Novel design techniques are presented for load-resonant and quasi-resonant converters for use in, for example, arc-welding and fan-load power supplies. Both converters are capable of very high switching frequency over a wide range of output power, with high efficiency and the presentation of near-unity power factor to the primary power supply. Previous work, by the author and his colleagues, has produced. a frequency-domain approach to produce circuit-designs for use in load-resonant converter applications. This design technique, although simple and straight-forward to understand, suffers by requiring some rather tedious trial-and-error algebraic and arithmetic manipulations albeit computer assisted. In this thesis, a systematic way of designing such circuits, based on Gr6bner Basis ideas, is explained, developed and compared with the previous best-practice design method. By employing the Gr6bner Basis techniques to synthesize electrical circuits, an entirely novel approach to the design of series-parallel load-resonant converter circuits is presented. This has led to the formulation of a new output-power control methodology in the design of the converters. These techniques produce output-power-control designs that have superior properties, compared with other established methods, in the sense of their simplicity, robustness and flexibility. It is found that the methodology can be further extended to alter any resonant circuits and, hence enables multilevel-output power to be controlled without involving complex control and advanced mathematical theories, while still preserving the desirable characteristics of resonant switching. The technique is, in fact, far more generally useful in the circuit-design/synthesis arena than the specific load-resonant-converter application for which it was developed. The novel technique used to vary the speed of an induction motor is found to be promising. Various test results are presented based on an experimental system.

Though the hybrid stepping motor has a long and proven history, in terms of toughness, accuracy of position and the ability to operate in open loop, motor performance improvements can still be made in terms of the physical structure of the motor's components. It is impossible to build a complete solution of the hybrid stepping motor using simple analytical functions or equivalent circuit representations. This is due to the difficulties introduced by the motor's highly non-linear three dimensional magnetic structure, of which the doubly salient tooth structure, axial magnet, and back iron all complicate the situation. However, with the recent advances in three dimensional finite element software a comprehensive study of the motor has been achieved in this thesis. This has allowed improvements to simpler two dimensional based mathematical models, which allow faster computation of the motor's electromagnetic performance. To aid modelling, novel equations which accurately model today's high permeability steels have been developed. These are shown to be more accurate than the established Jiles-Atherton method. Inductance calculations of the steel's flux paths have been comprehensively improved by the use of elliptical functions. The thesis concludes with the design of two quite individual new machines. The first dramatically improves a motor's power output, smoothness, noise levels, and resonance by modifying the tooth structure. The second uses soft magnetic composite materials to provide an isotropic path for cross lamination flux which flows in a stator's back iron. Both new designs are shown to offer a significant improvements to the high speed torque capability of the hybrid stepping motor.

This thesis is concerned with the study of secure wireless routing protocols, which have been deployed for the purpose of exchanging information in an adhoc networking enviromnent. A discrete event simulator is developed, utilising an adaptive systems modelling approach and emergence that aims to assess networking protocols in the presence of adversarial behaviour. The model is used in conjunction with the characteristics that routing protocols have and also a number of cryptographic primitives that can be deployed in order to safeguard the information being exchanged. It is shown that both adversarial behaviour, as well as protocol descriptions can be described in a way that allows for them to be treated as input on the machine level. Within the system, the output generated selects the fittest protocol design capable of withstanding one or more particular type of attacks. As a result, a number of new and improved protocol specifications are presented and benchmarked against conventional metrics, such as throughput, latency and delivery criteria. From this process, an architecture for designing wireless routing protocols based on a number of security criteria is presented, whereupon the decision of using particular characteristics in a specification has been passed onto the machine level.

As the most frequently used waveguide type in microwave and millimeter wave integrated circuits, microstrip theory occupies an important position in understanding the behaviour of such structures. In this thesis it is shown that coupled microstrip, through its several degrees of freedom can be used to achieve predictable state of the art performance of active devices such as a multioctave medium power amplifier operating over the 2 - 18 GHz frequency band with +29 dBm saturated output power and 20 percent power added efficiency. Both quasi-static and full wave and analytical techniques are covered for coupled microstrip lines. In depth analysis of edge-coupled multiconductor suspended microstrip with tuning septum, as well as multiconductor broadside coupled lines with position dependent coupling coefficient, are presented. Important relationships between the mechanical dimensions and such parameters as coupling factor and phase velocity are also derived. The technique based on spectral domain analysis uses considerable analytical preprocessing to eliminate the need for the sophisticated computer facilities which would perviously be required to analyze such complex structures. Based on the above mentioned technique, novel broadband planar hybrids, magic-T's and matching structures are proposed and analyzed.

A power supply incorporating a series-parallel load-resonant converter, capable of very efficient operation over a wide range of output power is presented. The series-parallel load-resonant converter is shown to have three pairs of resonant frequencies. Operation of the circuit at each of these resonant frequencies maintains zero current switching and high frequency operation. Design mathematics is developed which allow series-parallel load-resonant converters to be designed with specific resonant frequencies and circuit resistances. A new method of power control for series-parallel load-resonant converters is presented; the power delivered to the circuit and hence the load is shown to var substantially depending on which resonant frequency the circuit is excited at. Two series-parallel load-resonant converters are designed simulated, constructed and tested. There is good agreement between the simulation and experimental results. One of the circuits produces an output current of 200 A while the second demonstrates the new power control technique pulsing between 55 A and 145 A while running at frequencies of 63 kHz and 100 kHz. The new power supply is particularly suited to arc-welding. It contains an active rectifier and draws near unity power factor.

Recent years have seen unprecedented growth in the popularity and deployment of mobile phones. As this continues, so the strain on existing mobile cellular radio network has also increased, leading to the need to investigate new technologies to relieve this pressure. The problem is being further exacerbated by the introduction of the 3rd generation of mobile communications, otherwise known as UMTS (Universal Mobile Telecommunications System), with the aim of offering multimedia services on pocket sized portable receivers. A major cost of the mobile radio network, in terms of both financial and social/environmental aspects, is the need apparent need for more base transceiver stations (BTS), due to the increased number of services, and the density of them. Therefore, judicious use of fewer, but more "intelligent" base stations, thereby reducing the overall system costs, and extra flexibility in the design of mobile cells would be desirable. This can be achieved by having the BTS antennas remotely positioned from the BTS by transmitting the radio signals down an optical fibre or, as in this project, over free space. The main application for this is in densely urban heavy use areas, where there is extensive reuse of both cell and cell cluster. This, along with building shadowing, would require a BTS on every corner, and where extra cell design flexibility would be desirable. Also, in remote rural areas, where various natural features, such as rivers or mountains can cause similar cell design problems, there is a need for this flexibility. The problem with this requirement is that the electrical to optical conversion process, involving a laser diode driver unit, is inherently non-linear, and, unless this is resolved, the desired signal will become unusable due to distortion. To overcome these nonlinearities, a novel correction may be used, based on an optical feedforward correction technique. The prototype system employs off-the-shelf components, and has one Fabry Perot laser diode (FP-LD) providing two signals (via a beam splitter), for a main path and one for the error path loop. The error path signal is detected by a receiver circuit, then mixed with a reference signal to produce a 'pure' error signal, which then modulates the second FP-LD. In contrast with previous fibre feedforward systems, where the two LD outputs are then combined in the optical fibre pre-reception, this system has to combine the signals post-reception. After the main signal and error signal are received and recombined, the non-linearities of the main path are predominantly cancelled by those present in the error path signal, leaving only the desired signal, free of non-linearities.

An analysis of acoustic noise in electronically controlled variable speed drives is presented. The causes of vibration and acoustic noise in switched reluctance motors are discussed and it is shown that brushless d.c. motors can produce resonant vibration and acoustic noise by similar mechanisms. The flux switching motor is introduced. This new class of reluctance motor is an advance on the established switched reluctance motor, retaining many of its benefits, but with a simpler and cheaper power electronic converter. The phase windings and method of flux control are different and tests are performed to quantify the effect on the acoustic noise produced. Measurements of acoustic noise are made on one flux switching motor and one 2-phase switched reluctance motor, made from the same laminations and mechanical components. It is shown that the flux switching motor produced 2dB less acoustic noise under the same conditions. Finite element analysis is used to calculate the radical force profiles of the two motors during normal rotation, and further analysis of this data provides evidence to support the experimental results. The experimental results go on to show how the acoustic noise from a second flux switching drive was found to be comparable to that of a split phase induction motor.

Broad-band microwave integrated circuit (MIC) amplifier design is a complex, multi-disciplinary process. This work focuses on three important aspects: the behaviour of microstrip transmission lines, discontinuities, and related structures; the accurate measurement of components and devices mounted in microstrip circuits; and the circuit design methodology. Techniques for microstrip quasi-static analysis are reviewed in order to identify methods suitable for extension to deal with the effects of substrate anisotropy. An integral equation method is described and the anisotropic Green's function derived using an extension to the method of partial images. Proposed transform methods are assessed and the preferred option implemented by adaption of a microstrip analysis computer program. A method, by which accurate measurements of microstrip properties may be made, is developed. Involving measurements of the resonant behaviour of half-wavelength short circuit resonators with two arbitrary coupling conditions, this technique allows the unloaded properties to be deduced. Results for microstrip on a sapphire substrate concur with the analysis. A pragmatic but effective approach to the calculation of the capacity component of microstrip discontinuities, and some other three dimensional MIC structures, is described and developed to allow existing data for isotropic substrates to be applied to the anisotropic situation. The computer corrected network analyser (CCNA) is a widely used microwave measurement tool. Weaknesses in popular correction strategies are identified and remedies developed. In particular, revised calibration equations that better accommodate test port mismatch variation with s-parameter selection, and a model for quadrature error are presented. A 2-port calibration scheme suitable for use with MIC transmission lines, using only simple standards, is described. The standards are partially self-calibrating;the values of propagation constant, loss, and end effect are deduced in the calibration process. An effective jig for use with microstrip is described and the results of measurements on microwave transistors presented. Conventionally microwave amplifiers are designed using reactive components both to achieve good port matches and compensate the frequency dependent gain of the active devices. The problems associated with this approach are enumerated and the alternatives reviewed. A methodology which combines the benefits of frequency dependant dissipative networks with the elegance of reactive network synthesis is described. The device gain slope is compensated by simple lumped or distributed circuits incorporating a resistive element to produce a composite `device' with a specififed (flat) maximum available gain frequency response. Reactive matching networks are then used to interface these gain blocks. By this structured approach the amplifier gain breakdown can be defined at the outset and preserved through the design process. Other advantages stemming from the use of dissipative compensation include improved tolerance to device parameter and component value scatter, reduced group delay variations and enhanced reverse isolation. The method is demonstrated by the design and characterisation of 4 to 9 GHz amplifier having a representative specification. The close conformance of the performance of the untrimmed amplifier to that predicted by computer simulation testifies to the inherent accuracy of the design method, the microstrip (and related structures) analysis techniques and the CCNA MIC calibration scheme.

This thesis is a description of integrated injection logic, a bipolar large scale integration technique. Various alternatives to integrated injection logic are discussed and the advantages and disadvantages of each are outlined. The integrated injection logic structure is introduced and its advantages over other solutions are described. Those device characteristics necessary for successful operation of the integrated injection logic gate are established, and the physical mechanisms controlling D. C. and A. C. performance are investigated theoretically. These theoretical investigations are compared with experimental observations. The behaviour of the I2L gate is shown to be very dependent on the characteristics of the epitaxial emitter and the intrinsic base region of the device. In an extension to the basic device theory it is shown that the device characteristics can be related to the transistor characteristics in the conventional mode of operation. As the technique is primarily for large scale integration a considerable effort has been placed on yield studies. These yield studies have included work on the following: parametric control, photoengraving and silicon crystallographic defects. Silicon crystallographic defects are shown to be a major yield hazard and appropriate actions to eliminate them as failure mechanisms are described.

In recent years, there has been considerable interest in the use of gas/vapour monitors and electronic nose instruments by the environmental, automotive and medical industries. These applications require low cost and low power sensors with high yield and high reproducibility, with an annual prospective market of 1 million pounds. Present device and sensor technologies suffer a major limitation, their incompatibility with a standard silicon CMOS process. These technologies have either operating/annealing temperatures unsuited for MOSFET operation or an inappropriate sensing mechanism. The aim of this research is the development of CMOS compatible gas/vapour sensors, with a low cost of fabrication, high device repeatability and, in the future, transducer sensor amalgamation. Two novel approaches have been applied, utilising bulk CMOS and SOI BiCMOS. The bulk CMOS designs use a MOSFET sensing structure, with an active polymeric gate material, operating at low temperatures (<100°C), based on an array device of four elements, with channel lengths of 10 μm or 5 μm. The SOI designs exploit a MOSFET heater with a chemoresistive or chemFET sensing structure, on a thin membrane formed by the epi-taxial layer. By applying SOI technology, the first use in gas sensor applications, operating temperatures of up to 300 °C can be achieved at a power cost of only 35 mW (simulated). Full characterisation of the bulk CMOS chemFET sensors has been performed using electrochemically deposited (e.g. poly(pyrrole)/BSA)) and composite polymers (e.g. poly(9-vinylcarbazole)) to ethanol and toluene vapour in air. In addition, environmental factors (humidity and temperature) on the response and baseline were investigated. This was carried out using a newly developed flow injection analysis test station, which conditions the test vapour to precise analyte (<15 PPM of toluene) and water concentrations at a fixed temperature (RT to 105°C +- 0.1), with the sensor characterised by either I-V or constant current instrumentation. N-channel chemFET sensors operated at constant current (10 μA) with electrochemically deposited and composite polymers showed sensitivities of up to 1.1 μV/PPM and 4.0 μV/PPM to toluene vapour and to 1.1 μV/PPM and 0.4 μV/PPM for ethanol vapour, respectively, with detection limits of <20 PPM and <100 PPM to toluene and <20 PPM and 10+ PPM to ethanol vapour (limited by baseline noise), respectively. These responses followed either a power law (composite polymers) or a modified Langmuir isotherm model (electrochemically deposited polymers) with analyte concentration. It is proposed that this reaction-rate limited response is due to an alteration in polymers work function by either a partial charge transfer from the analyte or a swelling effect (polymer expansion). Increasing humidity caused, in nearly all cases a reduction in relative baseline, possible by dipole formation at the gate oxide surface. For the response, increasing humidity had no effect on sensors with composite polymers and an increase for sensors with electrochemically-deposited polymers. Higher temperatures caused a reduction in baseline signal, by a thermal expansion of the polymer, and a reduction in response explained by the analyte boiling point model describing a reduction in the bulk solubility of the polymer. Electrical and thermal characterisation of the SOI heaters, fabricated by the MATRA process, has been performed. I-V measurements show a reduction in drain current for a MOSFET after back-etching, by a degradation of the carrier mobility. Dynamic measurement showed a two stage thermal response (dual exponential), as the membrane reaching equilibrium (100-200 ms) followed by the bulk (1-2 s). A temperature coefficient of 8 mW/°C was measured, this was significantly higher than expected from simulations, explained by the membrane being only partially formed. Diode and resistive temperature sensors showed detection limits under 0.1°C and shown to measure a modulated heater output of less than 1°C at frequencies higher than 10Hz. The principal research objectives have been achieved, although further work on the SOI device is required. The results and theories presented in this study should provide a useful contribution to this research area.