the
band and Fig. 8.18 shows free space angular performance. Full 3 dB beam-
widths are 63 degrees (H-plane) and 53 degrees (E-plane).

Designs For 28.5 MHz

I again choose a basic cylindrical element 7/8 inches in diameter with
a taper schedule shown in Table 8.22. Two initial (half-length) sections
are shown; the first caused by the central mounting plate, which I have
chosen to be four inches square, and 1/4 inches thick. I choose a boom
two inches diameter resulting in a small correction of 0.06 inches. A second
initial section is a short piece of reinforcing one inch diameter tubing.
This taper schedule results in the taper equation:

I = 1.02100/_q— 0.96835 (8.5)

Three-element Beam
A three-element 28 MHz beam can be derived by the same technique
used above. The result is shown in Table 8.23 and its performance over
the 28 MHz band shown in Table 8.24. Fig. 8.19 shows plots of gain
and F/B performance over the band and Fig. 8.20 shows the free space
angular performance. Full 3 dB beamwidths are 98 degrees (H-plane) and
68 degrees (E-plane).

Six-element Beam
A six-element 28 MHz beam can be similarly specified. Table 8.25 shows
the final design and Table 8.26 the performance over the 28 MHz band.
Fig. 8.21 shows plots of gain and F/B over the 28 MHz band and Fig. 8.22
shows free space angular performance. Full 3 dB beamwidths are
63 degrees (H-plane) and 53 degrees (E-plane).

Eight-element Beam
In the same way a long eight-element 28 MHz beam can be specified.
Table 8.27 shows this design, and Table 8.28 shows the performance over
the 28 MHz band. Figure 8.23 shows plots of gain and F/B over the
28 MHz band and Fig. 8.24 shows free space angular performance. Full
3 dB beamwidths are 53 degrees (H-plane) and 47 degrees (E-plane).

Summary
A number of practical Yagi designs have been developed for the
Amateur Radio bands: 7, 14, 21 and 28 MHz. Theoretical performance
in free space is shown for each design. A substantial number of tables
and figures have been prepared.
 Practical Amateur Yagi Antennas 8-9

Although these designs are all theoretical, I believe they should give
excellent performance in practice. Minor modifications might be needed
to compensate for electrical interaction with ground or with other adja-
cent antennas; such modifications will probably be inconsequential if the
Yagi is sufficiently high about ground (see Chapter 5) and if it is suffi-
ciently separated from adjacent antennas (Chapter 6).
8-10 Chapter 8

Design and Performance Information for a Two-Element 7 MHz Yagi

Table 8.1 — Taper schedule for 7 MHz

elements.
length diameter overlap
section (inches) (inches) (inches)
0 11.91 3.134
1 228.0 2.000 12
2 132.0 1.000 12
3 variable 0.875 >8

Table 8.2 — Element positions and lengths for a two-element 7 MHz Yagi.
Element diameters are two inches (radius = 0.00060537 wavelengths).
position leg /
element (inches) () (inches) () (inches)
reflector 0 0.0 815.24 0.49352 847.24
driver 232 0.1405 776.44 0.47003 805.81

Table 8.3 — Performance of a two-element 7 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) | (dBi) (dB) (ohms) (ohms)
exe ts) -5.6 23.6 — 89.4
6.80
3.18 -3.7 21.2 — 76.6
6.85
4.73 -1.7 19.9 — 63.3
6.90
6.95 5.88 0.3 19.7 — 49.8

7.00 6.59 2.3 20.8 — 36.4

7:05 6.94 4:2 22.9 -—23.5
AO 7.02 6.1 26.0 — 11.4
6.93 7.8 29.8 -0.3
7.15
6.76 9.2 33.8 9.9
7.20
10.3 37.4 19.2
We25 6.56
7.30 6.34 10.8 41.9 27.6

7.35 6.12 10.9 45.6 35.4

7.40 5.92 10.6 49.0 42.6
7.45 5.73 10.1 5241 49.4
7.50 5.55 9.6 54.8 55.9
 Practical Amateur Yagi Antennas 8-11

Fig. 8.1 — Free space

F/B (dB) performance of a two-
t element 7 MHz beam.
|
|

6.8 7.0 7.2 7.4

FREQUENCY (MHz)

0 dB - 6.9 cBi

Fig. 8.2 — Two-

element beam H-plane
and E-plane patterns
at 7.15 MHz.
8-12 Chapter 8

Design and Performance Information for a Three-Element 7 MHz Yagi

Table 8.4 — Element positions and lengths for a three-element 7 MHz Yagi.
Element diameters are two inches (radius = 0.00060537 wavelengths).
position leg /
element (inches) W (inches) (\) (inches)
reflector 0 0.0 819.20 0.49589 851.47
driver 260 0.15740 807.01 0.48854 838.45
director 472 0.28573 770.26 0.46629 799.21

Table 8.5 — Performance of a three-element 7 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (dBi) (dB) (ohms) (ohms)
5.64 2.7 30.2 -— 55.2
6.80
685 639 48 30.6 ~ 46.4
6.91 THA 30.7 — 38.6
6.90
7.25 9.6 30.0 -31.5
6.95
7.00 7.47 12.5 28.3 — 24.5
7.05 7.64 16.2 25.6 -17.1
WlOMe ocr. 80 21.8 22.3 -9.0
7.15 7.99 29.9 18.7 0.0
8.21 21.4 15.2 9.9
7.20
8.46 15.4 12.0 20.7
7.25
8.69 11.4 9.3 32.2
7.30
7.35 8.79 8.1 7.3 44.2
8.56 5.3 6.0 56.7
7.40
7.45 7.76 2.8 5.7 69.3
7.50 6.36 0.6 6.3 81.8
 Practical Amateur Yagi Antennas 8-13

GAIN (dBi)

bee FG Ta aS

Fig. 8.3 — Free space

performance of a
three-element 7 MHz
beam.

6.8 7.0 72 7.4

FREQUENCY (MHz)

0 dB - 8.0 GBi

Fig. 8.4 — Three-

element beam H-plane
and E-plane patterns
at 7.15 MHz.
8-14 Chapter 8

Design and Performance Information for a Three-Element 14 MHz Yagi

‘Table 8.6 — Taper schedule for 14 MHz elements.

heavy taper light taper
section length diameter length diameter
(inches) (inches) (inches) (inches)
0 3.91 HAY & 3.91 2.548
1 32.0 1.250 20.0 1.125
2 50.0 1.125 48.0 1.000
3 44.0 0.875 64.0 0.875
4 32.0 0.625 40.0 0.750
5 variable 0.500 variable 0.625

Table 8.7 — Element positions and lengths for a three-element 14 MHz Yagi.
Element diameters are 7/8 inches (radius = 0.00052599 wavelengths).
position leg hy I
element (inches) () (inches) () (inches) (inches)
reflector 0 0.0 413.02 0.49656 431.48 423.14
driver 128 0.15389 407.01 0.48933 425.05 416.85
‘director 232 0.27893 388.96 0.46763 405.75 397.96

Table 8.8 — Performance of a three-element 14 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (aBi) (cB) (ohms) (ohms)
13.80 . 7.21 9.2 30.5 —31.7
13.85 7.33 10.6 29.8 — 28.9
13.90 7.43 12.2 28.8 — 24.8
13.95 7.51 13.9 27.5 -—21.8
14.00 7.59 15.8 25.9 -17.5
14.05 7.67 18.1 24.2 -14.1
14.10 7115 2 teil 22.3 -9.3
14.15 7.84 24.8 20.4 -5.4
14.20 7.94 27.3 18.5 -0.0
14.25 8.04 24.5 16.6 5
14.30 8.16 20.6 14.8 10.2
14.35 8.29 1725 13.0 16.2
14.40 8.42 14.9 11.4 21.4
14.45 8.54 12.7 9.9 27.8
14.50 8.65 10.8 8.6 33.3
14.55 8.72 9.1 7.5 40.0
14.60 8.72 7.5 6.7 45.7
 Practical Amateur Yagi Antennas 8-15

Fig. 8.5 — Free space

performance of a
three-element 14 MHz
beam.

()
13.8 14.0 14.2 14.4 14.6
FREQUENCY (MHz)

0 dB — 8.0 GBi

Fig. 8.6 — Three-

element beam H-plane
and E-plane patterns
at 14.2 MHz.
8-16 Chapter 8

Design and Performance Information for a Short-Boom Four-Element

14 MHz Yagi

Table 8.9 — Element positions and lengths for a short-boom four-element

14 MHz Yagi. Element diameters are 7/8 inches (radius = 0.00052599
wavelengths).
position leq hy I
element (inches) () (inches) () (inches) (inches)
reflector 0.0 0.0 411.66 0.49493 430.03 421.72
driver 77.33 0.09297 404.73 0.48659 422.61 414.46
D1 154.67 0.18595 387.68 0.46609 404.38 396.62
D2 232.00 0.27893 387.68 0.46609 404.38 396.62

Table 8.10 — Performance of a short-boom four-element 14 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (dBi) (aB) (ohms) (ohms)
13.80 2.79 -0.7 9.3 — 46.4
13.85 4.05 1.5 8.5 — 41.0
13.90 5.10 3.7 8.0 — 34.5
13.95 5.95 6.0 7.5 -—29.2
14.00 6.64 8.4 (2 _ 22.9
14.05 7.19 451 6.9 -17.7
14.10 7.64 14.0 6.7 -11.4
14.15 7.99 17.0 6.6 -6.2
14.20 8.27 18.5 6.5 0.1
14.25 8.49 Vat 6.5 6.4
14.30 8.64 14.8 6.6 11.7
14.35 8.74 12.7 6.8 18.2
14.40 8.78 11.0 7.1 23.7
14.45 8.787 Bea 91S rere 30.6
14.50 8.74 8.4 8.5 36.7
14.55 8.66 7.4 9.9 44.3
14.60 8.55 6.7 12.2 51.7
Le
eeeitieperepeeieememeeeenteermemeemmeneeneeeimemeemmeemmemeeieemieemnemememmeememmenetenmimemmmememmeemnammememmmmmestamemmenammmmmemmmmemmemmimmmmenmememememmmeeetiieeeeneee
caeannarne i ate mene
 Practical Amateur Yagi Antennas 8-17

ee ee
ale
oe eae
Fig. 8.7 — Free space
performance of a
short-boom four-
element 14 MHz

ag a
beam.

ee tos [Fel ee
13.8 14.0 14.2
FREQUENCY (MHz)
14.4 14.6

0 dB - 8.1 dBi

Fig. 8.8 — Short-

boom four-element
beam H-plane and E-
plane patterns at
14.2 MHz.
 8-18 Chapter 8

Design and Performance Information for a Long-Boom Four-Element

14 MHz Yagi

Table 8.11 — Element positions and lengths for a long-boom four-element

14 MHz Yagi. Element diameters are 7/8 inches (radius = 0.00052599
wavelengths).
position leg My I
element (inches) (\) (inches) () (inches) (inches)
reflector 0.0 0.0 407.26 0.48964 425.32 417.11
driver 213.33 0.25648 397.54 0.47795 414.93 406.94
D1 426.67 0.51297 383.54 0.46112 399.85 392.28
D2 640.00 0.76945 383.54 0.46112 399.95 392.28

_ Table 8.12 — Performance of a long-boom four-element 14 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (aBi) (dB) (ohms) (ohms)
13.80 9.37 8.4 50.4 -27.9
13.85 9:51 9.4 50.0 — 25.8
13.90 9.64 10.5 49.1 — 22.6
13.95 9.77 11.9 47.6 — 20.6
14.00 9.91 13.6 45.4 * =-17.3
14.05 10.06 15.8 42.6 —14.6
14.10 10.23 18.9 39.2 -10.3
14.15 1043 242 35.3 -6.3
14.20 1064 34.2 31.2 -0.2
14.25 1086 23.5 A 7.0
14.30 11.07 17.3 23.2 14.2
14.35 11.21 13.3 20.0 23.7
14.40 11.18 10.2 Wed 33.0
14.45 10.87 7.7 16.3 44.2
14.50 10.16 5:5 16.4 54.7
14.55 9.06 3.6 18.1 66.6
14.60 7.68 19 21.2 77.0
 Practical Amateur Yagi Antennas 8-19

Pes
‘ai
|

Fig. 8.9 — Free space

performance of a
long-boom four-
element 14 MHz
beam.

Fig. 8.10 — Long-

boom four-element
beam H-plane and E-
plane patterns at
14.2 MHz.
8-20 Chapter 8

Design and Performance Information for a Five-Element 14 MHz Yagi

Table 8.13 — Element positions and lengths for a five-element 14 MHz Yagi.
Element diameters are 7/8 inches (radius = 0.00052599 wavelengths).
position leg My I
element (inches) () (inches) () (inches) (inches)
reflector 0 0.0 414.00 0.49774 432.53 424.17
driver 160 0.19236 398.66 0.47930 416.12 408.11
D1 320 0.38473 374.57 0.45033 390.36 382.90
D2 480 0.57709 374.57 0.45033 390.36 382.90
D3 640 0.76945 374.57 0.45033 390.36 382.90

Table 8.14 — Performance of a five-element 14 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (ABi) (dB) (ohms) (ohms)
13.80 9.79 10.6 38.5 -31.2
13.85 9.90 11.9 38.5 — 28.2
13.90 9.99 13.3 38.3 — 24.1
13.95 10.08 14.9 37.8 —21.1
14.00 10.17 16.9 37.0 - 16.8
14.05 10.25 19.4 35.9 -13.5
14.10 10.34 22.9 34.7 : -8.9
14.15 10.43 28.7 33.3 -5.2
14.20 1052 46.5 31.9 -0.0
14.25 10.61 28.4 30.4 5.5
14.30 1069 22.4 29.0 10.2
14.35 10.77 18.7 27.7 16.4
14.40 10.83 16.1 26.5 21.8
14.45 1086 14.0 25.6 28.6
14.50 1085 123 25.0 34.7
14.55 10.79 10.8 24.9 42.3
14.60 10.68 9.5 25.3 49.0
 Practical Amateur Yagi Antennas 8-21

aaa

et |
ee
Fig. 8.11 — Free
space performance of

aortic Aao
a five-element 14 MHz
beam.
‘ Pak
|

e piesa [eee
()
13.8 14.0 14.2 14.4 14.6
FREQUENCY (MHz)

0 dB — 10.5 dBi

Fig. 8.12 — Five-

element beam H-plane
and E-plane patterns
at 14.2 MHz.
8-22 Chapter 8

Design and Performance Information for a Six-Element 14 MHz Yagi

Table 8.15 — Element positions and lengths for a six-element 14 MHz Yagi.
Element diameters are 7/8 inches (radius = 0.00052599 wavelengths).
position leg Ny I
element (inches) () (inches) () (inches) (inches)
reflector 0 0.0 410.47 0.49349 428.75 420.47
driver 128 0.15389 398.96 0.47966 416.44 408.42
D1 256 0.30778 371.37 0.44649 386.94 379.55
D2 384 0.46167 371.37 0.44649 386.94 379.55
D3 512 0.61556 371.37 0.44649 386.94 379.55
D4 640 0.76945 371.37 0.44649 386.94 379.55

Table 8.16 — Performance of a six-element 14 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (Bi) (AB) (ohms) (ohms)
13.80 9.29 8.3 24.0 — 41.4
13.85 9.58 9.7 23.7 — 36.9
13.90 9.83 11.3, 23.5 — 31.3
13.95 10.05 13.0 23.2 — 26.8
14.00 10.23 oe 22.9 -—21.2
14.05 10.39 17.6 22.5 _ — 16.7
14.10 10.54 20.9 22.1 -—11.0
14:15 10.665, 25:5 21.7 -6.2
14.20 10.77 29.6 Zilles -0.2
[Link] 10:87 = 25.6 21.0 5.9
14.30 10.94 21.2 20.9 of
14.35 10.99 18.1 20.8 17.6
14.40 11.01 15.8 21.0 23.2
14.45 11.01 13.9 Zao 30.0
14.50 10.97 12.5 22.4 36.0
14.55 10.89 Vite 23.7 43.2
14.60. 10.78 10.1 25.6 49.6
 Practical Amateur Yagi Antennas 8-23

|
| GAIN (dBi)

Fig. 8.13 — Free

space performance of
a six-element 14 MHz
F/B (dB)
ae beam.

Fig. 8.14 — Six-

element beam H-plane
and E-plane patterns
at 14.2 MHz.
8-24 Chapter 8

Design and Performance Information for a Three-Element 21 MHz Yagi

Table 8.17 — Taper schedule

for 21 MHz elements.
length diameter
section (inches) (inches)
00 2.91 ~ 2.206
0 3.0 1.000
1 66.0 0.875
2 40.0 0.750
3 variable 0.625

Table 8.18 — Element positions and lengths for a three-element 21 MHz Yagi.
Element diameters are 7/8 inches (radius = 0.00078899 wavelengths).
position Iq ]
element (inches) () (inches) () (inches)
reflector 0 0.0 274.08 0.49427 279.51
driver 81 0.14608 269.31 0.48567 274.56
director 162 0.29215 256.93 0.46335 261.70

Table 8.19 — Performance of a three-element 21 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (aBi) (dB) (ohms) (ohms)
20.80 7.38 10.5 22:7. — 26.6
20.85 7.48 11.6 22.3 — 24.3
20.90 7.57 12.7 21.8 — 22.0
20.95 7.66 14.0 21.3 — 18.6
21.00 7.75 ike) 20.6 - 16.3
21.05 7.82 17.1 20.0 -— 13.9
21.10 7.90 19.1 19.3 — 10.4
21.15 7.98 21.6 18.5 —8.0
21.20 8.05 25.0 17.7 -5.4
21.25 8.13 30.6 16.9 -1.8
21.30 8.21 38.3 16.1 0.9
21.35 8.28 30.5 15.2 3.6
21.40 8.37 24.9 14.4 AS)
21.45 8.45 21.5 13.5 10.4
21.50 8.53 18.9 12:7, 13.3
21.55 8.61 16.9 11.9 17.4,
21.60 8.69 15.2 11.2 20.5
21.65 8.76 13.8 10.4 23.6
21.70 8.83 12.4 9.8 27.9
21.75 8.88 11.3 9.1 31.1
21.80 8.92 10.2 8.5 34.5
 Practical Amateur Yagi Antennas 8-25

Fig. 8.15 — Three-

element 21 MHz beam
free space per-
formance.
f F/B (dB)
3

)
20.8 21.0 21.2 21.4 21.6 21.8
FREQUENCY (MHz)

0 dB - 8.3 dBi

Fig. 8.16 — Three-

element beam H-plane
and E-plane patterns
at 21.3 MHz.
8-26 Chapter 8

Design and Performance Information for a Six-Element 21 MHz Yagi

Table 8.20 — Element positions and lengths for a six-element 21 MHz Yagi.
Element diameters are 7/8 inches (radius = 0.00078899 wavelengths).
position leg
element (inches) () (inches) () (inches)
reflector 0.0 0.0 273.41 0.49307 278.82
driver 85.2 0.15365 265.23 0.47832 270.32
D1 170.4 0.30730 245.55 0.44282 249.88
D2 255.6 0.46095 245.55 0.44282 249.88
D3 340.8 0.61460 245.55 0.44282 249.88
D4 426.0 0.76825 245.55 0.44282 249.88

Table 8.21 — Performance of a six-element 21 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (dBi) (dB) (ohms) (ohms)
9.78 10.9 23.6 -— 31.9
20.80
9.92 11.9 23.4 - 29.1
20.85
10.05 13.1 23.2 — 26.4
20.90
1OS/, 14.3 23.0 — 22.5
20.95
10.27 15.7 22.7 - 19.8
21.00
10.37 17.3 22.5 . —16.9
21.05
10.47 19.2 22.2 - 13.0
21.10
10.55 21.5 22.0 - 10.1
21.15
10.63 24.3 Pat -7.2
21.20
10.71 27.7 Pails) -3.1
21.25
10.77 29.8 21.2 -—0.1
21.30
10.83 27.8 21.0 3.0
21.35
21.40 10.89 24.4 20.9 7.3
10.93 PANT 20.8 10.5
21.45
10.97 19.5 20.7 13.8
10.99 17.8 20.8 18.3
5 Pe
11.01 16.3 20.9 QAR
21.60
11.02 15.0 21.1 25.2
21.65
11.01 13.9 Failte) 29.9
21.70
10.99 13.0 21.9 33.6
Pala TAs
21.80 10.95 12.1 22.6 37.3
 Practical Amateur Yagi Antennas 8-27

GAIN (dBi)

Fig. 8.18 — Six-

element beam H-plane
and E-plane patterns
F/B (dB)
at 21.3 MHz.
| 3

()
20.8 21.0 21.2 21.4 21.6 21.8
FREQUENCY (MHz)

0 dB - 10.8 dBi

Fig. 8.17 — Six-

element 21 MHz beam
free space per-
formance.
8-28 Chapter 8

Design and Performance Information for a Three-Element 28 MHz Yagi

Table 8.22 — Taper schedule

for 28 MHz elements.
length diameter
section (inches) (inches)
00 1.94 1.902
0 1.0 1.000
1 69.0 0.875
2 variable 0.750

Table 8.23 — Element positions and lengths for a three-element 28 MHz Yagi.
Element diameters are 7/8 inches (radius = 0.0010557 wavelengths).
position leg /
element (inches) () (inches) () (inches)
reflector 0 0.0 205.27 0.49531 207.64
driver 57 0.13754 201.40 0.48598 203.69
director 114 0.27508 191.70 0.46257 193.79

Table 8.24 — Performance of a three-element 28 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (Bi) (AB) (ohms) (ohms)
27.80 7.33 10.5 23.0 - 27.3
27.90 7.46 12.1 22.4 * —23.2
28.00 7.57 13.8 21.6 - 20.1
28.10 7.67 15.9 20.6 -— 15.9
28.20 Tl 18.4 19.5 - 12.6
28.30 7.86 21.8 18.4 -8.2
28.40 7.96 26.9 17.1 -4.6
28.50 8.06 33.7 15.8 0.1
28.60 8.17 PA oF 14.5 4.9
28.70 8.27 21.8 13.3 9.0
28.80 8.39 18.3 12.0 14.2
28.90 8.50 15.7 10.9 18.5
29.00 8.61 13.5 9.8 23.9
29.10 8.70 UE 8.8 28.5
29.20 8.77 10.1 7.9 34.2
29.30 8.79 8.6 Wee 39.0
29.40 8.76 7.2 6.6 44.8
29.50 8.65 6.0 6.1 49.8 ©
29.60 8.44 4.8 5.8 55.8
29.70 8.11 3.6 5.6 60.8
29.80. 7.66 2.6 5.6 66.8
29.90 7.08 1.5 5.8 GO
 Practical Amateur Yagi Antennas 8-29

Fig. 8.19 — Three-

element 28 MHz beam
free space per-
formance.

0 dB — 8.1 dBi

Fig. 8.20 — Three-

element beam H-plane
and E-plane patterns
at 28.5 MHz.
8-30 Chapter 8

Design and Performance Information for a Six-Element 28 MHz Yagi

Table 8.25 — Element positions and lengths for a six-element 28 MHz Yagi.
Element diameters are 7/8 inches (radius = 0.0010557 wavelengths).
position leg /
element (inches) () (inches) () (inches)
reflector 0.0 0.0 204.20 0.49273 206.55
driver 63.6 0.15347 197.74 0.47715 199.96
D1 127.2 0.30693 182.27 0.43981 184.16
D2 190.8 0.46040 182.27 0.43981 184.16
D3 254.4 0.61387 182.27 0.43981 184.16
D4 318.0 0.76733 182.27 0.43981 184.16

Table 8.26 — Performance of a six-element 28 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (ABi) (AB) (ohms) (ohms)
27.80 9.78 10.9 23.6 — 32.7
27.90 9.98 12.4 23.3 — 27.8
28.00 10.15 14.1 23.0 — 23.8
28.10 10.30 16.2 22.6 - 18.9
28.20 10.44 18.6 22.2 - 14.8
28.30 10.56 21.9 21.9 -9.7
28.40 10.67 26.3 21.5 5-5
28.50 10.77 30.1 ane -0.2
28.60 1085 26.3 20.9 5.3
28.70 1092 22.1 20.7 9.9
28.80 10.97 19.0 PAT 15.6
28.90 11.01 16.7 20.8 20.4
29.00. 11.02 14.9 24-1 26.4
29.10 11.00 13.4 Path 31.6
29.20 10.96 5 22.6 37.9
29.30 10.88 se 23.8 43.4
29.40 10.79 10.2 25.6 49.9
29.50. 10.66 9.4 28.1 55.6
29.60 10.52 8.7 31.6 62.3
29.70 10.36 8.2 36.2 67.8
29.80 10.20 7.8 42.8 73.7
29.90 10.03 7.5 51.6 77.0
 Practical Amateur Yagi Antennas 8-31

aco
iiss
ae Fig. 8.21 — Six-
element 28 MHz beam
free space per-

ia icine
|

28.0 28.3 28.8 29.3 29,7

FREQUENCY (MHz)

0 dB — 10.8 dBi

Fig. 8.22 — Six-

element beam H-plane
and E-plane patterns
at 28.5 MHz.
8-32 Chapter 8

Design and Performance Information for an Eight-Element 28 MHz

Yagi

Table 8.27 — Element positions and lengths for an eight-element 28 MHz Yagi.
Element diameters are 7/8 inches (radius = 0.0010557 wavelengths).
position leg /
element (inches) (\) (inches) ) (inches)
reflector 0.0 0.0 205.14 0.49501 207.51
driver 76.3 0.18408 195.86 0.47261 198.04
D1 152.6 0.36816 175.70 0.42396 177.45
D2 228.9 0.55224 175.70 0.42396 177.45
D3 305.1 0.73631 175.70 0.42396 177.45
D4 381.4 0.92039 175.70 0.42396 177.45
D5 457.7 1.10447 175.70 0.42396 177.45
D6 534.0 1.28855 175.70 0.42396 177.45

Table 8.28 — Performance of an eight-element 28 MHz Yagi.

freq. gain F/B resistance reactance
(MHz) (ABi) (dB) (ohms) (ohms)
27.80 11.31 13.0 34.4 -29.1
27.90 11.42 14.5 34.6 — 24.6
28.00 11.52 16.1 34.7 -—21.2
28:10 , |11.61 18.0 34.7 -— 16.7
28:20) e707 20:3 34.7 -— 13.1
28.30 11.77 . 233 34.6 . —8.5
28.40 11.84 26.8 34.5 -4.7
28.50 11.90 29.1 34.4 0.1
28.60 11.96 26.7 34.4 Sit
23:70) 201 23.3 34.4 9.2
28.80 12.04 20.5 34.6 14.4
28.90 12.07 18.4 34.9 18.8
29.00 12.08 16.6 35.5 24.2
29.10 12.07 15.2 36.2 28.8
29.20 12.05 14.0 37.3 34.4
29.30 12.01 13.0 38.7 39.0
29.40. 11.95 12u8 40.6 44.6
29.50 11.87 11.3 42.8 49.0
29.60 11.78 10.7 45.7 54.2
29.70 11.67 10.1 49.2 58.1
29.80 11.56 9.7 53.4 62.2
29.90 11.43 9.3 58.1 64.5
 Practical Amateur Yagi Antennas 8-33

p Fig. 8.23 — Eight-

£/8 (68) x element 28 MHz beam
\ free space per-
formance.

28.0 28,3 28.8 29.3 29.7

FREQUENCY (MHz)

0 dB —- 11.9 dBi

Fig. 8.24 — Eight-

element beam H-plane
and E-plane patterns
at 28.5 MHz.
 Hoe
ee bdraminsed ety Lesa a an sae ha) a

ie ane aman ae tettoen!mp inte

Ap Eee &

Mt £
¥
“hey
ae
i luli
ART
ssi
shai te
ud hi a Mion a
-
nanremane
ee ee
B i wid inte Ap ; iF
or '

‘9G whege pete a nye ;

Mery Apomeminenat -
Oar, 4

= © ad re
1 7 :
7 i) eer
}
Z a i hee i et te 7

t - wTc ”
i ’ Aiea Se rs

aay
i

-®
f.

. if
\
Re
*

: allies } ven
'
- 4 ®
4 0 wy ard TLRS ih: 87

ve jai i .
noe f+ 4 6
4 >,
. i

<9
< ;
.
i»
a,
£ / 7

—~ rc }
b e = ——

sagmeee "he 2°a =a Ihe Ay ae

La at
asralg ee ee eit, 7 :
lwp
aie ee | “hd ary “eetig 7
a bs Pale
A tM, " 7% he
- ¥ y
ee lo ‘ ’ ,
Pe ¥s
? - |
ud ae \
ay fay) aes

f- i x :
oe Fe | >
5 7 art be

ae ’ 7

ae | a er ee
od ¥
’ 1 b

aS a

. "
: a]
‘
; ' \
a
’ ry mS A
i Ay
j es) He
)* & J fs Ps eg
 n

ere

'y
ue

Ce

'

iv

“18
:
i
ge
ry}

vi
uy

hs
a
]
 ARRL MEMBERS
This proof of pur-
chase may be used
as a $1.50 credit
on your next ARRL purchase or
renewal 1 credit per member.
Validate by entering your
membership number — the first 7
digits on your QST label —
below:
spnconse
ahs
 aANT

Rhea: ae 7 ae ie OA ae Mae:My yu he et Boreas, 4 Bau)

ip) : aad ie ity ay wihye Wig Vaan ha aa i a va a

Ae te

P 2

Ww

7 4 , ait ok yA ok

c fey) 4 le Fade oartar ek

i ti p

ARG NattmihaclMy
Fie Cu ny “his itp
iy iesoNie ore :
“tj

af | ag
wie" alt

Wie ene). iteihe:

vi , ! stan et Xa wei i)
+ ! ai fea rnp :

4)
ee
ae rh

wT i ?

ai
 [male]
Fela
Please use this form to give us your comments on this book. Tell us
what you liked best about the book, and what improvements you would
like to see us make in future editions.

Name
Call sign

Daytime Phone ( ) interne

Rete Re en home LACS

Address

City, State, Zip

How long have you been licensed?

 4

(
From Please affix (
¢ postage. Post
‘
Office will not
f
deliver without
f
sufficient postage. ‘
f
‘
‘
‘
f
,
‘
(
‘
(
(
(
(
(
t
Editor, Yagi Antenna Design, ‘

First Edition t
(
American Radio Relay League f

225 Main Street t

,
Newington, CT USA 06111 ‘
‘
:
:
‘
‘

;
secre ccccccccecccccceecccescscvcccveees Please fold and tape e@eoeeeeeoeeoeeaeoeeeeeoeeeoeeeoeed

'
 {

; : *
aN : 4 a
‘

, wt oh -. Deki emer tere hk ie WH

ant
BY
q ih? fy
oye i +>.
Wire st es Mn | hy (hapyad 4
a ic Cy La eay " y y :
; hae ty Loldabigateg ttef oe Whigsyue ‘ih
Lea p wf epyta ba
a ‘ i af
; Pha ‘ ; hy i }

} Wak oy ab eae
nhs ‘ aM Ay a |, ee AD
Be) , Po a ,( Me
. WV M.
. 5 j Ag
: a ral ee eee a a
fdas‘
\
inca
Ky Ay ‘
sy iad oe PPE © tahoe
ot
Regs all
i
y
ve de i i SHEDS x
P * ; i : is a OO Fay a : ie
if ‘ ' pee _
i |
Ad :
‘ '
My miyne
ae ef
se : 5

,
Ke ig » Tom |
4 > of un: th el Bt ty yd Pilea,
ba tin
Ae
PB a -
, 1 ry; ue Eautie a ercall hia
vf ie . 4 we
Se st)
' ( a,
ikaagiaiia 6° So
st
ls
j

Aisegarealinick GON hul

il ht

1 ‘4
ies hie has ge : ‘
f
a ra
wr! ‘ 2
Z) (i : ) 4. Ty
: fit 1 AAS iia ai
 The subject of antennas is always a hot
topic among radio amateurs, and volumes
have been written about each type. The
expert on the subject of Yagi Antenna
Design was Dr. James L. Lawson, W2PV;
and his presentation on this subject at a
convention was sure to ‘“‘pack them in.”
These talks culminated in a series of
articles on Yagis in ham radio magazine.
The material from these articles that is
presented here was polished and
expanded by Dr. Lawson before his
death. Yagi Antenna Design presents both
the theory and practical information you
need to understand this popular antenna.

. em