## Postgrads I have dubiously supervised.## Alan Robert Clark## October 11, 2012 |

**Design of a smart HF antenna with multiple pattern
capabilities using characteristic modes**

(M.Sc. by Dissertation)

*Project Proposal—*This research proposal is for the study of a method of analysis of antennas
called the theory of Characteristic Modes. Characteristic modes are a set
of fundamental orthogonal current distributions on an antenna, which are a
function of the geometry and are expressed independent of excitation. It
is proposed that the method is studied and then implemented in software, to
allow the modes of an antenna to be calculated. Using the software, a
demonstration antenna will be designed by analysis of its characteristic
modes. The antenna will be simulated in the conventional way to evaluate
its effectiveness. A thesis will be written on the theory and practice of
characteristic modes, and the design and performance of the demonstration
antenna.

**Computational Electromagnetics on Beowulf Clusters**

(M.Sc. by Dissertation)

*Project Proposal—*As is the trend with scientific modelling, electromagnetic simulations are
now being conducted on parallel computing clusters. This both decreases the
required computation time as well as increases the size of problems that
can be tackled. Of particular interest in the field is the use of
inexpensive Beowulf-class clusters. The student proposes that a comparison
between competing electromagnetic simulation packages FEKO and
SuperNEC should be done on such a cluster. The comparison should
include performance differences and recommendations for when to use each.
Prior to this, SuperNEC requires to be updated for the cluster and
its performance improved through the use of newer libraries and compilers.
There are also proposed improvements to the cluster which include the use
of different networking topologies and intra-node communication
optimisation. Using the knowledge gained, recommendations will be made for
future clusters.

**Modelling and development of ultra-wide-band receivers for
general radio astronomy systems**

(Ph.D.)

*Project Proposal—*Radio astronomical systems have very high sensitivity requirements, and
often require cooled receiver systems (at a minimum the LNA). The trend
towards multiple antenna elements (SKA) and wideband future geodesy
(VLBI2010), favour, economically, ultra wideband feed systems. The
multitude of available LNAs, feed systems (primary and secondary) and
feeding networks make the optimal choice complex. It is not possible to
build and test all the possible combinations, so reliable models are
required. The project proposed is to develop, model and test an ultra-wide
band system, along with modelling various feeds and feeding systems.

**Design and Automation of a Pre-Compliance EMC Test Facility**

(M.Sc. by Dissertation)

*Project Proposal—*Electromagnetic compatibility is of particular importance in the modern
electronic engineering industry. Strict standards have been imposed by
governments and international organisations to ensure that the limited
radio frequency spectrum is not swamped by aggressors. At the same time,
new electronic devices developed must be able to function correctly in
their intended environment. The only way to ensure products meet the
standards and can operate as intended, is to test them. This paper
presents methods to implement a highly accurate 3m pre-compliance test
facility, by making use of a fully anechoic chamber, a suitably optimized
antenna for use from 30–3000 MHz, a vector network analyser and
appropriate automation systems. Sources emitting a known field strength are
used to determine the error margin of the system.

**Parametrisation and Design of Quadrifilar Helices for use in
S-band Satellite Communications**

(M.Sc. by Dissertation), available at
`http://wiredspace.wits.ac.za/bitstream/handle/10539/8364/Msc_fraser_diss.pdf`

*Abstract—*This paper is a discussion on the Multiturn Quadrifilar Helix Antenna (QHA)
with particular focus on its application as a ground station antenna for
S-band communications with a Low Earth Orbit Satellite. A ground station
antenna without a tracking system requires a “saddle” shaped, circularly
polarised radiation pattern in order to compensate for the change in
distance between it and the antenna on the satellite. The Multiturn
Quadrifilar Helix can provide this radiation pattern with correct setting
of the parameters of pitch, radius and number of turns. The QHA was
simulated according to the adjustment of these parameters and the results
were assessed. The most suitable results were found for the antennas with
low to mid range of number of turns, radii less than 0.22λ and
pitch less than 0.6λ. A QHA with 3 turns, pitch of 0.6λ and
radius of 0.034λ was suitable for satellite communications.
Simulations showed it to have a gain of 6.16dB at 52^{∘} and −2.25dB
at 0^{∘}. Three separate feed networks: a corporate feed network,
90^{∘}−180^{∘} Hybrid combination and Wilkinson splitter feed
network, for the QHA were designed. The antenna was constructed for each
feed network and tested. The constructed antennas all had gains less than
predicted by simulation. The QHA using the Corporate Feed network had a
gain of approximately 10dB less than expected. The QHA using the 90^{∘}− 180^{∘} Hybrid combination feed network had a gain of approximately
8dB less than expected. The best performing QHA was fed by the Wilkinson
splitter feed network. It showed good comparison to the shape of the
pattern found in simulation but a gain of approximately 6dB lower than
expected.

**New Manufacturing Techniques for Antennas**

(M.Sc. by Dissertation), available at
`http://wiredspace.wits.ac.za/bitstream/handle/10539/7397/Msc_nash_diss.pdf`

*Abstract—*This paper is a discussion of new manufacturing techniques for antennas.
Many existing manufacturing methods, including conductive ink printing,
plastic plating, hot foil printing, etching, sintering and die cutting, are
investigated to determine their usefulness in the manufacture of antennas.
The advantages and disadvantages of each are discussed, and the most
promising method—plating on plastic—is further analysed. The method of
adapting the plating technique so that it can be used for antennas is
discussed. Two prototype antennas (a PIFA and omni antenna) were
manufactured to test the plating method’s effectiveness as a manufacturing
technique for antennas. Results showed a frequency shifted VSWR pattern for
the PIFA antenna of 10% on each notch. The gain plot for the omni antenna
showed a higher gain for the plated antenna at a frequency shifted by
approximately 0.4 GHz. A cost analysis was also performed to complete the
investigation of the new manufacturing method. A saving of up to 4 000%
can be realised on the substrate material, and the metal costs can be
lowered by 700% for each PIFA antenna.

**Formulations for Analysis of Probe-Fed
Printed Antennas in ****SuperNEC**

(M.Sc. by Dissertation), available from
`http://wiredspace.wits.ac.za/bitstream/handle/10539/6841/msc_dissertation_me_mathekga.pdf`

*Abstract—*Formulations for analysis of printed antenna structures are derived
and compared, to determine one to implemented in SuperNEC based on
the efficiency of its numerical solution in terms of memory usage and
solution time. SuperNEC is a software application for computing
the response of electromagnetic structures to electromagnetic fields.
SuperNEC cannot be used for simulation of printed antenna structures. This
is because the formulation that is implemented in SuperNEC does not account
for the effect of the substrates that the radiating elements of the antenna
structure are printed on, and it is also not intended for antenna
structures whose radiating elements are surfaces. Two MoM (Method of
Moments) formulations and a FEM (Finite Element Method)-MoM formulation are
presented, together with different models for the antenna feed. The FEM-MoM
formulation is selected for implementation in SuperNEC because it
is argued that it is likely to be more memory efficient when compared to
the MoM formulations, and also that less time is required to fill the
matrices resulting from the numerical solution of the formulation. The
formulation is implemented in a stand alone software application, which
will be integrated into SuperNEC. Numerical results that are
computed using the software application are presented to illustrate correct
implementation of the formulation. The results are compared to: an exact
solution, results from another publication, and results computed using a
different formulation. Good agreement is obtained in each case.

**Optimised Ray Tracing for the SuperNEC Implementation of the
Uniform Theory of Diffraction**

(M.Sc. by Dissertation), available from
`http://wiredspace.wits.ac.za/bitstream/handle/10539/2121/MasterThesis.pdf`

*Abstract—*Geometric optimisations are presented for the UTD in SuperNEC which is a
commercial electromagnetic software package. Path finding optimisations
rapidly find propagation paths of electromagnetic waves by using back face
culling to determine the visible plates of polyhedral structures and by
using reflection and diffraction zones which use image theory and the law
of diffraction to determine illuminated spatial regions. An octree reduces
the number of intersections during the shadow tests. Numerical results show
that overall the optimisations halve the run time of the software for
models which consist of plates and cylinders. The path finding
optimisations do not scale with model size, are limited to plates and
introduce errors. The mean absolute error due to the path finding
optimisations is on average 0.02 dB for first order rays and 0.17 dB for
second order rays. The octree optimisation scales with model size, can be
used with any geometry and any type of ray and does not cause errors.

**Analysing Lightning Data from Two Spatially Separate Magnetic Direction Finders**

(Co-supervised M.Sc. by Dissertation), available at
`http://wiredspace.wits.ac.za/bitstream/handle/10539/6776/DavidRice_MSc.pdf`

*Abstract—*Two lightning detectors, of the magnetic direction finder type, form part
of a two station system for determining the position of lightning strikes.
The detectors are on a baseline of approximately 600 m, and the ultimate
aim of the system is to accurately detect and map lightning within a radius
of 30 km. Although no real time capability is present, the archive data
collected from each separate station is used to find the offset errors in
the azimuthal orientation, as well as in time (using processes described in
Appendix A). The relative offset errors are determined by shifting the time
and azimuthal information for one station’s data and calculating the
maximum possible matching records (within certain time and azimuth
criteria) for each incremental shift. An analysis of the peaks in total
matching records, when plotted against the relevant shift increments, is
performed in order to obtain the values of the offset errors. Between the
two individual stations, the relative offset in orientation is found to be
24.5 degrees, and in time to be 0.001305 days (112.75 seconds). The
individual stations, as well as the triangulated data calculated from
matching records, can also be calibrated using data from the South African
Weather Service Lightning Detection Network (SAWSLDN). Individual station
calibration indicated an offset of +6.4 degrees and 0.00575 days (496.8
seconds) for Station 1, with the offsets for Station 2 determined as +29.4
degrees and −0.000105 days (9.07 seconds). Comparison of triangulated data
to SAWSLDN data yields unexpected results with regard to resultant shifts,
which may point to an error or anomaly in the triangulation calculations. A
detailed analysis of the storm data is contained in Appendix B of the
dissertation.

(M.Sc. by Dissertation), available at
`http://wiredspace.wits.ac.za/bitstream/handle/10539/5877/NEIL_OLEARY_MSC_DISSERTATION.PDF`

*Abstract—*SuperNEC is a method of moments (MoM) electromagnetic field solver
based on the Numerical Electromagnetics Code (NEC). Much of the simulation
time can be attributed to the filling of the impedance matrix, which is
performed at each frequency point of interest. Impedance matrix
interpolation methods have been implemented in SuperNEC to reduce
the computational time required to fill the impedance matrix [Z]. Elements
in [Z] vary predictably over frequency and can be approximated by a second
order polynomial. A second improved method is implemented where the
dominant frequency variation term is removed prior to calculating the
fitting function. A method of determining the optimum sample range relative
to simulation range and maximum interaction distance has been developed.
Given the correct choice of sample range the mean error in the MoM solution
is less than 10% over the frequency range and the input impedance can be
reproduced with good agreement over a wide bandwidth. Improvement in the
simulation electromagnetic efficiency of 1.7 times can be expected if
sufficient frequency points are of interest to account for the
computational time required to sample the matrix and determine fitting
function coefficients. This method has been applied to a dipole antenna, an
LPDA and a horn antenna. To increase the simulation bandwidth and retain an
acceptable level of accuracy, the bandwidth is split into multiple
sub-bands.

**Antenna Optimization using Particle Swarm Optimization**

(M.Sc. by Coursework and Research Report)

*Abstract—*The feasibility of using the particle swarm optimisation
(PSO) algorithm in SuperNEC, an antenna simulation software application, is
investigated in this paper. Previous research shows that the use of PSO in
optimising antenna design is promising for specific applications. A program
that uses this optimisation method is developed. The effect of using
different population size, social and cognitive factors, weights
implementation, velocity and boundaries on the algorithm performance are
investigated. The algorithm is then compared with the previously developed
genetic algorithm (GA) and simulated annealing (SA) optimiser. The
comparison shows that the PSO optimised better than the other algorithms on
average by at least 20%.

**Holographic
measurement of the 26m HartRAO telescope**

(M.Sc. by Dissertation), available from
`http://wiredspace.wits.ac.za/bitstream/handle/10539/6936/Benjamin_Klein_Msc_Dissertation.pdf`

*Abstract—*Microwave holography is a well established method of using the Fourier
relationship between an antenna’s current distribution and its complex
beam-pattern to produce surface maps of large parabolic antennas. As the
final part of a surface upgrade, a holographic map of the HartRAO 26 m
telescope was produced. This showed that the surface has an RMS error of
0.45 mm. The measurement used a small reference dish to correlate against
and retrieve amplitude and phase values. Due to system phase instabilities,
this dish had to be attached to the measured antenna in order to enable
sharing a high frequency local oscillator (LO). The movement was modelled
and corrected for. However, a slight distortion remained. It is recommended
that, either the LO distribution system is stabilised by using multiple
PLLs or amplifiers and low loss cables are used to enable moving the
reference antenna to a stationary position.

**Waveguide-based Antennas**

(M.Sc. by Dissertation)

*Abstract—*Investigations of slotted waveguide antenna arrays operating at 2.45GHz and
their applications to wireless local area networks (WLAN) are presented in
this paper. Requirements, considerations, and limitations associated with
the design process are discussed and presented. Various antenna parameters
were simulated using MATLAB and SuperNEC software simulation programs, and
were applied to a sensitivity analysis of antenna design. End-fed and
centre-fed antennas were designed, built, and measured at WLAN frequencies.
The antennas had high gain above 19dBi, broad beam around the azimuth, and
high efficiency, but were limited by their impedance dependency and narrow
bandwidth. The centre-fed antenna had 3dBi higher gain than the 19dBi gain
of the end-fed antenna. The VSWR ratio of both antennas was less than 1:1.5
at the operating frequency. The centre-fed antenna had broader azimuth and
elevation patterns by 40^{∘} and 10^{∘}, respectively. The end-fed
antenna had more stable gain and VSWR, 50% wider VSWR bandwidth, and more
directional elevation pattern. Measurements and test results were
consistent with theoretical and simulated antenna characteristics, and they
validated the analysis of the antenna performance.

**Wireless LAN modelling tool**

(M.Sc. by Dissertation)

*Abstract—*For the effective implementation of a wireless local area network (WLAN)
within a building, a complete understanding of indoor signal propagation is
required. This paper compares three empirical propagation prediction models
with regards to efficiency and accuracy. To achieve this, a software
prediction tool was developed using C++ which allows an end user to quickly
draw a building floor plan using user specified drawing materials. It also
has the ability to calculate the required empirical parameters from entered
measurements but this was found to produce results similar to that when
theoretical empirical parameters were used. The accuracy of the prediction
tool was gauged by comparing its outputs, using the different empirical
models, to measurements. In doing so it was determined that two of the
models produced functional levels of accuracy in which 93% and 82% of the
simulated results were within 15 and 10 dB of the measured results
respectively for the most accurate of the models used. All three empirical
models were found to have computational times low enough, less than 5
minutes for an average building, as to allow for interactive WLAN design.

**Low Cost Direction Finding with the Electronically Steerable
Parasitic Array Radiator (ESPAR) Antenna**

(M.Sc. by Dissertation)

*Abstract—*In this paper, the Electronically Steerable Parasitic Array Radiator
(ESPAR) antenna, developed by the Advanced Telecommunications Research
Institute (ATR) in Japan, was analysed to determine its feasibility as a
low cost direction finding (DF) system. Simulations of the antenna were
performed in SuperNEC and Matlab was used to determine the direction of
arrival (DOA) using the Reactance Domain multiple signal classification
(MUSIC) algorithm. Results show the ideal configuration has 6 parasitic
elements with a diameter of 0.5λ. Up to 5 periodic, uncorrelated
signals spread 360^{∘} in azimuth and above 45^{∘} elevation
produce sharp peaks in the MUSIC spectra. Azimuth separations of only
2^{∘} at 40dB are resolvable while signals arriving with 25% full
power are still detectable. For the DOA to be resolved the radiation
pattern should be asymmetrical and hence the reactance set should have a
range of unequal values. Comparative results show that the 6 element ESPAR
offers excellent overall performance despite the reduction in cost and is
comparable in performance to the 6 element uniform linear array.

**An Investigation and Study into Circularly Polarised Wide-Bandwidth
Antennas**

(M.Sc. by Dissertation)

*Abstract—*With the increase in wireless data communication over the last few years,
there has been an increased need for wide-bandwidth capabilities in antenna
and hardware systems. The quasi-taper axial mode helical antenna and
bifilar Archimedian spiral antenna both exhibit wide bandwidth capabilities
with a 2:1 bandwidth ratio shown by the helical antenna and well over 10:1
capable with the spiral antenna. The investigations completed on the
antennas attempted to extract design guidelines and characteristics that
could be incorporated into a design wizard running on a Method-of-Moments
simulation package. These guidelines would be used to predict properties
such as gain, bandwidth and VSWR of the respective antennas. While the gain
observed for the spiral was relatively static and difficult to vary, that
of the helical antenna could be varied easily with a change in axial
length. The bandwidth of the spiral was found to be extensive, chiefly
dependant on the inner feed and outermost radii; while that of the helix
was limited to a maximum of 2:1 when a ratio of 1.41 for the radius of
upper to lower helical sections was used. With both antennas being
travelling-wave structures, they both exhibit a purely real input impedance
which can be matched effectively with a microstrip transmission line.

**Patch Antenna Design Wizard **

(M.Sc. by Dissertation)

*Abstract—*This paper describes the formulation of a design procedure for the
rectangular, probe-fed and rectangular, electromagnetically coupled patch
antennas. The design procedure is presented in the form of design curves.
The design curve data was generated by solving a bandwidth optimization
problem subject to a specified minimum gain constraint. Given a particular
performance requirement, an appropriate antenna geometry can be found using
the design curves. An unavoidable tradeoff between the gain and impedance
bandwidth performance parameters is evident from the design curves. The
bandwidth and minimum gain configurations for design of the probe-fed patch
range from 6.2% and 9.9dBi to 13.2% and 6.1dBi respectively. The
bandwidth and minimum gain configurations for design of the
electromagnetically coupled patch range from 9.5% and 9.7dBi to 44% and
6.0dBi respectively.

**An Algorithm for the Automatic Resolution of the position,
Orientation and Dipole Moment of a Magnetic Dipole Antenna Buried in Rock.**

(M.Sc. by Coursework and Project Report)

*Abstract—*In this paper, an algorithm for resolving the position, orientation and
dipole moment of a transmitting dipole antenna buried in rock is presented.
The algorithm is based on simplified radiation equations that are shown to
be valid within a region of space in the extreme near field surrounding the
antenna termed quasi-static region. Within this region, medium dependent
propagation effects are negligible, allowing accurate, medium independent
resolution of position, orientation and dipole moment.

The magnetic field magnitude and direction values observed at a number of arbitrarily located points in space are used by the algorithm to generate a system of nonlinear equations. This system of equations is solved using a simultaneous multi-variable Newton-Raphson solver with line searches and backtracking providing a measure of global convergence. An additional method, based on an iterated random search, further improves the global convergence capability of the algorithm.

If the dipole moment of the transmitting antenna is known, measurements from two observer points are sufficient to resolve the position and orientation of the dipole. If the dipole moment is unknown, three observer points are required to solve for the unknown position, orientation and dipole moment. The algorithm is able to find an exact solution using exact theoretical measurements, and a minimised least squares solution where measurements are subject to noise.

The theory relating to the algorithm is discussed including distance-frequency relations for the quasi-static region of a number of common rock types. Probabilistic modelling, simulation and test results of the algorithm are also included.

**Centre of Rotation Determination for ISAR Imaging.**

(M.Sc. by Dissertation)

*Project Proposal—*This investigation proposes a solution to the problem of centre of
rotation determination of ISAR images. This problem is essentially an
inverse problem whereby the 2d ISAR image of an observed target
undergoing some arbitrary 3d motion in space has to be used in finding
the centre of rotation of the image. A solution would aid
non-cooperative target recognition (NCTR) algorithms minimize their
search area when trying to compare images of a current target against a
target database. The problem required that translational motion
compensation and minimum entropy methods be incorporated in order to
auto-focus the ISAR images so that the point scatterer features could be
extracted. The point scatterer histories were then processed to
calculate the COR of the ISAR image. The computational intensity of
many of the processing blocks recommend optimization strategies such as
advanced search algorithms and parallelization.

**Optimizing Algorithms for Antenna Design**

(M.Sc. by Dissertation)

*Abstract—*In this paper two global optimisation techniques namely, the Genetic Algorithm and Simulated Annealing were investigated. These techniques were programmed in and
implemented in SuperNEC which is an Electromagnetics Method of Moments
software simulation program. The optimisation techniques were then applied
to the 5, 6, 12 and 15 element yagi antennas by firstly varying only
element lengths and then by varying element lengths and the spacing between
them. The optimisation techniques were also applied to the 5, 15 and 35
turn axial mode helix antennas by varying the spacing between turns and the
radius of the turns. The results showed improvements to the classical
designs in both cases for both optimisation techniques. Comparisons
between the two optimisation techniques showed that the Genetic Algorithm worked more
effectively in terms of results obtained and efficiently in terms of run
time. The tests also indicated that the Genetic Algorithm works effectively up to at
least a problem with 2.43× 10^{12} possible solutions, while Simulated Annealing could only consistently handle a problem of about half this size.

**An RF and Antenna Systems Toolbox**

(M.Sc. by Dissertation)

*Project Proposal—*Radio frequency (RF) and antenna system toolbox is to be designed and
implemented for the MATLAB environment. The toolbox will contain several
antenna design wizards, based on various design parameters. The design
parameters have to be user-interfaced and the research component of this
project is based on the investigation of the user interface required for
the implementation of the antenna design wizards. The research outcome must
define the best method of interfacing the design parameters to the user.
For example, design curves may be suitable for the design of certain
antennas, whereas ordinary parameter fields will be suitable in some cases.
The toolbox must be implemented and tested against actual measurement
results done on certain antennas and also against ease-of-use. The toolbox
must also have access to the structure assemblies in SuperNEC and enable
the user to simulate antennas without direct access to SuperNEC (SuperNEC
is an antenna analysis program).

(M.Sc. by Dissertation, converted to PhD registration, converted to Part
Time), available at
`http://wiredspace.wits.ac.za/bitstream/handle/10539/7508/RLDreyer_Thesis.pdf`

*Abstract—*The aim of this research was to improve the matrix solution methods for
SuperNEC MoM problems, which is an electromagnetic simulation software
package used to model antennas, and develop a new preconditioner for the
iterative method BICGSTAB(L). This was achieved by firstly implementing the
ATLAS BLAS library optimised for a specific computer architecture. The
ATLAS code primarily makes use of code generation to build and optimise
applications. Comparisons show that the matrix solution times using LU
decomposition optimised by ATLAS is improved by between 4.1 and 4.6 times,
providing a good coding platform from which to compare other techniques.
Secondly the BICGSTAB iterative solution method in SuperNEC was improved by
making use of an alternative algorithm BICGSTAB(L). Systems of equations
that converged slowly or not at all using BICGSTAB, converged more quickly
when using BICGSTAB(L) with L set to 4, despite the high condition numbers
in the coefficient matrices. Thirdly a domain decomposition method, Simply
Sparse, was characterised. Investigations showed that Simply Sparse is a
good compression technique for SuperNEC MoM matrices. The custom Simply
Sparse solver also solves large matrix problems more quickly than LU
decomposition and scales well with increased problem sizes. LU
decomposition is still however quicker for problems smaller than 7000
unknowns as the overheads in compressing the coefficient matrices dominate
the Simply Sparse method for small problems. Lastly a new preconditioner
for BICGSTAB(L) was developed using a modified form of the Simply Sparse
matrix. This was achieved by considering the Simply Sparse matrix to be
equivalent to the full coefficient matrix [A] . The largest 1% to 2% of
the Simply Sparse elements was selected to form the basis of the
preconditioning matrix. These elements were further modified by multiplying
them by a large constant i.e. 1× 10^{7} . The system of equations was
then solved using BICGSTAB(L) with L set to 4. The new preconditioned
BICGSTAB(L) algorithm is quicker than both LU decomposition and the custom
Simply Sparse solution method for problems larger than 5000 unknowns.

**Investigations of Soft Magnetic Materials and a Simulation
Technique to Develop Conformal Antennas in the HF and V/UHF Frequency Range**

(M.Sc. by Dissertation)

*Abstract—*This paper presents a novel conformal antenna concept. Soft magnetic
materials are used to suppress currents on conductive surfaces in order to
realise conformal antennas. An initial material selection is undertaken to
identify suitable materials. A numerical modelling technique is developed
to analyse antennas of this kind. The modelling technique is verified by
comparing simulated and measured antenna characteristics. An initial
feasibility assessment is performed by comparing conventional antennas to
antennas utilising the new antenna concept. Vitroperm 500 F is identified
as the most suitable material. Good comparison between measured and
simulated results was obtained, verifying the simulation technique that was
developed. In addition to conformity the new antenna promises up to 15dB
higher radiation efficiency in the frequency band from 30 to 88MHz when
compared to commercially-available aircraft antennas in this frequency
band.

**
CHARACTERISTICS OF THE LOG PERIODIC DIPOLE ARRAY**

(Multi-supervised, M.Sc. by Dissertation), available at
`http://wiredspace.wits.ac.za/bitstream/handle/10539/2164/0Abstract.pdf`

*Abstract—*The performance of the Log Periodic dipole array antenna has been
characterized, in the form of parametric curves available in most antenna
design handbooks and other relevant literature. These characteristic curves
are often limiting in scope, as for instance they do not contain parametric
curves giving the relationship between the boom-length ’L’ and the number
of dipole element ’N’ for any given bandwidth, even when it is known that
these two parameters are the main cost determinants of a LPDA Antenna. The
concept of convergence is introduced to aid cost optimization of the LPDA
Antenna in terms of number of dipole element ’N’. Although ’N’ is used as
the minimization criterion, the criteria for establishing convergence
encompass all the main electrical characteristics of the LPDA Antenna, such
as VSWR, gain and radiation patterns. Lastly, the effects of boom impedance
’Zo’ and length to diameter ration ’Ln/Dn’, on the performance
characteristics of the LPDA Antenna was investigated with the view to
determining if neglecting the effects of these two parameters were
responsible for the disparity in the directive gain values obtained by R. L
Carrel compared to those obtained by later researchers. The investigation
indicates that if an LPDA Antenna is converged, then the effects of Zo and
Ln/Dn ratio though significant can not alone account for the fairly large
disparity in the gain values. In other to perform these investigations, a
modern scientific tool in the form of numerical modelling by method of
moments based, Super Numerical electromagnetic code version2 was utilized.
The numerical modelling tool was first validated by agreement between
measured values and the values as predicted by the modelling tool. Next,
simulation of the performance of LPDA antennas under variations of their
number of elements was done. Thereafter, the means and standard deviations
of the gain were extracted from the simulated numerical models. Trends in
the pattern of variation of the means and standard deviations of the gain
are used as the basis for deciding the value of number of element at which
the antenna can yield acceptable performance (convergence criteria). These
are presented as convergence curves, which gives for any given boom-length
and operating bandwidth, the minimum number of elements required for the
antenna to yield acceptable performance. Finally, the effect of length to
diameter ratio and boom-impedance on the gain of optimized LPDA antennas
are presented as parametric curves.

**Re-engineering CASED, a machines and drives modelling and
simulation environment**

(Co-supervised, M.Sc. by Dissertation)

*Abstract—*CASED, a drive simulation tool, is re-engineered to allow non-specialist
users to model and analyse complex drive systems. A comprehensive analysis
of the original system is performed to identify the system’s components and
their interrelationships. The strengths and weaknesses are assessed in
order to determine which components of the old system must be replaced and
which can be reused. The shortfalls of the CASED system are overcome by
re-engineering it to resemble the machines and drives domain with the use
of visual interactive modelling while retaining the inherent strengths of
the CASED simulation engine. A graphical, icon-based module interface
called Visual CASED was developed. The simulation model is developed and
displayed in a schematic format that resembles the physical layout of the
drive system itself.

A novel approach has been taken in the software design and development in which the application is advanced through successive working versions of the program evolving from its predecessor. In this way the program grows with user expectations. The software approach used differs from conventional methods because each stage of the software development produces a functional prototype.

Visual CASED’s ease of use has been demonstrated through its use as a teaching aid for a Generalised Machines course at the Department of Electrical Engineering, University of the Witwatersrand. As an example of its use, the transient effects of a resistance starter on a 75kW wound-rotor induction motor driving a conveyor load are investigated.

**SuperNEC Method of Moments and Geometric Theory of Diffraction
Hybrid Evaluation**

(M.Sc. by Coursework and Project Report)

*Abstract—*Super Numerical Electromagnetics Code (SuperNEC) is a MOM/UTD
hybrid program with the MOM component based on the well known Numerical
Electromagnetics Code (NEC) version 2. SuperNEC differs from other
hybrids in that structures can be evaluated using either MOM or UTD, as
well as combining both MOM and UTD in a hybrid at the interaction matrix
level. Tests on a dipole and reflector plate using only MOM, then using a
UTD plate and comparing the results to physical measurements are all in
excellent agreement with differences of less than 5Ω (6%) for the
real part of the impedance and 10Ω (12%) for the imaginary part. A
Hybrid using both MOM and UTD on a single structure is also possible
whereby large parts of the structure are constructed using UTD and smaller
more detailed areas constructed using MOM. A hybrid reflector plate was
tested and compared to the MOM reflector plate. The error was indirectly
proportional to the size of the UTD plate, confirming that the UTD rule
still holds, whereby for accurate results the UTD plate lengths must be
larger than a wavelength in size. A hybrid structure has been shown to be
possible as long as the part of the structure to be modelled using UTD is
larger than a wavelength in size.

This document was translated from L^{A}T_{E}X byH^{E}V^{E}A.