Antenna

i

THE UNIVERSITY OF QUEENSLAND

School of Information Technology and Electrical Engineering

Design of Conformal Antennas

for Telephone Handsets

Bachelor of Engineering

Honours Thesis

By

Andrew James Causley

2002

ii

2/48 Beatrice St,

Taringa, QLD 4068.

4th October 2002.

The Dean,

Faculty of Engineering,

The University of Queensland,

St Lucia, QLD 4072.

Dear Sir,

In accordance with the requirements for the Degree of Bachelor of Engineering

(Honours), I hereby submit for your consideration the following thesis entitled “Design

of Conformal Antennas for Telephone Handsets”. This work was performed under the

supervision of Associate Professor Marek E. Bialkowski. I declare that the work

submitted in this thesis is my own, except as acknowledged in the text and footnotes, and

has not been previously submitted for a degree at the University of Queensland or any

other institution.

Yours Faithfully,

Andrew Causley

iii

Acknowledgments

I would like to extend my appreciation to everybody who has assisted me with

this thesis throughout the course of the year. I am most grateful to Associate Professor

Bialkowski for his continual guidance and support needed in achieving this thesis.

Special thanks also goes to fellow undergraduate microwave thesis students, (who are

also under supervision by Professor Bialkowski,) in particular Mr. Eng Hoe Yap and Mr.

Khoo Choon Wee for their timely but nevertheless vital assistance. Finally, I would like

to thank my family and friends, for their invaluable support during the past four years.

iv

Abstract

Modern cellular phone systems currently operate at a number of frequency bands,

the most common being 900MHz, 1.8GHz and 2.0GHz. As mobile phones are becoming

smaller it is not feasible to simply equip the handset with an array of antennas, each tuned

to a specific frequency band. This has resulted in a demand for antennas that can operate

at multiple bands without the need for multiple antennas. The Inverted-F Antenna (IFA)

has been demonstrated to be capable of operating at multiple frequencies with the initial

results very encouraging.

Another approach available in dealing with the ever-diminishing space available

on handsets is to conform the antenna. This involves wrapping the microstrip antenna

around a cylindrical surface. The results of this approach using the IFA are encouraging,

especially in terms of space minimization, bandwidth enhancement and return loss

reduction.

The purpose of this thesis is to design a conformal antenna for a mobile handset.

The first step was to choose a suitable planar triple-band antenna capable of operating at

900MHz, 1.8GHz and 2.0GHz. The antenna that was selected for this thesis is the E-

shaped planar IFA. This antenna can easily be tuned for correct frequency operation and

possess a compact design. The design was created using FEKO, an electromagnetic

simulation software package. Once the planar version of the antenna was simulated, the

antenna was made conformal.

v

By using the optimised dimensions obtained from the planar IFA and applying

them to the conformal IFA, it was found that there was a reduction in quality of the far-

field patterns and the return loss values for similar resonant frequencies. Therefore further

work is required to refine the conformal IFA.

Table of Contents

vi

Table of Contents

CHAPTER 1: INTRODUCTION AND OVERVIEW .............................................12

1.1 Introduction .....................................................................................................12

1.2 Objectives.........................................................................................................16

1.3 Thesis Overview ...............................................................................................17

CHAPTER 2 LITERATURE REVIEW.................................................................19

2.1 Introduction .....................................................................................................19

2.2 Review of relevant articles...............................................................................20

2.2.1Analysis of microstrip antennas on a curved surface using the conformal

grids FD-TD method ................................................................................20

2.2.2 The Effect of Conformality on the Electrical Properties of Small

Antennas...................................................................................................22

2.2.3 Triple band planar inverted F antennas for handheld devices..............24

2.2.4 E-Shaped planar inverted-F antenna with low-absorption for triple-

band operation in cellular phones ...........................................................26

Table of Contents

vii

CHAPTER 3: THE PLANAR INVERTED-F ANTENNA......................................29

3.1 Introduction .....................................................................................................29

3.2 The Derivation of the PIFA .............................................................................30

3.3 Spatial Network Method (SNM) analysis of the PIFA ...................................33

3.3.1 The Electric Field and Current Distributions of the PIFA....................33

3.4 Resonant Frequency ........................................................................................37

3.5 The Bandwidth.................................................................................................39

3.6 The Triple Band E-shaped PIFA Antenna......................................................40

3.6.1 The E-PIFA and shorting pins................................................................41

3.6.2 The E-PIFA and slot lengths...................................................................42

CHAPTER 4: METHODOLOGY.........................................................................44

4.1 Introduction .....................................................................................................44

4.2 The E-shaped PIFA and the simulations.........................................................45

4.3 The Simulation Software: FEKO ....................................................................48

4.3.1 The Components Of FEKO ....................................................................49

4.4 The Experimental Process ...............................................................................52

4.4.1 Development of the planar E-shaped IFA..............................................52

4.4.2 Development of the conformal E-shaped IFA........................................56

Table of Contents

viii

CHAPTER 5: RESULTS AND DISCUSSION.....................................................59

5.1 Introduction .....................................................................................................59

5.2 The Planar E-shaped IFA................................................................................60

5.3 The Conformal E-shaped IFA of radius 100 mm. ..........................................68

5.4 The Conformal E-shaped IFA of radius 150 mm. ..........................................75

5.5 Discussion .........................................................................................................82

5.5.1 The Return Losses...................................................................................82

5.5.2 The Three-dimensional Far Field Patterns ............................................82

5.5.3 The Two-dimensional Far Field Patterns...............................................86

CHAPTER 6: CONCLUSIONS...........................................................................89

REFERENCES ...................................................................................................92

APPENDIX..........................................................................................................94

The conformal antenna development process using FEKO .......................................94

A.1. Initialising the nessercary variables ................................................................95

A.2. Creating the radiating element......................................................................103

A.3. Creating the ground plane.............................................................................112

A.4 Adding the shorting posts and the feed probe ..............................................115

A.5 Compiling the code and implementing the simulation .................................116

List of Figures

ix

List of Figures

Figure 3.1: The Inverted-L Antenna...............................................................................30

Figure 3.2: The Inverted-F Antenna...............................................................................31

Figure 3.3: The Planar Inverted-F Antenna. ...................................................................32

Figure 3.4: The structure of the PIFA as used in the SNM analysis. ...............................34

Figure 3.5: The Electric Field Distribution of the PIFA at the x-y plane.........................35

Figure 3.6: Surface Current distributions for varying short circuit plate widths. .............36

Figure 3.7: The effect of the short circuit plate width with relation to the frequency. .....37

Figure 3.8: The effect of the planar element width with relation to the frequency...........38

Figure 3.9: The effect of the antenna height on the bandwidth. ......................................39

Figure 3.10: Top view of the E-shaped PIFA.................................................................40

Figure 3.11: Return losses versus shorting positions. L1 =38.2 mm, L2=37.8 mm. .........41

Figure 3.12: Return losses versus slot length L1. ............................................................42

Figure 3.13: Return losses versus slot length L2. ............................................................43

Figure 4.1: The Far-Field Radiation patterns of the E-shaped PIFA in the horizontal plane

with (a) representing 900 MHz, (b) 1800 MHz and (c) 2000 MHZ.........................47

Figure 4.2: The WinFEKO operating environment. ........................................................51

Figure 4.3: The Planar E-shaped PIFA with the initial dimensions. ................................53

Figure 4.4: The Optimised Planar E-shaped IFA............................................................55

Figure 4.5: The Conformal E-shaped IFA with a radius of 100 mm. ..............................57

Figure 4.6: The Conformal E-shaped IFA with a radius of 150 mm. ..............................58

List of Figures

x

Figure 5.1: The simulated return losses for the Planar E-shaped IFA. ............................60

Figure 5.2: The simulated 3D Far Field Radiation patterns for the Planar E-shaped IFA at

950 MHz. ..............................................................................................................62


1775 MHz..............................................................................................................63


2000 MHz..............................................................................................................64


950 MHz. ..............................................................................................................65


1775 MHz..............................................................................................................66


2000 MHz..............................................................................................................67

Figure 5.8: The simulated return losses for the Conformal E-shaped IFA of radius 100

mm. .......................................................................................................................68

Figure 5.9: The simulated 3D Far Field Radiation patterns for the Conformal E-shaped

IFA of radius 100 mm at 950 MHz. .......................................................................69


IFA of radius 100 mm at 1775 MHz. .....................................................................70

Figure 5.11: The simulated 3D Far Field Radiation patterns or the Conformal E-shaped


List of Figures

xi







Figure 5.15: The simulated return losses for the Conformal E-shaped IFA of radius 150

mm. .......................................................................................................................75













Chapter 1: Introduction and Overview

12

CHAPTER 1

Introduction and Overview

1.1 Introduction

The foundations for the wireless communication industry were established in 1864,

when James Maxwell noted that the engagement of electric and magnetic waves could

allow energy to be transported through materials and space at a finite velocity [1].

Heinrich Hertz demonstrated Maxwell’s theory of electromagnetic radiation in 1888 by

launching electromagnetic waves using a spark transmitter and a Rhumkoff inductor.

Hertz’s apparatus demonstrated the first transmission of regulated radio waves [2].

Guglielmo Marconi, dubbed the father of the wireless communications industry,

took the discoveries of Maxwell and Hertz further. It was in 1897 that Marconi

demonstrated the practical applications of wireless communication, when he established

continuous radio contact between the shore and ships travelling in the English Channel

[3]. By mid December in 1901, Marconi took a much greater step by performing the first

transcontinental wireless communication, in this case between England and Canada. This

achievement created the potential for worldwide wireless communication [3].


13

Over the next hundred years, the wireless communications industry has gone from

strength to strength, with a continuous stream of discoveries advancing the industry at a

tremendous pace. From 1904 to 1906, the development of the vacuum tube diode

occurred, establishing the foundations for RF detectors, oscillator circuits and amplifiers,

which are vital for the radio transmission and receiver circuitry. In this same period, the

heterodyne receiver circuit was invented, paving the way for speech and music radio

transmission [4].

For the first half of the 20th century the mobile communication industry was strictly

limited by the size and weight of the early mobile phone systems [2]. Such phones,

normally based on the simple Amplitude Modulation (AM) techniques, were limited to

use by the military and emergency services. However, the demands of World War II

catalysed the development of mobile communications and it was the development of the

transistor in the early 1960’s that greatly reduced the power and space demand of mobile

radio systems. This trend of shrinking transceiver size, established by the advent of the

transistor, has continued unabated to the present day [1]. Coupled to this trend has been

the unrelenting boom in the personal mobile communications industry. In Australia for

example, the number of mobile phones in this country now exceeds the total number of

fixed landlines.

As the mobile phone continues its downward spiral in terms of size, it is worth

noting that there is also an increasing demand on the mobile phone’s performance. The

modern mobile phone now operates digitally, thus radically improving noise immunity,

offering improved multiplexing capacity, error checking, source encoding, encryption,


14

equalisation as well a host of other powerful functions [3]. Complex signal processing

techniques and the cellular system have also addressed the issue of frequency congestion

[2]. However, the one main issue concerning mobile communications that has only

recently been examined is the performance of the mobile phone’s antenna.

Traditionally most mobile phones and handsets have been equipped with the

monopole antenna. Monopole antennas are very simple in design and construction and

are well suited to mobile communication applications. The most common λ/4 monopole

antenna is the whip antenna, which can operate at a range of frequencies and can deal

with most environmental conditions better than other monopole antennas. However, the

monopole antenna possesses a number of drawbacks. Monopole antennas are relatively

large in size and protrude from the handset case in an awkward way. This problem with

the monopole’s obstructive and space demanding structure also complicate any efforts

taken to equip a handset with several antennas to enable multiband operation. Monopole

antennas also lack any built-in shielding mechanisms, to direct any radiating waves away

from the user’s body, thus increasing the potential risk of producing cancerous tumour

growth in the user’s head and reducing the antenna efficiency.

There are a wide variety of methods being investigated to deal with the deficiencies

of the common λ/4 monopole antenna, many of these methods being based on microstrip

antenna designs. One such promising design is the Inverted-F Antenna (IFA), a distant

derivative of the monopole antenna. The IFA utilises a modified Inverted-L low profile

structure, as has frequently been used for aerospace applications. The common IFA

possesses a rectangular element with an omnidirectional radiation pattern and has


15

exhibited a reasonably high gain. The bandwidth of the IFA is also broad enough for

mobile operation, and is also highly sensitive to both vertically and horizontally polarized

radio waves, thus making the IFA ideally suited to mobile applications.

As well as the benefits described before, the IFA was chosen as the basis of this

thesis, because of its ability to facilitate multiband operation [6], [7]. Cellular phone

systems currently operate at a number of frequency bands (such as 900MHz, 1.8GHz and

2.0GHz), thus there being a demand for antennas that can operate at multiple bands. As

mobile phones are becoming smaller with time, it is not feasible for separate antenna

elements to be used to facilitate multiband operation. The PIFA has been demonstrated to

be able to operate at multiple frequencies and the results to date have been quite

encouraging.

As the available space for antennas on mobile handsets continues to be minimised,

other approaches for minimising space demand must be examined. One approach is the

utilisation of conformal antenna design. This involves wrapping the microstrip antenna

around a cylindrical surface [8]. The results of this approach so far, using the conformal

IFA are encouraging, in terms of space minimisation, bandwidth enhancement and return

loss reduction.


16

1.2 Objectives

The fundamental aim of this thesis is to design a conformal antenna suitable for

telephone handsets. By using a conformal antenna, the space demand of the antenna as

part of a telephone handset can be minimised, thus reducing the obtrusiveness of the

handset’s appearance.

This thesis has three primary goals:

1. Select and design an efficient, low profile and realisable antenna capable of

operating at a number of frequency bands (900 MHz, 1.8 GHz and 2.0 GHz).

2. Verify the operation of the antenna at the prescribed frequencies in terms of input

impedance and field patterns, using the antenna design software package FEKO.

3. Investigate the effects of conformality on the antenna in terms of performance.

In order to achieve the first goal as set out above, a comprehensive literature is

required to obtain an antenna that requires minimal modification to suit the requirements

of this thesis. As the process of optimising an antenna’s dimensions to meet a set of

specifications is highly rigorous, finding an antenna that operates efficiently at the three

required frequencies, as well being compact and having a low profile, is very much

desired.


17

1.3 Thesis Overview

The literature review is performed in Chapter 2 of this thesis. The articles that are

reviewed in this chapter deal with triple-band planar IFAs and the effects of conformality.

Three triple-band planar IFAs in total are proposed in two of the articles and their

performance is gauged in terms of input impedance, field patterns and gain. A review of a

number of papers explaining the effects of conformality on planar IFAs and other

microstrip antennas is also presented with brief summaries.

The theory and fundamentals behind the planar IFA (PIFA) are analysed in Chapter

3. An extensive examination of the PIFA is performed as well as a study into the

proposed triple-band planar IFA that will form the basis of this thesis. The majority of the

theory presented as part of this chapter is from the work of Dou [7] and Reidy [9].

The methodology, used as part of this thesis for the simulation of results of the

proposed antennas, is provided in Chapter 4. A brief introduction into the

Electromagnetic simulation software, FEKO, is included in this chapter together with an

examination of the antenna configurations simulated as part of this thesis.


18

Chapter 5 consists of the results derived from the methods explained in Chapter 4.

The return loss characteristics and the far field patterns are presented in this chapter. The

far field patterns are displayed in this chapter in both two-dimensional and three-

dimensional graphs. A discussion of the results is also provided towards the end of this

chapter.

Chapter 6 forms the conclusion of the said thesis.

Chapter 2: Literature Review

19

CHAPTER 2

Literature Review

2.1 Introduction

This section of the thesis reviews a number of papers, which have formed the basis

for the research component of this thesis. The first two papers provide an insight into the

background of conformality and its effects on microstrip antennas. The final two papers

detail a number of triple band microstrip antennas, which establishes a formative step into

the analysis of such antennas. Of particular importance is the paper by W. Dou and

M.Y.W. Chia, detailing a triple band planar inverted-F antenna. This particular antenna

has formed the basis for this thesis and the paper by Dou and Chia provides a detailed

insight into the design and performance of the antenna.


20

2.2 Review of relevant articles

2.2.1 Analysis of microstrip antennas on a curved surface using the conformal

grids FD-TD method

T. Kashiwa, T. Onishi, I. Fukai

The need for small, portable antennas for mobile communications has spurred the

study of microstrip antennas. Microstrip antennas are quite flexible and have been used as

conformal antennas on arbitrary curved surfaces. The characteristics of conformal

microstrip antennas can be expected to differ from those of planar models. One technique

that offers versatility when analysing conformal microstrip antennas is the finite-

difference time domain method (FD-TD). This method utilises the direct discretisation of

Maxwell’s equations in the time and space domains and is well suited to the analysis of

microstrip antennas. However, when irregularly shaped structures, such as conformal

structures are to be analysed, the FD-TD method demands large amounts of time and

memory to deal with such structures. This problem has been overcome by developing

generalised derivatives of the FD-TD method to operate with curvilinear coordinates as

found in conformal microstrip antennas. These generalised FD-TD methods suited for

curvilinear coordinates can fit the complex boundary shape with more accuracy than

conventional FD-TD methods and can also be applied to CAD applications in


21

conjunction with a mesh generator. This paper examined the analysis of curved

microstrip antennas by the curvilinear FD-TD method.

This paper utilised a generic planar microstrip antenna for obtaining the simulation

results to be used as a reference. The same microstrip antenna was then simulated using

the curvilinear FD-TD techniques and the cylindrical radius of curvature was set at two

values: 30 mm and 60 mm.

The results of the comparison between the planar antenna and the two curved

antennas found that as the radius of curvature was reduced, the antenna conformed more

closely to the free space impedance. Also, the bandwidth appeared to widen. With respect

to the radiation patterns for the above antennas, the paper found that, as the patched

antenna became increasingly curved, the directivity of the antenna became more

predominant in the direction away from the radiating element side of the antenna.


22

2.2.2 The Effect of Conformality on the Electrical Properties of Small Antennas

Pawel Kabicik, Andrzej Kucharshki

Within the current mobile and wireless communications market, there is a strong

consumer push for handsets to possess a range of available functions and styled cases.

Modern multimedia handsets also call for multiband antennas that should feature

optimised radiation properties and must be scarce in space demand. As a result, it would

be favourable for antenna designers to create antennas to utilise any arbitrary space

available in small handsets, specifically in the upper parts of the handset. One technique

to deal with this problem, as discussed in this paper, is to use conformally shaped

antennas, which can be integrated into hand held terminals in a more flexible way.

This paper proposed two types of conformality: terminal antennas with ground and

radiating plates that are smoothly curved over the dominant surface area, or antennas bent

anywhere by a right angle or slightly smaller angle (curvature having a relatively small

radius). All simulations carried out as part of this paper utilised Concept software, which

uses the Moments Method (explained in Chapter 4). The paper focussed on the 900 MHz

frequency band antennas. Planar antennas were simulated as reference against conformal

antennas of identical configuration. The ground plane cylindrical radius approached

60mm for the simulations.


23

In all cases presented in this paper, the return losses for the conformal antenna were

improved over the reference antenna by 1 or 2 dB. However, in the bent configuration

antenna case, the results were mixed although it was found that the impedance matching

improved when the antenna bending occurred close to the feed probe. This paper also

found that the antenna performance was particularly sensitive to the ground plane size

when conformality is applied.

The adoption of conformality was also found to influence the shape of the radiation

pattern although little attention was paid to this issue in this paper as the actual

concentration of radiated energy is affected by scattering phenomena. This paper also

found that there are no particular losses attributable solely to the conformal shape of an

antenna.

This paper concluded with the discovery that the inclusion of conformality in the

design of an antenna does not result in a marginal effect on the antenna bandwidth and in

some cases actually accounts for the broadening of the bandwidth.


24

2.2.3 Triple band planar inverted F antennas for handheld devices

C.T.P. Song, P.S. Hall, H. Ghafouri-Shiraz and D. Wake

The recent explosion in the communications market has resulted in the emergence of

a number of systems operating at different frequencies. The planar inverted-F antenna

(PIFA), which can be implemented in a small space, is an attractive candidate for the

incorporation of multi-band operation. This paper describes two novel methods for

designing a compact, triple-band, dual-feed PIFA. Both antennas have been tuned to

operate at centre frequencies of 0.9. 1.89 and 2.44 GHz and are both capable of being

implemented as mobile terminals.

The first triple band PIFA proposed by this paper utilises a dual-band single-feed

PIFA as proposed earlier, modified to a triple band configuration. This antenna has been

shown to be capable of matching to a frequency ratio of 1.45 – 3.25, without having to

vary the feed position. When the simulation of this antenna was performed, it was found

that there was an isolation of –20dB achieved between the two feed ports. The radiation

patterns produced demonstrate high vertical and horizontal field elements, and hence the

antenna is capable of both polarisations.

The second design was achieved by embedding two single resonance PIFAs within

a quarter-wave patch antenna. Unlike the first antenna, this second antenna consists of 3


25

feeds and possesses a different radiating element configuration. An isolation of -15dB

was achieved between the three ports when the antenna was simulated. The radiation

patterns obtained for this antenna showed a slight squint due to the proximity of other

radiating elements.

This paper also noted the important fact that the positioning and the dimensions of

the ground plane can influence the radiation patterns of the above antennas. Both

directional and omni-directional applications can be tailored by suitably adjusting this

ground plane.


26

2.2.4 E-Shaped planar inverted-F antenna with low-absorption for triple-band

operation in cellular phones

W. Dou, M.Y.W. Chia

With the advent and popularity of portable communication devices, health issues

regarding the use of radiation waves within close proximity to users have been of grave

concern. There is concern that the radiation power dissipated by contemporary antennas,

notably the whip antennas, may cause potential harm to the users health. The planar

inverted-F antenna (PIFA) has been proposed to address these health concerns. The

PIFAs metallic ground plane acts as a shield to prevent the electromagnetic (EM) waves

from radiating to the user, thus minimising energy dissipation in the user’s biological

tissues. Worth mentioning also is the fact that it is compact, low cost, robust and its low

profile structure is preferable for portable applications.

In order to design a compact, single feed low profile antenna for triple-band

operation, this paper proposed a novel slotted PIFA. The design of this antenna consisted

of two narrow linear slots etched into the radiating element of the patch. As the resulting

slotted antenna looks like the letter E, this paper referred to the antenna as an E-shaped

PIFA. The antenna was tuned to operate at GSM 900 MHz, 1.8 GHz and 1.9 GHz. Two

shorting strips, located close to the feed, were also included in the design for improved


27

matching at the triple bands. The E-shaped PIFA also has the same size as a

conventional 900 MHz PIFA while being able to operate at three resonant frequencies.

The paper examined the effects on the performance of the antenna by adjusting the

two linear slots. It was discovered that, by altering the lengths of these slots, the antenna

could be tuned into the second and third centre frequencies (1800 MHz and 1900 MHz).

This is particularly useful as it allows the antenna to be easily adaptable to operate at

other centre frequencies. The first centre frequency (900 MHz) is roughly governed by

the length and width of the overall radiating element of the antenna.

Once the optimisation of the antennas dimensions was completed, the antenna in

the paper had it’s operating frequencies centred at 902 MHz, 1.84 GHz and 1.94 GHz

with bandwidths of 48.8 MHz, 36.6 MHz and 61.0 MHz respectively. These results were

similar to the results obtained by Song and his triple band antenna that utilised two or

three feeds.

The far field patterns for E-shaped PIFA for each operating frequency demonstrated

that more energy radiated in the opposite direction of the ground plane. With the higher

centre frequencies, even less energy was radiated towards the ground plane, indicating

that the ground effect is stronger at higher frequencies. Regardless, the radiation patterns

as described in this paper indicate an approximate omnidirectional pattern at each of the

three frequencies. Therefore the E-shaped PIFA is very versatile for portable applications


28

where a large broad beamwidth is essential. At the front bore sight of the antenna, the

gains obtained are as follows: 4.7 dBi at 902 MHz, 1.3 dBi at 1.84 Hz and 0.8 dBi at 1.94

GHz.

Chapter 3: The Planar Inverted-F Antenna

29

CHAPTER 3

The Planar Inverted-F Antenna

3.1 Introduction

The planar inverted-F antenna is a microstrip antenna design that shows much

promise in dealing with the shortfalls of the λ/4 monopole antenna in mobile

communication applications. The planar inverted-F antenna has exhibited a reasonably

high gain and possesses a bandwidth that is broad enough for mobile operation. The

antenna also has a high degree of sensitivity to both vertically and horizontally polarized

radio waves, thus making the planar inverted-F antenna ideally suited to mobile

applications.

The objective of this chapter is to explain the theory and fundamentals of the planar

inverted-F antenna. Although the planar E-shaped IFA is considerably more sophisticated

in design and operation compared to generic planar inverted-F antenna, it must still obey

the same fundamental design principles, as outlined in this Chapter. Furthermore, some of

the design issues specific to the planar E-shaped IFA are provided in latter sections of this

chapter.


30

3.2 The Derivation of the PIFA

The basis for the planar inverted-F antenna originated from the inverted-L antenna

(ILA) [10]. This antenna, illustrated in Figure 3.1, is essentially a microstrip coaxial

connection with the inner conductor in free space and bent at a right angle in such a way

that the inner conductor end forms a horizontal element that is parallel with the ground

plane. The outer conductor is extended to form the ground plane. The resulting structure

possesses a short monopole element as a vertical element and a wire horizontal element,

which is attached to the end of the monopole.

Figure 3.1: The Inverted-L Antenna.

As the height of the vertical element is restricted to only a fraction of a wavelength,

the ILA has a very low profile structure. The horizontal element is normally set to a

quarter of a wavelength. As a result, the ILA has low impedance with a magnitude similar

to a short monopole. In order to increase the impedance of the ILA, another inverted-L


31

element must be incorporated into the original antenna, with one end of the inverted-L

element connected to the bend of the original ILA and the other end shorted to the ground

plane. As shown in Figure 3.2, the new antenna now appears to have an inverted-F

configuration: hence the name inverted-F antenna (IFA).

Figure 3.2: The Inverted-F Antenna.

By including the second inverted-L element into the resulting IFA, the input

impedance of the antenna can be set to match the load impedance of the antenna [10].

This feature of the IFA allows any additional matching circuitry of the ILA to be

discarded, resulting in the IFA being a more practical antenna. The low profile of the

IFA, as well as its performance with two polarisations facilitates the antenna for urban

environmental use, especially in conjunction with mobile applications.

One major drawback of the IFA is the lack of bandwidth. This issue can be resolved

by converting the horizontal wire element into a plate. The resulting antenna, shown in

Figure 3.3, is referred to as the planar inverted-F antenna (PIFA).


32

Figure 3.3: The Planar Inverted-F Antenna.


33

3.3 Spatial Network Method (SNM) analysis of the PIFA

To understand the operation of the PIFA, the Spatial Network Method (SNM) is

introduced. SNM is a three-dimensional electromagnetic field time-domain analysis

method, which can be used to explain some of the operational aspects of the antenna,

such as the effect of the short circuit plate and planar element widths on the antennas

resonant frequency. However, while the SNM has had minimal influence on the final

PIFA design of this thesis, it provides an understanding of the operation of the generic

PIFA. A more in depth study of the SNM is provided in Hirasawa [11].

3.3.1 The Electric Field and Current Distributions of the PIFA

The configuration of the PIFA used in the SNM is illustrated in Figure 3.4. The

dimensions used as part of this analysis are L1 = L2 = 16∆d and H = 4∆d, where ∆d is the

distance between similar nodes of the SNM. The width of the short circuit plate is set at a

variable with the following values: 2∆d, 4∆d, 8∆d, 12∆d and 16∆d. A TEM transmission

line feeds the PIFA from the back of the ground plane to the radiating patch. Conductive

and dielectric losses are ignored as part of this analysis.


34

Figure 3.4: The structure of the PIFA as used in the SNM analysis.

The reflection wave that results when a pulse wave is inserted into a TEM

transmission line can be viewed, when observed in the time domain. The next step is to

perform a Fourier transform of the time domain wave into the frequency domain, thus

allowing the reflection coefficient to be determined. The resonant frequency and the

bandwidth characteristics of the antenna can now be obtained by locating where the

minimum reflection coefficient occurs (in this case, at the feed point). From here a

sinusoidal wave of the resonant frequency is fed into the TEM transmission line. The

resulting amplitude of the electric node leads to the electric field distribution, while the

amplitude of the magnetic node component determines the current distribution.


35

Figure 3.5: The Electric Field Distribution of the PIFA at the x-y plane.

The electric field distribution of Ex, Ey and Ez for the PIFA is shown in Figure 3.5.

Worth noting from this figure is the zero value of the dominant component of the electric

field Ez at the short circuit plate in contrast to the much larger values of Ez on the

opposite edge of the planar element. Also of importance is the effect of when the short

circuit plate width is less than the planar element width. The resulting electric fields Ey

and Ez are generated at the open circuit edges, resulting in two polarisations.


36

Figure 3.6 depicts the distribution of surface currents for the PIFA at resonance

with a range of short circuit plate widths. The surface current on the upper surface of the

planar element is demonstrated on the first three distributions, while the surface current

of the underside of the planar element and the ground plane is provided in the second and

last distributions.

Figure 3.6: Surface Current distributions for varying short circuit plate widths.


37

3.4 Resonant Frequency

Probably the most significant factor that controls the resonant frequency of a PIFA

is the ratio of the short circuit plate width to the planar element width [10]. However,

what must not be discounted is the role that many other factors play in influencing the

radiating frequency and performance of the PIFA. The graph in Figure 3.7 demonstrates

this relationship, with a reduction of the short circuit plate width relative to the planar

element width resulting in a fall in resonant frequency. Also worth noting in Figure 3.7 is

the approximately 3% difference between the calculated and the measured frequency due

to the discrete positioning of the feed probe. To solve this problem, the feed probe must

be positioned at the electric nodes of the interval ∆d.

Figure 3.7: The effect of the short circuit plate width with relation to the frequency.


38

The resonant frequency can also be influenced by the ratio of the planar element

width to the planar element length, as demonstrated in Figure 3.8. Only L1, the planar

element width, is varied so that the overall size ratio can change in accordance. As

evident in the graph, if planar element width is reduced relative to the planar element

length, the resonant frequency is reduced.

Figure 3.8: The effect of the planar element width with relation to the frequency.


39

3.5 The Bandwidth

While there are many factors in PIFA design that influence the bandwidth, the

antenna height h is the most dominant factor. As illustrated in Figure 3.9, the antenna

bandwidth increases proportionally with the antenna height, relative to the wavelength.

However, while it is tempting to simply just increase the antenna height, an increase in

antenna volume and, hence, handset volume, will be incurred. Therefore, other means of

enhancing the antenna’s bandwidth must be utilised. It has been found that by increasing

the planar element width relative to the planar element length, the bandwidth is improved

[10]. It has also been found that by increasing the short circuit plate width relative to the

planar element width, the bandwidth can be enhanced even further [10]. Other bandwidth

enhancement techniques involve the use of shorting pins.

Figure 3.9: The effect of the antenna height on the bandwidth.


40

3.6 The Triple Band E-shaped PIFA Antenna

As noted in Chapter 2, the choice of antenna to form the basis of this thesis is the

design proposed by Dou and Chia [7]. This antenna calls for a derivation of the PIFA for

triple band operation. The resulting antenna is a compact design with a single feed and,

like other PIFAs, possesses a low profile. Two narrow linear slots are etched into the

radiating element side of the patch to give it an E shape, hence the name E-shaped PIFA.

Two shorting strips are included in the design to improve the impedance matching at the

triple bands. The proposed antenna allows operation at GSM (900 MHz), DCS (1800

MHz) and 2000 MHz. The design, as proposed by Dou and Chia, is illustrated in Figure

3.10.

Figure 3.10: Top view of the E-shaped PIFA


41

3.6.1 The E-PIFA and shorting pins

For multi-band PIFAs there are two methods available to facilitate good impedance

matching. One technique is to short the patch edge, allowing the matching to be

controlled by the distance from the feedline. Another technique is to use multiple shorting

pins. While the later technique is more flexible, too many pins present difficulty when the

antenna is to be fabricated. The method used by Dou and Chia is to use two metallic

strips to short the triple-band antenna to improve matching.

Figure 3.11: Return losses versus shorting positions. L1 =38.2 mm, L2=37.8 mm.

Figure 3.11 demonstrates the return losses for the E-shaped PIFA for three different

shorting pin configurations. Case A utilises only a single shorting strip at G1 (refer to

Figure 3.10), case B utilises a single shorting strip at G2, while case C uses a pair of

shorting strips at G1 and G2. As can be seen from figure 3.11, the best matching of the

antenna occurs for case C. Also, the three nulls for the return losses at case C are all less

than –10 dB.


42

The shorting pin width also plays a major factor in influencing return losses [7].

Dou and Chia found that by using a shorting pin width of 1 mm, the first resonant

frequency of 900 MHz exhibits poor return losses. When the strip width is increased to 3

mm, the matching at 900 MHz improves but the third resonant frequency 200 MHz

becomes significantly narrower. As a trade-off, the width of d = 2 mm was found to

produce the best compromise.

3.6.2 The E-PIFA and slot lengths

The two linear slot lengths were found to have a critical effect on the performance

of the antenna [7]. Figure 3.12 demonstrates how changes in the length of slot 1 influence

the return losses. As the slot length L1 is decreased, there is no shift in the second

resonant frequency, although the third resonant frequency is correspondingly increased.

Therefore, it is concluded that the slot length L1 can be adjusted to set the third resonant

frequency.

Figure 3.12: Return losses versus slot length L1.


43

Similarly, Figure 3.13 demonstrates how changes in the length of slot 2 influence

the return losses. As the slot length L2 is decreased, there is no shift in the third resonant

frequency, although the second resonant frequency is correspondingly increased.

Therefore it is concluded that the slot length L2 can be adjusted to set the second resonant

frequency.

Figure 3.13: Return losses versus slot length L2.

As well as the above investigations, Dou and Chia have examined the slot width

effect and discovered that there is minimal influence by the slot width on the resonant

frequencies of the E-shaped PIFA. In summary, the dimensions of the patch and the

lengths of the two slots can provide some degree of control over the three resonant

frequencies of the E-shaped PIFA.

Chapter 4: Methodology

44

CHAPTER 4

Methodology

4.1 Introduction

With a suitable antenna chosen for further development, the next step is to verify

the operation of the antenna at the prescribed frequencies in terms of input impedance

and field patterns. However, the selected antenna, the E-shaped IFA as proposed by

W.Dou and M.Chia [7] needs to be further developed and refined before any testing can

proceed. To achieve this, an antenna design software package, called FEKO, has been

selected to develop and to also simulate this antenna.

This section of the thesis will further review the work by W.Dou and M.Chia [7]

into the development of the planar E-shaped IFA. Furthermore, a comprehensive study of

FEKO will be presented in Section 4.2. Finally, a detailed analysis of the development

process, from the selected antenna to the resulting three antennas, will be provided in

Section 4.3.


45

4.2 The E-shaped PIFA and the simulations

As mentioned earlier in Chapter 2, the E-shaped PIFA was chosen to form the

basis of this thesis. The E-shaped PIFA, as proposed by Dou And Chia [7], meets the

antenna design requirements of this thesis, as outlined in Chapter 1. The antenna is a

triple band design and is much simpler to tune to the correct resonant frequencies

compared to other triple band PIFA designs, such as the antenna proposed by Song and

Hall [7]. The E-shaped PIFA possesses the qualities that make the PIFA a more

favourable antenna for mobile communications, such as being small and compact, fully

self-matching, capable of dual polarisation in all directions as well as being simple to

fabricate. The antenna has also been optimised by Dou and Chia to operate at two of the

three resonant frequencies, as required for this thesis. Additionally, the E-shaped PIFA

also has a 1-gram Specific Absorption Rate (SAR) that is much lower than the US

Federal Communications Commission (FCC) requirements [7]. This last point is very

important, considering the concern in the wireless communications industry relating to

the potential harm of the mobile handset user’s health, due to antenna radiation power

dissipation.

Dou and Chia optimised the dimensions of the E-shaped PIFA by the use of the

FDTD method based software. They designed the patch to measure Ly x Wx = 44 mm x

40 mm. The patch height was set at 8 mm and the ground plane had the dimensions Gy x

Gx = 100 mm x 50 mm. The two narrow linear slots S1 and S2 were etched into the patch,

resulting in the patch being divvied up into three parts, A, B and C. The resulting widths


46

of the divisions were set at WA = 23.5 mm, WB = 8.8 mm and WC = 7.7 mm. The slots S1

and S2 are each 2 mm wide. Both the metallic strips at G1 and G2 have an equal width of 2

mm and are located close to the feed position F. A single 50 Ω coaxial connection feeds

the triple band antenna and the inner conductor has a radius of 0.655 mm. The inner

conductor is also soldered onto point F and the outer conductor is soldered onto the

ground plane. Figure 3.10 of Chapter 3 provides a schematic of this antenna.

The results that Dou and Chia found are very encouraging. The three frequency

bands that the antenna was tuned to operate at are centred at 902 MHz, 1840 MHz and

1940 MHz with bandwidths for S11< -10 dB of 48.8 MHz, 36.6 MHz and 612 MHz

respectively. The resulting bandwidths are 5.41% for 900 MHz, 1.99% for 1800 MHz and

3.14% for 1900 MHz [7].

The far-field patterns obtained by Dou and Chia for the E-shaped PIFA in the

horizontal plane are illustrated in Figure 4.1 [7]. The radiation patterns depict the total

electric field strength for each of the three resonant frequencies. It is clear from Figure

4.1 that, for all three resonant frequencies, the electric field is stronger in the direction to

where the ground plane positioned (opposite to where θ =180°). As the resonant

frequency is increased, there is a corresponding reduction in electric field strength in the

ground plane direction of the antenna, indicating that the operating frequency influences

the effect of the ground plane on the far field patterns. Regardless, the radiation patterns

for the E-shaped PIFA are approximately omnidirectional at each frequency,

demonstrating that the antenna is ideally suited for mobile applications. Also worth


47

noting is that the field patterns seem to decrease at θ =90°, where the shorting pins are

located on the E-shaped PIFA.

Figure 4.1: The Far-Field Radiation patterns of the E-shaped PIFA in the horizontal

plane with (a) representing 900 MHz, (b) 1800 MHz and (c) 2000 MHZ.


48

4.3 The Simulation Software: FEKO

The University of Queensland has adopted FEKO as the principle electromagnetic

simulation software package for use with antenna design. FEKO has the ability to deal

with a wide variety of electromagnetic field analysis, utilising objects with a range of

configurations and properties. The time required for FEKO to simulate a design is quite

low compared to its rivals (for example, HFSS, another software package used by the

University, takes several hours for a single basic simulation), thus improving efficiency

when it comes to simulation. Most importantly, FEKO can perform simulations of objects

that possess curved structures, with one of the main goals of this thesis being to

investigate the performance of the triple-band antenna selected when subjected to the

effects of bending. For the above reason Ensemble, one of the other primary

electromagnetic software packages that the University uses, was not considered for the

task of antenna simulation in this thesis.

The technique that FEKO utilises for simulation of electromagnetic problems is the

Method of Moments (MoM). By calculation of the electric surface currents that exist on

conducting structures, the electromagnetic fields can be ascertained. Furthermore,

parameters such as the near and far fields, directivity, input impedance can be extracted

from the program if the user requests so.

In order to solve electrically large problems, the Physical Optics (PO)

approximation or the Uniform Theory of Diffraction (UTD) are normally hybridised with


49

the MoM in order to be utilised by FEKO. By performing this hybridisation, FEKO has

the capacity to solve electromagnetic problems where the object under consideration is

too large (in terms of wavelength) to be solved with the MoM but too small to apply the

asymptotic UTD approximation with any accuracy. By utilising the hybrid MoM/PO or

hybrid MoM/UTD techniques, any critical region of a structure can be evaluated using

the MoM while the remaining regions, that may be larger, flat or curved metallic surface,

can be analysed using the PO approximation or UTD. The most recent version of FEKO

only supports time domain harmonic sources, resulting in all calculations being

performed in the frequency domain.

4.3.1 The Components Of FEKO

The software package FEKO is actually a combination of several software modules.

Each module performs a certain vital task in the simulation process. The main interface

module is WinFEKO, which is used to control the solution of the problem. From

WinFEKO most other modules can be accessed. WinFEKO displays the three-

dimensional rendering of the antenna’s geometry, which allows the user to view the

antenna structure from all angles and, hence, check that the geometry is correct.

WinFEKO can also be used to display other output information, such as current densities

and three-dimensional field patterns.

To create the geometry of the antenna there are two programs required: EditFEKO

and PreFEKO. EditFEKO is a customised text editor, which uses geometry and control


50

cards to construct and manage the antenna and the simulation. Although EditFEKO

allows the user to have more control over the geometry and simulation process compared

to other electromagnetic software simulation packages (such as HFSS), there is still the

major drawback in that the structure of the antenna cannot be created graphically. This

results in a much more complicated design procedure. To process the resulting file,

PreFEKO is used to generate the mesh for the WinFEKO environment. By using

PreFEKO, complex geometries can be created, such as flat, spherical and cylindrical

plates as well as wire and dielectric structures.

Once the antenna is created and the resulting mesh is generated, the antenna can be

viewed by using WinFEKO, as illustrated in Figure 4.2. From here GraphFEKO can be

used to view the simulation output results. GraphFEKO can generate polar and linear

plots of the S-parameters, 3D far fields, 2D and 3D near fields. The plots can also be viewed

in GraphFEKO when a frequency sweep has been implemented in EditFEKO.


51

Figure 4.2: The WinFEKO operating environment.


52

4.4 The Experimental Process

4.4.1 Development of the planar E-shaped IFA

Now that a suitable triple-band microstrip antenna has been selected, the next step

was to create the antenna in the FEKO environment. The planar E-shaped IFA possesses

a very complicated geometry as the antenna has a large number of edges and the radiating

element utilises a complex shape. As the user also had no previous experience with

FEKO, an already difficult task became a very challenging task. The antenna dimensions

from Dou and Chia were used during the construction of this antenna. The resulting

antenna, though impressive in structure when viewed in the WinFEKO environment (see

Figure 4.3), performed very poorly. The return losses for each of the three frequencies

were very high with values of –1.0 dB for 900 MHz, -2.1 dB for 1600 MHz and –2.0 dB

for 1950 MHz. Clearly these results were very poor compared to the values obtained by

Dou and Chia.


53

Figure 4.3: The Planar E-shaped PIFA with the initial dimensions.

Initially there was no clear reason as to why this antenna was performing so poorly.

However advice was sought and it was discovered that, because Dou and Chia created

their planar E-shaped IFA using a software package very different to FEKO, the

dimensions used might not work as effectively for the identical antenna designed in the

FEKO environment. Also, a variation of the antenna’s dimensions of just a millimetre can

radically alter the performance of an antenna. With these aspects in mind, it was decided

to start from scratch with the dimensions of the E-shaped IFA, while still adopting the

same structure.


54

Throughout the process of optimisation, the work of Dou and Chia proved

indispensable. Chapter 3.6 details the achievements of Dou and Chia as to how the

dimensions of the planar E-shaped IFA affected the performance of the antenna. Of note

is that while the shorting pin positions greatly affected the antenna performance, as

outlined in the paper by Dou and Chia, the slot lengths did not directly determine the

operating frequencies of the E-shaped IFA, contrary to the findings of Dou and Chia. By

altering the dimensions of the slots S1 and S2, the second and third operating frequencies

exhibited only minor shifting. The predominant effect of changing slots S1 and S2 was the

major changes recorded in the return losses. Therefore it was found that by adjusting the

two slot lengths, the antenna could be optimised in terms of minimising return losses.

Another finding that was worth noting was the effect of the ground plane on the antenna

performance. The ground plane played a major part in determining the return loss

performance of the antenna.

The optimisation process of the planar E-shaped IFA was quite involved, due to the

number of variables that made up the dimensions of the antenna. However, the time

between each simulation was approximately fifteen seconds, allowing the optimisation

process to progress rapidly. As one of the antennas dimensions was optimised and fixed

temporarily, another dimension was optimised and fixed. This process cycled through all

of the antennas dimensions several times until the return losses were minimised, while

ensuring the resonant frequencies corresponded with those specified in the objectives of

this thesis. The following optimised dimensions were attained:


55

Gx = 52 mm

Gy = 100 mm

Wx = 39 mm

Ly = 43 mm

L1 = 34.7 mm

L2 = 35.2 mm

WA = 21.9 mm

WB = 8.0 mm

WC = 7.7 mm

S1 = S2 = 2 mm

D1 = 10.5 mm

D = 2 mm

H = 8 mm

With the dimensions of the planar E-shaped IFA completed, the antenna was ready

for simulation. The structure of the optimised planar E-shaped IFA is illustrated in Figure

4.4.

Figure 4.4: The Optimised Planar E-shaped IFA.


56

4.3.1 Development of the conformal E-shaped IFA

With the planar E-shaped IFA optimised and meeting the specifications of this

thesis, the next step was to create a conformal E-shaped IFA. The basic structure of the

antenna was to be identical to the planar E-shaped IFA but would be bound around a

cylindrical surface, as proposed by Kabicik and Kucharshki [8]. The length dimension of

the antenna will be the dimension that will be curved while the width of the antenna will

remain linear (i.e. parallel with the y-direction).

While the planar E-shaped IFA proved to be a complex antenna to create, the

conformal E-shaped IFA was exponentially more difficult to implement in FEKO. The

first and by far the most challenging problem to overcome was how to render the

conformal antenna accurately and efficiently with the geometry cards available in

EditFEKO. Once this was determined the planar antenna had to be converted to the radial

coordinate system to simulate the curved surface. Once the required calculations and

offsets were factored in, the antenna then had to be converted back to the Cartesian

coordinate system as FEKO can only work with Cartesian coordinates. The above-

mentioned geometric offsets proved to be difficult to create. Offsets were required to

ensure that the antenna faced away from the x-y plane and also to “stretch” the radiating

element relative to the ground plane so that the shorting pins and feed probe connected to

the ground plane and the radiating element correctly. Details of the above conformal

antenna creation procedure is closely examined in the Appendix.


57

With the conformal E-shaped IFA created and possessing the identical dimensions

from the planar E-shaped IFA, the antenna was ready for simulation. However, the radius

of curvature had to be determined. To match the approximate radius of the human head, it

was decided to create two conformal antennas, each with a different radius of curvature.

The radii chosen were 100 mm and 150 mm. To simulate conformal antennas of larger

radii, further work is required on the FEKO code, as there was an unexplainable

geometry error in the code that occurred for radii of more than 150 mm. Figure 4.5

illustrates the structure of the conformal E-shaped IFA with a radius of 100mm, while the

conformal E-shaped IFA with a radius of 150mm is depicted in Figure 4.6.

Figure 4.5: The Conformal E-shaped IFA with a radius of 100 mm.


58

Figure 4.6: The Conformal E-shaped IFA with a radius of 150 mm.

Chapter 5: Results and Discussion

59

CHAPTER 5

Results and Discussion 5.1 Introduction

With the earlier chapters establishing the theory, the design and the development

process of the E-shaped inverted-F antennas, the results for the simulations can now be

presented and discussed. This chapter will present the results obtained from the three

antennas developed to meet the objectives of this thesis. The first three sections will

provide the software simulation results obtained from FEKO for each of the antennas.

This chapter also includes a discussion of the results. The three antennas will be

compared in terms of input impedance as well as two and three-dimensional electric far

field patterns.


60

5.2 The Planar E-shaped IFA

The first antenna to be simulated using FEKO was the planar E-shaped IFA. The

dimensions of this antenna are provided in Chapter 4.3.1 and the structure is illustrated in

Figure 4.4. The return losses for the antenna at the three operating frequencies can be

found below in Figure 5.1.

Figure 5.1: The simulated return losses for the Planar E-shaped IFA.


61

As can be seen from the graph, the resonant frequencies of the antenna are 950,

1770 and 2080 MHz. The return losses of the antenna are –13.5 dB for 950 MHz, -8.7

dB for 1770 MHz and –8.9 for 2080 MHz. While the return losses are not as low as those

obtained by Dou and Chia in Figure 3.11, they are sufficiently small enough to meet the

first objective in Chapter 1.2 of this thesis. The resonant frequencies are also close

enough to the specified frequencies, as outlined in the objectives, to be acceptable.

Figures 5.2, 5.3 and 5.4 illustrate the three dimensional electric far field radiation

patterns for the planar IFA at 950, 1775 and 2000 MHz respectively. It is worth noting

for all three cases that the far field patterns are approximately omnidirectional, a desired

property for mobile communications. In particular, the antenna operating at 950 MHz

yielded very impressive results, with an almost perfect field pattern.


62

Figure 5.2: The simulated 3D Far Field Radiation patterns

for the Planar E-shaped IFA at 950 MHz.


63




64




65

Figures 5.5, 5.6 and 5.7 illustrate the two dimensional electric far field radiation

patterns for the planar IFA at 950, 1775 and 2000 MHz respectively. Note that the field

patterns were taken in the x-y plane.




66




67




68

5.3 The Conformal E-shaped IFA of radius 100 mm.

With the planar E-shaped IFA optimised after a long and tedious design process,

the next step was to transfer the optimised dimensions from the planar antenna to the

conformal antenna and compare the results. For this first antenna, the radius of curvature

was set at 100 mm. The structure of the conformal 100 mm E-shaped IFA is illustrated in

Figure 4.6. The return losses for the antenna at the three operating frequencies can be

found in Figure 5.8.

Figure 5.8: The simulated return losses for the Conformal E-shaped IFA of radius

100 mm.


69

It can be clearly seen from the above figure that the performance of the conformal

antenna is significantly different to that of the planar antenna. The resulting resonant

frequencies are 800, 1770 and 2000 MHz. The return losses of the antenna are –3 dB for

800 MHz, -4 dB for 1770 MHz and –4.75 for 2000 MHz. Therefore, it can be clearly

observed that further work is required on the conformal antenna to ensure that the

specifications are sufficiently met.

Figures 5.9, 5.10 and 5.11 illustrate the three dimensional far field radiation

patterns for the 100 mm Conformal IFA at 800, 1770 and 2000 MHz respectively.


for the Conformal E-shaped IFA of radius 100 mm at 950 MHz.


70




71




72

Figures 5.12, 5.13 and 5.14 illustrate the two-dimensional far field radiation

patterns for the 100 mm-conformal IFA at 800, 1770 and 2000 MHz respectively. Note

that the field patterns were taken in the x-y plane.




73




74




75

5.4 The Conformal E-shaped IFA of radius 150 mm.

With the Conformal 100 mm E-shaped IFA simulated, the next step was to test the

second conformal antenna. The radius of curvature was set to 150 mm. The structure of

the conformal 150 mm E-shaped IFA is illustrated in Figure 4.7. The return losses for the

antenna at the three operating frequencies can be found in Figure 5.15

Figure 5.15: The simulated return losses for the Conformal E-shaped IFA of radius

150 mm.


76

The resonant frequencies obtained from the 150 mm conformal E-shaped IFA are

870, 1770 and 2000 MHz. The return losses of the antenna are –2.7 dB for 870 MHz, -3.8

dB for 1770 MHz and –4.75 for 2000 MHz. It is worth noting that the 150 mm conformal

antenna possesses less return losses compared to the 100 mm conformal antenna. Again,

further work is nessercary for this antenna.

Figures 5.16, 5.17 and 5.18 illustrate the three dimensional far field radiation

patterns for the planar IFA at 950, 1770 and 2000 MHz respectively.




77




78




79

Figures 5.19, 5.20 and 5.21 illustrate the two dimensional far field radiation

patterns for the conformal IFA at 950, 1770 and 2000 MHz respectively. Note that the

field patterns were taken in the x-y plane.




80




81




82

5.5 Discussion

5.5.1 The Return Losses

As illustrated in Figure 5.1 of the results section, the planar E-shaped IFA yielded

suitable return losses. However, when the optimised dimensions were transferred from

the planar antenna to the two conformal antennas, the return losses failed to decline any

further. This result was a stark contrast to the expected result. In fact there was, on

average, a 6 dB increase in return losses in the case of both the conformal antennas. Most

disappointing was the resonant frequency shift that occurred when implementing the 100

mm conformal antenna, as shown in Figure 5.8. While the third resonant frequency was

shifted to 200 MHz, as desired, the first resonant frequency shifted from 950 MHz to 800

MHz, a poor outcome. However, it is worth mentioning that the 150 mm conformal

antenna yielded resonant frequencies that were actually an improvement compared to the

resonant frequencies of the planar antenna. The results for the 150 mm conformal antenna

are provided in Figure 5.15.

5.5.2 The Three-dimensional Far Field Patterns

While consideration of the antennas performance in terms of efficiency (i.e. return

losses) is important, one of the main antenna properties that must be analysed when

considering the viability of the antenna in future handsets is the radiation pattern. The


83

following will discuss the three-dimensional field pattern performance of the three

antennas.

Looking at the three-dimensional field patterns generated from the planar antenna,

it is evident that the best result came from when the antenna was operating at 900 MHz

(see Figure 5.2), an almost perfect outcome. The far-field patterns were strongly

omnidirectional and the electric field magnitudes were very high and consistent for most

directions. There was a very slight drop in the electric field strength in the direction of the

two ends of the antenna. These drops resulted in a pair of nulls at either end of the

antenna, an expected result with microstrip antennas. Regardless, the far field

performance of the planar antenna at 900 MHz was very satisfactory.

When the planar antenna operated at the second resonant frequency, 1775 MHz,

there were a number of notable differences, compared to the 900 MHz field pattern (see

Figure 5.3). Firstly, the electric field strength of the antenna while operating at 1775 MHz

was more predominant in the x-y plane, compared to the antenna operating at 900 MHz

that had field strength more predominant in the z-direction. Secondly, while the antenna

operating at 1775 MHz possessed two nulls at either end of the ground plane, the null that

occurred closest to the radiating patch end of the ground plane was very dominant in

reducing the electric field strength, while the second null at the opposite end of the

antenna only was very mild in its effect. Thirdly, the electric field patterns seemed to be

more prevalent in the direction above the antenna, compared to the direction underneath

the ground plane. Fourthly, the null located at the radiating patch end of the antenna


84

appeared to have shifted to a position further underneath the ground plane. These last

three points indicate that the ground plane had a direct influence over the field patterns.

The third resonant frequency had an even more notable effect on the field patterns.

The results were basically an extrapolation of the trend established by the antenna when

operating at the first two frequencies. The field patterns became again more predominant

in the x-y as well as the field patterns becoming stronger in the direction above the

antenna relative to the direction underneath the ground plane. The null at the radiating

patch end of the antenna became more dominant and also shifted further underneath the

antenna. As well as the above observations, it was also noted that the field patterns

seemed to weaken in the corner of the antenna where the shorting pins and feed probe

were located. All the above points validated the idea that the ground plane had a stronger

effect at higher frequencies. However, as with the previous case, the field patterns were

still essentially omnidirectional.

The 100 mm conformal antenna, when operating at 950 MHz, indicated a strong

similarity in overall far field pattern compared to the planar antenna when operating at

the same frequency. The field pattern was very much omnidirectional and the two nulls

existed at either end of the antenna, on both ends of the y-axis. The magnitude of the

electric field in virtually all directions was also very high.

When this same antenna was operating at 1775 MHz, there was a major change in

the field patterns. The result was a “peanut-shaped” pattern that was considerably


85

different to the field pattern for the same antenna at 950 MHz. The null that was

originally located at the radiating element end of the antenna was now located directly

underneath the antenna. The overall electric field strength of the antenna appeared to

have been reduced. While all far field patterns to this point possessed high electric field

strength in the direction above the antenna, this antenna operating at 1775 MHz had poor

field strength in this direction.

Again this ”peanut-shaped” pattern and its associated field effects were prevalent in

the 100mm conformal antenna operating at 2000 MHz. It appeared that the field pattern

for the 2000 MHz case was slightly more degraded compared to the same antenna

operating at 1775 MHz.

In the case where the conformal E-shaped IFA possessing a radius of curvature of

150 mm was operating at 950 MHz, the far-field pattern appeared to be virtually identical

to those of the two before mentioned antennas. Also, when the antenna was operating at

the second and third resonant frequencies, the field patterns possessed the “peanut

shape”, as exhibited by the 100 mm conformal antenna also operating at the second and

third resonant frequencies. However, it appeared that the field patterns for the 150 mm

antenna at the two upper resonant frequencies were, on average, of a stronger electric

field magnitude compared to the two cases with the 100mm conformal antenna.

While the planar E-shaped IFA appeared to produce sound far field patterns, there

was reasonable doubt concerning the quality of results obtained from the two conformal


86

antennas, especially when both the antennas were operating at the two upper resonant

frequencies.

5.5.3 The Two-dimensional Far Field Patterns

While the three-dimensional field patterns indicated the electric field strength

coverage of the antenna from all directions around the antenna, it must be considered that

an antenna used for mobile application cannot be guaranteed to be positioned in a

particular orientation. For example, an antenna mounted on a vehicle or a house can

generally be assumed to be positioned perpendicular from the surface of the earth. With

mobile applications, the antenna is expected to operate efficiently while being placed at

most random orientations. Mobile antennas are also expected to operate in cluttered

mobile environments where signal polarisation is frequently randomised by reflections.

Therefore, the performance of the antennas in terms of both polarisations (i.e. the e-phi

polarisation and the e-theta polarisation) was considered.

From the results it appeared that, for all three antennas operating at the first

resonant frequency, the resulting far-field patterns in the horizontal plane were as

expected from a typical microstrip antenna. The e-phi polarisation was omnidirectional

while the e-theta polarisation possessed the two lobes 180° apart with the two nulls 90°

from each of the lobes. However, for the two conformal antennas operating at the first

resonant frequency, the two nulls that occurred for the e-theta polarisation were several

decibels less in magnitude than for the equivalent using the planar antenna. This


87

occurrence indicated the degradation in performance of the conformal antennas compared

to the planar antenna.

For the case where the three antennas were simulated at the two upper resonant

frequencies, the results were quite varied in quality. The planar antenna at the second

resonant frequency lost one of the nulls for the e-theta polarisation but had a degraded e-

phi polarisation at 255°. However, when the planar antenna was operating at the third

frequency, both the polarisations, though almost identical to the case for the second

resonant frequency, were slightly degraded. Therefore, it can be concluded that dual

polarisation performance of the planar E-shaped IFA improves as the resonant frequency

is increased. Overall, however, the results were quite impressive and indicated that the

planar antenna is well suited to mobile applications.

When the conformal antenna of a radius of 100 mm was operating at the second

resonant frequency, both polarisations became nearly identical but offset at 105° from

each other. The e-theta polarisation also had a stronger magnitude on average compared

to the e-phi polarisation. Both polarisations had nulls of –18 dB. When the resonant

frequency was increased to 2000 MHz, the e-theta polarisation became stronger and

omnidirectional, while the e-phi polarisation developed two nulls, although one of the

lobes was of a much less magnitude compared to the other lobe. Therefore, it can be

concluded that, as the resonant frequency is increased, the antenna switches from being e-

phi polarisation dominant to e-theta polarisation dominant.


88

By setting the radius of the conformal antenna to 150 mm and operating it at the

second resonant frequencies, the polarisation patterns were identical to the case when the

antenna was operating at the first resonant frequency. In other words, the antenna had

polarisation patterns that are expected of a typical microstrip antenna for the first two

frequencies. Once the resonant frequency was set to 2000 MHz, the e-theta field

improved, becoming highly omnidirectional except for one null of –12 dB at 160°. The e-

phi field significantly deteriorated with a null at 250° of –30 dB. Therefore, it can be

concluded that the conformal E-shaped IFA behaves like a typical microstrip antenna for

the first two resonant frequencies but, once the third resonant frequency is reached, the

antenna performance switches from being e-theta dominant to e-phi dominant, depending

on the angle on the x-y plane at which the electric field strength is measured.

In summary, the planar antenna possessed the most suitable far-field patterns for

mobile applications. The two conformal antennas seemed to reverse their polarisation

dominance as the operating frequency was increased.

Chapter 6: Conclusions

89

CHAPTER 6

Conclusions

With the wireless communications industry expanding at a rate almost

unprecedented by any other technical field, there is a tremendous push to improve all

possible areas of the performance of mobile phones. Most elements of the mobile phone

have been able to keep shoulder to shoulder with this push. However, the antenna of the

mobile phone has basically been left untouched throughout the years. As mobile phones

continue to shrink in size, there is an overwhelming need to miniaturise and improve the

performance of the antenna. Microstrip antennas, and in particular, the planar inverted-F

antenna, help to address the above concerns. However, antenna design techniques, such

as developing an antenna that can operate at multiple frequencies as well as being

conformal in design, offer some potential in further dealing with the deficiencies of the

modern mobile phone antenna. The purpose of this thesis was to develop a multiband

antenna and investigate the effects of conformality on the antenna.

The first and second objectives of this thesis were comprehensively accomplished.

The first aim was to select and design an efficient, low profile and realisable antenna

capable of operating at a number of frequency bands (900 MHz, 1.8 GHz and 2.0 GHz).

This has been successfully achieved by altering the E-shaped IFA to operate effectively at

the prescribed frequencies. Both the creation and optimisation process involving this


90

antenna were very time consuming, but the end result was quite rewarding with the

second objective of verifying the operation of the antenna successfully attained. The

planar E-shaped IFA produced satisfactory return loss results and the far-field patterns

indicated that the E-shaped IFA has significant potential for wireless communication

application.

The investigation that was performed concerning the third objective of this thesis,

the effects of conformality on the performance of the selected antenna, yielded interesting

results. When the dimensions obtained from the optimised E-shaped IFA were transferred

to same antenna of a conformal structure, the results were degraded. This finding is in

contrast to the research performed by Pawel Kabicik and Kucharshki [8]. Each simulation

of the conformal antenna consumed over 20 minutes, while the planar antenna of the

same structure took only 15 seconds to be simulated. As it was nessercary to perform

over 400 simulations to optimise the planar antenna, it was not feasible with the time

remaining to directly optimise the conformal antenna. Therefore, it is recommended that,

in the future, further work should be performed in optimising the conformal antenna to

ensure that it adequately meets the basic requirements of an antenna used for mobile

applications.

There is much work to be performed in the future with developing both the planar

and conformal multiband mobile antennas. This is especially true with the case of the

conformal antenna. Fortunately, this thesis has been able to develop techniques to build

the conformal E-shaped IFA using FEKO. As a result, any future work performed at this


91

university in order to design conformal antennas using FEKO should be a much more

straightforward task. One of the major obstacles that must be overcome with developing

an effective conformal antenna is the optimisation process. As demonstrated above, the

conformal optimisation process is an extremely time consuming process, especially if

FEKO is to be used. Unfortunately, FEKO is probably the only electromagnetic analysis

software in university possession that can best deal with the design of conformal

microstrip antennas. All other software packages are either ill equipped to deal with

creating conformal antennas or require excessive lengths of time for each simulation.

I found this thesis to be quite challenging, time consuming and, indeed, very

frustrating at times throughout this year. Regardless, the subject of this thesis proved to

be highly fascinating and the rewards at the finale of this thesis have given me a great

feeling of accomplishment.

References

92

REFERENCES

[1] K.Fujimoto and J.R.James, Mobile Antenna Systems Handbook, Artech House,

1994.

[2] M.E Bialkowski, Wireless: From Marconi – The Way Ahead, IWTS 1997, Shah

Alum, Malaysia 1997.

[3] T.S. Rappaport, Wireless Communications, Principles and Practice, Prentice

Hall, 1996.

[4] J.D. Kraus, Antennas Since Hertz and Marconi, IEEE Transactions on Antennas

and Propagation, Vol.33, No.2, Feb.1985.

[5] A. Ando, Y. Honma, K. Kagoshima, Performance of a Novel built-in Antenna

installed in Personal Handy-phone System Units, Proceedings of ISAP ’96,

Chiba, Japan, 1996.

[6] C.T.P. Song, P.S. Hall, H. Ghafouri-Shiraz, D. Wake, Triple band planar

inverted-F antennas for handheld devices, Electronic Letters, Vol.36, No.2,

pp112-114, Jan. 2000.

References

93

[7] W. Dou, M.Y.W. Chia, E-Shaped planar inverted-F antenna with low-absorption

for triple-band operation in cellular phones, International Journal of Electronics,

Vol.88, No.5, pp575-585.

[8] Pawel Kabicik, Andrzej Kucharshki, The Effect of Conformality on the Electrical

Properties of Small Antennas, Wroclaw University of Technology.

[9] M. Reidy, The Planar Inverted-F Antenna, Thesis, pp.58-79, University of

Queensland, 1997.

[10] S.Y. Lee, Behavior of Handheld Transceivers in the presence of the Human Body,

Thesis, pp.21-23, University of Queensland, 2000.

[11] K. Hirasawa and M. Haeishi, Analysis, Design, and Measurement of Small and

Low Profile Antennas, Artech House, 1992.

Appendix: Antenna Development Process

94

APPENDIX

The Conformal Antenna Development Process using FEKO

As stated in Chapter 4, the development and simulation management of an antenna

in the FEKO environment is performed using EditFEKO. The antenna is created using

geometry cards that represent the antenna as a mesh structure. This mesh structure can be

created using wire elements, triangles, quadrilaterals, as well as cubic elements. The

structure can also incorporate dielectric elements, with properties such as the dielectric

constant capable of being defined. Once the structure is created, the control cards are

used to specify the frequencies of operation as well as specifying the output that is

required (i.e. far-field patterns, surface currents).

The following provides a technical explanation as to how the generic conformal

antenna was created using EditFEKO. While the focus of this discussion is on how the

structure was derived in the EditFEKO environment, the steps involved in extracting the

far-field patterns and the return losses are also explained.


95

A.1. Initialising the nessercary variables

The first step required for the antenna creation in EditFEKO was to define any of

the variables that are be required throughout the EditFEKO code. Firstly, a scaling factor

to scale all the dimensions from metres to millimetres was implemented, using the SF

card. From here, the IP card was used to define the segmentation parameters. The next

step was to define the variable values for all the conformal antenna’s dimensions. The

following are the variables and the dimensions used:

#gx = 52 **Ground plane length

#gy = 100 **Ground plane width

#wx = 39 **Conducting plane length

#ly = 43 **Conducting plane width

#l1 = 34.7 **This length controls f2 ( =1900 MHz)

#l2 = 35.2 **This length controls f1 ( =1800 MHz)

#wa = 21.9 **Antenna A width

#wb = 8.0 **Antenna B width

#wc = 7.7 **Antenna C width

#s1 = 2 **Antenna A-B slot gap

#s2 = 2 **Antenna B-C slot gap

#d1 = 12.5 **Shorting pin distance from edge

#d = 2 **Shorting pin length

#h = 8 **Antenna height


96

With the main dimensional variables defined, the next set of variables that needed

to be defined were the geometric offsets and the radius of curvature. The variable #r was

used to define the radius of curvature from y-axis to the ground plane. The variable

#totrad was created in order to find the radius of curvature from the y-axis to the antenna

element. Put simply, #totrad is the sum of #r and #h, the gap thickness between the

ground plane and the antenna element. Figure A1 illustrates the radii-related variables.

Figure A1: The radii variables

With the various radii defined, the next procedure was to define where the radiating

element is positioned with relation to the ground plane using the variables, #aoffx and

#aoffy. These were used to optimise the performance of the antenna. How these two

variables relate to the structure of the antenna can be seen in Figure A2. It must be

pointed out, however, that these two variables are in the radial perspective. In other

words, they provided the offset of the radiating element over a curved surface.


97

Figure A2: The antenna offset variables.

The next step was to obtain the angle subtended by the ground plane relative to the

y-axis (as shown in Figure A3). The length of an arc, L, is equal to the product of the

radius of the arc and the angle formed by the arc. In other words,

L = rθ

Using the above equation, the angle formed by the ground plane was obtained by

the following line of code:

#anglegnd = 180*#gy/(#r*#pi)


98

Figure A3: Calculation of the ground plane angle

With the radiating offset relative to the ground plane defined, the next step was to

define the ground plane offset, relative to the z-axis. While the variables, #aoffx and

#aoffy, related to the radiating element offset over a curved surface, the variables, #offx

and #offy, provided the true offset between the antenna as a whole and the x-y axis.

Figure A4 illustrates this process. For the conformal antenna, the overall offsets were set

in such a way so as to ensure the antenna was centred through the z-axis.

Figure A4: Derivation of the overall offsets.


99

With all the preliminary offsets established, the next step was to consider the need

to increase the length of the radiating element relative to the dimensions used on a planar

antenna. Let us take an arc of length L1 with a radius of curvature of R1 with a resulting

angle formed from the equation, L = rθ. Now if the radius of curvature was increased to

R2 but the length L1 remains constant, the angle formed will decrease. Lets take a second

case where the length L1 is still constant, the radius of curvature is R3 and the resulting

angle is even smaller than the other above case. Refer to Figure A5 for a graphical

representation of the problem.

Figure A5: The offset problem with conformal antennas

Now let us say that the arc of radius R2 is the ground plane and the arc of radius R3

is the radiating element. Both arcs are of equal length but the arc L2 appears longer. If we

connect a shorting pin to each of the arc, as shown in Figure A5, the shorting pin will not

appear perpendicular to the surfaces of which it is being joined to. As a planar antenna


100

has the shorting pins joining the two surfaces at angles perpendicular to each other, it is

ideal to maintain this 90° between the shorting pin and adjacent surfaces. To achieve this

in the case above, the arc of R3 must be lengthened on either end, by an appropriate

amount in order to ensure than both arcs are subtended by the same angle. Using the

equation, L = rθ, it is clear that, if the radius is increased but the angle is kept constant,

the length must increase in proportion to the radius in order to ensure the two arcs

correctly align. Therefore, all curved dimensions of the radiating element must be

multiplied by the total radius and then divided by the radius of the ground plane. In

relation to the problem in FEKO with conformal antenna design, the following line of

code was included,

#r_offx = #totrad / #r

Using this variable, it was now possible to calculate the true antenna offsets, as

shown in Figure A6. This was achieved by using the following lines of code,

#t_offx = #a_offx * #r_offx

#t_offy = #offy + #a_offy

It was nessercary to multiply #a_offx by #r_offx to take into account the change in

radius problem, as outlined above.


101

Figure A6: Calculating the true antenna offsets

The final step in this section was to define the ground plane offset from the y-axis,

as illustrated in Figure A6. This variable was essential as it dictated where the ground

plane edge must start and also leaded to where the radiating element edge must start by

simply adding #t_offx to #gnd_off. The role of this variable was to also ensure that the

ground plane was symmetrical around the z-axis, thus ensuring the antenna was

positioned correctly for simulation purposes. The derivation process of #gnd_off was

quite simple. As shown in Figure A&, #gnd_off was be derived by subtracting

#angle_gnd from 180° and then dividing the result by 2 to give #gnd_off.


102

Figure A7: Calculating the ground offset


103

A.2. Creating the radiating element

With all the necessary variables dealing with the positioning of both the ground

plane and radiating element created the next step was to create the radiating element. The

processes involved in creating this element and the ground plane took months to perfect.

Compared to other microstrip antenna designs that have been simulated in previous

theses using FEKO, the planar E-shaped IFA design was already a major challenge to

create in the planar perspective. However, creating the conformal E-shaped IFA made the

task far more challenging.

The first step was to develop a technique to create the conformal surface with

minimal difficulty while still adhering to the rules of antenna design using FEKO. The

structure of the antenna had to be broken up into pieces that would allow them to connect

properly at each of their nodes. After months of contemplation and reflection, an effective

method emerged that would be relatively simple to implement and would accurately

resemble a conformal E-shaped IFA.

The method of implementing the conformal E-shaped IFA revolved around the ZY

geometry card, which facilitated the implementation of curved rectangular structures in

EditFEKO. The radiating element had to be broken up into modules to facilitate the

implementation of the conformal antenna. Figure A6 demonstrates how the conformal E-

shaped IFA was broken up for construction purposes.


104

Figure A8: The modular breakup of the conformal antenna

By implementing the modular breakup, as shown in Figure A8, the radiating

element was assembled using 26 individual cylindrical surfaces implemented using the

ZY geometry card. The structure in Figure A8 also allowed the edges to align correctly,

resulting in the wire mesh that connecting correctly at all surface edges.

To implement the above solution, the first step was to define the distances from

each point along both the x-axis and the y-axis, for example, the distance from AAJ to


105

AAK, from AC to AF etc. This procedure set out to find all the lengths and widths of

each of the rectangles formed by the above method. These dimensions were obtained by

using the antenna dimensions that were declared in the variables component of the code.

For example, the distance from point AAM to AAO was be defined as #d1 - #d – (#wx -

#l1). The resulting variable was referred to as #dxaam_aao. This procedure started for

#dxaa_ab and ends with #dxaao_aap. The same process was undertaken for all

dimensions in the radial-y dimension. However, all the radial-y dimensions were required

to be multiplied by the variable #r_offx to offset for the apparent decrease in the curved

length of the antenna relative to the ground plane. For example, to define the distance

from AA to AD, the variable #dyaa_ad was created and given the value (#ly-#wa-#s1-

#wb-#s2) * #r_offx. Note all the resulting dimensions were indeed distances along the

curved surface and were not always straight-line displacements.

With the radiating element dimensions determined for the antenna, the next step

was to find the true x-y-z coordinates for each of the points. The first step was to take into

account all the offsets specified earlier, thus leading to the true x values for each of the

radiating points. This was be achieved simply by adding #t_offx to each of the radial x-

axis dimensions as defined earlier. For example, #dxac was set to equal #t_offx, while

#dxab equalled the sum of #t_offx and dxaac_ab. This process continued for all 26 points,

from #dxaa to #dxaap.

With the true x-coordinate found for each of the radiating element points, the next

step was to obtain the y and z-coordinates for each of the points. Unlike finding the x-


106

coordinates, the y and z-coordinates were directly affected by the effects of conformality.

Figure A9 graphically defines this problem.

Figure A9: Finding the true (x-y-z) coordinates

The first step in finding the two remaining coordinates was to convert the radial-y

dimensions defined earlier into angles in the y-z plane. These angles provided the offset

of each of the radiating element points through the y-z plane. To achieve this the

following formula was again used,

L = rθ

In this case, L was the radial-y dimension, r was equal to the variable #totrad and θ

was the unknown. For example, in the case of #dyaa_ad,


107

#a_aa_ad = 180*#dyaa_ad/(#totrad*#pi)

Where #a_aa_ad equalled angle formed by the point AA. This process of angle

definition started from #a_aa_ad through to #a_aac_aai for each of the radial y-

coordinate dimensions.

However, the resulting angles are only formed between adjacent radiating element

points. To find the total angle for each of the points from the y-axis, it was necessary to

sum all the angles together before the point, including the ground offset and the total

offset. The result was the true angle of elevation from the y-axis. For example, to find the

angle created by the point AA, the following line of code was required,

#a_aa = 180*#t_offx/(#totrad*#pi) + #gnd_off

To find the angle created by the point AD it was simply a matter of adding the

angle between AA and AD to the angle created by the point AA, as shown below,

#a_ad = #a_aa + #a_aa_ad

This process was performed for all points from AA to AAI.

With the angles formed between the radiating points defined, the next step was to

utilise some basic trigonometry available in FEKO to find the true y-z coordinates of the


108

radiating points. For example, to find the y and z-coordinates for the point AA, the

following code was needed,

#dxaa = #totrad*cos(rad(#a_aa)) #dzaa = #totrad*sin(rad(#a_aa))

Again, this process was performed for each of the points, from AA to AAI.

Now that the true x-y-z coordinates for the radiating element points have been

defined, the points could now be declared in a format that FEKO could interpret. For

example, the point AA was defined using the following code,

DP AA #dyaa #dxaa #dzaa

Each of the points was defined using the DP card in the above way, from AA to

AAI.

With each of the radiating element points defined and ready for use in FEKO, the

last step was simply to use the ZY geometry card to create each of the 26 surfaces that

made up the radiating element. However, a number of important steps were required in

order to use the ZY card. So far all the points that make up the corners of all the

rectangular modules of the radiating element have been created. What was now

nessercary for the ZY card to be implemented was to declare a number of points that


109

were placed on the x-axis in order for the ZY card to work. Figure A10 illustrates each of

the parameters nessercary for the ZY card to be implemented. The points S1 and S2 were

these points that had to be placed on the x-axis. To achieve this, it was simply a matter of

defining a number of points that equalled the number of possible x-coordinate values

used in both the construction of the radiating element and the ground plane and assigning

the y and z values of zero. For example, the first point was created in the following way,

DP CA #dxca 0 0

The result was 9 points, each represented every x-coordinate dimension used in the

radiating and ground plane points, but with the y and z-coordinates set to zero.

Figure A10: The ZY geometry card in EditFEKO


110

With all the necessary coordinates for using the ZY command defined, the last step

was simply to specify the maximum edge lengths for both the straight edges and the

conformal edges of the radiating element, as shown in Figure A11 (for simplistic reasons,

the conformal E-shaped IFA radiating element is depicted as a conformal rectangular

patch). It was nessercary to ensure that adjacent rectangular elements shared the same

nodes in the wire grid. Thus if the two adjacent cylindrical surfaces have been defined

using the ZY geometry card, the two joining edges must have the same maximum edge

lengths, whether the two joining edges were curved or straight. The variables defining the

maximum edge lengths were set to obey the above conditions.

Figure A11: The ZY geometry card in EditFEKO


111

Now that the two above steps have been undertaken, it was now possible to

implement the ZY card, as shown in Figure A10. Care was taken here to ensure that the

correct values and variables were placed in the correct parameter registers of the ZY card.

An example of using the ZY card to create module 1 (refer to Figure A8) is shown below,

ZY CB CG AA 0 #a_aa_ad #acv_len

Where #a_aa_ad defined the angle between the two adjacent corners of the surface,

AA and AD, while #acv_len defined the maximum curved edge length of the surface (this

variable was implemented in the initialisation of variables part of the code). The

maximum straight edge length of the surface was defined at the start of the code, using

the IP card.

Each of the 26 cylindrical surfaces was created using the above technique, with the

resulting surface representing the conformal E-shaped radiating element.


112

A.3. Creating the ground plane

The ground plane was constructed using the same techniques as involved with the

radiating element but with a few differences. Figure A12 illustrates the breakup of the

ground plane structure into the nodal points and the rectangular modules. The result was

20 rectangular modules and 30 points from GA to GAD.

Figure A12: The modular breakup of the ground plane


113

The first step undertaken was to obtain the true x-values for each of the ground

plane points. The first column of points (i.e. GF, GL, GR, GX and GAD) were set to

equal the variable #offy. The resulting notation for the x-values was #dygf, #dygl, etc.

This process continued for all of the ground plane points, from GA to GAD.

The next step was to obtain the radial x-direction differences in adjacent points

along the x-direction. This was achieved by referring to the structural layout of the

ground plane, as provided in Chapter 4. From here, the values for the variables could be

determined. For example, to find the difference between GE and GD, the structural layout

was referred to, yielding the expression #dl - #d. The resulting radial x-direction

difference, #dxge_gd, was set to equal the expression #dl - #d.

From here, the radial y-directional difference in adjacent points needed to be

calculated. Again, this was achieved by referring to the structural layout of the ground

plane that was provided in Chapter 4. From here, the values for the variables could be

determined. For example, to find the difference between GA and GG, the structural

layout was referred to, yielding the expression #a_offx + #ly - #dl. The resulting

difference, #dyga_gg, was set to equal the expression #a_offx + #ly - #dl.

With all the differences between the ground plane points in the radial perspective

defined and the true x-directional point values found, the following step was to find the

true y-z coordinate values for each of the points. This process was similar to how the

coordinates for the radiating element were found, but the radius that was taken in all


114

calculations was #r, not #totrad as in the case for the radiating element. The angles

between points were obtained, whereby allowing the true angles are calculated. The

variable #gnd_off was used to find the angle of the first row of points (GA, GB, GC, GD,

GE and GF). From here, the following code provides an example of how the true y and z-

coordinate magnitudes were realised (in this case, point GA was considered),

#dyga = #r*cos(rad(#a_ga))

#dzga = #r*sin(rad(#a_ga))

The next step was to define the points for the ground plane using the DP command,

enabling the ZY card to be used, and creating the 20 rectangular modules that comprised

the ground plane. For example, module 1 was implemented using the following code,

ZY CA CC GA 1 #a_ga_gg #acv_len

Care was taken to ensure that the maximum edge length where the surface connects

to the shorting pins equalled the maximum edge lengths for the connecting surfaces of the

radiating element. With the ZY card implementing the 20 rectangular modules, the

ground plane was now complete.


115

A.4 Adding the shorting posts and the feed probe

With the construction of the radiating element and the ground plane complete, the

last geometric features that needed to be added were both the two rectangular shorting

posts and the feed probe. The creation of the shorting posts was performed with the

following two lines,

BQ GV GU AAM AAN 2 2 2 2

BQ GH GN AAC AV 2 2 2 2

Note that the maximum edge length for each of the four sides of the two shorting

posts was set to 2 mm. This had to be considered when setting the maximum edge lengths

of both the ground plane and the radiating element.

The feed probe was implemented by first ensuring that the wire radius was set to

0.655mm by the IP card earlier. Next, the LA card was used assign a label to the feed

probe, ensuring that the feed probe would be element to be excited with a voltage source.

Finally, the feed probe was constructed by using the following code,

BL GW AAO

With the shorting posts and the feed probe implemented the structure of the antenna

was fully complete and ready for compiling and simulation.


116

A.5 Compiling the code and implementing the simulation

This section of the code completed the design process of the conformal antenna by

allowing the user to select and control what output from the antenna was to be produced.

The first step was to use the following code, which basically writes the results to an

output file,

EG 1 0 0 0 0

PS 0 0 1 0

The next step was to impress a voltage source onto the feed probe by use of the

following code,

A2 1 1 1 0

With the antenna energised, the frequencies of operation were specified. The

following code implemented a frequency sweep across 18 frequencies, from 800 MHz, in

steps of 75 MHz,

FR 18 0 800e6 75e6

To enable the return losses to be viewed when using GraphFEKO, the following

code, which assigns input impedance to the feed probe, was used,

SP 50


117

The far-field radiation patterns could be viewed in GraphFEKO and the output file

produced by Win FEKO by the implementation of the following code,

FF 1 72 72 1 0.0 0.0 5.0 5.0

This code performed a field sweep in divisions of 5° in both the x-y plane and the

y-z plane, allowing the user of GraphFEKO to view the far-field radiation patterns from

all directions around the antenna.

Finally, using the EN command completed the code.

Antenna

Documents

Transcript of Antenna