Over the last decade there has been a growing increase in research in the field of vibrational energy harvesting - devices which convert ambient vibrational energy into electrical energy. The major application area for such devices is as power sources for wireless sensors, thereby replacing currently used batteries which suffer from a finite lifespan and pose environmental issues during disposal. The vast majority of designs are cantilever beams comprising of piezoelectric layers having coverage identical to the substrate layer. It is evident from the literature that rudimentary work has been performed on design optimisation, with reliable and extensive parametric studies on geometry, especially piezoelectric layer coverage, being overlooked. As a result of this, outcomes from previous research are yet to be seen in designs for practical applications. In this work a versatile linear model is developed which can accurately predict the performance of cantilever piezoelectric energy harvesters. An integral part of the model uses a transfer matrix method to accommodate the difference in structural dynamics of both uniform and non-uniform structures with model validation provided through extensive experimental work. The linear model developed is used to carry out parametric studies on the geometry of three distinct energy harvester cases thereby providing comprehensive knowledge on key variables and geometrical changes which can improve performance. In one of the cases examined, an improvement in performance of over 100% is predicted by solely altering piezoelectric layer coverage. However, the load resistance, i.e. electrical condition, has a significant effect on the trends in generated power which led to work directed toward harvester optimisation in a more realistic electrical scenario. Investigation on harvester geometry whilst utilising an electrical scenario comprising of an energy storage medium is undertaken in this work. The developed model ensures the effects of electro-mechanical coupling remain and provides a solid basis from which users can readily apply model extensions through inclusion of further electrical components to resemble practical circuitry. Theoretically, for all examined case studies, improvements in performance were realised through alterations to piezoelectric layer dimensions with the most notable result indicating an improvement of over 200% during optimisation of piezoelectric layer length. In conjunction to theoretical findings, outcomes of extensive experimental work are provided in order to highlight the accuracy and reliability of the presented theoretical models in both electrical scenarios. Variation in mechanical damping magnitude plays a pivotal role throughout experimental testing and is one key factor in explaining why devices comprising of shorter piezoelectric layers have high performance. A methodology behind unbiased design comparisons is also provided in this work, and involves comparing devices with identical fundamental frequencies. The reasoning behind this approach is to allow for each device to perform as efficiently as possible in the same excitation scenario. Systematic alterations to multiple geometric parameters are used to achieve this. Geometric parameters such as the substrate thickness are observed to provide adequate frequency control. Using this approach, performance improvements from adjustments to piezoelectric coverage still remain. The occurrence of non-linearity in piezoelectric materials is a widely known phenomena and so lastly, a more robust model is provided which incorporates material and geometric non-linearity. This model is useful in determining dynamical responses of uniform and non-uniform piezoelectric energy harvesters when subjected to moderate-to-high acceleration levels. A thorough validation of the theoretical model is achieved using extensive experimental data obtained from a range of samples. For the harvester composition tested in this work, the occurrence of mild non-linearity at base acceleration levels as low as 1 meter per second squared is witnessed with softening behaviour causing the resonant frequency to decrease with base acceleration. In order to avoid reduced efficiency in the final application, the prediction of possible frequency shifts is vital during the design process.

The work contained within this thesis seeks to address the dielectric and magnetic properties of bismuth substituted barium hexaferrite (composition BaBixFe12-xO19 where x = 0.0, 0.2, 0.5, 0.8, 1.0 and 1.5) across a broad frequency spectrum. This material is potentially of interest in antenna applications (specifically a dielectric resonator antenna) where high permittivity and permeability materials can be exploited to physically minimize the size of antenna devices. At low frequencies (20Hz - 3MHz) LCR meter based analysis was used to investigate the capacitance and inductance of the materials. The highest measured permittivity at 1MHz was Er = 86.18 - 9.910j, tanδ = 0.11. High frequency (45MHz - 20GHz) permittivity was investigated using a vector network analyser (VNA) and coaxial probe. The highest recorded permittivity at 2.45GHz was Er = 14.69 - 1.664j, tanδ = 0:08. A link between substitution level and permittivity and substitution level and inductance was made at either 1MHz or 2.45GHz. The homogeneity of the samples was explored using a near- field permittivity sensor. This showed small localized variations of permittivity across the surface but not significant enough to adversely effect bulk measurements. A dielectric resonator antenna was fabricated using a bismuth substituted barium hexaferrite sample. Resonant frequencies were identified at 6.6, 9.55 and 13.6GHz and radiation patterns for the system showed agreement with published theoretical results. This confirmed the suitability of the material in this application as a high permittivity material with relatively low loss. This thesis contributes to scientific knowledge by characterising a broad range of bismuth composition materials across a wide frequency range, investigating the link between permittivity/inductance and doping level and presenting full results. The material has also been characterised using a near-field permittivity technique and used to fabricated a dielectric resonator antenna which have not previously been undertaken.

Many problems in electrical engineering are associated with the way in which electric and magnetic fields propagate and distribute themselves in various media. Maxwell's equations provide a concise description for the interaction of fields with themselves and with the various boundaries of a problem. Therefore, a numerical procedure for the solution of these equations is an important consideration. This thesis shows how a general three-dimensional medium may be represented by an interconnection of continuous ideal two-wire transmission-lines made up of generalised two-dimensional nodes which are introduced in the earlier chapters. It is then shown how this model may be used for the numerical solution of the electric and magnetic vector fields within the medium. This is the TLM method of numerical analysis. A universal three-dimensional computer program based on the method is also introduced. This program has been written in only 110 lines of FORTRAN including the subroutines. The ease of application, versatility and accuracy of the TLM method is demonstrated by analysing a wide variety of microwave resonators using this program. The surface mode phenomenon of microstrip is also investigated.

Nowadays, power electronic converters based exclusively on IGBTs seem to have achieved excellent load side performance up to megawatt powers range in the low voltage range (200-690Vrms) and are steadily gaining good performance in the medium voltage range as well. However, the medium and high voltage/high power range remains dominated by converters using naturally commutated thyristors, such as line-commutated cycloconverters, line-commutated current source inverters, which provide comparatively poorer output side performance. The purpose of this thesis is to investigate both the conventional cycloconverter, which will be referred as standard cycloconverter in the thesis, and the new hybrid cycloconverter topologies, which are capable of improving the performance of the standard cycloconverter by adding an auxiliary forced commutated inverter with reduced installed power. It will be shown that the new topology is not only able to improve the quality of the output voltage, but also to enhance the control over the circulating current and therefore, for some of the standard cycloconverter arrangements, to improve the input power quality. To realize the evaluation of the standard cycloconverter and validate the feasibility of the new hybrid cycloconverter in both circulating current and circulating current-free mode, SABER simulation models are developed in the first place to perform the initial analysis. A configurable three-phase input to three-phase output cycloconverter prototype which can be easily changed via a switch box to test four different cycloconverter topologies (standard and hybrid) is designed and implemented in the laboratory. Finally, the whole system is debugged and tested. All the relevant results obtained from both the simulation and experiment will be thoroughly analyzed in the thesis.

Digital filtering (DF) techniques are receiving significant interest, because they can represent fine features such as vias, thin-panels and thin-wires in full-field solutions of electromagnetic problems with significant savings in computational costs. However, a limitation of this technique is that DF can only represent a fine feature as a plane or as an internal boundary. In other words, an internal boundary can represent the electromagnetic properties of a fine feature in one dimension or two directions. The DF technique is usually involved with time domain solvers such as the Finite-difference time-domain (FDTD) and the Transmission Line Modeling (TLM) methods. Both of them are commonly used to investigate the electromagnetic fields in the problem spaces. Here the TLM method is selected for demonstrating the DF technique. This thesis presents the formulation of TLM in three-dimensions in order to investigate the limitations of the DF technique and the solutions. As a result, new techniques have been developed. These techniques can be applied to the three dimensional TLM method in order to represent the fine features in three-dimensions appropriately. The developed techniques were demonstrated using some examples of three-dimensional embedded objects, such as conducting volumes and dielectrics. Their accuracy and efficiency are compared with the standard TLM method in the time and frequency-domain. The results show good agreement between these techniques and the standard TLM method.

With the advantages of low cost, low power consumption, high reliability and potential for large scale integration, CMOS monolithically integrated active pixel chips have significant application in optical sensing systems. The optical front end presented in this thesis will have application in Optical Scanning Acoustic Microscope System (O-SAM), which involves a totally non-contact method of acquiring images of the interaction between surface acoustic waves (SAWs) and a solid material to be characterized. In this work, an ultra fast optical front-end using improved regulated cascade scheme is developed based on AMS 0.35mm CMOS technology. The receiver consists of an integrated photodiode, a transimpedance amplifier, a mixer, an IF amplifier and an output buffer. By treating the n-well in standard CMOS technology as a screening terminal to block the slow photo-generated bulk carriers and interdigitizing shallow p+ junctions as the active region, the integrated photodiode operates up to 4.9 GHz with no process modification. Its responsivity was measured to be 0.016 A/W. With multi-inductive-series peaking technique, the improved ReGulated-Cascade (RGC) transimpedance amplifier achieves an experimentally measured -3dB bandwidth of more than 6 GHz and a transimpedance gain of 51 dBW, which is the fastest reported TIA in CMOS 0.35mm technology. The 5 GHz Gilbert cell mixer produces a conversion gain of 11 dB, which greatly minimized the noise contribution from the IF stage. The noise figure and input IIP3 of the mixer were measured to be 15.7 dB and 1.5 dBm, respectively. The IF amplifier and output buffer pick up and further amplify the signal for post processing. The optical front end demonstrates a typical equivalent input noise current of 35 pA=pHz at 5 GHz, and a total transimpedance gain of 83 dB ohm whileconsuming a total current of 40 mA from 3.3 V power supply. The -3 dB bandwidth for the optical front end was measured to be 4.9 GHz. All the prototype chips, including the optical front end, and the individual circuits including the photodiode, TIA, mixer were probe-tested and all the measurements were taken with Anritsu VNA 37397D and Anritsu spectrum analyser MS2721A.

Characterization of electromagnetic emissions from printed circuit boards (PCBs) is an important issue in electromagnetic compatibility (EMC) design and analysis of modern electronic systems. This thesis is focused on the development of a novel modelling and characterization methodology for predicting the electromagnetic emissions from PCBs in both free space and closed environment. The basic idea of this work is to model the actual PCB radiating source with a dipole-based equivalence found from near-field scanning. A fully automatic near-field scanning system and scanning methodology are developed that provide reliable and sufficient data for the construction of equivalent emission models of PCB structures. The model of PCB emissions is developed that uses an array of equivalent dipoles deduced from magnetic near-field scans. Guidelines are proposed for setting the modelling configuration and parameters. The modelling accuracy can be improved by either improving the measurement efforts or using the mathematical regularization technique. An optimization procedure based on genetic algorithms is developed which addresses the optimal configuration of the model. For applications in closed environments, the equivalent model is extended to account for the interactions between the PCB and the enclosure. The extension comprises a dielectric layer and a ground plane which explicitly represent the necessary electromagnetic passive properties of a PCB. This is referred to as the dipole-dielectric-conducting plane (DDC) model and provides a completely general representation which can be incorporated into electromagnetic simulation or analysis tools. The modelling and characterization methodology provides a useful tool for efficient analysis of issues related to EMC design of systems with PCBs as regards predicting electromagnetic emissions in both free space and closed environment. The proposed method has significant advantages in tackling realistic problems because the equivalent models greatly reduce the computational costs and do no rely on the knowledge of detailed PCB structure.

The aim of the work presented in this thesis is to improve the operational reliability of a power module and increase the efficiency of its associated cooling system by integrating the design of the cooler as part of the module. Power modules are increasingly used in a variety of applications ranging from aircraft and mass transport systems, to motor control and power conversion in the home. Reliability of the power module is very important in aerospace applications where the highest levels of safety and robustness are required while keeping the volume and mass of the module as low as possible. Certain parts of the power module such as the solder layer beneath the silicon device and the substrate are prone to failure with thermal cycling. The layer of thermal grease between the baseplate of the module and the heatsink significantly increases the thermal resistance between the electronic devices and the coolant fluid. The power module can be constructed so that some of the interfaces within the module which are prone to failure are improved or completely removed from the assembly greatly reducing the thermal resistance from junction to ambient. The research identified cooling methods which are able to cope with the increasingly high heat fluxes produced by power electronic devices. Jet impingement cooling was selected for testing and further development. An initial series of tests confirmed that liquid jet impingement can be used to generate high heat transfer coefficients for the efficient cooling of power modules. Results from experimental tests showed that directly cooling the substrate tile with jet impingement resulted in the devices being cooled more effectively compared to the commonly used serpentine coldplate and a direct-baseplate cooled jet impingement system. It was postulated that more efficient cooling can be achieved by targeting the hotspots on the substrate beneath each device with a carefully designed impingement array. A test apparatus was constructed to test a variety of jet impingement arrays to confirm the hypothesis. A second test apparatus was constructed to characterise the performance of the jet arrays in more detail using a thermal imaging camera to monitor the surface temperature of a single device. An optimal jet configuration was found for the efficient cooling of a single device. The work concluded that an improvement in efficiency and reliability can be gained by constructing power modules with integrated jet impingement arrays direct-substrate cooling the hotspots beneath the devices.

Nowadays, there has been an increasing interest in Terahertz (THz) radiation for application across scientific disciplines including atmospheric sensing, medical diagnosis, security screening and explosive detection. The limitation of THz generators and detectors has gained interest from scientists and engineers to explore the development of both sources and detectors. With the advantages of low cost, low power consumption, high reliability and potential for large-scale integration, sub-THz generator and detector can be developed using CMOS process technology. In this thesis, an IMPATT diode acts as a sub-THz generator, HEMTs and MOSFETs act as sub-THz detectors, which are developed in AMS 0.35 μm CMOS technology and UMC, 0.18 μm CMOS technology. The size of the IMPATT diode was 120 μm x 50 μm with the target resonant frequency at 30 GHz. The experiment results show that the operating frequency of the IMPATT diode was between 12 GHz up to 14 GHz. Then by using HEMTs with 0.2 μm gate length and 200 μm gate widths, sub-THz radiation detection has been demonstrated. Experimental results show that the photoresponse depends on the drain current and the gate to source voltage VGS. In addition, photoresponse also depends on varying frequencies up to 220 GHz and fixed the drain current. Furthermore, the HEMT also give an indication of response by varying the input power of microwave extender. MOSFETs from two types of CMOS technology; AMS 0.35μm and UMC 0.18 μm technology with different gate length ranging from 180 nm up 350 nm were demonstrated. These results provide evidence that the photoresponse increases with the drain current and the RF input power, but inversely to the frequencies. These results also provide evidence that the MOSFETs are able to work as low cost and sensitive sub-THz detector.

The trajectory of the Taiwanese economy over past decades has reflected transitions in global geo-economy towards a vertical specialisation of global trade, knowledge-based competition, variation of industrialisation in the Third World, regional trisection of the world economy, etc. Its industry, making remarkable progress based on a long-term national assistance, is involved in the outward direct investment whereby overseas production is arranged. Such strategic actions of industrial capital slice through national boundaries and, meanwhile, incorporate state-business relationships on a broader scale at a national level. The thesis seeks to portray the transnationalisation of the Taiwanese IC sector through its specific organisational processes and spatial dynamics with an aim to understanding the way that indigenous firms are associated with the home state and the convergence of IC production systems on subnational, transnational and global scales. This work finds that the outward expansion of the IC industrial capital reflects a spatial trend converging on China, the Yangtze River Delta in particular, owing to a reshuffle of the global electronic production chain. However, flow of capital and material along the chain across the Taiwan Strait move along a circuitous route. It also finds that a persistent inter-state feud accounts for the domestic debates over the westward investment of the IC capital. There is tension between the neo-liberalist logic of business practices and national intervention in a guided capitalist stance. It underlines the conflicts of imperative territoriality. On the one hand, the firms recognise the necessity of stretching industrial territories to the Mainland in consideration of sectoral competitiveness. On the other hand, what the state is concerned with lies not only in the impacts of industrial de-territorialisation upon domestic industries but also in the result of over-dependence upon China that would incorporate national economy into economic territories of the Great China Circle.