Are you sure?
This action might not be possible to undo. Are you sure you want to continue?

ARM SPIRAL ANTENNAS
A thesis submitted in partial fulfillment
of the requirements for the degree of
Master of Science in Engineering
By
DOUGLAS JEREMY GLASS
B.S.E.P., Wright State University, 2003
2007
Wright State University
WRIGHT STATE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
June 29, 2007
I HEREBY RECOMMEND THAT THE THESIS PREPARED UNDER MY
SUPERVISION BY Douglas Jeremy Glass ENTITLED Investigation of Cylindrically
Conformed FourArm Spiral Antennas BE ACCEPTED IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF Master of Science in Engineering.
Committee on
Final Examination
_________________________________
Ronald Riechers, Ph.D.
_________________________________
Fred Garber, Ph.D.
_________________________________
Marian Kazimierczuk, Ph.D.
_________________________________
Joseph F. Thomas, Jr., Ph.D.
Dean, School of Graduate Studies
_________________________________
Ronald Riechers, Ph.D.
Thesis Director
_________________________________
Fred Garber, Ph.D.
Department Chair
iii
ABSTRACT
Glass, Douglas Jeremy, M.S. Egr., Department of Electrical Engineering, Wright State
University, 2007. Investigation of CylindricallyConformed FourArm Spiral Antennas.
A fourarm spiral antenna offers broadband frequency response, wide beamwidths,
reduced size compared to other antenna designs, and the ability to determine the relative
direction of an incident signal with appropriate modeforming. The reduced overall area
projection of the fourarm spiral antenna compared to other antenna designs and the
ability to be manufactured in a planar format allows the antenna to reside within an
Unmanned Air Vehicle (UAV) fuselage. This thesis investigates the effects of
cylindricallyconforming two different designs of a fourarm spiral antenna to reside
within the fuselage of a mediumsized UAV. Theoretical predictions of antenna
performance were created using the Numerical Electromagnetics Code (NEC) package
and compared to measured results of flat and cylindricallyconformed fourarm spiral
antennas with and without ground plane apertures.
iv
Table of Contents
Page
1. Introduction………………………………………………………………..…………1
1.1. Thesis Motivation…………………………………………………………...……1
1.2. Thesis Objectives…………………………………………………………………1
2. FrequencyIndependent Antennas……………………………………...…………..2
2.1. Theory of Frequency Independence…………………………………………..….2
2.2. Types of FrequencyIndependent Antennas…………………………………...…5
2.2.1. Biconical Antennas……………………………………………………..…5
2.2.2. LogPeriodic Antennas……………………………………............…..…..8
2.2.3. Spiral Antennas……………………………………………………….….12
2.3. Modal Behavior of the Spiral Antenna…………………………………….....…19
3. Antenna System Development……………………………………….….………....26
3.1. Spiral Antenna Development………………………………………………...….26
3.1.1. Spiral Antenna Design……………………………………………...……26
3.1.2. Spiral Antenna Fabrication………………………………………...…….28
3.2. ModeFormer Development…………………………………………………….35
3.2.1. ModeFormer Design ……………………………………………………35
3.2.2. ModeFormer Fabrication ……………………………………………….36
4. Theoretical Model……………………………………………………….……….…39
4.1. Maxwell’s Equations……………………………………………………...…….39
4.2. Numerical Electromagnetics Code Discussion…………………………...……..41
4.2.1. Integral Equations……………………………………………………..…41
4.2.2. Thin Wire Approximation…………………………………………..……44
4.2.3. Method of Moments Solution Method……………………………….…..48
4.3. Numerical Electromagnetics Code Results…………………………………..…54
4.3.1. Archimedean Spiral……………………………………………….……..55
4.3.2. CylindricallyConformed Archimedean Spiral………………………..…66
4.3.3. Square Spiral…………………………………………………………..…76
4.3.4. CylindricallyConformed Square Spiral……………………………..…..86
4.3.5. Radius of Curvature Analysis of the Archimedean Spiral…………….…96
5. Measured Results…………………………………………………………..…..….102
5.1. Measurement Method………………………………………………………….102
5.1.1. Vector Network Analyzer Discussion………………………………….103
5.1.2. Reference Antenna Discussion…………………………………………104
5.1.3. Desktop Antenna Measurement System Discussion……………………105
5.2. ModeFormer Results………………………………………………………….107
v
5.2.1. Power Pattern Phase Balance Comparison……………………………..113
5.3. Spiral Antenna Results………………………………………………...………119
5.3.1. Archimedean Spiral……………………………………………...……..120
5.3.2. CylindricallyConformed Archimedean Spiral…………………………125
5.3.3. Square Spiral……………………………………………………...…….130
5.3.4. CylindricallyConformed Square Spiral………………………………..135
5.3.5. Square Spiral within Ground Plane Aperture……………………….….140
5.3.6. CylindricallyConformed Archimedean Spiral within Ground Plane
Aperture…………………………………………………………….…..148
6. Conclusions…………………………………………………………….…………..158
6.1. Summary of Work……………………………………………………….…….158
6.2. Recommendations for Further Study…………………………………..………159
References………………………………………………………………..……….........160
vi
Acknowledgements
I would like to begin by thanking Dr. Ronald Riechers for all the time that he has
spent guiding and motivating me to complete this task. I would also like to thank Dr.
Riechers for instilling a passion for electromagnetics that has only grown over the many
years that I have had the honor of being one of his students.
I would also like to thank the personnel and the measurement capabilities of
Spectral Energetics Incorporated. Without it, I would not have had any measured data
with which to compare my theoretical model.
I would also like to thank Dr. Fred Garber and the Electrical Engineering
Department of Wright State University for providing me the opportunity to attain a
Master of Science in Engineering. I would also like to thank Dr. Marian Kazimierczuk
for taking the time to serve on my committee. Finally, I would also like to thank Mr.
Barry Woods. Without his guidance, my undergraduate and graduate experience would
not have been as enjoyable.
My final thank you goes out to all my friends and family who encouraged me
throughout my education, especially my friend Jason and my oldest brother John. I would
also like to thank my friends Mark Stewart, Lamar Westbrook, and Edward Dininger for
the time they spent helping me take antenna pattern measurements and my friend Dr.
Chad Hagedorn for his time spent editing my thesis. Finally, to my wife Sharon, without
your love, patience, and understanding, “Sweets,” none of this would have been possible.
1
1. INTRODUCTION
1.1. Thesis Motivation
Developing a lowcost (disposable) and lightweight directionfinding system for
use with an Unmanned Air Vehicle (UAV) is the motivation of this thesis. As the
battlefield evolves around the world, the ability to locate communications of enemy
combatants is one of the most relevant tasks for supporting the warfighter. Developing
an antenna for a directionfinding system with an emphasis on modularity for different
mission types and broad frequency response would demonstrate the utility of such a
system to the military.
1.2. Thesis Objectives
The objectives of this thesis are the following:
Define a suitable antenna as defined by the thesis motivation.
Describe the characteristics of the antenna.
Discuss the development of an antenna capable of directionfinding.
Define a theoretical model capable of describing the antenna.
Provide a comparison of the theoretical and measured results.
2
2. FREQUENCYINDEPENDENT ANTENNAS
2.1 Theory of Frequency Independence
Due to the everexpanding utilization of the electromagnetic spectrum, antennas
that can operate over wide frequency ranges have become increasingly desirable. As
stated in Balanis [1], prior to the 1950s, antenna technology could only provide
bandwidths of 2:1 or less. Afterwards though, the concept of specifying antenna
geometries purely by angles instead of characteristic lengths allowed for bandwidths of
40:1 or more. If an antenna is only defined by angles, attributes such as the characteristic
input impedance, field pattern, and polarization remain constant as the size of the antenna
is scaled.
The theory of frequencyindependence, as described by the work of Rumsey [2]
and Elliot [3], begins by assuming an antenna geometry within spherical coordinates
(ρ.θ.φ), where each terminal (feed point) of the antenna surface is infinitesimally close to
the origin and is confined to rotation about the θ = 0 axis. To further simplify the
problem, surfaces are assumed to be perfectly conducting, and the region surrounding the
antenna is assumed to be infinite in extent, homogenous, and isotropic. Such a surface is
described by a single curve (for a wire) or a set of curves (to define the edges of a strip)
( ) φ θ, F r = (2.11)
3
where r represents the radial distance along the curve. To define the property of scaling,
start by assuming an antenna were to be scaled to operate at a frequency that is K times
lower than the original antenna. The physical size of the antenna would need to be
increased by a factor of K. The scaled curve is then described by
( ) φ θ, KF r = ′ (2.12)
where both surfaces superimpose one another after the second curve is rotated by an
angle φ
c
about the θ = 0 axis. This is shown by the following
( ) ( ) φ θ φ φ θ , , KF F
c
= + (2.13)
where the rotated angle φ
c
depends only on K. The following figure depicts three spirals
that correspond to a singlearm spiral, a scaled singlearm spiral, and a scaled singlearm
spiral that has been rotated to visualize congruence of the two spirals.
Figure 2.11 Comparison of Scaled Spirals
4
To prove congruence between the two curves, begin by differentiating both sides
of equation (2.13) with respect to both φ
c
and φ.
( )   ( )   ( )  
c
c c c
F F
d
dK
KF
d
d
φ φ θ
φ
φ θ
φ
φ θ
φ
+
∂
∂
= = , , , (2.14)
( )  
( )
( )  
c
c c
F KF
d
d
φ φ θ
φ φ
φ θ
φ
+
+ ∂
∂
= , , (2.15)
( )   ( )   ( )  
c
F F K KF φ φ θ
φ
φ θ
φ
φ θ
φ
+
∂
∂
=
∂
∂
=
∂
∂
, , , (2.16)
( )  
( )
( )  
c
c
F KF φ φ θ
φ φ
φ θ
φ
+
+ ∂
∂
=
∂
∂
, , (2.17)
Equating (2.15) and (2.17) results in
( )   ( )   φ θ
φ
φ θ
φ
, , F K F
d
dK
c
∂
∂
= . (2.18)
Substituting (2.11) into (2.18) gives
φ φ ∂
∂
=
r
r d
dK
K
c
1 1
. (2.19)
Due to the spatial independence of the lefthand side of (2.17), the general solution of
the curve becomes
( ) ( ) θ φ θ
αφ
f e F r = = , (2.110)
where α is
c
d
dK
K φ
α
1
= (2.111)
and f(θ) is a completely arbitrary function that describes the curve or curves that define
the surface of the antenna.
5
2.2 Types of FrequencyIndependent Antennas
Frequencyindependent antennas within the context of this thesis can be
categorized into three types: biconical, logperiodic, and spiral. The following sections
describe the variants, attributes, limitations, and the basic antenna theory behind each
type.
2.2.1 Biconical Antennas
Biconical antennas have several geometric variants such as the planar biconical
(bowtie), finite biconical, and the discone. This section focuses on the infinite biconical
antenna to describe the general theory. Although it is impractical to assume the existence
of a biconical antenna of infinite extent, it can be shown that the current diminishes to a
negligible amount after a finite distance. It is suitable to define the length of a finite
biconical antenna by this truncated distance, where the performance of the antenna
approaches the infinite case. To describe the theory of biconical antennas, we start with
the well known fact that the bandwidth of a dipole antenna can be increased by enlarging
the diameter of the conductors. Expanding on this concept, if we instead flared the
diameter of the conductors outward as the distance from the feed point increases, the
geometry of a biconical antenna is defined. The structure can be analyzed as a
transmission line where the voltage source produces a current distribution that flows
radially outwards. This current distribution creates a magnetic field that encircles each
cone. The electric field, which is perpendicular to the magnetic field, is θ directed. The
following figure depicts the geometry of the infinite biconical antenna.
6
Figure 2.2.11 Geometry of the Infinite Biconical Antenna (From Stutzman and Thiele [4])
From Stutzman and Thiele [4], the normalized field pattern of the antenna is
( )
( )
( ) θ
θ
θ
sin
sin
h
F = (2.2.11)
where θ
h
is the cone half angle and θ is the angle from the zaxis. From this relation, it
can be determined that the larger the cone angle, the greater the field strength will be
within the available freespace. The following figure depicts the normalized field pattern
of the infinite biconical antenna.
7
Figure 2.2.12 Normalized Field Pattern of the Infinite Biconical Antenna for Different θ
h
The input impedance of the antenna is defined as


¹

\


¹

\

=
2
cot ln 120
h
in
Z
θ
(2.2.12)
which has the unique property of being invariant with respect to frequency. The same
frequency invariant properties can also be extended to the field pattern. The following
figure depicts the input impedance versus the cone half angle. It should be noted that for
the case of the finite biconical antenna, the ends of the cones cause a reflection of the
incident waves produced by the voltage source to produce a standing wave within the
antenna structure. This will create a complex input impedance at the antenna terminals.
8
Figure 2.2.13 Normalized Impedance of the Biconical Antenna for Different θ
h
As a candidate for the antenna system, the biconical antenna would need to be
utilized in its planar form, the bowtie, to satisfy the geometry requirements. The bowtie
antenna would also need to be a part of a phased array to provide the modal
characteristics necessary to provide directionfinding capabilities. These inadequacies
preclude the biconical antenna from being considered for the antenna system defined by
the motivation of this thesis.
2.2.2 LogPeriodic Antennas
The second type of frequencyindependent antenna is the logperiodic. As with
the biconical antenna, there are several geometric variants of the logperiodic, such as the
toothed planar, the toothed wedge, the trapezoid wire, the zigzag wire, and the dipole
array. The following section will focus on the logperiodic dipole array to describe the
9
general theory. The geometry of the logperiodic dipole array antenna, like the biconical,
is based on the dipole antenna. We begin by defining the logperiodic dipole array as a
linear distribution of dipoles with increasing length that are connected together by
alternating the feed polarity to each dipole. The following figure depicts the geometry of
the logperiodic dipole array.
Figure 2.2.21 Geometry of the LogPeriodic Dipole Array Antenna (From Stutzman and Thiele [4])
It can be seen from the geometry of the logperiodic antenna that it cannot be solely
described by angles as prescribed by the characteristics of the frequencyindependent
antenna. It should be noted, though, that the performance of the logperiodic antenna
approaches the characteristics of the frequencyindependent antenna.
As derived in Stutzman and Thiele [4], the bandwidth of the logperiodic dipole
array is defined by the following equations
2
1
ency LowerFrequ
L
λ
≈ and
2
ency UpperFrequ
N
L
λ
≈ (2.2.21)
10
where L
1
and L
N
are the longest and shortest dipoles respectively. The following
equations define the design parameters used to produce logperiodic dipole arrays of
varying performance.
n
n
n
n
n
n
d
d
L
L
R
R
1 1 1 + + +
= = = τ (2.2.22)
Where τ is the geometric ratio of the antenna, it defines the length growth rate for each of
the successive dipoles used in the array.
n
n
L
d
2
= σ (2.2.23)
Where σ is the spacing factor, it describes the distance between each successive dipole
within the array.

¹

\
 −
=
−
σ
τ
α
4
1
tan 2
1
(2.2.24)
Finally, α is the angle that is formed from the intersection of the lines extending from
both ends of the dipoles along the array. The following figure depicts design curves for
different logperiodic dipole arrays and their respective maximum gains.
11
Figure 2.2.22 Maximum Gain Curves of the LogPeriodic Dipole Array Antenna with Respect to Spacing
and Scaling Factors (From Balanis [1])
As seen in the figure, the best performing antennas have a very slow transition in length
from one dipole to the next. The active region, or where the current is concentrated on the
antenna, is around the dipole where the length is approximately one half of a wavelength
of the operating frequency. It should also be noted that, in practice, dipoles are often
added to either end of the antenna to ensure the strength of the active region at either end
of the antenna. The impedance, like other frequencyindependent antennas is nearly
constant over frequency. The following figure depicts a general impedance variation with
respect to frequency for a logperiodic dipole array.
12
Figure 2.2.23 Representative Impedance Variation with Frequency for a LogPeriodic Dipole Array
Antenna (From Balanis [1])
As a candidate for use with the antenna system, the logperiodic antenna would
need to be utilized in any of its planar forms to satisfy the geometry requirements. The
logperiodic antenna would also need to be a part of a phased array to provide the modal
characteristics necessary to provide directionfinding capabilities. These inadequacies
preclude the logperiodic antenna from being considered for the antenna system defined
by the motivation of this thesis.
2.2.3 Spiral Antennas
The final type of frequencyindependent antenna to describe within this
discussion is the spiral. Like the biconical and logperiodic antennas, the spiral also has
several geometric variations in design. The variants can be categorized into three types:
equiangular, logarithmic, and Archimedean. Each spiral variant also has a planar, conical,
13
and spherical form. Previously, the conical form was the most prevalent where the planar
form has become the popular choice for research due to its ability to be manufactured by
printed circuit techniques. To describe the general theory of each type of spiral antenna,
we will focus on the twoarm planar versions of each. The twoarm versions of the spiral
antenna can be thought of as a long dipole whose conductors have been wound around
the feed axis.
The first type of spiral to examine is the equiangular; the following figure depicts
the geometry of the equiangular spiral.
Figure 2.2.31 Geometry of the Planar Equiangular Spiral Antenna (From Stutzman and Thiele [4])
As stated above, all spiral antennas can be defined purely by angles. We begin by
defining the design equations from Stutzman and Thiele [4]. There are four equations that
define the radius of the inner and outer curves of each conductor’s surface.
φ a
o
e r r =
1
(2.2.31)
14
( ) δ φ−
=
a
o
e r r
2
(2.2.32)
( ) π φ−
=
a
o
e r r
3
(2.2.33)
( ) δ π φ − −
=
a
o
e r r
4
(2.2.34)
The flare rate a is more easily represented by the expansion ratio
π
ε
2 a
e = . (2.2.35)
When designing equiangular spirals, it has been experimentally shown that the best
performing designs have one and onehalf turns (φ = 0 to 3π) and have a flare rate equal
to 0.221. The typical bandwidth of this design is 8:1, where r
o
is equal to λ
UpperFrequency
/4,
and the maximum radius R is equal to λ
LowerFrequency
/4. It should be noted that the previous
figure can also be described as a selfcomplementary antenna. Selfcomplementary
structures are comprised of a conductor surface whose compliment is identical to the
original surface. This occurs when the conductor surface occupies the same area as the
free space surface. Selfcomplementary antennas have the remarkable property of having
a nearly constant impedance across their bandwidth regardless of shape. The twoarm
selfcomplementary antenna has a theoretical input impedance of 189 ohms where
experimentation has determined that this number should be reduced to approximately 164
ohms.
Other forms of the equiangular spiral are shown is the following figures, the first
is the conical spiral, which was one of the first spiral antennas geometries to be evaluated.
15
Figure 2.2.32 Geometry of the Conical Equiangular Spiral Antenna (From Stutzman and Thiele [4])
The other geometric conformation is the spherical spiral. Practically speaking though,
this specific spiral would need to be cut in half along the equatorial plane to create a
viable antenna. It does provide additional application possibilities, such as conforming to
the nose of a guided munition.
16
Figure 2.2.33 Geometry of the Spherical Equiangular Spiral Antenna (From Gazale [6])
The second type of spiral to examine is the logarithmic; the following figure
depicts the geometry.
Figure 2.2.34 Geometry of the Planar Logarithmic Spiral Antenna (From Johnson [5])
17
Like the equiangular spiral, the radius of each arm is given by an exponential. The radius
equations as defined by Johnson [5] are
φ a
o
e r r =
1
(2.2.36)
( ) δ φ−
=
a
o
e r r
2
. (2.2.37)
The other radii, r
3
and r
4
, can be created from r
1
and r
2
by subtracting π from the
exponent. The pitch angle ψ is related to the flare rate a by
a
1
tan = ψ (2.2.38)
and the design ratio τ is the ratio of the radius of the spiral arm after one turn.


¹

\

−
−
= =
ψ
π
π
τ
tan
2
2
e e
a
(2.2.39)
The final type of spiral, which will be discussed throughout the remainder of this
thesis, is the Archimedean spiral. The following figure depicts the geometry of the planar
form.
18
Figure 2.2.35 Geometry of the Planar Archimedean Spiral Antenna (From Johnson [5])
As defined by Johnson [5], the arms of the Archimedean spiral can be described by the
centerline curve
φ a r = (2.2.310)
where either side of the strip is defined as

¹

\

± =
2
δ
φ a r (2.2.311)
and the second conductor can be described by rotating the first arm by π.
The width of the strip is defined as
δ a W = (2.2.312)
and the spacing between the centerlines of each strip is
a S π 2 = . (2.2.313)
19
It should be noted that the selfcomplementary case occurs when W/S is equal to 0.25.
Because the Archimedean spiral is only defined by a constant instead of an angle, it does
not truly conform to the frequencyindependent principle defined earlier. This should not
diminish the fact that the Archimedean spiral performs extremely well compared to other
spiral designs and dominates in terms of the level of research that has been devoted to its
design.
2.3 Modal Behavior of the Spiral Antenna
The property of multiarm spiral antennas that is appealing to the motivation of
this thesis is the ability to provide a directionfinding capability to the antenna system.
This directionfinding capability is made possible by feeding the spiral antenna in
different configurations or modes. Unlike the biconical or logperiodic antenna, the multi
arm spiral allows other feed configurations to be used rather than the balanced line feed
of the twoarm antenna. To understand the principles of feed configurations, we begin by
discussing the general theory of modeforming and the properties of the twoarm antenna
feed. A twoarm antenna radiates by feeding each arm with an excitation that is 180
degrees out of phase with one another to create what is called a balanced feed. A
balanced feed is defined as an antenna feed structure where the vector sum of all of the
arm excitations is equal to zero. It should be noted that a balanced feed is different from a
balanced line, where a balanced line is defined as a transmission line that has equal and
opposite currents traveling along the two conductors. Extending this principle to an
antenna with four arms as described by Corzine and Mosko [7], a balanced feed would
20
consist of a set of feed excitations whose relative phases would be 0, 90, 180, and 270
degrees with respect to the first terminal.
As described by the two and fourarm antenna cases, the total phase progression
around the feed points of the antenna is 360 degrees. To understand the mode of an
antenna, it is possible to allow the total phase progression around the feed points of an
antenna be any integer multiple of 360 degrees, where the integer multiplier is known as
the mode number. An antenna that has a phase progression of 360 degrees or 2π radians
is said to be operating in mode one, where a phase progression of 720 degrees or 4π
radians is operating in mode two. The number of modes an antenna is capable of
supporting is
( ) 1 − =
Arms Modes
N N (2.31)
where N
Arms
is the number of antenna arms. The following equation describes the feed
excitation phase progression for any mode.
( )
( )
M
N
N
N P
Arms
o


¹

\
 −
=
1
360 (2.32)
N is an integer between 1 and N
Arms
, P is the phase of the N
th
arm in degrees, and M is the
mode number. The following figure depicts what the feed excitation phase progressions
are for each mode of a spiral antenna that has up to six arms. It should be noted that the
following phase excitations are not exclusively for a spiral antenna; any antenna
geometry with multiple arms will suffice.
21
Figure 2.31 Feed Excitation Phase Progression for each Mode of a Multiarm Antenna (From Corzine and
Mosko [7])
The fundamental property of directionfinding is the ability to exploit the
distinctive field patterns and phasing of the different modes. The following figure depicts
a vertical cut of the normalized field pattern of a generic fourarm spiral antenna.
22
Figure 2.32 Normalized Field Patterns for Different Modes of a FourArm Spiral Antenna (From Penno
and Pasala [8])
The mode one field pattern is always a single lobe oriented along the boresight of the
antenna and has a typical beamwidth of 70 degrees or more. Mode two causes a null
along the boresight of the antenna that creates a toroidlike field pattern that has a
maximum gain at approximately 38 degrees off of boresight. Mode three creates a similar
pattern to mode two, but the maximum gain occurs approximately 45 degrees off of the
boresight. As described by Penno and Pasala [8], the field pattern of each mode can be
reproduced by replacing the spiral with an infinite set of current loops that range in
circumference from the feed point loop to the outer loop of the spiral. Each loop of
current has a circumference of one wavelength for mode one, two wavelengths for mode
two, and so on. The phase progression on each loop goes from 0 to 360 degrees as it
traverses around the loop for mode one, 0 to 720 degrees for mode two and so on. This
principle can be extended to understand that the minimum frequency that a mode two
23
field pattern can exist is at least twice the frequency of the lowest mode one field pattern.
Therefore, for a fourarm spiral antenna, the lowest frequency that a mode three field
pattern can exist is at least three times the frequency of the lowest mode one field pattern.
To understand the directionfinding properties of the modeformer, we start by
defining the requirements of directionfinding. At a minimum, an elevation and azimuth
angle measured from the boresight of the antenna is needed to define the relative
orientation of a received signal to the antenna. The elevation angle of the detected signal
with respect to the boresight of the antenna is found by comparing the amplitudes of the
mode one and mode two output signals, where every elevation angle provides a unique
set of output amplitudes for each of the modes. Examining figure 2.32, it is apparent that
at elevation angles near boresight, the mode two field pattern strength is considerably
smaller than the mode one field pattern strength. The azimuth angle of the detected signal
with respect to the boresight of the antenna is found by comparing the phase of the mode
one and mode two output signals. The following figure illustrates how the azimuth angle
is determined by subtracting the mode one and mode two output phases. It should be
noted that the antenna surface is in the xy plane and the mode one ring has a
circumference of one wavelength and the mode two ring has a circumference of two
wavelengths.
24
Figure 2.33 Azimuth Determination from Modal Phase Difference
It can be concluded from the figure that the mode two output phase is double that of the
mode one phase. This difference between the two output phases is unique at every
azimuth point. The following figure depicts the modal phase difference as a function of
azimuth angle.
25
Figure 2.33 Azimuth Determination by Modal Phase Difference (From Corzine and Mosko [7])
26
3. ANTENNA SYSTEM DEVELOPMENT
3.1 Spiral Antenna Development
The following sections provide the details of the design and fabrication of the
square and Archimedean fourarm spiral antennas developed for this thesis. As noted
previously, the focus of this thesis is the Archimedean spiral, but a square spiral was also
developed to understand the effects of different spiral designs on the performance of the
antenna system.
3.1.1 Spiral Antenna Design
Defining the upper and lower frequencies of operation for the spiral antenna
provides a general starting point for calculating the overall dimensions. To achieve the
best results, the antennas were etched on the thinnest microwave substrate available.
Microwave substrate is similar to printed circuit board material, but is designed for radio
and microwave frequency applications where the relative permittivity and thickness of
the dielectric material sandwiched between the two layers of copper has to be known to
create circuits that operate properly at these higher frequencies. Due to the dielectric core,
the thinner the substrate, the less the pattern would be affected due to the rise of the
effective permittivity caused by adding a dielectric to the free space surrounding the
27
antenna. The limitations on antenna size were set by the thinnest substrate available,
which was 0.010 inches thick and approximately 7.500 inches wide. To provide enough
room around the edges of the antenna to prevent any coupling between the antenna and
the enclosure, the maximum size of the antenna to be etched on the microwave substrate
was limited to just under six inches. The following equation provides a relation between
the outer diameter of the spiral and the lowest operating frequency
Outer
Lower
D
c
f
π
= (3.1.11)
where D
Outer
is the outer diameter of the spiral in meters, c is the speed of light within a
vacuum in meters per second, and f
Lower
is the lower frequency in Hertz. The upper
frequency of the antenna is generally limited by the diameter of the coaxial cables used to
feed the antenna. If a bundle of four coaxial lines are used to feed a fourarm antenna, the
distance between the center conductors of each coaxial line is limited by the outer
diameter of the coaxial line. The maximum operating frequency of a fourarm spiral
antenna feed by a bundle of coaxial lines is limited by the following equation
Inner
Upper
D
c
f
2 π
= (3.1.12)
where D
Inner
is the diameter of the coaxial cable in meters and f
Upper
is the upper
frequency in Hertz. A general design rule for determining the distance between feed
points at the center of the spiral dictates the gap between arm surfaces should be equal to
the arm width.
To determine the number of rotations a spiral arm will make as each arm is drawn
out, the spiral strip width and arm spacing must be calculated beforehand. For self
complementary antennas, the strip width is one half of the arm spacing. This creates a
28
characteristic input impedance of 95 ohms for mode one of a fourarm antenna. To
reduce the magnitude of the impedance mismatch of a 50 ohm transmission line
terminating into a 95 ohm antenna, the strip width can be increased to reduce the
impedance of the antenna. As described by Caswell [9], the effect of increasing the strip
width does come with the penalty of causing the completely realvalued impedance to
acquire a negative reactive component where narrowing the strip width causes an
increase in the real part of the impedance and a slight variation to the reactive component.
Even though widening the strip width causes the impedance to change, the overall
magnitude is still less than the selfcomplementary case. The previous design equations
for the Archimedean spiral hold for the square spiral; the only difference between the two
designs is the square spiral does not have an analytic function that defines the arc of each
arm. Each arm of the square spiral has to be drawn out by hand or by drafting software
suxh as AutoCAD to create. The following parameters were used to design the square
and Archimedean spiral antennas.
Spiral Antenna Type Archimedean Square
Outer Diameter (in) 5.684 5.684
Lower Frequency (GHz) 0.661 0.661
Inner Diameter (in) 0.282 0.282
Upper Frequency (GHz) 13.323 13.323
Strip Width (in) 0.141 0.141
Arm Spacing (in) 0.203 0.203
Number of Rotations 3.25 3.375
Table 3.1.11 Design Parameters for Archimedean and Square FourArm Spiral Antennas
3.1.2 Spiral Antenna Fabrication
The fabrication of the antenna structures started with creating AutoCAD drawings
of the square and Archimedean spirals defined by table (3.1.11). After the drawings were
29
created, each spiral was printed out in actual size on clear 8½ by 11 inch sheets of plastic
designed to be used with the Dalpro Benchtop Dry Film Etching System. The spirals
were etched on 0.010 inch thick microwave substrate with a ε
r
= 3.2. The opposite
copper layer was also etched off to produce a printedwire antenna without a ground
plane on the back of the dielectric. After the spirals were etched, they were immersed in a
tinning solution to protect the copper from corrosion.
The next step in fabricating the antenna was to create two enclosures to shield the
rearward radiation and provide support to the antennas. The two enclosures were
constructed with twenty gauge aluminum sheet to form a cavity that was 7.25 inches
square and 5 inches deep. One enclosure was fabricated with a flat face and the other
with a cylindricallyconformed face that had a 7.874 inch (20 cm) radius of curvature.
The top edge of each enclosure was formed with a lip to provide enough room to allow
holes to be drilled and tapped for number eight nylon screws. These nylon screws were
used to attach the antenna to the front of the enclosure. To provide attenuation to the
rearward traveling radiation from the antenna, which reduces interference between the
waves reflecting from the back wall of the enclosure, Emerson & Cuming AN77
microwave absorber was used to line the back wall of the cavity. The AN77 microwave
absorber provides reflectance attenuation of 20 dB or more at frequencies above 1.2 GHz.
The following figure shows a finished flatfaced enclosure with absorber lining.
30
Figure 3.1.21 Antenna Enclosure with Flat Face
The feed structure for the antenna was constructed with a bundle of four 0.141
inch copper coaxial lines with bulkhead SMA connectors on one end of each of the
cables to allow the coaxial lines to be attached to the enclosure and provide a connection
point for the antenna through the enclosure. The four coaxial lines were soldered together
by two pieces of 0.050 inch thick microwave substrate with holes drilled in each to align
the coaxial lines for each antenna. This substrate acted as a support structure for the
bundle and also provided a grounding mechanism for the coaxial bundle as they extended
away from the enclosure back wall. The coaxial lines were cut to length to align with the
top of the enclosure and connect to the etched antenna. The outer conductor of each
coaxial line was removed from the last 0.070 inches to provide a better impedance match
between the 50 ohm coaxial lines and the antenna surface. The following figure shows
the finished feed structure attached to the Archimedean spiral surface.
31
Figure 3.1.22 FourArm Coaxial Feed Bundle Attached to Antenna Surface
The following figure shows a finished Archimedean spiral mounted in an enclosure with
a cylindricallyconformed face. The entire assembly is being held by a frame used to
attach the antenna to the measurement system that will be described in a measured data
chapter.
32
Figure 3.1.23 Archimedean Spiral Mounted in Antenna Enclosure with CylindricallyConformed Face
The final step in fabricating the antenna was constructing two ground plane
apertures: one flat and one cylindricallyconformed with the purpose of holding the
antenna enclosures during testing. The ground plane structures were constructed using
0.500 inch by 0.500 inch thick pine studs for the frame and 20 gauge aluminum sheet for
the ground plane. To allow enough room for the cabling on the back of each antenna,
each ground plane structure was 12 inches tall. The flat ground plane structure was 24
inches long by 24 inches wide and the curved structure was 24 inches long by 16 inches
wide. The shorter width on the curved structure is due to the ground plane wrapping
around the curvature of the structure. The following figures show the finished flat and
cylindricallyconformed ground planes apertures with the antenna enclosures mounted
within them.
33
Figure 3.1.24 Flat Ground Plane Aperture Mounted with Empty Antenna Enclosure
Figure 3.1.25 CylindricallyConformed Ground Plane Aperture Mounted with Empty Antenna Enclosure
34
Figure 3.1.26 Flat Ground Plane Aperture Mounted with the Square Spiral
Figure 3.1.27 CylindricallyConformed Ground Plane Aperture Mounted with the Archimedean Spiral
35
The following figure is a closer view of the square spiral mounted within the flat ground
plane aperture to show the copper tape used to provide an electrical bond between the
ground plane and the antenna enclosure.
Figure 3.1.28 Closer View of Square Spiral Mounted in Flat Ground Plane Aperture
3.2 ModeFormer Development
The following sections provide the details of the design and fabrication of the
modeformers developed for this thesis.
3.2.1 ModeFormer Design
The purpose of a modeformer for a fourarm antenna is to produce a set of four
feed excitations of equal amplitude and phase progressions of 360 degrees for mode one,
and 720 degrees for mode two. The design of the modeformer began with producing
36
four equalamplitude excitations. A Wilkinson power divider was realized in microstrip
lines with a corporate feed arrangement to provide the four excitations. The next step was
to produce the phase progression needed for each mode. For simplicity and due to the fact
that the motivation of this thesis was to produce an antenna and not the modeformer, a
fixed line length design was utilized to provide an inexpensive, but narrowbanded
solution to phasing the excitations. The phasing of the feed arms was calculated to
provide proper phase progression at a frequency of 1.75 GHz. The wavelength can be
used to determine the phase length by noting that a quarter wavelength is equal to 90
degrees, a half wavelength is equal to 180 degrees and so on. After the phasing of each
arm was calculated, the additional line lengths were calculated for microstrip
transmission lines. It should be noted that an ideal modeformer provides a phase
progression that is constant across all frequencies. This modeformer design only
provides the proper phasing at one frequency, where the phase error is proportional to the
difference between the measurement frequency and the design frequency. The frequency
was chosen due to the proximity of the PCS cellular frequency band as a possible
detection signal for directionfinding.
3.2.2 ModeFormer Fabrication
The fabrication of the two modeformers started with creating AutoCAD
drawings of each. As before, the drawings were printed out in actual size on clear plastic
for use with the etching system. The microwave substrate used for the modeformers was
a 0.050 inch thick RO3010 series highfrequency laminate from Rogers Corporation.
After the modeformers were etched, SMA endlaunchers were attached to both sides of
37
the board to provide connection points between the measurement system and the mode
former and between the modeformer and the antenna enclosure. The following figures
show the finished modeformers.
Figure 3.2.21 Mode 1 ModeFormer
Figure 3.2.22 Mode 2 ModeFormer
38
As the figures show, the left side of each modeformer provides a connection point to the
measurement system. This connection point is split into four arms where additional
microstrip line has been added to each arm to provide the proper phase progression at
1.75 GHz. The four arms are numbered in ascending order starting from the top of each
picture. The figures also depict the handtuning of the connection points on the right side
of the modeformers to ensure that phase angle of each arm was within +/0.5 degrees of
the design and to provide compensation for the four (not phase matched) cables that were
used to connect the modeformer to the antenna. The measured results chapter provides
the phase error versus frequency for both of the modeformers.
39
4. THEORETICAL MODEL
4.1 Maxwell’s Equations
As a starting point to understanding the theoretical model, the magnetic vector
potential will be derived from Maxwell’s equations to define one of the possible solution
methods of Maxwell’s equations. We begin with Maxwell’s equations in the time
harmonic form within a homogeneous medium as defined by Balanis [10]
B j E
v v
ω − = × ∇ (4.11)
D j J H
v v v
ω + = × ∇ (4.12)
ρ = ⋅ ∇ D
v
(4.13)
0 = ⋅ ∇ B
v
(4.14)
with the following constitutive parameters
E D
v v
ε = (4.15)
H B
v v
u = . (4.16)
Based on the vector identity
( ) 0 = Ω × ∇ ⋅ ∇
v
(4.17)
where Ω
v
is an arbitrary vector, the magnetic field strength is defined as
A H
v v
× ∇ =
u
1
. (4.18)
40
Substituting (4.16) and (4.18) into (4.11) results in
( ) A j E
v v
× ∇ − = × ∇ ω (4.19)
where bringing both terms to one side results in
( ) 0 = + × ∇ A j E
v v
ω . (4.110)
Using the vector identity
( ) 0 = ∇ − × ∇ g (4.111)
where g is a scalar function, (4.110) becomes
φ ω ∇ − − = A j E
v v
(4.112)
where φ is the electric scalar potential. Inserting (4.18) and (4.112) into (4.12) results
in
( ) ( ) φ ω ωuε u ∇ − − + = × ∇ × ∇ A j j J A
v v v
(4.113)
and using the vector identity
( ) ( ) Ω ∇ − Ω ⋅ ∇ ∇ = Ω × ∇ × ∇
v v v
2
(4.114)
reduces (4.113) to
( ) ωεuφ u εu ω j A J A A + ⋅ ∇ ∇ + − = + ∇
v v v v
2 2
. (4.115)
Using the Lorentz Gauge
ωεuφ j A − = ⋅ ∇
v
(4.116)
now φ can be written as
A
j
v
⋅ ∇ − =
ωεu
φ
1
(4.117)
and (4.115) reduces to the inhomogeneous vector Helmholtz Equation for the magnetic
vector potential
41
J A A
v v v
u εu ω − = + ∇
2 2
(4.118)
where E
v
can now be written as
( ) A j A j E
v v v
⋅ ∇ ∇ − − =
ωεu
ω
1
. (4.119)
4.2 Numerical Electromagnetics Code Discussion
The theoretical model utilized within this thesis is the Numerical
Electromagnetics Code (NEC) package developed by the Lawrence Livermore
Laboratory. NEC utilizes the Method of Moments (MoM) solution method to solve for
the current distribution on a wire antenna structure that has been segmented. The
calculated current distribution can then be used to find the field pattern, impedance, and
polarization of the antenna. The following sections provide the general theory necessary
to understand the operation of NEC.
4.2.1 Integral Equations
The following section describes the formulation of the electric field integral
equation from the inhomogeneous vector Helmholtz Equation that NEC uses to solve for
the currents on each wire segment that comprises a wire antenna structure. We begin by
simplifying the problem by assuming a source with a current density J
z
is located at the
origin of the coordinate system. Due to the infinitesimally small size of the source, it can
be thought of as a delta function source as defined by Balanis [10], and Stutzman and
Thiele [4]. We can convert (4.118) to
42
z z z
J A A u β = + ∇
2 2
(4.2.11)
where the substitution for β is
2 2
β εu ω = . (4.2.12)
If the source is assumed to be a delta function, any point other than the origin of the
coordinate system will be equal to zero. Assuming this delta function source allows the
wave equation to be solved as a homogeneous equation for any point not at the origin.
0
2 2
= + ∇
z z
A A β (4.2.13)
This is known as the scalar Helmholtz’s equation and has a solution in spherical
coordinates of the form
r
e
A
r j
z
π
β
4
−
= (4.2.14)
where r is the distance from the source point (at the origin) to the observation point. If we
allow the source point to be at an arbitrary position other than the origin, the distance
parameter R is defined as the distance between the source point and the observation
points.
R
e
A
R j
z
π
β
4
−
= (4.2.15)
As defined by the Green’s function method, if we assume a collection of source points,
then the total contribution is equal to the integral of the source point responses due to
(4.2.15) and weighted by the total distribution of the current density over a volume V.
∫∫∫
−
=
V
R j
z z
dV
R
e
J A
π
u
β
4
(4.2.16)
If each of the other components are considered, the general solution for A
v
is defined as
43
( ) ( ) dV
R
e
r J r A
V
R j
∫∫∫
−
′ =
π
u
β
4
v
v
v
v
(4.2.17)
where
r
v
is the vector from the origin to the observation point
r
v
′ is the vector from the origin to the source point
r r R
v v
′ − = is the distance from the source point to the observation point.
At this point, the free space Green’s Function can be defined as
( )
R
e
r r G
R j
π
β
4
,
−
= ′
v v
(4.2.18)
reducing (4.2.17) down to
( ) ( ) ( )dV r r G r J r A
V
∫∫∫
′ ′ =
v v v
v
v
v
, u . (4.2.19)
If we assume that the current density is confined to a wire of radius a that is perfectly
electrically conducting (PEC), the current will only reside on the surface. The volume
integral can then be converted into a surface integral that operates over the surface A of
the wire.
( ) ( ) ( )dA r r G r J r A
A
s
∫∫
′ ′ =
v v v
v
v
v
, u (4.2.110)
If we define that the antenna surface is constructed with an arrangement of PEC wires,
the total tangential electric field on the surface of the wire is equal to zero. If the problem
is thought of in terms of electromagnetic scattering, the total tangential field can be
equated to the summation of scattered and incident waves at the surface of the antenna
0 = + =
S
t
I
t
T
t
E E E
v v v
(4.2.111)
( ) ( )
s
S
t s
I
t
r r E r r E
v v
v
v v
v
= − = = (4.2.112)
44
where
s
r
v
is the observation point that is now restricted to the surface of the wire. The
incident field can be thought of inducing a current density. If reciprocity holds, a negative
scattered field is then induced by the same current density. Substituting (4.2.110) into
(4.119) and utilizing the reciprocity principle of (4.2.112), the electric field can be
written as
( ) ( ) ( ) ( ) ( )
t
A
s s
A
s s s
I
t
dA r r G r J j dA r r G r J j r r E
′ ′ ⋅ ∇ ∇ + ′ ′ = =
∫∫ ∫∫
v v v
v
v v v
v
v v
v
,
1
,
ωε
ωu (4.2.113)
where the subscript t is the tangential component.
4.2.2 Thin Wire Approximation
The following sections provide a broad overview of the utilization of the electric
field integral equation tailored for use with wire antennas as described in detail by Mittra
[13]. The electric field integral equation of (4.2.113) can be simplified further by
utilizing the thinwire approximation. The first step as described by Caswell [12], is to
bring the divergence and gradient within the integral after simplification,
( ) ( ) ( ) ( ) ( ) ( )
t
A
s s
A
s s s
I
t
dA r r G r J j dA r r G r J j r r E
′ ∇ ′ ⋅ ∇′ + ′ ′ = =
∫∫ ∫∫
v v v
v
v v v
v
v v
v
,
1
,
ωε
ωu (4.2.21)
where
z
z
y
y
x
x
′
′ ∂
∂
+ ′
′ ∂
∂
+ ′
′ ∂
∂
= ∇′ ˆ ˆ ˆ
and ∇′ operates only on the source coordinates. With the integral simplified, the thinwire
approximation can be applied. The thinwire approximation is defined by the following
assumptions:
45
The current only flows in the direction of the wire axis
The current can be represented by a filament on the wire axis
The boundary condition on the electric field needs to be enforced in the axial
direction only
The following figure illustrates the geometry of the thin wire approximation.
Figure 4.2.21 Equivalent Source Conversion (from Stutzman and Thiele [4])
The thinwire approximation allows the current density on the surface ,
s
J
v
to be
represented as an infinitely narrow current filament of length L that has been divided by
the circumference of a wire. This conversion is allowed due to the fact that the current
remains azimuthally invariant around the circumference of the wire and is directed along
the axis of the wire. As the figure shows, the observation points can either be on the axis,
like (b) and (c), or on the radius of the wire.
( ) ( ) ( )   r s s I
a
r
s
J ′ ′ ′ = ′
v v
v
ˆ
2
1
π
(4.2.22)
46
The following figure describes the local axes for the source and observation points.
Figure 4.2.22 Local Axes Description
Taking the tangential component of the electric field can now be defined as
( ) ( ) ( ) ( ) ( ) ( ) ( )
′ ∇ ′ ⋅ ∇′ + ′ ′ ⋅ = ⋅
∫∫ ∫∫
dA r r G r J j dA r r G r J j r s E r s
A
s
A
s
I
v v v
v
v v v
v
v
v
v
,
1
, ˆ ˆ
ωε
ωu . (4.2.23)
If the surface integral is separated into its components of one integral operating around
the circumference of the wire and the other integral operating along the length of the wire,
we can substitute the thinwire approximation into the integrals resulting in
47
( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) s d ad r r G
a s s
s I j
s d ad r r G
a
r s r s s I j E r s
L
L
I
′
′ ′
′ ∂ ∂
∂
′ +
′
′ ′ ′ ′ ⋅ ′ = ⋅
∫ ∫
∫ ∫
π
π
ψ
π ωε
ψ
π
ωu
2
0
2
2
0
,
2
1 1
,
2
1
ˆ ˆ ˆ
v v
v v v v
v
v
(4.2.24)
where
( ) ( ) ( ) r r G
s
r r G r s ′
∂
∂
= ′ ∇ ⋅
v v v v v
, , ˆ (4.2.25)
( )
s
r s
′ ∂
∂
′ ′ = ∇′
v
ˆ . (4.2.26)
If we evaluate the circumferential integral, all of the constants cancel to reveal the
remaining Green’s function that is now confined to the wire segment instead of free space.
The new Green’s Function, which operates on the loop of radius a around the equivalent
thinwire current element
( ) ( ) r r G r r G ′ = ′
v v v v
, ,
ψ
(4.2.27)
where R is now restricted to the wire segment
( )
2 2
a s s R + ′ − =
then (4.2.24) reduces to
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) s d r r G
s s
s I j s d r r G r s r s s I j E r s
L L
I
′ ′
′ ∂ ∂
∂
′ + ′ ′ ′ ′ ⋅ ′ = ⋅
∫ ∫
v v v v v v
v
v
,
1
, ˆ ˆ ˆ
2
ψ ψ
ωε
ωu (4.2.28)
which defines the electric field integral equation used by NEC as described by Burke and
Poggio [11].
48
4.2.3 Method of Moments Solution Method
The following section defines the general theory of the Method of Moments for
solving operator equations. We begin by defining the linear operator equation
e Lf = (4.2.31)
where L is a linear operator (which in the case of the electric field integral equation is an
integral operator), f is the unknown response, and e is a known excitation. The unknown
response can be expanded into a sum of constant coefficients multiplied by a basis
function f
n
. The choice of basis function depends on the problem being solved. When the
form of the solution is unknown, the piecewise constant (pulse) is the most
straightforward. If the form of the solution is known beforehand, the basis function is
chosen to match the response of the solution such as the triangular and sinusoidal basis
functions. These particular functions are more capable of handling solutions that have a
traveling wave form, such as antennas.
∑
=
≈
N
n
n n
f a f
1
(4.2.32)
A set of equations for the unknown coefficient a
n
is created by taking the inner product of
(4.2.31) with a set of weighting functions
e w Lf w
m m
, , = ; N m ,..., 1 = . (4.2.33)
where the inner product, which for this geometry, is not weighted, is defined as
( ) ( )dA r v r u v u
A
∫∫
′ =
v v
, . (4.2.34)
Combining (4.2.33) and (4.2.32) produces
e w Lf w a
m n m
N
n
n
, ,
1
=
∑
=
; N m ,..., 1 = (4.2.35)
49
which can be written in matrix form to produce Ohm’s Law
     V Z A = (4.2.36)
where A is a vector containing the unknown coefficients. Z is a matrix of the summed
inner products of the integral operator and basis function with the weighting function and
V is a vector of the inner products between the weighting function and the excitation. To
calculate the values A, the Z matrix is inverted
      V Z A
1 −
= . (4.2.37)
The weighting function, like the basis function, can be chosen for the problem type. The
weighting and basis functions can be different depending on the problem geometry. The
other possibility is if the weighting and basis functions are the same, where this is defined
as Galerkin’s Method.
The following equations describe how NEC utilizes the Method of Moments to
solve for the current of (4.2.28). Starting with an expansion for the current on each of the
n segments of wire that defines the antenna surface
( ) ( )
∑
=
′ ≈ ′
N
n
n n
s f a s I
1
(4.2.38)
and substituting into (4.2.28) results in
( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) s d r r G
s s
s f a j
s d r r G r s r s s f a j E r s
L
N
n
n n
L
N
n
n n
I
′ ′
′ ∂ ∂
∂
′ +
′ ′ ′ ′ ⋅ ′ = ⋅
∫
∑
∫
∑
=
=
v v
v v v v
v
v
,
1
, ˆ ˆ ˆ
2
1
1
ψ
ψ
ωε
ωu
. (4.2.39)
The next step is to employ the weighting and basis functions. NEC uses
( )
m m
s s w − = δ (4.2.310)
( ) ( ) ( ) ( ) ( )
n n n n n n
s s C s s B A s f − ′ + − ′ + = ′ β β cos sin (4.2.311)
50
where the weighting function effectively samples the integral at the midpoint of each
source location s
m
of segment length L
m
and the basis function operates over the midpoint
of each segment s
n
. Substituting (4.2.310) into (4.2.39) results in the following
equations that constitute the matrix elements and vectors of (4.2.37).
( ) ( ) ( )ds s s E r s V
m
L
I
m
m
− ⋅ =
∫
δ
v
v
ˆ (4.2.312)
( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ( ) ds s d r r G
s s
s f a s s j
ds s d r r G r s r s s f a s s j Z
L
N
n
n n
L
m
L
N
n
n n
L
m mn
m
m
′ ′
′ ∂ ∂
∂
′ − +
′ ′ ′ ′ ⋅ ′ − =
∫
∑
∫
∫
∑
∫
=
=
v v
v v v v
,
1
, ˆ ˆ
2
1
1
ψ
ψ
δ
ωε
δ ωu
(4.2.313)
Utilizing the definition of the integrated delta function, (4.2.312) and (4.2.313) reduce
to
( ) ( ) ( )
m
I
m m
s E s r s V
v
v
⋅ = ˆ (4.2.314)
( ) ( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ( ) s d r s r G
s s
s f j
s d r s r G r s s r s s f j Z
L
m n
L
m m n mn
′ ′
′ ∂ ∂
∂
′ +
′ ′ ′ ′ ⋅ ′ =
∫
∫
v v
v v v v
,
1
, ˆ ˆ
2
ψ
ψ
ωε
ωu
(4.2.315)
where
∑
=
=
N
n
m n mn
V a Z
1
; N m ,..., 1 = . (4.2.316)
The Z matrix will be diagonally dominant due to the source segment terms being the
largest component of the observation segment terms. This dominance is due to the
distance between each source and observation segment. The diagonal terms of the Z
matrix represent situations where the source segment is at the same location as the
observation segment. As the observation segment moves away from the source segment,
its location in the Z matrix moves further away from the diagonal therefore decreasing its
51
contribution to the overall calculation of the current. The V vector will be mostly sparse
except for the segments where there is a voltage source, which in the case of a fourarm
spiral antenna, will be four locations.
After the current for each segment has been found, the field pattern can be
calculated from the following magnetic vector potential farfield relations for the electric
field.
A j E
A
v v
ω − ≈ (4.2.317)
The farfield of the electric field occurs when the distance from the current source
becomes great enough that the difference between the elevation angle of a ray extending
from the center of the antenna to the observation point and a ray extending from a point
along the antenna to the observation point approaches zero. This principle defines a
region where the angular field response is nearly independent of the distance to the
antenna. If (4.2.317) is used to find the field of each current segment, all of the
individual field contributions can be vectorsummed to produce the field response of the
entire wire antenna structure. The following figure illustrates the antenna coordinate
geometry.
52
Figure 4.2.31 Antenna Coordinate System (from Balanis [1])
After the electric field is known, the power at each angle can be defined as
( )
*
2
2
1
) , ( E E
r
P
o
o
v v
⋅ =
u
ε
φ θ (4.2.318)
and the power gain as
( )
=
in
P
P
G
) , (
4 ,
φ θ
π φ θ (4.2.319)
where the P
in
is calculated from the current and voltage at the source
( )
*
Re VI P
in
= . (4.2.320)
Now that the theory of operation for NEC has been explained, understanding how
to create wire antenna models for NEC can be discussed. Due to the fact that NEC is a
wire antenna simulation, creating the input files for NEC starts with converting the flat
53
strips of the etched antenna into a wire with a specific radius. As shown in Caswell [9],
the widely accepted transformation between flat strips and wires is
s w
W r
4
1
= (4.31)
where r
w
is the wire radius and W
s
is the strip width. To create the wire model for each
antenna design, MATLAB was used to calculate the length and location of each wire
segment. The calculated wire segments were used to create an input file that was properly
formatted to be read by NEC. As a general design rule, the length of each segment should
be at least six times the radius of the wire to conform to the thinwire approximation.
Within NEC, each wire segment has an associated wire radius and a set of Cartesian
coordinates (x,y,z) that describe the starting and ending points of the wire. This
information is used by NEC to calculate the relative orientation of each wire segment and
to determine where the observation point due to wire radius is located on each wire
segment. It should be noted that the wire segments must be orientated so that the starting
point of the first segment is at the feed point of each antenna arm and the end point is
directed away from the origin. Each successive segment should use the end point of the
previous segment as the starting point, and the end point should follow the spiral arm as it
progresses away from the origin.
The final step to creating an NEC input file is to designate the feed excitations
and their relative phases. NEC allows the user to designate any single wire segment to
provide a location for the deltagap voltage source to exist on. The length of the wire
segment provides the value of delta needed to produce an electric field from a voltage
potential divided by a distance. Along with the ability to designate which segments
54
constitute the feed structure of the antenna, the phase angle and amplitude of each voltage
source can be modified to allow for different modes of the spiral antenna to be simulated.
4.3 Numerical Electromagnetics Code Results
The following sections provide the results obtained from the NEC simulations
that were performed using the 4NEC2X version of NEC. As a clarification, only the
vertically polarized power patterns will be shown, where the spiral antenna is generally
circularly polarized. This is to avoid confusion with the measured results chapter where
the measurement system was limited to vertically polarized results due to the vertically
polarized receive antenna.
The NEC results will be displayed in spherical coordinates (ρ.θ.φ) where the
angle θ is measured from the zaxis. A θ angle of zero degrees equates to the boresight of
the antenna and a positive angle corresponds to a counterclockwise rotation looking
from an observation point along the positive yaxis. The following figure illustrates the
local geometry used to display the power pattern plots of each antenna design.
55
Figure 4.31 Antenna Plotting Geometry Description
4.3.1 Archimedean Spiral
The first spiral to examine is the flat Archimedean design. The following figure
depicts the segmented spiral used by NEC.
56
Figure 4.3.11 NEC Geometry of Archimedean Spiral
The voltage sources of the spiral are designated by the four pink bands near the origin.
Each voltage source is located on the second segment in from the beginning of each arm.
The following figure depicts the mode one maximum vertically polarized gain of the
spiral antenna versus frequency.
57
Figure 4.3.12 Mode 1 Archimedean Spiral Maximum Vertically Polarized Gain versus Frequency
As the figure shows, the maximum gain of the antenna reaches a nearly steady value after
3 GHz. The following figures depict the power patterns of the spiral for mode one and
mode two. It should be noted that the power patterns have been plotted in a logarithmic
scale where the unit of dBi is the relative gain of the antenna compared to an isotropic
radiator.
58
Figure 4.3.13 Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 4.3.14 Mode 1 Archimedean Spiral Power Pattern (dBi) θ Cuts (1 GHz)
59
Figure 4.3.15 Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.16 Mode 1 Archimedean Spiral Power Pattern (dBi) θ Cuts (1.75 GHz)
60
Figure 4.3.17 Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 4.3.18 Mode 1 Archimedean Spiral Power Pattern (dBi) θ Cuts (2.5 GHz)
61
Figure 4.3.19 Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.110 Mode 1 Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz)
62
Figure 4.3.111 Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 4.3.112 Mode 2 Archimedean Spiral Power Pattern (dBi) θ Cuts (1 GHz)
63
Figure 4.3.113 Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.114 Mode 2 Archimedean Spiral Power Pattern (dBi) θ Cuts (1.75 GHz)
64
Figure 4.3.115 Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 4.3.116 Mode 2 Archimedean Spiral Power Pattern (dBi) θ Cuts (2.5 GHz)
65
Figure 4.3.117 Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.118 Mode 2 Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz)
66
The previous figures illustrate that the mode one beamwidth decreases as the frequency
increases. There is also a slight “squaring” of the mode one θ cuts which is most likely
due to the antenna becoming electrically large enough to support the beginning of a mode
three power pattern. The figures also depict that the power pattern for mode two does not
fully develop until after 2.5 GHz where the θ cuts become nearly concentric. At 5 GHz,
the mode two power pattern is nearly perfect, agreeing with the work of Penno and Pasala.
As a final point, the effects of scaling that were discussed in chapter two can now by seen
as the rotation of the power pattern around the zaxis as the frequency is increased.
4.3.2 CylindricallyConformed Archimedean Spiral
The following section characterizes the results of the cylindricallyconformed
Archimedean spiral. The following figure depicts the segmented spiral used by NEC.
Figure 4.3.21 NEC Geometry of CylindricallyConformed Archimedean Spiral
67
The following figure depicts the mode one maximum vertically polarized gain of the
antenna versus frequency.
Figure 4.3.22 Mode 1 CylindricallyConformed Archimedean Spiral Maximum Vertically Polarized Gain
versus Frequency
As with the flat case, the maximum gain versus frequency approaches a nearly steady
solution after 3 GHz. The following figures depict the power patterns of the spiral for
mode one and mode two.
68
Figure 4.3.23 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 4.3.24 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1 GHz)
69
Figure 4.3.25 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75
GHz)
Figure 4.3.26 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1.75
GHz)
70
Figure 4.3.27 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 4.3.28 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (2.5 GHz)
71
Figure 4.3.29 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.210 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz)
72
Figure 4.3.211 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 4.3.212 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1 GHz)
73
Figure 4.3.213 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75
GHz)
Figure 4.3.214 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1.75
GHz)
74
Figure 4.3.215 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5
GHz)
Figure 4.3.216 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (2.5
GHz)
75
Figure 4.3.217 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.218 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz)
76
As with the flat Archimedean spiral, the previous figures illustrate that the mode one
beamwidth decreases as the frequency increases. The difference between the flat and
conformed case appears to be the warping of the θ cuts to produce an ellipse instead of a
circle. Also, the circularly polarized antenna in the flat case has changed to an elliptically
polarized antenna seen by the φ cuts not connecting at boresight. Mode two appears to
also suffer from conforming the antenna where the power pattern has warped slightly.
4.3.3 Square Spiral
The following section characterizes the results of the square spiral. The following
figure depicts the segmented spiral used by NEC.
Figure 4.3.31 NEC Geometry of Square Spiral
The following figure depicts the mode one maximum vertically polarized gain of the
antenna versus frequency.
77
Figure 4.3.32 Mode 1 Square Spiral Maximum Vertically Polarized Gain versus Frequency
Differing from the Archimedean spiral, the maximum gain versus frequency begins to
oscillate after 3 GHz. The following figures depict the power patterns of the spiral for
mode one and mode two.
78
Figure 4.3.33 Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 4.3.34 Mode 1 Square Spiral Power Pattern (dBi) θ Cuts (1 GHz)
79
Figure 4.3.35 Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.36 Mode 1 Square Spiral Power Pattern (dBi) θ Cuts (1.75 GHz)
80
Figure 4.3.37 Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 4.3.38 Mode 1 Square Spiral Power Pattern (dBi) θ Cuts (2.5 GHz)
81
Figure 4.3.39 Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.310 Mode 1 Square Spiral Power Pattern (dBi) θ Cuts (5 GHz)
82
Figure 4.3.311 Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 4.3.312 Mode 2 Square Spiral Power Pattern (dBi) θ Cuts (1 GHz)
83
Figure 4.3.313 Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.314 Mode 2 Square Spiral Power Pattern (dBi) θ Cuts (1.75 GHz)
84
Figure 4.3.315 Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 4.3.316 Mode 2 Square Spiral Power Pattern (dBi) θ Cuts (2.5 GHz)
85
Figure 4.3.317 Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.318 Mode 2 Square Spiral Power Pattern (dBi) θ Cuts (5 GHz)
86
The previous figures illustrate that the mode one power pattern of the square spiral agrees
with the Archimedean spiral up to 5 GHz where the power pattern begins to look like the
head of a Phillips screwdriver. Mode two power patterns of the square spiral also agree
with the results of the Archimedean spiral where at 5 GHz, the square spiral is nearly
concentric.
4.3.4 CylindricallyConformed Square Spiral
The following section characterizes the results of the cylindricallyconformed
square spiral. The following figure depicts the segmented spiral used by NEC.
Figure 4.3.41 NEC Geometry of CylindricallyConformed Square Spiral
The following figure depicts the mode one maximum vertically polarized gain of the
antenna versus frequency.
87
Figure 4.3.42 Mode 1 CylindricallyConformed Square Spiral Maximum Vertically Polarized Gain versus
Frequency
The frequency response of the conformed square spiral agrees with the flat where the
same oscillation of gain is seen after 3 GHz. The following figures depict the power
patterns of the spiral for mode one and mode two.
88
Figure 4.3.43 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 4.3.44 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (1 GHz)
89
Figure 4.3.45 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.46 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (1.75 GHz)
90
Figure 4.3.47 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 4.3.48 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (2.5 GHz)
91
Figure 4.3.49 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.410 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (5 GHz)
92
Figure 4.3.411 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 4.3.412 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (1 GHz)
93
Figure 4.3.413 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.414 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (1.75 GHz)
94
Figure 4.3.415 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 4.3.416 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (2.5 GHz)
95
Figure 4.3.417 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.418 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (5 GHz)
96
The previous figures illustrate that the mode one power patterns of the conformed square
spiral agrees with the conformed Archimedean spiral results. In both designs, the pattern
is warped creating a slight elliptical polarization. Mode two power patterns of the
conformed square spiral also agree with the results of the conformed Archimedean spiral,
where the θ cuts of the pattern settle into a nearly concentric circle even though the
surface has been conformed.
4.3.5 Radius of Curvature Analysis of the Archimedean Spiral
The following section compares the results of conforming the Archimedean spiral
at different radiuses of curvature. The purpose of this analysis is to determine the
maximum radius of curvature that still permits the use of the antenna for direction
finding. The following figures provide the mode one and mode two power patterns of an
Archimedean spiral antenna that has been conformed at the following radiuses of
curvature: infinity (flat), 80 cm (31.5 inches), 40 cm (16 inches), 20 cm (7.9 inches), 14
cm (5.5 inches), and 6 cm (2.4 inches). These radiuses of curvature correspond to air
platforms of different sizes ranging from the 707 to a Predator down to one of the
smallest UAVs used by the Army, the Raven.
97
Figure 4.3.51 Radius of Curvature Analysis Mode 1 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.52 Radius of Curvature Analysis Mode 1 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) θ Cuts (1.75 GHz)
98
Figure 4.3.53 Radius of Curvature Analysis Mode 1 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.54 Radius of Curvature Analysis Mode 1 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) θ Cuts (5 GHz)
99
Figure 4.3.55 Radius of Curvature Analysis Mode 2 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.56 Radius of Curvature Analysis Mode 2 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) θ Cuts (1.75 GHz)
100
Figure 4.3.57 Radius of Curvature Analysis Mode 2 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.58 Radius of Curvature Analysis Mode 2 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) θ Cuts (5 GHz)
101
The previous figures illustrate the near invariance of the power pattern to the radius of
curvature. Even when the antenna is conformed at a radius of curvature of 14 cm, the
pattern shape is almost indistinguishable from the flat case. Only when the pattern is
conformed to a radius of curvature equal to 6 cm does the power pattern shape begin to
change. Even with this slight change in antenna power pattern shape, if the antenna
pattern is known, the directionfinding calculations could be compensated to make up for
these distortions.
102
5. MEASURED RESULTS
5.1 Measurement Method
The following sections describe the components of the antenna measurement
system that was used to gather the measured results of this chapter. The following figure
provides an overview of the components comprising the antenna measurement system.
Figure 5.11 Antenna Measurement System Component Layout
103
5.1.1 Vector Network Analyzer Discussion
The Vector Network Analyzer (VNA) used to provide the measured results of this
thesis was the twoport Anritsu 37347C which has an operating frequency range from
40MHz to 20 GHz. The twoport VNA provides simultaneous transmission and reflection
measurements at and between each of the ports. The transmission and reflection of the
device under test is found by measuring of the ratio of the transmitted and received
voltages at each of the ports. The transmission and reflection terms are categorized as
scattering parameters, or sparameters. Each of the four possible measurements are given
by the following sparameters:
S
11
is the ratio of the received voltage at port one to the transmitted voltage from
port one
S
22
is the ratio of the received voltage at port two to the transmitted voltage from
port two
S
21
is the ratio of the received voltage at port two to the transmitted voltage from
port one
S
12
is the ratio of the received voltage at port one to the transmitted voltage from
port two
Each of the sparameters can be used to provide information about the frequency
response of a device under test such as the insertion loss (IL), voltage standing wave ratio
(VSWR), and group delay. For the antenna measurement system, the VNA is used to
measure the transmission, or S
21
, across a range of frequencies between the two antennas.
104
5.1.2 Reference Antenna Discussion
The reference antenna used by the antenna measurement system was the Antenna
Design & Manufacturing Corporation DHR118/A 118 GHz Double Ridged Horn.
Within the antenna measurement system, the reference antenna acts as both the receive
antenna during a measurement and is also the calibrated antenna used to determine the
final gain of the antenna being measured. The following figures provide the maximum
gain versus frequency and antenna power pattern for the reference antenna.
Figure 5.1.21 Reference Antenna Maximum Vertically Polarized Gain versus Frequency
105
Figure 5.1.22 Reference Antenna Power Pattern (dBi) φ Cuts
5.1.3 Desktop Antenna Measurement System Discussion
The antenna measurement system utilized for measured results of this thesis was
the Diamond Engineering Desktop Antenna Measurement System (DAMS) that operates
from 0 to 18 GHz. The DAMS system is comprised of a software application that
controls the VNA and the rotation platform controller. Along with the software, the
system provides a tripod to hold the receive antenna and a second tripod to hold the
rotation platform. The rotation platform is designed to hold the antenna under test while it
is rotated across a set of azimuth angles and/or elevation angles.
The measurement of an antenna under test starts with mounting two identical
reference antennas to the tripods supplied by the DAMS system. The reference antennas
are then aligned so that the boresights of each antenna are pointing directly at each other.
106
After alignment, the reference antennas are then connected to the VNA where the port
two reference antenna acts as the receive antenna and the port one reference antenna acts
as the transmit antenna. Once both of the antennas have been connected to the VNA, a
frequency range is designated and the VNA measures the transmission between the two
antennas. This reference antenna transmission measurement is used to calculate the
isotropic antenna transmission between the two antennas by subtracting the known gain
with respect to frequency of one of the reference antennas from the transmission
measurement. After the reference antenna transmission has been taken, the distance
between the two antennas is recorded and the reference antenna that was mounted to the
rotation platform is removed.
The antenna under test is then mounted to the rotation platform and is situated to
be at the same distance from the receive antenna as the transmit antenna was. This allows
the free space path loss differences between the reference antenna transmission and the
antenna under test transmission to be ignored. After both antennas have been mounted
and aligned again, the DAMS software is used to control the rotation platform and VNA
to record the angular and frequency dependent transmission results of the antenna under
test. As the DAMS software moves the rotation platform around one azimuth or elevation
step at a time, the VNA measures the transmission of the two antennas across a frequency
range. Each angle is then recorded by the DAMS software and stored to memory. After
all of the angles and frequencies of interest have been measured, the data is exported into
a text file that can be read by MATLAB or any spreadsheet software. After the data has
been exported, the transmission measurements of the antenna under test is compared to
107
the reference antenna transmission results to calculate the relative gain of the antenna
under test to the known gain of the reference antenna.
5.2 ModeFormer Results
The following section provides the results of the modeformers utilized to create
the different power patterns of the antennas. The following figure depicts the
transmission loss of each arm of the mode one modeformer with respect to frequency.
Figure 5.21 Mode 1 ModeFormer Transmission Loss versus Frequency
An ideal modeformer would provide only 6 dB of loss, but due to the utilization of the
Wilkinson power divider, some of the transmission is lost due to the bridging resistors.
The following figure depicts the maximum variation in transmission between each arm of
the mode one modeformer.
108
Figure 5.22 Mode 1 ModeFormer Maximum Transmission Loss Variation versus Frequency
The figure shows that up to 2 GHz, the transmission balance between all of the arms of
the modeformer is within 0.3 dB. The following figure shows the relative phase
difference between each arm of the mode one modeformer.
109
Figure 5.23 Mode 1 ModeFormer Arm Phase versus Frequency
The following figures depict the relative phase error with respect to frequency of the
mode one modeformer compared to an ideal mode one modeformer.
Figure 5.24 Mode 1 ModeFormer Arm Phase Error versus Frequency
110
Figure 5.25 Mode 1 ModeFormer Arm Phase Error versus Frequency (Expanded View)
The following figures depict the transmission loss and maximum variation in
transmission loss between of each arm of the mode two modeformer with respect to
frequency.
111
Figure 5.26 Mode 2 ModeFormer Transmission Loss versus Frequency
Figure 5.27 Mode 2 ModeFormer Maximum Transmission Loss Variation versus Frequency
The following figure shows the relative phase difference between each arm of the mode
two modeformer.
112
Figure 5.28 Mode 2 ModeFormer Arm Phase versus Frequency
The following figures provide the relative phase error with respect to frequency of the
modeformer relative to an ideal mode one modeformer.
Figure 5.29 Mode 2 ModeFormer Arm Phase Error versus Frequency
113
Figure 5.210 Mode 2 ModeFormer Arm Phase Error versus Frequency (Expanded View)
The results show that the modeformers perform well around 1.75 GHz. The phase error
begins to degrade the quality of the modeformer after a frequency of 50 MHz on either
side of 1.75 GHz. As a lowcost solution, several of these modeformers could be
fabricated to operate at different frequencies to provide a modular solution to the
frequency coverage of spiral antennas.
5.2.1 Power Pattern Phase Balance Comparison
The following section compares the results of the NEC model that has been feed with the
actual phase errors of each modeformer to the measured results of the Archimedean
spiral antenna. Only φ cuts were measured on antennas that were not mounted within a
ground plane aperture. The following figures depict measured and simulated φ cuts of
each modeformer across a set of frequencies.
114
Figure 5.2.11 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.25
GHz) (NEC is Phase and Amplitude Compensated)
Figure 5.2.12 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.5 GHz)
(NEC is Phase and Amplitude Compensated)
115
Figure 5.2.13 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75
GHz) (NEC is Phase and Amplitude Compensated)
Figure 5.2.14 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (2 GHz)
(NEC is Phase and Amplitude Compensated)
116
Figure 5.2.15 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.25
GHz) (NEC is Phase and Amplitude Compensated)
Figure 5.2.16 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.25
GHz) (NEC is Phase and Amplitude Compensated)
117
Figure 5.2.17 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.5 GHz)
(NEC is Phase and Amplitude Compensated)
Figure 5.2.18 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75
GHz) (NEC is Phase and Amplitude Compensated)
118
Figure 5.2.19 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (2 GHz)
(NEC is Phase and Amplitude Compensated)
Figure 5.2.110 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.25
GHz) (NEC is Phase and Amplitude Compensated)
119
The previous figures illustrate the increased elliptical polarization that occurs when the
modeformer is not providing the ideal phase progression to the antenna. The mode one
simulated results loosely agree with the measured results for frequencies at and below
1.75 GHz. Mode two begins with a stronger power pattern than what NEC predicts but
settles into a closer agreement by 1.75 GHz.
5.3 Spiral Antenna Results
The following sections describe the results of each antenna that was fabricated for
this thesis. All of the measured results have been plotted with the results of the NEC
simulation (with perfect phase) to provide a comparison between the measured and
theoretical results. It should be noted that the NEC results are for an antenna in free space
as depicted in the theoretical section. The best situation for comparing the theoretical
results to the measured results would be to construct a model that includes strips printed
on a dielectric sheet instead of wires, an enclosure lined with microwave absorber instead
of free space, and the ability to create a ground plane aperture. The availability of the
electromagnetics software due to cost or licensing rules prevented these characteristics
from being modeled and therefore is not a part of the comparison. It should also be noted
that only φ cuts were measured for antennas not mounted in ground plane apertures.
120
5.3.1 Archimedean Spiral
The following section compares the measured results of the Archimedean spiral to
the NEC simulated results. The following figure depicts the Archimedean spiral mounted
on the DAMS platform.
Figure 5.3.11 Archimedean Spiral Mounted in Test Fixture
The following figure compares the mode one maximum vertically polarized gain of the
measured antenna and the NEC simulated antenna versus frequency.
121
Figure 5.3.12 Measured and Simulated Mode 1 Archimedean Spiral Maximum Vertically Polarized Gain
versus Frequency
As the figure shows, the maximum gain of the antenna oscillates about the theoretical
curve up to 2 GHz and then falls off afterward. This is most likely due to the Standing
Wave Ratio (SWR) of the input impedance oscillating around 50 ohms and the depth of
the antenna enclosure favoring a lower frequency response. The following figures depict
the power patterns of the spiral for mode one and mode two.
122
Figure 5.3.13 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 5.3.14 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75
GHz)
123
Figure 5.3.15 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 5.3.16 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz)
124
Figure 5.3.17 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75
GHz)
Figure 5.3.18 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
125
The previous figures illustrate the increased elliptical polarization at lower frequencies
for the mode one power pattern. Not surprisingly, the strongest agreement between
measured and theoretical results occurred at 1.75 GHz.
5.3.2 CylindricallyConformed Archimedean Spiral
The following section compares the measured results of the cylindrically
conformed Archimedean spiral to the NEC simulated results. The following figure
depicts the cylindricallyconformed Archimedean spiral mounted on the DAMS platform.
Figure 5.3.21 Archimedean Spiral Mounted in Test Fixture
The following figure compares the mode one maximum vertically polarized gain of the
measured antenna and the NEC simulated antenna versus frequency.
126
Figure 5.3.22 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral Maximum
Vertically Polarized Gain versus Frequency
Like the flat geometry, the maximum gain of the antenna oscillates about the theoretical
curve up to 2 GHz and then falls off afterward. The following figures depict the power
patterns of the spiral for mode one and mode two.
127
Figure 5.3.23 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (1 GHz)
Figure 5.3.24 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (1.75 GHz)
128
Figure 5.3.25 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (2.5 GHz)
Figure 5.3.26 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (1 GHz)
129
Figure 5.3.27 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (1.75 GHz)
Figure 5.3.28 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral Power
Pattern (dBi) φ Cuts (2.5 GHz)
130
The previous figures again illustrate the strong agreement between measured and
theoretical results at 1.75 GHz. It is also seen that, like the theoretical model predicted,
the power patterns between the flat and conformed cases have not been altered.
5.3.3 Square Spiral
The following section compares the measured results of the square spiral to the
NEC simulated results. The following figure depicts the square spiral mounted on the
DAMS platform.
Figure 5.3.31 Square Spiral Mounted in Test Fixture
The following figure compares the mode one maximum vertically polarized gain of the
measured antenna and the NEC simulated antenna versus frequency.
131
Figure 5.3.32 Measured and Simulated Mode 1 Square Spiral Maximum Vertically Polarized Gain versus
Frequency
Like the previous antenna geometries, the maximum gain of the antenna oscillates about
the theoretical curve up to 2 GHz and then falls off afterward. The following figures
depict the power patterns of the spiral for mode one and mode two.
132
Figure 5.3.33 Measured and Simulated Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (1 GHz)
Figure 5.3.34 Measured and Simulated Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
133
Figure 5.3.35 Measured and Simulated Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 5.3.36 Measured and Simulated Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (1 GHz)
134
Figure 5.3.37 Measured and Simulated Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 5.3.38 Measured and Simulated Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
135
The previous figures again illustrate the strong agreement between measured and
theoretical results at 1.75 GHz, where the square spiral has surprisingly provided the
most concentric mode one power pattern of all of the antenna geometries.
5.3.4 CylindricallyConformed Square Spiral
The following section compares the measured results of the cylindrically
conformed square spiral to the NEC simulated results. The following figure depicts the
cylindricallyconformed square spiral mounted on the DAMS platform.
Figure 5.3.41 CylindricallyConformed Square Spiral Mounted in Test Fixture
The following figure compares the mode one maximum vertically polarized gain of the
measured antenna and the NEC simulated antenna versus frequency.
136
Figure 5.3.42 Measured and Simulated Mode 1 CylindricallyConformed Square Spiral Maximum
Vertically Polarized Gain versus Frequency
The shape of the maximum gain curve of the antenna with respect to frequency remains
unchanged compared to the other measured antenna results which show that, for at least
lower frequencies, the square and Archimedean spiral generally perform equally well.
This also happens to be the conclusion found within the theoretical results. The following
figures depict the power patterns of the spiral for mode one and mode two.
137
Figure 5.3.43 Measured and Simulated Mode 1 CylindricallyConformed Square Spiral Power Pattern
(dBi) φ Cuts (1 GHz)
Figure 5.3.44 Measured and Simulated Mode 1 CylindricallyConformed Square Spiral Power Pattern
(dBi) φ Cuts (1.75 GHz)
138
Figure 5.3.45 Measured and Simulated Mode 1 CylindricallyConformed Square Spiral Power Pattern
(dBi) φ Cuts (2.5 GHz)
Figure 5.3.46 Measured and Simulated Mode 2 CylindricallyConformed Square Spiral Power Pattern
(dBi) φ Cuts (1 GHz)
139
Figure 5.3.47 Measured and Simulated Mode 2 CylindricallyConformed Square Spiral Power Pattern
(dBi) φ Cuts (1.75 GHz)
Figure 5.3.48 Measured and Simulated Mode 2 CylindricallyConformed Square Spiral Power Pattern
(dBi) φ Cuts (2.5 GHz)
140
The previous figures again illustrate the strong agreement between measured and
theoretical results at 1.75 GHz, although the conforming of the antenna has created a
slightly elliptical polarization on the mode one pattern and the maximum gain at
boresight is reduced.
5.3.5 Square Spiral within Ground Plane Aperture
The following section compares the measured results of the square spiral mounted
within a ground plane aperture to the NEC simulated results. The following figure depicts
the square spiral mounted within the ground plane aperture.
Figure 5.3.51 Square Spiral Mounted within Ground Plane Aperture
The following figure compares the mode one maximum vertically polarized gain of the
measured antenna and the NEC simulated antenna versus frequency.
141
Figure 5.3.52 Measured and Simulated Mode 1 Square Spiral Maximum Vertically Polarized Gain versus
Frequency
The shape of the maximum gain curve has smoothed out during the ground plane aperture
case which suggests that the enclosure is affecting the maximum gain power patterns of
the previous antenna geometries. The following figures depict the power patterns of the
spiral for mode one and mode two.
142
Figure 5.3.53 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern
(dBi) φ Cuts (1 GHz)
Figure 5.3.54 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern
(dBi) θ Cuts (1 GHz)
143
Figure 5.3.55 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern
(dBi) φ Cuts (1.75 GHz)
Figure 5.3.56 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern
(dBi) θ Cuts (1.75 GHz)
144
Figure 5.3.57 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern
(dBi) φ Cuts (2.5 GHz)
Figure 5.3.58 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern
(dBi) θ Cuts (2.5 GHz)
145
Figure 5.3.59 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power Pattern
(dBi) φ Cuts (1 GHz)
Figure 5.3.510 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power
Pattern (dBi) θ Cuts (1 GHz)
146
Figure 5.3.511 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power
Pattern (dBi) φ Cuts (1.75 GHz)
Figure 5.3.512 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power
Pattern (dBi) θ Cuts (1.75 GHz)
147
Figure 5.3.513 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power
Pattern (dBi) φ Cuts (2.5 GHz)
Figure 5.3.514 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power
Pattern (dBi) θ Cuts (2.5 GHz)
148
The previous figures illustrate that along with smoothing out the maximum gain with
respect to frequency, the ground plane aperture has smoothed out the power patterns at
each frequency and mode. The pointed tip at boresight for mode one is caused by the
antenna not being rotated when the boresight measurement was made. This forced a
single measurement to be used on the φ cuts where there should have been separate
measurements for each angle to depict the polarization difference between each cut.
5.3.6 CylindricallyConformed Archimedean Spiral within Ground
Plane Aperture
The following section compares the measured results of the cylindrically
conformed Archimedean spiral mounted within a ground plane aperture to the NEC
simulated results. The following figure depicts the cylindricallyconformed Archimedean
spiral mounted within the ground plane aperture.
149
Figure 5.3.61 CylindricallyConformed Archimedean Spiral Mounted within Ground Plane Aperture
The following figure depicts the mode one maximum vertically polarized gain of the
measured antenna and the NEC simulated antenna versus frequency.
150
Figure 5.3.62 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Maximum Vertically Polarized Gain versus Frequency
The shape of the maximum gain curve agrees with the square spiral case which further
suggests that the ground plane aperture drives the measured results. The following figures
depict the power patterns of the spiral for mode one and mode two.
151
Figure 5.3.63 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) φ Cuts (1 GHz)
Figure 5.3.64 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) θ Cuts (1 GHz)
152
Figure 5.3.65 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 5.3.66 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) θ Cuts (1.75 GHz)
153
Figure 5.3.67 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 5.3.68 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) θ Cuts (2.5 GHz)
154
Figure 5.3.69 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) φ Cuts (1 GHz)
Figure 5.3.610 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) θ Cuts (1 GHz)
155
Figure 5.3.611 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 5.3.612 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) θ Cuts (1.75 GHz)
156
Figure 5.3.613 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 5.3.614 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within
Ground Plane Aperture Power Pattern (dBi) θ Cuts (2.5 GHz)
157
The previous figures illustrate that at frequencies below 1.75 GHz, the measured antenna
power pattern trends closer to the simulated case in each geometry that was discussed
within this thesis.
158
6. CONCLUSIONS
6.1 Summary of Work
Within this thesis, different types of frequencyindependent antennas were
discussed along with the theory of frequencyindependence. It was found that a fourarm
spiral antenna offered the best design choice for being integrated into a directionfinding
system. Two different spiral types were used to design antennas that had a materials cost
of less than $300 and could be fabricated within a week by a technician. If a higher
quality modeformer was designed to be used in conjunction with this antenna design, a
viable lowcost and lightweight solution would be realized.
Each of the antenna geometries were modeled in NEC to show that the difference
between spiral types does not change the power pattern of either antenna at lower
frequencies. On the other hand, at higher frequencies, the square spiral begins to diverge
away from the results of the Archimedean spiral. The theoretical model also proved that
changing the radius of curvature for each spiral type does not change the shape of the
power pattern until a very small radius of curvature (6 cm) is used. The theoretical model
did show that the power pattern changes more rapidly when the phase progression at the
feed of the antenna is not ideal.
After fabricating each antenna type, power patterns of each antenna geometry was
measured. These measured results demonstrated that an antenna provides better results
159
when a modeformer is producing a phase progression that is too short for a given
frequency. At higher frequencies where the phase was much longer than needed, the
power pattern quickly deteriorates. This conclusion agrees with the theoretical model to
suggest that spiral type and conformation of the antenna surface does not change the
overall power pattern as much as connecting the antenna to a less than an ideal mode
former.
6.2 Recommendations for Further Study
As with any antenna study, the recommendations for further study are almost
limitless. One possible path of research might be to create antennas of different radiuses
of curvature to continue proving the invariance of spiral antennas to most conformations.
Another path of research might be to explore different antenna geometry types, such as
the logarithmic or equiangular spiral. Along with different antenna types, the number of
arms could be explored along with different frequencies of operation. One particular
antenna of interest to the author was the eightarm selfcomplementary spiral antenna
which has a characteristic input impedance of nearly 50 ohms. Such an antenna design
would not need an impedance matching circuit and therefore offer an excellent SWR
response with respect to frequency.
160
REFERENCES
[1] Balanis, C. A., Antenna Theory: Analysis and Design, 3
rd
Edition, New Jersey: John
Wiley & Sons Inc., 2005; pages 497652.
[2] Rumsey, V. H., Frequency Independent Antennas, 1957 IRE National Convention
Record, Part 1, 1957; pages 114118.
[3] Elliot, R. S., A View of Frequency Independent Antennas, Microwave Journal,
December 1962; pages 6168.
[4] Stutzman, W. L. and Thiele, G. A., Antenna Theory and Design, 1
st
Edition, New
York: John Wiley & Sons Inc., 1981, pages 260305.
[5] Johnson, R. C., Antenna Engineering Handbook, 3
rd
Edition, New York: McGraw
Hill Inc., 1993, pages 141  1468.
[6] Gazale, M. J., Gnomon: From Pharaohs to Fractals, New Jersey: Princeton
University Press, 1999, pages 151186.
[7] Corzine, R. G. and Mosko, J. A., FourArm Spiral Antennas, Massachusetts: Artech
House, 1990, pages 179.
[8] Penno, R. P. and Pasala, K. M., Theory of Angle Estimation Using a Multiarm
Spiral Antenna, IEEE Transactions on Aerospace and Electronic Systems, Volume
37, Number 1, January 2001, pages 123133.
[9] Caswell, E. D., Design and Analysis of Star Spiral with Application to Wideband
Arrays with Variable Element Sizes, Dissertation, Virginia Polytechnic Institute,
Blacksburg Virginia, 2001.
[10] Balanis, C. A., Advanced Engineering Electromagnetics, New Jersey: John Wiley
& Sons Inc., 1998, pages 254309 and 670742.
[11] Burke, G. J. and Poggio, A. J., Numerical Electromagnetics Code (NEC) Program
Description  Theory, Lawrence Livermore Laboratory Report, January 1981.
161
[12] Caswell, E. D., Analysis of a Helix Antenna Using a Moment Method Approach
With Curved Basis and Testing Functions, M.S. Thesis, Virginia Polytechnic
Institute, Blacksburg Virginia, 1998.
[13] Mittra, R., Computer Techniques for Electromagnetics, New York: Pergamon, 1973.
WRIGHT STATE UNIVERSITY SCHOOL OF GRADUATE STUDIES
June 29, 2007
I HEREBY RECOMMEND THAT THE THESIS PREPARED UNDER MY SUPERVISION BY Douglas Jeremy Glass ENTITLED Investigation of CylindricallyConformed FourArm Spiral Antennas BE ACCEPTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Master of Science in Engineering.
_________________________________ Ronald Riechers, Ph.D. Thesis Director
_________________________________ Fred Garber, Ph.D. Department Chair
Committee on Final Examination
_________________________________ Ronald Riechers, Ph.D.
_________________________________ Fred Garber, Ph.D.
_________________________________ Marian Kazimierczuk, Ph.D.
_________________________________ Joseph F. Thomas, Jr., Ph.D. Dean, School of Graduate Studies
ABSTRACT
Glass, Douglas Jeremy, M.S. Egr., Department of Electrical Engineering, Wright State University, 2007. Investigation of CylindricallyConformed FourArm Spiral Antennas.
A fourarm spiral antenna offers broadband frequency response, wide beamwidths, reduced size compared to other antenna designs, and the ability to determine the relative direction of an incident signal with appropriate modeforming. The reduced overall area projection of the fourarm spiral antenna compared to other antenna designs and the ability to be manufactured in a planar format allows the antenna to reside within an Unmanned Air Vehicle (UAV) fuselage. This thesis investigates the effects of cylindricallyconforming two different designs of a fourarm spiral antenna to reside within the fuselage of a mediumsized UAV. Theoretical predictions of antenna performance were created using the Numerical Electromagnetics Code (NEC) package and compared to measured results of flat and cylindricallyconformed fourarm spiral antennas with and without ground plane apertures.
iii
103 5. Desktop Antenna Measurement System Discussion……………………105 5.…96 5.... CylindricallyConformed Archimedean Spiral……………………….28 3.…..….1.2.1.2..2.2..1.1.. ModeFormer Development……………………………………………………. Introduction……………………………………………………………….…. Numerical Electromagnetics Code Discussion………………………….2.……. Integral Equations…………………………………………………….12 2. Numerical Electromagnetics Code Results………………………………….…19 3. Square Spiral…………………………………………………………..…5 2.2.. Measurement Method………………………………………………………….1..5..2.. Spiral Antennas………………………………………………………. Modal Behavior of the Spiral Antenna…………………………………….…….….2..1.1.….102 5. Spiral Antenna Design……………………………………………. Thesis Motivation………………………………………………………….8 2.102 5..…5 2..2.…41 4.3. Spiral Antenna Development………………………………………………. Theory of Frequency Independence………………………………………….. Maxwell’s Equations…………………………………………………….1.……1 1.3. LogPeriodic Antennas……………………………………. Archimedean Spiral……………………………………………….…... FrequencyIndependent Antennas…………………………………….…………. Thin Wire Approximation…………………………………………. Spiral Antenna Fabrication……………………………………….3.3.3.1..2 2.107 iv .3.. ModeFormer Design ……………………………………………………35 3...….……26 3.2. Biconical Antennas…………………………………………………….. CylindricallyConformed Square Spiral…………………………….…76 4.1....2.2. ModeFormer Fabrication ………………………………………………..Table of Contents Page 1.……..…39 4. Vector Network Analyzer Discussion………………………………….….2.26 3. Reference Antenna Discussion…………………………………………104 5.. Types of FrequencyIndependent Antennas………………………………….3.35 3..41 4..3.…….2.3...1..….………..3. Radius of Curvature Analysis of the Archimedean Spiral…………….86 4.……….1....……44 4.55 4.39 4.. Antenna System Development………………………………………. ModeFormer Results………………………………………………………….2.…………1 1...2.…...36 4..1.. Theoretical Model………………………………………………………..…54 4.2..2.1.1.4.48 4.2..26 3. Method of Moments Solution Method………………………………....…66 4.2 2.1.3. Measured Results…………………………………………………………. Thesis Objectives…………………………………………………………………1 2..
.120 5.……. CylindricallyConformed Archimedean Spiral within Ground Plane Aperture…………………………………………………………….160 v .5. Square Spiral……………………………………………………..….3.3.148 6.1. Square Spiral within Ground Plane Aperture……………………….1.….…….113 5.3..2.. Conclusions…………………………………………………………….. Summary of Work……………………………………………………….. Spiral Antenna Results……………………………………………….. CylindricallyConformed Square Spiral………………………………..3.3...………119 5..………159 References………………………………………………………………..140 5..130 5.6. Power Pattern Phase Balance Comparison…………………………….………….5..158 6.……....1.135 5.4. Recommendations for Further Study………………………………….3.. CylindricallyConformed Archimedean Spiral…………………………125 5.3.3.2.2.………..158 6. Archimedean Spiral……………………………………………...
vi . Marian Kazimierczuk for taking the time to serve on my committee. I would also like to thank the personnel and the measurement capabilities of Spectral Energetics Incorporated. and understanding. Without his guidance. patience. Without it. and Edward Dininger for the time they spent helping me take antenna pattern measurements and my friend Dr. especially my friend Jason and my oldest brother John. Finally. My final thank you goes out to all my friends and family who encouraged me throughout my education. Lamar Westbrook. “Sweets. I would also like to thank Dr. Ronald Riechers for all the time that he has spent guiding and motivating me to complete this task. I would also like to thank Mr. Riechers for instilling a passion for electromagnetics that has only grown over the many years that I have had the honor of being one of his students. Finally. I would also like to thank Dr. Barry Woods.” none of this would have been possible. Fred Garber and the Electrical Engineering Department of Wright State University for providing me the opportunity to attain a Master of Science in Engineering. my undergraduate and graduate experience would not have been as enjoyable.Acknowledgements I would like to begin by thanking Dr. I would also like to thank my friends Mark Stewart. Chad Hagedorn for his time spent editing my thesis. I would not have had any measured data with which to compare my theoretical model. I would also like to thank Dr. without your love. to my wife Sharon.
1.2. the ability to locate communications of enemy combatants is one of the most relevant tasks for supporting the warfighter. Provide a comparison of the theoretical and measured results. INTRODUCTION 1. Thesis Motivation Developing a lowcost (disposable) and lightweight directionfinding system for use with an Unmanned Air Vehicle (UAV) is the motivation of this thesis. 1 . Discuss the development of an antenna capable of directionfinding. Define a theoretical model capable of describing the antenna. Describe the characteristics of the antenna. Thesis Objectives The objectives of this thesis are the following: Define a suitable antenna as defined by the thesis motivation. As the battlefield evolves around the world. Developing an antenna for a directionfinding system with an emphasis on modularity for different mission types and broad frequency response would demonstrate the utility of such a system to the military. 1.1.
where each terminal (feed point) of the antenna surface is infinitesimally close to the origin and is confined to rotation about the θ = 0 axis. The theory of frequencyindependence. and the region surrounding the antenna is assumed to be infinite in extent. the concept of specifying antenna geometries purely by angles instead of characteristic lengths allowed for bandwidths of 40:1 or more. as described by the work of Rumsey [2] and Elliot [3].1 Theory of Frequency Independence Due to the everexpanding utilization of the electromagnetic spectrum. field pattern. and polarization remain constant as the size of the antenna is scaled. prior to the 1950s. homogenous. FREQUENCYINDEPENDENT ANTENNAS 2. and isotropic.θ. To further simplify the problem. If an antenna is only defined by angles. antennas that can operate over wide frequency ranges have become increasingly desirable.φ ) (2.2. surfaces are assumed to be perfectly conducting. Such a surface is described by a single curve (for a wire) or a set of curves (to define the edges of a strip) r = F (θ . Afterwards though. begins by assuming an antenna geometry within spherical coordinates (ρ.φ).11) 2 . antenna technology could only provide bandwidths of 2:1 or less. attributes such as the characteristic input impedance. As stated in Balanis [1].
The scaled curve is then described by r ′ = KF (θ . a scaled singlearm spiral. φ ) (2. φ ) (2. To define the property of scaling. This is shown by the following F (θ . and a scaled singlearm spiral that has been rotated to visualize congruence of the two spirals. start by assuming an antenna were to be scaled to operate at a frequency that is K times lower than the original antenna.13) where the rotated angle φc depends only on K. Figure 2. The physical size of the antenna would need to be increased by a factor of K.where r represents the radial distance along the curve.12) where both surfaces superimpose one another after the second curve is rotated by an angle φc about the θ = 0 axis.11 Comparison of Scaled Spirals 3 . The following figure depicts three spirals that correspond to a singlearm spiral. φ + φc ) = KF (θ .
φ ) = eαφ f (θ ) where α is (2. d [KF (θ .19) (2.φ + φc )] ∂φ ∂ (φ + φc ) Equating (2.13) with respect to both φc and φ.φ )] = ∂ [F (θ . the general solution of the curve becomes r = F (θ .14) (2.φ )] = dK [F (θ .φ )] = ∂ [F (θ .17) results in dK [F (θ . K dφc r ∂φ (2.To prove congruence between the two curves. 4 .110) α= 1 dK K dφc (2.φ )] = K ∂ [F (θ .φ )]. φ + φc )] dφ c ∂ (φ + φc ) ∂ [KF (θ . φ )] = ∂ [F (θ .φ + φc )] ∂φ ∂φ ∂φ ∂ [KF (θ .φ + φc )] d φc dφc ∂φc d [KF (θ .111) and f(θ) is a completely arbitrary function that describes the curve or curves that define the surface of the antenna.15) and (2.15) (2.18) (2.11) into (2. d φc ∂φ Substituting (2.φ )] = K ∂ [F (θ .17) Due to the spatial independence of the lefthand side of (2.16) (2.17).φ )] = ∂ [F (θ .18) gives 1 dK 1 ∂r = . begin by differentiating both sides of equation (2.
finite biconical.2. is θ directed. The electric field. It is suitable to define the length of a finite biconical antenna by this truncated distance. This current distribution creates a magnetic field that encircles each cone. Although it is impractical to assume the existence of a biconical antenna of infinite extent. and the basic antenna theory behind each type. 2.2 Types of FrequencyIndependent Antennas Frequencyindependent antennas within the context of this thesis can be categorized into three types: biconical. and spiral.1 Biconical Antennas Biconical antennas have several geometric variants such as the planar biconical (bowtie). attributes. This section focuses on the infinite biconical antenna to describe the general theory. logperiodic. where the performance of the antenna approaches the infinite case. 5 . it can be shown that the current diminishes to a negligible amount after a finite distance. limitations. The following sections describe the variants. Expanding on this concept. we start with the well known fact that the bandwidth of a dipole antenna can be increased by enlarging the diameter of the conductors. if we instead flared the diameter of the conductors outward as the distance from the feed point increases. the geometry of a biconical antenna is defined.2. To describe the theory of biconical antennas. which is perpendicular to the magnetic field. The following figure depicts the geometry of the infinite biconical antenna. and the discone. The structure can be analyzed as a transmission line where the voltage source produces a current distribution that flows radially outwards.
6 .11) where θh is the cone half angle and θ is the angle from the zaxis. the greater the field strength will be within the available freespace. The following figure depicts the normalized field pattern of the infinite biconical antenna. the normalized field pattern of the antenna is F (θ ) = sin (θ h ) sin (θ ) (2.Figure 2.2.11 Geometry of the Infinite Biconical Antenna (From Stutzman and Thiele [4]) From Stutzman and Thiele [4]. From this relation. it can be determined that the larger the cone angle.2.
7 . The following figure depicts the input impedance versus the cone half angle.12 Normalized Field Pattern of the Infinite Biconical Antenna for Different θh The input impedance of the antenna is defined as θ Z in = 120 ln cot h 2 (2.2. It should be noted that for the case of the finite biconical antenna.Figure 2. the ends of the cones cause a reflection of the incident waves produced by the voltage source to produce a standing wave within the antenna structure. This will create a complex input impedance at the antenna terminals. The same frequency invariant properties can also be extended to the field pattern.2.12) which has the unique property of being invariant with respect to frequency.
Figure 2.2.13 Normalized Impedance of the Biconical Antenna for Different θh
As a candidate for the antenna system, the biconical antenna would need to be utilized in its planar form, the bowtie, to satisfy the geometry requirements. The bowtie antenna would also need to be a part of a phased array to provide the modal characteristics necessary to provide directionfinding capabilities. These inadequacies preclude the biconical antenna from being considered for the antenna system defined by the motivation of this thesis.
2.2.2 LogPeriodic Antennas
The second type of frequencyindependent antenna is the logperiodic. As with the biconical antenna, there are several geometric variants of the logperiodic, such as the toothed planar, the toothed wedge, the trapezoid wire, the zigzag wire, and the dipole array. The following section will focus on the logperiodic dipole array to describe the 8
general theory. The geometry of the logperiodic dipole array antenna, like the biconical, is based on the dipole antenna. We begin by defining the logperiodic dipole array as a linear distribution of dipoles with increasing length that are connected together by alternating the feed polarity to each dipole. The following figure depicts the geometry of the logperiodic dipole array.
Figure 2.2.21 Geometry of the LogPeriodic Dipole Array Antenna (From Stutzman and Thiele [4])
It can be seen from the geometry of the logperiodic antenna that it cannot be solely described by angles as prescribed by the characteristics of the frequencyindependent antenna. It should be noted, though, that the performance of the logperiodic antenna approaches the characteristics of the frequencyindependent antenna. As derived in Stutzman and Thiele [4], the bandwidth of the logperiodic dipole array is defined by the following equations L1 ≈
λLowerFrequency
2
and LN ≈
λUpperFrequency
2
(2.2.21)
9
where L1 and LN are the longest and shortest dipoles respectively. The following equations define the design parameters used to produce logperiodic dipole arrays of varying performance.
τ=
Rn+1 Ln+1 d n+1 = = Rn Ln dn
(2.2.22)
Where τ is the geometric ratio of the antenna, it defines the length growth rate for each of the successive dipoles used in the array.
σ=
dn 2 Ln
(2.2.23)
Where σ is the spacing factor, it describes the distance between each successive dipole within the array.
α = 2 tan −1
1 −τ 4σ
(2.2.24)
Finally, α is the angle that is formed from the intersection of the lines extending from both ends of the dipoles along the array. The following figure depicts design curves for different logperiodic dipole arrays and their respective maximum gains.
10
11 . The active region. dipoles are often added to either end of the antenna to ensure the strength of the active region at either end of the antenna. The following figure depicts a general impedance variation with respect to frequency for a logperiodic dipole array. like other frequencyindependent antennas is nearly constant over frequency. The impedance. It should also be noted that. or where the current is concentrated on the antenna. is around the dipole where the length is approximately one half of a wavelength of the operating frequency.Figure 2.2. in practice. the best performing antennas have a very slow transition in length from one dipole to the next.22 Maximum Gain Curves of the LogPeriodic Dipole Array Antenna with Respect to Spacing and Scaling Factors (From Balanis [1]) As seen in the figure.
and Archimedean.3 Spiral Antennas The final type of frequencyindependent antenna to describe within this discussion is the spiral. These inadequacies preclude the logperiodic antenna from being considered for the antenna system defined by the motivation of this thesis. conical. 2. Like the biconical and logperiodic antennas.2. logarithmic.Figure 2. Each spiral variant also has a planar. The logperiodic antenna would also need to be a part of a phased array to provide the modal characteristics necessary to provide directionfinding capabilities. The variants can be categorized into three types: equiangular.2.23 Representative Impedance Variation with Frequency for a LogPeriodic Dipole Array Antenna (From Balanis [1]) As a candidate for use with the antenna system. the spiral also has several geometric variations in design. 12 . the logperiodic antenna would need to be utilized in any of its planar forms to satisfy the geometry requirements.
We begin by defining the design equations from Stutzman and Thiele [4].2.31 Geometry of the Planar Equiangular Spiral Antenna (From Stutzman and Thiele [4]) As stated above. Previously. all spiral antennas can be defined purely by angles. The twoarm versions of the spiral antenna can be thought of as a long dipole whose conductors have been wound around the feed axis. Figure 2. To describe the general theory of each type of spiral antenna. The first type of spiral to examine is the equiangular. we will focus on the twoarm planar versions of each. the following figure depicts the geometry of the equiangular spiral. There are four equations that define the radius of the inner and outer curves of each conductor’s surface. the conical form was the most prevalent where the planar form has become the popular choice for research due to its ability to be manufactured by printed circuit techniques. r1 = ro e aφ (2.31) 13 .and spherical form.2.
14 . The twoarm selfcomplementary antenna has a theoretical input impedance of 189 ohms where experimentation has determined that this number should be reduced to approximately 164 ohms. (2. which was one of the first spiral antennas geometries to be evaluated.2.35) When designing equiangular spirals. and the maximum radius R is equal to λLowerFrequency/4. the first is the conical spiral. where ro is equal to λUpperFrequency/4.r2 = ro e a (φ −δ ) r3 = ro e a (φ −π ) r4 = ro e a (φ −π −δ ) (2. Selfcomplementary antennas have the remarkable property of having a nearly constant impedance across their bandwidth regardless of shape. It should be noted that the previous figure can also be described as a selfcomplementary antenna. This occurs when the conductor surface occupies the same area as the free space surface. Other forms of the equiangular spiral are shown is the following figures.33) (2.2. it has been experimentally shown that the best performing designs have one and onehalf turns (φ = 0 to 3π) and have a flare rate equal to 0.2. The typical bandwidth of this design is 8:1. Selfcomplementary structures are comprised of a conductor surface whose compliment is identical to the original surface.2.32) (2.221.34) The flare rate a is more easily represented by the expansion ratio ε = e a 2π .
32 Geometry of the Conical Equiangular Spiral Antenna (From Stutzman and Thiele [4]) The other geometric conformation is the spherical spiral.Figure 2. Practically speaking though. 15 .2. such as conforming to the nose of a guided munition. It does provide additional application possibilities. this specific spiral would need to be cut in half along the equatorial plane to create a viable antenna.
Figure 2.33 Geometry of the Spherical Equiangular Spiral Antenna (From Gazale [6]) The second type of spiral to examine is the logarithmic.2.Figure 2.34 Geometry of the Planar Logarithmic Spiral Antenna (From Johnson [5]) 16 .2. the following figure depicts the geometry.
−2π a 2π − tan ψ τ =e =e (2. The pitch angle ψ is related to the flare rate a by tanψ = 1 a (2. r3 and r4. can be created from r1 and r2 by subtracting π from the exponent.2.2.39) The final type of spiral.38) and the design ratio τ is the ratio of the radius of the spiral arm after one turn. the radius of each arm is given by an exponential. 17 . which will be discussed throughout the remainder of this thesis.2.37) The other radii. (2.Like the equiangular spiral. is the Archimedean spiral. The following figure depicts the geometry of the planar form.36) (2.2. The radius equations as defined by Johnson [5] are r1 = ro e aφ r2 = ro e a (φ −δ ) .
(2.Figure 2.2. The width of the strip is defined as W = aδ (2.2.2. the arms of the Archimedean spiral can be described by the centerline curve r = aφ where either side of the strip is defined as (2.310) δ r = a φ ± 2 and the second conductor can be described by rotating the first arm by π.311) (2.312) and the spacing between the centerlines of each strip is S = 2πa .35 Geometry of the Planar Archimedean Spiral Antenna (From Johnson [5]) As defined by Johnson [5].313) 18 .2.2.
A balanced feed is defined as an antenna feed structure where the vector sum of all of the arm excitations is equal to zero. A twoarm antenna radiates by feeding each arm with an excitation that is 180 degrees out of phase with one another to create what is called a balanced feed. the multiarm spiral allows other feed configurations to be used rather than the balanced line feed of the twoarm antenna.It should be noted that the selfcomplementary case occurs when W/S is equal to 0. This directionfinding capability is made possible by feeding the spiral antenna in different configurations or modes.25. 2.3 Modal Behavior of the Spiral Antenna The property of multiarm spiral antennas that is appealing to the motivation of this thesis is the ability to provide a directionfinding capability to the antenna system. Because the Archimedean spiral is only defined by a constant instead of an angle. Extending this principle to an antenna with four arms as described by Corzine and Mosko [7]. It should be noted that a balanced feed is different from a balanced line. To understand the principles of feed configurations. it does not truly conform to the frequencyindependent principle defined earlier. we begin by discussing the general theory of modeforming and the properties of the twoarm antenna feed. This should not diminish the fact that the Archimedean spiral performs extremely well compared to other spiral designs and dominates in terms of the level of research that has been devoted to its design. where a balanced line is defined as a transmission line that has equal and opposite currents traveling along the two conductors. a balanced feed would 19 . Unlike the biconical or logperiodic antenna.
and M is the mode number. The following equation describes the feed excitation phase progression for any mode. An antenna that has a phase progression of 360 degrees or 2π radians is said to be operating in mode one. To understand the mode of an antenna. where the integer multiplier is known as the mode number. any antenna geometry with multiple arms will suffice. As described by the two and fourarm antenna cases. It should be noted that the following phase excitations are not exclusively for a spiral antenna. The number of modes an antenna is capable of supporting is N Modes = ( N Arms − 1) (2. it is possible to allow the total phase progression around the feed points of an antenna be any integer multiple of 360 degrees. The following figure depicts what the feed excitation phase progressions are for each mode of a spiral antenna that has up to six arms. ( N − 1) M P( N ) = 360o N Arms (2. and 270 degrees with respect to the first terminal. P is the phase of the Nth arm in degrees. where a phase progression of 720 degrees or 4π radians is operating in mode two. the total phase progression around the feed points of the antenna is 360 degrees.32) N is an integer between 1 and NArms.consist of a set of feed excitations whose relative phases would be 0. 20 .31) where NArms is the number of antenna arms. 90. 180.
Figure 2. The following figure depicts a vertical cut of the normalized field pattern of a generic fourarm spiral antenna.31 Feed Excitation Phase Progression for each Mode of a Multiarm Antenna (From Corzine and Mosko [7]) The fundamental property of directionfinding is the ability to exploit the distinctive field patterns and phasing of the different modes. 21 .
and so on. Mode two causes a null along the boresight of the antenna that creates a toroidlike field pattern that has a maximum gain at approximately 38 degrees off of boresight.Figure 2. This principle can be extended to understand that the minimum frequency that a mode two 22 . two wavelengths for mode two. Mode three creates a similar pattern to mode two. The phase progression on each loop goes from 0 to 360 degrees as it traverses around the loop for mode one. Each loop of current has a circumference of one wavelength for mode one. As described by Penno and Pasala [8].32 Normalized Field Patterns for Different Modes of a FourArm Spiral Antenna (From Penno and Pasala [8]) The mode one field pattern is always a single lobe oriented along the boresight of the antenna and has a typical beamwidth of 70 degrees or more. the field pattern of each mode can be reproduced by replacing the spiral with an infinite set of current loops that range in circumference from the feed point loop to the outer loop of the spiral. but the maximum gain occurs approximately 45 degrees off of the boresight. 0 to 720 degrees for mode two and so on.
The following figure illustrates how the azimuth angle is determined by subtracting the mode one and mode two output phases. It should be noted that the antenna surface is in the xy plane and the mode one ring has a circumference of one wavelength and the mode two ring has a circumference of two wavelengths. At a minimum. To understand the directionfinding properties of the modeformer. The elevation angle of the detected signal with respect to the boresight of the antenna is found by comparing the amplitudes of the mode one and mode two output signals.field pattern can exist is at least twice the frequency of the lowest mode one field pattern. we start by defining the requirements of directionfinding. Examining figure 2. it is apparent that at elevation angles near boresight. 23 . where every elevation angle provides a unique set of output amplitudes for each of the modes. for a fourarm spiral antenna. The azimuth angle of the detected signal with respect to the boresight of the antenna is found by comparing the phase of the mode one and mode two output signals. the lowest frequency that a mode three field pattern can exist is at least three times the frequency of the lowest mode one field pattern.32. the mode two field pattern strength is considerably smaller than the mode one field pattern strength. Therefore. an elevation and azimuth angle measured from the boresight of the antenna is needed to define the relative orientation of a received signal to the antenna.
This difference between the two output phases is unique at every azimuth point. 24 .33 Azimuth Determination from Modal Phase Difference It can be concluded from the figure that the mode two output phase is double that of the mode one phase. The following figure depicts the modal phase difference as a function of azimuth angle.Figure 2.
Figure 2.33 Azimuth Determination by Modal Phase Difference (From Corzine and Mosko [7]) 25 .
3. ANTENNA SYSTEM DEVELOPMENT 3.1 Spiral Antenna Design Defining the upper and lower frequencies of operation for the spiral antenna provides a general starting point for calculating the overall dimensions. but a square spiral was also developed to understand the effects of different spiral designs on the performance of the antenna system. the less the pattern would be affected due to the rise of the effective permittivity caused by adding a dielectric to the free space surrounding the 26 . As noted previously. the antennas were etched on the thinnest microwave substrate available. Microwave substrate is similar to printed circuit board material. the thinner the substrate. To achieve the best results.1 Spiral Antenna Development The following sections provide the details of the design and fabrication of the square and Archimedean fourarm spiral antennas developed for this thesis.1. but is designed for radio and microwave frequency applications where the relative permittivity and thickness of the dielectric material sandwiched between the two layers of copper has to be known to create circuits that operate properly at these higher frequencies. 3. the focus of this thesis is the Archimedean spiral. Due to the dielectric core.
which was 0. The following equation provides a relation between the outer diameter of the spiral and the lowest operating frequency f Lower = c πDOuter (3.11) where DOuter is the outer diameter of the spiral in meters. the distance between the center conductors of each coaxial line is limited by the outer diameter of the coaxial line. To provide enough room around the edges of the antenna to prevent any coupling between the antenna and the enclosure. If a bundle of four coaxial lines are used to feed a fourarm antenna.010 inches thick and approximately 7. c is the speed of light within a vacuum in meters per second. To determine the number of rotations a spiral arm will make as each arm is drawn out. The maximum operating frequency of a fourarm spiral antenna feed by a bundle of coaxial lines is limited by the following equation fUpper = c π 2 DInner (3.500 inches wide. and fLower is the lower frequency in Hertz.1. The limitations on antenna size were set by the thinnest substrate available. The upper frequency of the antenna is generally limited by the diameter of the coaxial cables used to feed the antenna.1. the strip width is one half of the arm spacing. A general design rule for determining the distance between feed points at the center of the spiral dictates the gap between arm surfaces should be equal to the arm width. This creates a 27 .antenna. For selfcomplementary antennas.12) where DInner is the diameter of the coaxial cable in meters and fUpper is the upper frequency in Hertz. the spiral strip width and arm spacing must be calculated beforehand. the maximum size of the antenna to be etched on the microwave substrate was limited to just under six inches.
323 0.1.661 0. the effect of increasing the strip width does come with the penalty of causing the completely realvalued impedance to acquire a negative reactive component where narrowing the strip width causes an increase in the real part of the impedance and a slight variation to the reactive component. Each arm of the square spiral has to be drawn out by hand or by drafting software suxh as AutoCAD to create.684 0. After the drawings were 28 .375 Table 3. the only difference between the two designs is the square spiral does not have an analytic function that defines the arc of each arm.323 0.11 Design Parameters for Archimedean and Square FourArm Spiral Antennas 3.141 0.2 Spiral Antenna Fabrication The fabrication of the antenna structures started with creating AutoCAD drawings of the square and Archimedean spirals defined by table (3. To reduce the magnitude of the impedance mismatch of a 50 ohm transmission line terminating into a 95 ohm antenna.203 3.11).661 0. As described by Caswell [9]. the overall magnitude is still less than the selfcomplementary case. Spiral Antenna Type Outer Diameter (in) Lower Frequency (GHz) Inner Diameter (in) Upper Frequency (GHz) Strip Width (in) Arm Spacing (in) Number of Rotations Archimedean 5.282 13.25 Square 5.1.141 0. The previous design equations for the Archimedean spiral hold for the square spiral. the strip width can be increased to reduce the impedance of the antenna.characteristic input impedance of 95 ohms for mode one of a fourarm antenna.203 3.1.684 0.282 13. Even though widening the strip width causes the impedance to change. The following parameters were used to design the square and Archimedean spiral antennas.
The two enclosures were constructed with twenty gauge aluminum sheet to form a cavity that was 7. Emerson & Cuming AN77 microwave absorber was used to line the back wall of the cavity. One enclosure was fabricated with a flat face and the other with a cylindricallyconformed face that had a 7.created.010 inch thick microwave substrate with a εr = 3.25 inches square and 5 inches deep. After the spirals were etched. The spirals were etched on 0. which reduces interference between the waves reflecting from the back wall of the enclosure. To provide attenuation to the rearward traveling radiation from the antenna.2.874 inch (20 cm) radius of curvature. they were immersed in a tinning solution to protect the copper from corrosion. The AN77 microwave absorber provides reflectance attenuation of 20 dB or more at frequencies above 1. The opposite copper layer was also etched off to produce a printedwire antenna without a ground plane on the back of the dielectric.2 GHz. The next step in fabricating the antenna was to create two enclosures to shield the rearward radiation and provide support to the antennas. The following figure shows a finished flatfaced enclosure with absorber lining. These nylon screws were used to attach the antenna to the front of the enclosure. The top edge of each enclosure was formed with a lip to provide enough room to allow holes to be drilled and tapped for number eight nylon screws. each spiral was printed out in actual size on clear 8½ by 11 inch sheets of plastic designed to be used with the Dalpro Benchtop Dry Film Etching System. 29 .
050 inch thick microwave substrate with holes drilled in each to align the coaxial lines for each antenna.1. The following figure shows the finished feed structure attached to the Archimedean spiral surface.21 Antenna Enclosure with Flat Face The feed structure for the antenna was constructed with a bundle of four 0.Figure 3. This substrate acted as a support structure for the bundle and also provided a grounding mechanism for the coaxial bundle as they extended away from the enclosure back wall. The coaxial lines were cut to length to align with the top of the enclosure and connect to the etched antenna. The outer conductor of each coaxial line was removed from the last 0. The four coaxial lines were soldered together by two pieces of 0. 30 .070 inches to provide a better impedance match between the 50 ohm coaxial lines and the antenna surface.141 inch copper coaxial lines with bulkhead SMA connectors on one end of each of the cables to allow the coaxial lines to be attached to the enclosure and provide a connection point for the antenna through the enclosure.
The entire assembly is being held by a frame used to attach the antenna to the measurement system that will be described in a measured data chapter.22 FourArm Coaxial Feed Bundle Attached to Antenna Surface The following figure shows a finished Archimedean spiral mounted in an enclosure with a cylindricallyconformed face. 31 .Figure 3.1.
The flat ground plane structure was 24 inches long by 24 inches wide and the curved structure was 24 inches long by 16 inches wide.1. each ground plane structure was 12 inches tall. The following figures show the finished flat and cylindricallyconformed ground planes apertures with the antenna enclosures mounted within them.Figure 3.23 Archimedean Spiral Mounted in Antenna Enclosure with CylindricallyConformed Face The final step in fabricating the antenna was constructing two ground plane apertures: one flat and one cylindricallyconformed with the purpose of holding the antenna enclosures during testing. The ground plane structures were constructed using 0. The shorter width on the curved structure is due to the ground plane wrapping around the curvature of the structure. 32 .500 inch by 0. To allow enough room for the cabling on the back of each antenna.500 inch thick pine studs for the frame and 20 gauge aluminum sheet for the ground plane.
Figure 3.24 Flat Ground Plane Aperture Mounted with Empty Antenna Enclosure Figure 3.25 CylindricallyConformed Ground Plane Aperture Mounted with Empty Antenna Enclosure 33 .1.1.
Figure 3.26 Flat Ground Plane Aperture Mounted with the Square Spiral Figure 3.27 CylindricallyConformed Ground Plane Aperture Mounted with the Archimedean Spiral 34 .1.1.
The following figure is a closer view of the square spiral mounted within the flat ground plane aperture to show the copper tape used to provide an electrical bond between the ground plane and the antenna enclosure.2. Figure 3.2 ModeFormer Development The following sections provide the details of the design and fabrication of the modeformers developed for this thesis.1. The design of the modeformer began with producing 35 . 3.1 ModeFormer Design The purpose of a modeformer for a fourarm antenna is to produce a set of four feed excitations of equal amplitude and phase progressions of 360 degrees for mode one. and 720 degrees for mode two.28 Closer View of Square Spiral Mounted in Flat Ground Plane Aperture 3.
2 ModeFormer Fabrication The fabrication of the two modeformers started with creating AutoCAD drawings of each.050 inch thick RO3010 series highfrequency laminate from Rogers Corporation. the additional line lengths were calculated for microstrip transmission lines. a fixed line length design was utilized to provide an inexpensive. a half wavelength is equal to 180 degrees and so on. SMA endlaunchers were attached to both sides of 36 . A Wilkinson power divider was realized in microstrip lines with a corporate feed arrangement to provide the four excitations. The wavelength can be used to determine the phase length by noting that a quarter wavelength is equal to 90 degrees. This modeformer design only provides the proper phasing at one frequency. It should be noted that an ideal modeformer provides a phase progression that is constant across all frequencies. The next step was to produce the phase progression needed for each mode.2. The frequency was chosen due to the proximity of the PCS cellular frequency band as a possible detection signal for directionfinding.four equalamplitude excitations. For simplicity and due to the fact that the motivation of this thesis was to produce an antenna and not the modeformer. The microwave substrate used for the modeformers was a 0. After the modeformers were etched. The phasing of the feed arms was calculated to provide proper phase progression at a frequency of 1. 3. After the phasing of each arm was calculated.75 GHz. the drawings were printed out in actual size on clear plastic for use with the etching system. where the phase error is proportional to the difference between the measurement frequency and the design frequency. As before. but narrowbanded solution to phasing the excitations.
2.21 Mode 1 ModeFormer Figure 3.2.the board to provide connection points between the measurement system and the modeformer and between the modeformer and the antenna enclosure. The following figures show the finished modeformers. Figure 3.22 Mode 2 ModeFormer 37 .
38 .As the figures show.75 GHz.5 degrees of the design and to provide compensation for the four (not phase matched) cables that were used to connect the modeformer to the antenna. The four arms are numbered in ascending order starting from the top of each picture. the left side of each modeformer provides a connection point to the measurement system. This connection point is split into four arms where additional microstrip line has been added to each arm to provide the proper phase progression at 1. The measured results chapter provides the phase error versus frequency for both of the modeformers. The figures also depict the handtuning of the connection points on the right side of the modeformers to ensure that phase angle of each arm was within +/0.
17) µ (4.1 Maxwell’s Equations As a starting point to understanding the theoretical model. the magnetic field strength is defined as v v 1 H = ∇× A.15) (4.13) (4. the magnetic vector potential will be derived from Maxwell’s equations to define one of the possible solution methods of Maxwell’s equations. THEORETICAL MODEL 4.11) (4.12) (4.4. (4.16) (4.18) 39 . Based on the vector identity v ∇ ⋅ (∇ × Ω ) = 0 v where Ω is an arbitrary vector. We begin with Maxwell’s equations in the timeharmonic form within a homogeneous medium as defined by Balanis [10] v v ∇ × E = − jω B v v v ∇ × H = J + jωD v ∇⋅D = ρ v ∇⋅B = 0 with the following constitutive parameters v v D = εE v v B = µH .14) (4.
(4.114) ( ) (4.Substituting (4. (4.12) results in v v v ∇ × ∇ × A = µJ + jωµε − jωA − ∇φ ( ) ( ) (4. Inserting (4.112) where φ is the electric scalar potential.11) results in v v ∇ × E = − jω ∇ × A ( ) (4.113) to v v v v ∇ 2 A + ω 2εµA = − µJ + ∇ ∇ ⋅ A + jωεµφ .18) and (4.112) into (4.110) becomes v v E = − jωA − ∇φ (4.111) (4.113) and using the vector identity v v v ∇ × (∇ × Ω ) = ∇(∇ ⋅ Ω ) − ∇ 2Ω reduces (4.116) φ =− 1 jωεµ v ∇⋅ A (4.115) reduces to the inhomogeneous vector Helmholtz Equation for the magnetic vector potential 40 .19) where bringing both terms to one side results in v v ∇ × E + jω A = 0 .16) and (4.115) Using the Lorentz Gauge v ∇ ⋅ A = − jωεµφ now φ can be written as (4. ( ) (4.117) and (4.110) Using the vector identity ∇ × (− ∇g ) = 0 where g is a scalar function.18) into (4.
it can be thought of as a delta function source as defined by Balanis [10].v v v ∇ 2 A + ω 2εµA = − µJ v where E can now be written as v v v 1 E = − jω A − j ∇ ∇⋅ A .119) 4. and Stutzman and Thiele [4]. The calculated current distribution can then be used to find the field pattern. 4.118) ωεµ ( ) (4.2 Numerical Electromagnetics Code Discussion The theoretical model utilized within this thesis is the Numerical Electromagnetics Code (NEC) package developed by the Lawrence Livermore Laboratory.2. and polarization of the antenna. (4. NEC utilizes the Method of Moments (MoM) solution method to solve for the current distribution on a wire antenna structure that has been segmented. Due to the infinitesimally small size of the source.1 Integral Equations The following section describes the formulation of the electric field integral equation from the inhomogeneous vector Helmholtz Equation that NEC uses to solve for the currents on each wire segment that comprises a wire antenna structure. impedance.118) to 41 . We can convert (4. We begin by simplifying the problem by assuming a source with a current density Jz is located at the origin of the coordinate system. The following sections provide the general theory necessary to understand the operation of NEC.
2.12) If the source is assumed to be a delta function.2. Az = e − jβR 4πR (4.15) and weighted by the total distribution of the current density over a volume V. if we assume a collection of source points.2. If we allow the source point to be at an arbitrary position other than the origin. (4. ∇ 2 Az + β 2 Az = 0 (4.15) As defined by the Green’s function method.14) where r is the distance from the source point (at the origin) to the observation point. then the total contribution is equal to the integral of the source point responses due to (4. the general solution for A is defined as 42 . any point other than the origin of the coordinate system will be equal to zero. Assuming this delta function source allows the wave equation to be solved as a homogeneous equation for any point not at the origin.2.2.11) where the substitution for β is ω 2εµ = β 2 .∇ 2 Az + β 2 Az = µJ z (4.2.13) This is known as the scalar Helmholtz’s equation and has a solution in spherical coordinates of the form Az = e − jβr 4πr (4.2.16) v If each of the other components are considered. Az = µ ∫∫∫ J z V e − jβR dV 4πR (4. the distance parameter R is defined as the distance between the source point and the observation points.
2.v v v v e − jβR A(r ) = µ ∫∫∫ J (r ′) dV 4πR V where v r is the vector from the origin to the observation point v r ′ is the vector from the origin to the source point (4. v v v v v v A(r ) = µ ∫∫ J s (r ′)G (r . V (4.18) reducing (4.2. the total tangential field can be equated to the summation of scattered and incident waves at the surface of the antenna vT vI vS Et = Et + Et = 0 v v v v v v EtI (r = rs ) = − EtS (r = rs ) (4.2. the total tangential electric field on the surface of the wire is equal to zero.2.2.17) v v R = r − r ′ is the distance from the source point to the observation point. The volume integral can then be converted into a surface integral that operates over the surface A of the wire. If the problem is thought of in terms of electromagnetic scattering.17) down to v v v v v v A(r ) = µ ∫∫∫ J (r ′)G (r . r ′)dA A (4. r ′)dV .110) If we define that the antenna surface is constructed with an arrangement of PEC wires.2. the free space Green’s Function can be defined as e − jβR v v G (r .19) If we assume that the current density is confined to a wire of radius a that is perfectly electrically conducting (PEC). r ′) = 4πR (4. the current will only reside on the surface. At this point.112) 43 .111) (4.2.
2. With the integral simplified. is to bring the divergence and gradient within the integral after simplification.110) into (4. (4. The first step as described by Caswell [12]. r ′)dA + j ∇ ∇ ⋅ ∫∫ J s (r ′)G (rs .112). If reciprocity holds.119) and utilizing the reciprocity principle of (4. the thinwire approximation can be applied. r ′)dA v v s s A v v t (4. v v v v v 1 v v EtI (r = rs ) = jωµ ∫∫ J s (r ′)G (rs .113) can be simplified further by utilizing the thinwire approximation.2. r ′)dA + j ωε A ∫∫ (∇′ ⋅ J (r ′))∇G (r .2.2 Thin Wire Approximation The following sections provide a broad overview of the utilization of the electric field integral equation tailored for use with wire antennas as described in detail by Mittra [13]. The thinwire approximation is defined by the following assumptions: 44 . The incident field can be thought of inducing a current density.21) where ∇′ = ∂ ∂ ∂ ˆ ˆ ˆ x′ + y′ + z′ ∂x′ ∂y′ ∂z′ and ∇′ operates only on the source coordinates.v where rs is the observation point that is now restricted to the surface of the wire. Substituting (4. r ′)dA ωε A A t where the subscript t is the tangential component. a negative scattered field is then induced by the same current density. the electric field can be written as v v v v v v v 1 v v v v EtI (r = rs ) = jωµ ∫∫ J s (r ′)G (rs .2.2.113) 4. The electric field integral equation of (4.2.
This conversion is allowed due to the fact that the current remains azimuthally invariant around the circumference of the wire and is directed along the axis of the wire. like (b) and (c). to be represented as an infinitely narrow current filament of length L that has been divided by the circumference of a wire.2.2. v v 1 v ˆ J (r ′) = I (s′)[s′(r ′)] s 2πa (4.21 Equivalent Source Conversion (from Stutzman and Thiele [4]) v The thinwire approximation allows the current density on the surface J s . the observation points can either be on the axis. Figure 4. or on the radius of the wire. As the figure shows.The current only flows in the direction of the wire axis The current can be represented by a filament on the wire axis The boundary condition on the electric field needs to be enforced in the axial direction only The following figure illustrates the geometry of the thin wire approximation.22) 45 .
r ′)dA . v v s A v v (4.2.23) If the surface integral is separated into its components of one integral operating around the circumference of the wire and the other integral operating along the length of the wire. Figure 4. r ′)dA + j ωε A ∫∫ (∇′ ⋅ J (r ′))∇G(r . we can substitute the thinwire approximation into the integrals resulting in 46 .The following figure describes the local axes for the source and observation points.22 Local Axes Description Taking the tangential component of the electric field can now be defined as v v 1 v v v v v ˆ ˆ s (r ) ⋅ E I = s(r ) ⋅ jωµ ∫∫ J s (r ′)G (r .2.
r ′) = G (r . r ′)adψ ′ds′ ωε L 0 (4.2. r ′)ds′ ∂s∂s′ (4.24) where ∂ v v v v v ˆ s (r ) ⋅ ∇G (r .2π v v v v 1 v v ˆ ˆ ˆ s (r ) ⋅ E I = jωµ ∫ I (s′)(s(r ) ⋅ s′(r ′)) ∫ G (r .25) v ∂ ˆ ∇′ = s′(r ′) . r ′)ds′ + j L ωε ∫ I (s ′ ) L ∂2 v v Gψ (r .24) reduces to (s − s′)2 + a 2 1 v v v v v v ˆ ˆ ˆ s (r ) ⋅ E I = jωµ ∫ I (s′)(s (r ) ⋅ s′(r ′))Gψ (r . r ′) = G (r .27) R= then (4.2. all of the constants cancel to reveal the remaining Green’s function that is now confined to the wire segment instead of free space.2.2. r ′) where R is now restricted to the wire segment (4. ∂s′ (4. 47 .2.28) which defines the electric field integral equation used by NEC as described by Burke and Poggio [11]. r ′)adψ ′ds′ L 2πa 0 2π 2 1 ∂ 1 v v +j ∫ I (s′) ∂s∂s′ 2πa ∫ G (r .2.26) If we evaluate the circumferential integral. which operates on the loop of radius a around the equivalent thinwire current element v v v v Gψ (r . r ′) ∂s (4. The new Green’s Function.
2. A (4. is defined as (4. N (4. e .2. We begin by defining the linear operator equation Lf = e (4.31) where L is a linear operator (which in the case of the electric field integral equation is an integral operator).33) and (4. f is the unknown response. Lf n = wm . If the form of the solution is known beforehand. m = 1.2... which for this geometry..2. These particular functions are more capable of handling solutions that have a traveling wave form.2. The choice of basis function depends on the problem being solved.34) Combining (4. N . When the form of the solution is unknown. The unknown response can be expanded into a sum of constant coefficients multiplied by a basis function fn. where the inner product.2.32) produces ∑a n =1 N n wm . f ≈ ∑ an f n n =1 N (4.2.31) with a set of weighting functions wm . v = ∫∫ u (r ′)v(r )dA . the piecewise constant (pulse) is the most straightforward. e . and e is a known excitation.3 Method of Moments Solution Method The following section defines the general theory of the Method of Moments for solving operator equations.2.33) v v u .4..32) A set of equations for the unknown coefficient an is created by taking the inner product of (4. such as antennas.. m = 1.2.35) 48 . is not weighted.... the basis function is chosen to match the response of the solution such as the triangular and sinusoidal basis functions. Lf = wm .
The weighting and basis functions can be different depending on the problem geometry. can be chosen for the problem type.2.38) and substituting into (4.2. the Z matrix is inverted [A] = [Z ]−1[V ] .37) The weighting function.2.310) (4. NEC uses wm = δ (s − sm ) f n (s′) = An + Bn sin (β (s′ − s n )) + Cn cos(β (s′ − sn )) (4.36) where A is a vector containing the unknown coefficients.311) 49 .which can be written in matrix form to produce Ohm’s Law [A][Z ] = [V ] (4.2. The other possibility is if the weighting and basis functions are the same. r ′)ds′ L n =1 ∂2 v v +j an f n (s′) Gψ (r . r ′)ds′ ∫∑ ωε L n=1 ∂s∂s′ 1 N .2. Z is a matrix of the summed inner products of the integral operator and basis function with the weighting function and V is a vector of the inner products between the weighting function and the excitation.28).2.2. The following equations describe how NEC utilizes the Method of Moments to solve for the current of (4. (4. To calculate the values A. where this is defined as Galerkin’s Method. like the basis function.39) The next step is to employ the weighting and basis functions. Starting with an expansion for the current on each of the n segments of wire that defines the antenna surface I (s′) ≈ ∑ an f n (s′) n =1 N (4. (4.2.28) results in N v v v v v v ˆ ˆ ˆ s (r ) ⋅ E I = jωµ ∫ ∑ an f n (s′)(s (r ) ⋅ s′(r ′))Gψ (r .
312) and (4.2. Vm = v ˆ ∫ (s(r )⋅ E )δ (s − s )ds vI m Lm (4.310) into (4..2. N ..2.2.37). m = 1.2. This dominance is due to the distance between each source and observation segment.39) results in the following equations that constitute the matrix elements and vectors of (4.316) The Z matrix will be diagonally dominant due to the source segment terms being the largest component of the observation segment terms.313) reduce to v v ˆ Vm = s(r (sm )) ⋅ E I (sm ) v v v v ˆ ˆ Z mn = jωµ ∫ f n (s′)(s (r (sm )) ⋅ s′(r ′))Gψ (r (sm ). (4.where the weighting function effectively samples the integral at the midpoint of each source location sm of segment length Lm and the basis function operates over the midpoint of each segment sn. r ′)ds′ ∂s∂s′ (4. r ′)ds′ L (4.2.313) Utilizing the definition of the integrated delta function..2. (4. The diagonal terms of the Z matrix represent situations where the source segment is at the same location as the observation segment.312) N v v v v ˆ ˆ Z mn = jωµ ∫ δ (s − sm )∫ ∑ an f n (s′)(s (r ) ⋅ s′(r ′))Gψ (r .2. As the observation segment moves away from the source segment.314) +j where 1 ωε ∫ L ∂2 v v f n (s ′ ) Gψ (r (sm ). r ′)ds′ds Lm L n =1 ∂2 v v δ (s − sm )∫ ∑ an f n (s′) Gψ (r .2. r ′)ds′ds +j ∫ ∂s∂s′ ωε Lm L n =1 1 N (4.315) ∑Z n =1 N mn n a = Vm . its location in the Z matrix moves further away from the diagonal therefore decreasing its 50 . Substituting (4.2..
2. all of the individual field contributions can be vectorsummed to produce the field response of the entire wire antenna structure. v v E A ≈ − jω A (4. will be four locations.317) is used to find the field of each current segment. This principle defines a region where the angular field response is nearly independent of the distance to the antenna. The V vector will be mostly sparse except for the segments where there is a voltage source. 51 .317) The farfield of the electric field occurs when the distance from the current source becomes great enough that the difference between the elevation angle of a ray extending from the center of the antenna to the observation point and a ray extending from a point along the antenna to the observation point approaches zero. the field pattern can be calculated from the following magnetic vector potential farfield relations for the electric field. After the current for each segment has been found. If (4. which in the case of a fourarm spiral antenna.2.contribution to the overall calculation of the current. The following figure illustrates the antenna coordinate geometry.
creating the input files for NEC starts with converting the flat 52 .320) Now that the theory of operation for NEC has been explained.2. Due to the fact that NEC is a wire antenna simulation.319) (4.31 Antenna Coordinate System (from Balanis [1]) After the electric field is known.2. understanding how to create wire antenna models for NEC can be discussed.2. φ ) G (θ . the power at each angle can be defined as 2 1 εo r v v* P (θ .318) and the power gain as P(θ .2.Figure 4. φ ) = E⋅E 2 µo ( ) (4. (4. φ ) = 4π Pin where the Pin is calculated from the current and voltage at the source Pin = Re(VI * ) .
This information is used by NEC to calculate the relative orientation of each wire segment and to determine where the observation point due to wire radius is located on each wire segment. MATLAB was used to calculate the length and location of each wire segment.y. the widely accepted transformation between flat strips and wires is 1 rw = Ws 4 (4. Each successive segment should use the end point of the previous segment as the starting point. As shown in Caswell [9].strips of the etched antenna into a wire with a specific radius. Within NEC. The calculated wire segments were used to create an input file that was properly formatted to be read by NEC. The length of the wire segment provides the value of delta needed to produce an electric field from a voltage potential divided by a distance. NEC allows the user to designate any single wire segment to provide a location for the deltagap voltage source to exist on.31) where rw is the wire radius and Ws is the strip width.z) that describe the starting and ending points of the wire. Along with the ability to designate which segments 53 . the length of each segment should be at least six times the radius of the wire to conform to the thinwire approximation. and the end point should follow the spiral arm as it progresses away from the origin. each wire segment has an associated wire radius and a set of Cartesian coordinates (x. As a general design rule. The final step to creating an NEC input file is to designate the feed excitations and their relative phases. It should be noted that the wire segments must be orientated so that the starting point of the first segment is at the feed point of each antenna arm and the end point is directed away from the origin. To create the wire model for each antenna design.
φ) where the angle θ is measured from the zaxis. As a clarification. This is to avoid confusion with the measured results chapter where the measurement system was limited to vertically polarized results due to the vertically polarized receive antenna.constitute the feed structure of the antenna. 4. 54 .θ. A θ angle of zero degrees equates to the boresight of the antenna and a positive angle corresponds to a counterclockwise rotation looking from an observation point along the positive yaxis. The following figure illustrates the local geometry used to display the power pattern plots of each antenna design. the phase angle and amplitude of each voltage source can be modified to allow for different modes of the spiral antenna to be simulated.3 Numerical Electromagnetics Code Results The following sections provide the results obtained from the NEC simulations that were performed using the 4NEC2X version of NEC. The NEC results will be displayed in spherical coordinates (ρ. only the vertically polarized power patterns will be shown. where the spiral antenna is generally circularly polarized.
Figure 4.1 Archimedean Spiral The first spiral to examine is the flat Archimedean design. 55 .3.31 Antenna Plotting Geometry Description 4. The following figure depicts the segmented spiral used by NEC.
Figure 4.3. Each voltage source is located on the second segment in from the beginning of each arm. The following figure depicts the mode one maximum vertically polarized gain of the spiral antenna versus frequency.11 NEC Geometry of Archimedean Spiral The voltage sources of the spiral are designated by the four pink bands near the origin. 56 .
It should be noted that the power patterns have been plotted in a logarithmic scale where the unit of dBi is the relative gain of the antenna compared to an isotropic radiator. 57 . the maximum gain of the antenna reaches a nearly steady value after 3 GHz.Figure 4.12 Mode 1 Archimedean Spiral Maximum Vertically Polarized Gain versus Frequency As the figure shows.3. The following figures depict the power patterns of the spiral for mode one and mode two.
14 Mode 1 Archimedean Spiral Power Pattern (dBi) θ Cuts (1 GHz) 58 .13 Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 4.3.3.Figure 4.
15 Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz) 59 .Figure 4.3.3.75 GHz) Figure 4.16 Mode 1 Archimedean Spiral Power Pattern (dBi) θ Cuts (1.
3.18 Mode 1 Archimedean Spiral Power Pattern (dBi) θ Cuts (2.5 GHz) Figure 4.Figure 4.5 GHz) 60 .17 Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.3.
3.Figure 4.19 Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.110 Mode 1 Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz) 61 .3.
Figure 4.3.112 Mode 2 Archimedean Spiral Power Pattern (dBi) θ Cuts (1 GHz) 62 .3.111 Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 4.
Figure 4.3.113 Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 4.3.114 Mode 2 Archimedean Spiral Power Pattern (dBi) θ Cuts (1.75 GHz)
63
Figure 4.3.115 Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
Figure 4.3.116 Mode 2 Archimedean Spiral Power Pattern (dBi) θ Cuts (2.5 GHz)
64
Figure 4.3.117 Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz)
Figure 4.3.118 Mode 2 Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz)
65
3.21 NEC Geometry of CylindricallyConformed Archimedean Spiral 66 . 4.5 GHz where the θ cuts become nearly concentric. Figure 4. There is also a slight “squaring” of the mode one θ cuts which is most likely due to the antenna becoming electrically large enough to support the beginning of a mode three power pattern.The previous figures illustrate that the mode one beamwidth decreases as the frequency increases. agreeing with the work of Penno and Pasala. The following figure depicts the segmented spiral used by NEC. The figures also depict that the power pattern for mode two does not fully develop until after 2. As a final point.3.2 CylindricallyConformed Archimedean Spiral The following section characterizes the results of the cylindricallyconformed Archimedean spiral. the effects of scaling that were discussed in chapter two can now by seen as the rotation of the power pattern around the zaxis as the frequency is increased. the mode two power pattern is nearly perfect. At 5 GHz.
the maximum gain versus frequency approaches a nearly steady solution after 3 GHz. The following figures depict the power patterns of the spiral for mode one and mode two. Figure 4.22 Mode 1 CylindricallyConformed Archimedean Spiral Maximum Vertically Polarized Gain versus Frequency As with the flat case.The following figure depicts the mode one maximum vertically polarized gain of the antenna versus frequency. 67 .3.
Figure 4.23 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 4.3.3.24 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1 GHz) 68 .
26 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1.3.3.Figure 4.75 GHz) 69 .75 GHz) Figure 4.25 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1.
3.27 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (2.28 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (2.Figure 4.5 GHz) Figure 4.3.5 GHz) 70 .
29 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.3.Figure 4.3.210 Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz) 71 .
211 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 4.Figure 4.212 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1 GHz) 72 .3.3.
214 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1.75 GHz) Figure 4.Figure 4.3.213 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1.3.75 GHz) 73 .
Figure 4.3.3.216 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (2.5 GHz) Figure 4.215 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5 GHz) 74 .
218 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz) 75 .3.Figure 4.3.217 Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.
3.3. 4. Also. The following figure depicts the segmented spiral used by NEC.As with the flat Archimedean spiral. 76 . the previous figures illustrate that the mode one beamwidth decreases as the frequency increases. Mode two appears to also suffer from conforming the antenna where the power pattern has warped slightly.31 NEC Geometry of Square Spiral The following figure depicts the mode one maximum vertically polarized gain of the antenna versus frequency. The difference between the flat and conformed case appears to be the warping of the θ cuts to produce an ellipse instead of a circle. the circularly polarized antenna in the flat case has changed to an elliptically polarized antenna seen by the φ cuts not connecting at boresight. Figure 4.3 Square Spiral The following section characterizes the results of the square spiral.
77 .32 Mode 1 Square Spiral Maximum Vertically Polarized Gain versus Frequency Differing from the Archimedean spiral.Figure 4.3. the maximum gain versus frequency begins to oscillate after 3 GHz. The following figures depict the power patterns of the spiral for mode one and mode two.
33 Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 4.34 Mode 1 Square Spiral Power Pattern (dBi) θ Cuts (1 GHz) 78 .Figure 4.3.3.
75 GHz) Figure 4.3.75 GHz) 79 .36 Mode 1 Square Spiral Power Pattern (dBi) θ Cuts (1.Figure 4.3.35 Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (1.
37 Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (2.3.Figure 4.3.38 Mode 1 Square Spiral Power Pattern (dBi) θ Cuts (2.5 GHz) Figure 4.5 GHz) 80 .
3.Figure 4.310 Mode 1 Square Spiral Power Pattern (dBi) θ Cuts (5 GHz) 81 .39 Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.3.
311 Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 4.Figure 4.3.3.312 Mode 2 Square Spiral Power Pattern (dBi) θ Cuts (1 GHz) 82 .
Figure 4.75 GHz) Figure 4.314 Mode 2 Square Spiral Power Pattern (dBi) θ Cuts (1.3.75 GHz) 83 .313 Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (1.3.
315 Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (2.316 Mode 2 Square Spiral Power Pattern (dBi) θ Cuts (2.3.Figure 4.5 GHz) Figure 4.3.5 GHz) 84 .
3.Figure 4.3.317 Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.318 Mode 2 Square Spiral Power Pattern (dBi) θ Cuts (5 GHz) 85 .
41 NEC Geometry of CylindricallyConformed Square Spiral The following figure depicts the mode one maximum vertically polarized gain of the antenna versus frequency.3. The following figure depicts the segmented spiral used by NEC. 86 .The previous figures illustrate that the mode one power pattern of the square spiral agrees with the Archimedean spiral up to 5 GHz where the power pattern begins to look like the head of a Phillips screwdriver. Figure 4. Mode two power patterns of the square spiral also agree with the results of the Archimedean spiral where at 5 GHz.4 CylindricallyConformed Square Spiral The following section characterizes the results of the cylindricallyconformed square spiral.3. 4. the square spiral is nearly concentric.
3.42 Mode 1 CylindricallyConformed Square Spiral Maximum Vertically Polarized Gain versus Frequency The frequency response of the conformed square spiral agrees with the flat where the same oscillation of gain is seen after 3 GHz. The following figures depict the power patterns of the spiral for mode one and mode two.Figure 4. 87 .
3.3.44 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (1 GHz) 88 .43 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 4.Figure 4.
3.75 GHz) 89 .46 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (1.3.45 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1.Figure 4.75 GHz) Figure 4.
3.Figure 4.5 GHz) Figure 4.3.47 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz) 90 .48 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (2.
3.49 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.Figure 4.410 Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (5 GHz) 91 .3.
3.411 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 4.3.412 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (1 GHz) 92 .Figure 4.
75 GHz) 93 .75 GHz) Figure 4.3.414 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (1.Figure 4.413 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1.3.
3.3.416 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (2.Figure 4.415 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz) 94 .5 GHz) Figure 4.
418 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) θ Cuts (5 GHz) 95 .3.3.417 Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.Figure 4.
where the θ cuts of the pattern settle into a nearly concentric circle even though the surface has been conformed. the Raven. 80 cm (31.The previous figures illustrate that the mode one power patterns of the conformed square spiral agrees with the conformed Archimedean spiral results. 20 cm (7. The following figures provide the mode one and mode two power patterns of an Archimedean spiral antenna that has been conformed at the following radiuses of curvature: infinity (flat). 14 cm (5.3. 4. and 6 cm (2. 40 cm (16 inches). 96 . These radiuses of curvature correspond to air platforms of different sizes ranging from the 707 to a Predator down to one of the smallest UAVs used by the Army.5 Radius of Curvature Analysis of the Archimedean Spiral The following section compares the results of conforming the Archimedean spiral at different radiuses of curvature. the pattern is warped creating a slight elliptical polarization.5 inches).5 inches). Mode two power patterns of the conformed square spiral also agree with the results of the conformed Archimedean spiral. In both designs. The purpose of this analysis is to determine the maximum radius of curvature that still permits the use of the antenna for directionfinding.9 inches).4 inches).
Figure 4.3.3.51 Radius of Curvature Analysis Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz) Figure 4.75 GHz) 97 .52 Radius of Curvature Analysis Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1.
53 Radius of Curvature Analysis Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.Figure 4.3.54 Radius of Curvature Analysis Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz) 98 .3.
3.Figure 4.75 GHz) Figure 4.3.55 Radius of Curvature Analysis Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz) 99 .56 Radius of Curvature Analysis Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (1.
3.3.Figure 4.57 Radius of Curvature Analysis Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (5 GHz) Figure 4.58 Radius of Curvature Analysis Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) θ Cuts (5 GHz) 100 .
Even with this slight change in antenna power pattern shape. if the antenna pattern is known. Only when the pattern is conformed to a radius of curvature equal to 6 cm does the power pattern shape begin to change.The previous figures illustrate the near invariance of the power pattern to the radius of curvature. the pattern shape is almost indistinguishable from the flat case. the directionfinding calculations could be compensated to make up for these distortions. 101 . Even when the antenna is conformed at a radius of curvature of 14 cm.
5. MEASURED RESULTS 5.11 Antenna Measurement System Component Layout 102 .1 Measurement Method The following sections describe the components of the antenna measurement system that was used to gather the measured results of this chapter. Figure 5. The following figure provides an overview of the components comprising the antenna measurement system.
5. For the antenna measurement system.1. and group delay.1 Vector Network Analyzer Discussion The Vector Network Analyzer (VNA) used to provide the measured results of this thesis was the twoport Anritsu 37347C which has an operating frequency range from 40MHz to 20 GHz. or sparameters. or S21. the VNA is used to measure the transmission. 103 . The twoport VNA provides simultaneous transmission and reflection measurements at and between each of the ports. The transmission and reflection of the device under test is found by measuring of the ratio of the transmitted and received voltages at each of the ports. The transmission and reflection terms are categorized as scattering parameters. voltage standing wave ratio (VSWR). Each of the four possible measurements are given by the following sparameters: S11 is the ratio of the received voltage at port one to the transmitted voltage from port one S22 is the ratio of the received voltage at port two to the transmitted voltage from port two S21 is the ratio of the received voltage at port two to the transmitted voltage from port one S12 is the ratio of the received voltage at port one to the transmitted voltage from port two Each of the sparameters can be used to provide information about the frequency response of a device under test such as the insertion loss (IL). across a range of frequencies between the two antennas.
The following figures provide the maximum gain versus frequency and antenna power pattern for the reference antenna.2 Reference Antenna Discussion The reference antenna used by the antenna measurement system was the Antenna Design & Manufacturing Corporation DHR118/A 118 GHz Double Ridged Horn.5.1. the reference antenna acts as both the receive antenna during a measurement and is also the calibrated antenna used to determine the final gain of the antenna being measured. Within the antenna measurement system.21 Reference Antenna Maximum Vertically Polarized Gain versus Frequency 104 .1. Figure 5.
The DAMS system is comprised of a software application that controls the VNA and the rotation platform controller.1.22 Reference Antenna Power Pattern (dBi) φ Cuts 5.Figure 5. The rotation platform is designed to hold the antenna under test while it is rotated across a set of azimuth angles and/or elevation angles.3 Desktop Antenna Measurement System Discussion The antenna measurement system utilized for measured results of this thesis was the Diamond Engineering Desktop Antenna Measurement System (DAMS) that operates from 0 to 18 GHz. Along with the software. 105 . The reference antennas are then aligned so that the boresights of each antenna are pointing directly at each other.1. The measurement of an antenna under test starts with mounting two identical reference antennas to the tripods supplied by the DAMS system. the system provides a tripod to hold the receive antenna and a second tripod to hold the rotation platform.
the transmission measurements of the antenna under test is compared to 106 . The antenna under test is then mounted to the rotation platform and is situated to be at the same distance from the receive antenna as the transmit antenna was. After the data has been exported. Each angle is then recorded by the DAMS software and stored to memory. This allows the free space path loss differences between the reference antenna transmission and the antenna under test transmission to be ignored. the data is exported into a text file that can be read by MATLAB or any spreadsheet software. After all of the angles and frequencies of interest have been measured. Once both of the antennas have been connected to the VNA. the VNA measures the transmission of the two antennas across a frequency range. After the reference antenna transmission has been taken. As the DAMS software moves the rotation platform around one azimuth or elevation step at a time. the DAMS software is used to control the rotation platform and VNA to record the angular and frequency dependent transmission results of the antenna under test. the reference antennas are then connected to the VNA where the port two reference antenna acts as the receive antenna and the port one reference antenna acts as the transmit antenna. a frequency range is designated and the VNA measures the transmission between the two antennas.After alignment. the distance between the two antennas is recorded and the reference antenna that was mounted to the rotation platform is removed. After both antennas have been mounted and aligned again. This reference antenna transmission measurement is used to calculate the isotropic antenna transmission between the two antennas by subtracting the known gain with respect to frequency of one of the reference antennas from the transmission measurement.
the reference antenna transmission results to calculate the relative gain of the antenna under test to the known gain of the reference antenna.2 ModeFormer Results The following section provides the results of the modeformers utilized to create the different power patterns of the antennas. some of the transmission is lost due to the bridging resistors. 5. The following figure depicts the maximum variation in transmission between each arm of the mode one modeformer. Figure 5. 107 . The following figure depicts the transmission loss of each arm of the mode one modeformer with respect to frequency. but due to the utilization of the Wilkinson power divider.21 Mode 1 ModeFormer Transmission Loss versus Frequency An ideal modeformer would provide only 6 dB of loss.
3 dB. the transmission balance between all of the arms of the modeformer is within 0.22 Mode 1 ModeFormer Maximum Transmission Loss Variation versus Frequency The figure shows that up to 2 GHz. The following figure shows the relative phase difference between each arm of the mode one modeformer. 108 .Figure 5.
Figure 5.24 Mode 1 ModeFormer Arm Phase Error versus Frequency 109 .23 Mode 1 ModeFormer Arm Phase versus Frequency The following figures depict the relative phase error with respect to frequency of the mode one modeformer compared to an ideal mode one modeformer. Figure 5.
Figure 5. 110 .25 Mode 1 ModeFormer Arm Phase Error versus Frequency (Expanded View) The following figures depict the transmission loss and maximum variation in transmission loss between of each arm of the mode two modeformer with respect to frequency.
26 Mode 2 ModeFormer Transmission Loss versus Frequency Figure 5.Figure 5.27 Mode 2 ModeFormer Maximum Transmission Loss Variation versus Frequency The following figure shows the relative phase difference between each arm of the mode two modeformer. 111 .
Figure 5. Figure 5.28 Mode 2 ModeFormer Arm Phase versus Frequency The following figures provide the relative phase error with respect to frequency of the modeformer relative to an ideal mode one modeformer.29 Mode 2 ModeFormer Arm Phase Error versus Frequency 112 .
Only φ cuts were measured on antennas that were not mounted within a ground plane aperture. 113 . 5. several of these modeformers could be fabricated to operate at different frequencies to provide a modular solution to the frequency coverage of spiral antennas.75 GHz.Figure 5.210 Mode 2 ModeFormer Arm Phase Error versus Frequency (Expanded View) The results show that the modeformers perform well around 1. The phase error begins to degrade the quality of the modeformer after a frequency of 50 MHz on either side of 1. The following figures depict measured and simulated φ cuts of each modeformer across a set of frequencies.75 GHz. As a lowcost solution.1 Power Pattern Phase Balance Comparison The following section compares the results of the NEC model that has been feed with the actual phase errors of each modeformer to the measured results of the Archimedean spiral antenna.2.
25 GHz) (NEC is Phase and Amplitude Compensated) Figure 5.11 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.5 GHz) (NEC is Phase and Amplitude Compensated) 114 .2.2.12 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.Figure 5.
Figure 5.2.13 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz) (NEC is Phase and Amplitude Compensated)
Figure 5.2.14 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (2 GHz) (NEC is Phase and Amplitude Compensated)
115
Figure 5.2.15 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.25 GHz) (NEC is Phase and Amplitude Compensated)
Figure 5.2.16 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.25 GHz) (NEC is Phase and Amplitude Compensated)
116
Figure 5.2.17 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.5 GHz) (NEC is Phase and Amplitude Compensated)
Figure 5.2.18 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz) (NEC is Phase and Amplitude Compensated)
117
Figure 5.110 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.19 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (2 GHz) (NEC is Phase and Amplitude Compensated) Figure 5.25 GHz) (NEC is Phase and Amplitude Compensated) 118 .2.2.
3 Spiral Antenna Results The following sections describe the results of each antenna that was fabricated for this thesis. It should be noted that the NEC results are for an antenna in free space as depicted in the theoretical section. The mode one simulated results loosely agree with the measured results for frequencies at and below 1.75 GHz. 119 . The availability of the electromagnetics software due to cost or licensing rules prevented these characteristics from being modeled and therefore is not a part of the comparison. an enclosure lined with microwave absorber instead of free space. 5. Mode two begins with a stronger power pattern than what NEC predicts but settles into a closer agreement by 1. The best situation for comparing the theoretical results to the measured results would be to construct a model that includes strips printed on a dielectric sheet instead of wires. and the ability to create a ground plane aperture.The previous figures illustrate the increased elliptical polarization that occurs when the modeformer is not providing the ideal phase progression to the antenna. All of the measured results have been plotted with the results of the NEC simulation (with perfect phase) to provide a comparison between the measured and theoretical results. It should also be noted that only φ cuts were measured for antennas not mounted in ground plane apertures.75 GHz.
5. 120 . Figure 5.11 Archimedean Spiral Mounted in Test Fixture The following figure compares the mode one maximum vertically polarized gain of the measured antenna and the NEC simulated antenna versus frequency. The following figure depicts the Archimedean spiral mounted on the DAMS platform.3.1 Archimedean Spiral The following section compares the measured results of the Archimedean spiral to the NEC simulated results.3.
12 Measured and Simulated Mode 1 Archimedean Spiral Maximum Vertically Polarized Gain versus Frequency As the figure shows.3. the maximum gain of the antenna oscillates about the theoretical curve up to 2 GHz and then falls off afterward. 121 . This is most likely due to the Standing Wave Ratio (SWR) of the input impedance oscillating around 50 ohms and the depth of the antenna enclosure favoring a lower frequency response.Figure 5. The following figures depict the power patterns of the spiral for mode one and mode two.
Figure 5.3.75 GHz) 122 .13 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 5.14 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.3.
16 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz) 123 .Figure 5.15 Measured and Simulated Mode 1 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.3.3.5 GHz) Figure 5.
18 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (2.17 Measured and Simulated Mode 2 Archimedean Spiral Power Pattern (dBi) φ Cuts (1.3.75 GHz) Figure 5.3.Figure 5.5 GHz) 124 .
2 CylindricallyConformed Archimedean Spiral The following section compares the measured results of the cylindricallyconformed Archimedean spiral to the NEC simulated results.21 Archimedean Spiral Mounted in Test Fixture The following figure compares the mode one maximum vertically polarized gain of the measured antenna and the NEC simulated antenna versus frequency. The following figure depicts the cylindricallyconformed Archimedean spiral mounted on the DAMS platform.3. the strongest agreement between measured and theoretical results occurred at 1.3. 5.75 GHz. Not surprisingly. 125 . Figure 5.The previous figures illustrate the increased elliptical polarization at lower frequencies for the mode one power pattern.
Figure 5. the maximum gain of the antenna oscillates about the theoretical curve up to 2 GHz and then falls off afterward.3.22 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral Maximum Vertically Polarized Gain versus Frequency Like the flat geometry. 126 . The following figures depict the power patterns of the spiral for mode one and mode two.
24 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1.3.23 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 5.3.75 GHz) 127 .Figure 5.
5 GHz) Figure 5.25 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (2.Figure 5.3.3.26 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1 GHz) 128 .
Figure 5.3.27 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (1.75 GHz)
Figure 5.3.28 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral Power Pattern (dBi) φ Cuts (2.5 GHz)
129
The previous figures again illustrate the strong agreement between measured and theoretical results at 1.75 GHz. It is also seen that, like the theoretical model predicted, the power patterns between the flat and conformed cases have not been altered.
5.3.3 Square Spiral
The following section compares the measured results of the square spiral to the NEC simulated results. The following figure depicts the square spiral mounted on the DAMS platform.
Figure 5.3.31 Square Spiral Mounted in Test Fixture
The following figure compares the mode one maximum vertically polarized gain of the measured antenna and the NEC simulated antenna versus frequency.
130
Figure 5.3.32 Measured and Simulated Mode 1 Square Spiral Maximum Vertically Polarized Gain versus Frequency
Like the previous antenna geometries, the maximum gain of the antenna oscillates about the theoretical curve up to 2 GHz and then falls off afterward. The following figures depict the power patterns of the spiral for mode one and mode two.
131
3.75 GHz) 132 .34 Measured and Simulated Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (1.3.Figure 5.33 Measured and Simulated Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 5.
5 GHz) Figure 5.3.Figure 5.3.36 Measured and Simulated Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (1 GHz) 133 .35 Measured and Simulated Mode 1 Square Spiral Power Pattern (dBi) φ Cuts (2.
38 Measured and Simulated Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (2.37 Measured and Simulated Mode 2 Square Spiral Power Pattern (dBi) φ Cuts (1.75 GHz) Figure 5.5 GHz) 134 .Figure 5.3.3.
4 CylindricallyConformed Square Spiral The following section compares the measured results of the cylindricallyconformed square spiral to the NEC simulated results. where the square spiral has surprisingly provided the most concentric mode one power pattern of all of the antenna geometries. 5. 135 .The previous figures again illustrate the strong agreement between measured and theoretical results at 1.3.41 CylindricallyConformed Square Spiral Mounted in Test Fixture The following figure compares the mode one maximum vertically polarized gain of the measured antenna and the NEC simulated antenna versus frequency.3. Figure 5.75 GHz. The following figure depicts the cylindricallyconformed square spiral mounted on the DAMS platform.
The following figures depict the power patterns of the spiral for mode one and mode two.42 Measured and Simulated Mode 1 CylindricallyConformed Square Spiral Maximum Vertically Polarized Gain versus Frequency The shape of the maximum gain curve of the antenna with respect to frequency remains unchanged compared to the other measured antenna results which show that.Figure 5. 136 .3. for at least lower frequencies. This also happens to be the conclusion found within the theoretical results. the square and Archimedean spiral generally perform equally well.
43 Measured and Simulated Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1 GHz) Figure 5.75 GHz) 137 .Figure 5.3.44 Measured and Simulated Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1.3.
3.46 Measured and Simulated Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1 GHz) 138 .Figure 5.3.45 Measured and Simulated Mode 1 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (2.5 GHz) Figure 5.
Figure 5.48 Measured and Simulated Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (2.75 GHz) Figure 5.5 GHz) 139 .47 Measured and Simulated Mode 2 CylindricallyConformed Square Spiral Power Pattern (dBi) φ Cuts (1.3.3.
51 Square Spiral Mounted within Ground Plane Aperture The following figure compares the mode one maximum vertically polarized gain of the measured antenna and the NEC simulated antenna versus frequency.75 GHz.5 Square Spiral within Ground Plane Aperture The following section compares the measured results of the square spiral mounted within a ground plane aperture to the NEC simulated results. The following figure depicts the square spiral mounted within the ground plane aperture.The previous figures again illustrate the strong agreement between measured and theoretical results at 1.3. Figure 5. although the conforming of the antenna has created a slightly elliptical polarization on the mode one pattern and the maximum gain at boresight is reduced. 140 .3. 5.
Figure 5. 141 .3. The following figures depict the power patterns of the spiral for mode one and mode two.52 Measured and Simulated Mode 1 Square Spiral Maximum Vertically Polarized Gain versus Frequency The shape of the maximum gain curve has smoothed out during the ground plane aperture case which suggests that the enclosure is affecting the maximum gain power patterns of the previous antenna geometries.
54 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (1 GHz) 142 .3.53 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (1 GHz) Figure 5.Figure 5.3.
56 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (1.55 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (1.Figure 5.3.75 GHz) Figure 5.75 GHz) 143 .3.
5 GHz) 144 .57 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (2.3.5 GHz) Figure 5.58 Measured and Simulated Mode 1 Square Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (2.Figure 5.3.
3.510 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (1 GHz) 145 .Figure 5.59 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (1 GHz) Figure 5.3.
75 GHz) 146 .512 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (1.75 GHz) Figure 5.3.3.Figure 5.511 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (1.
3.513 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (2.Figure 5.5 GHz) 147 .3.5 GHz) Figure 5.514 Measured and Simulated Mode 2 Square Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (2.
This forced a single measurement to be used on the φ cuts where there should have been separate measurements for each angle to depict the polarization difference between each cut. the ground plane aperture has smoothed out the power patterns at each frequency and mode. The following figure depicts the cylindricallyconformed Archimedean spiral mounted within the ground plane aperture. The pointed tip at boresight for mode one is caused by the antenna not being rotated when the boresight measurement was made. 148 . 5.The previous figures illustrate that along with smoothing out the maximum gain with respect to frequency.3.6 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture The following section compares the measured results of the cylindricallyconformed Archimedean spiral mounted within a ground plane aperture to the NEC simulated results.
Figure 5.3. 149 .61 CylindricallyConformed Archimedean Spiral Mounted within Ground Plane Aperture The following figure depicts the mode one maximum vertically polarized gain of the measured antenna and the NEC simulated antenna versus frequency.
The following figures depict the power patterns of the spiral for mode one and mode two. 150 .62 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Maximum Vertically Polarized Gain versus Frequency The shape of the maximum gain curve agrees with the square spiral case which further suggests that the ground plane aperture drives the measured results.Figure 5.3.
63 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (1 GHz) Figure 5.Figure 5.3.3.64 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (1 GHz) 151 .
65 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (1.3.66 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (1.Figure 5.3.75 GHz) 152 .75 GHz) Figure 5.
3.3.5 GHz) Figure 5.5 GHz) 153 .67 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (2.Figure 5.68 Measured and Simulated Mode 1 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (2.
3.69 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (1 GHz) Figure 5.3.610 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (1 GHz) 154 .Figure 5.
611 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (1.612 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (1.75 GHz) 155 .3.Figure 5.75 GHz) Figure 5.3.
613 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) φ Cuts (2.614 Measured and Simulated Mode 2 CylindricallyConformed Archimedean Spiral within Ground Plane Aperture Power Pattern (dBi) θ Cuts (2.5 GHz) Figure 5.3.3.Figure 5.5 GHz) 156 .
157 .75 GHz.The previous figures illustrate that at frequencies below 1. the measured antenna power pattern trends closer to the simulated case in each geometry that was discussed within this thesis.
The theoretical model did show that the power pattern changes more rapidly when the phase progression at the feed of the antenna is not ideal. different types of frequencyindependent antennas were discussed along with the theory of frequencyindependence. at higher frequencies. power patterns of each antenna geometry was measured. CONCLUSIONS 6. If a higher quality modeformer was designed to be used in conjunction with this antenna design. the square spiral begins to diverge away from the results of the Archimedean spiral. On the other hand. After fabricating each antenna type. The theoretical model also proved that changing the radius of curvature for each spiral type does not change the shape of the power pattern until a very small radius of curvature (6 cm) is used.6. These measured results demonstrated that an antenna provides better results 158 . It was found that a fourarm spiral antenna offered the best design choice for being integrated into a directionfinding system. a viable lowcost and lightweight solution would be realized.1 Summary of Work Within this thesis. Two different spiral types were used to design antennas that had a materials cost of less than $300 and could be fabricated within a week by a technician. Each of the antenna geometries were modeled in NEC to show that the difference between spiral types does not change the power pattern of either antenna at lower frequencies.
This conclusion agrees with the theoretical model to suggest that spiral type and conformation of the antenna surface does not change the overall power pattern as much as connecting the antenna to a less than an ideal modeformer.when a modeformer is producing a phase progression that is too short for a given frequency. Such an antenna design would not need an impedance matching circuit and therefore offer an excellent SWR response with respect to frequency. Another path of research might be to explore different antenna geometry types. One possible path of research might be to create antennas of different radiuses of curvature to continue proving the invariance of spiral antennas to most conformations. such as the logarithmic or equiangular spiral. 159 . the recommendations for further study are almost limitless. Along with different antenna types. 6.2 Recommendations for Further Study As with any antenna study. the number of arms could be explored along with different frequencies of operation. One particular antenna of interest to the author was the eightarm selfcomplementary spiral antenna which has a characteristic input impedance of nearly 50 ohms. the power pattern quickly deteriorates. At higher frequencies where the phase was much longer than needed.
Virginia Polytechnic Institute.. New York: McGrawHill Inc. Stutzman. Massachusetts: Artech House.. 1957.. Blacksburg Virginia. W. and Pasala. J. M. A. Design and Analysis of Star Spiral with Application to Wideband Arrays with Variable Element Sizes. and Poggio. R. pages 497652.. J. Lawrence Livermore Laboratory Report. 2005. Rumsey.REFERENCES [1] Balanis. 1993.Theory.. pages 260305. R. 3rd Edition. Numerical Electromagnetics Code (NEC) Program Description .1468. V. Dissertation. A. December 1962. pages 114118.. pages 151186. K... January 1981. pages 123133. pages 254309 and 670742. Caswell. C. and Mosko.. New Jersey: John Wiley & Sons Inc. pages 179. 1981. A. New Jersey: John Wiley & Sons Inc. 3rd Edition. Number 1. 1990.. Elliot.. Johnson. 1999. Antenna Theory and Design. 2001. Penno. 1998. Antenna Engineering Handbook. D. IEEE Transactions on Aerospace and Electronic Systems.. Antenna Theory: Analysis and Design. pages 141 . FourArm Spiral Antennas. and Thiele. 160 . G. A View of Frequency Independent Antennas. 1st Edition. C. New Jersey: Princeton University Press. Microwave Journal. January 2001. C. Volume 37. [2] [3] [4] [5] [6] [7] [8] [9] [10] Balanis. R.. 1957 IRE National Convention Record. R. E. G. A. Theory of Angle Estimation Using a Multiarm Spiral Antenna. [11] Burke. S. Corzine.. L. Advanced Engineering Electromagnetics. G. J. A. P. Gnomon: From Pharaohs to Fractals. H. New York: John Wiley & Sons Inc. Gazale. Part 1. M. J. pages 6168.. Frequency Independent Antennas.
1973. R. [13] Mittra...S. D. M.[12] Caswell. Computer Techniques for Electromagnetics. E. 1998. Analysis of a Helix Antenna Using a Moment Method Approach With Curved Basis and Testing Functions. New York: Pergamon. Virginia Polytechnic Institute. Blacksburg Virginia. Thesis. 161 .
This action might not be possible to undo. Are you sure you want to continue?