Made available in DSpace on 2014-10-09T12:54:00Z (GMT). No. of bitstreams: 0 / Made available in DSpace on 2014-10-09T14:09:21Z (GMT). No. of bitstreams: 1 12436.pdf: 2191865 bytes, checksum: 12b10ebed00b6647c2de939f5f567b1d (MD5) / Tese (Doutoramento) / IPEN/T / Escola Politecnica, Universidade de Sao Paulo - POLI/USP

Made available in DSpace on 2014-10-09T12:54:04Z (GMT). No. of bitstreams: 0 / Made available in DSpace on 2014-10-09T14:09:09Z (GMT). No. of bitstreams: 1 12440.pdf: 3445926 bytes, checksum: 53acfb637b1d63432057f034c3999fda (MD5) / Dissertacao (Mestrado) / IPEN/D / Escola Politecnica, Universidade de Sao Paulo - POLI/USP

Estudos de transitórios eletromagnéticos são importantes fontes de informação para que os transformadores sejam dimensionados de maneira correta. No entanto, para que tais estudos sejam bem sucedidos, os modelos utilizados devem refletir com fidelidade o comportamento do equipamento. Este trabalho mostra como os elementos do modelo de um transformador são influenciados pelas dimensões geométricas de sua parte ativa. Também introduz uma formulação alternativa, para o transformador saturável (STC) do ATP, desenvolvida dentro do programa MATLAB. Os ramos RL foram representados usando o Método de Integração Trapezoidal e a magnetização foi equacionada pelo Método da Compensação. Uma das contribuições que esta dissertação oferece é a possibilidade de identificar erros numéricos que ocorrem em simulações do ATP, bem como permitir a interpretação de resultados que apresentem oscilações numéricas. / Electromagnetic transient studies are an important source of information to develop transformer dimensioning. But, for the success of that purpose, it is important the models which are being used reflect with fidelity the behavior of the machine. This lecture presents how the transformer model elements are influenced by the active part geometrical dimensions. It also introduces an alternative formulation for the ATP saturable transformer (STC), written inside the MATLAB program. The RL branches are represented using the Trapezoidal Rule and the magnetization by the Compensation Method. One of the contributions of this dissertation is the possibility to identify numerical errors that occur in ATP simulations, and also permit numerical oscillatory results interpretation.

D.Ing. / In the field of power electronics, the relentless demand for higher efficiencies, lower costs and ever-decreasing volume and profile have driven the development of many innovative technologies. Planarization and hybridization have become a substantial part of present system integration methodology. With the subsequent size reduction, the effects of layout and component parasitics are becoming vital issues in the development of innovative structures. The component ‘parasitics’ can be considered dimensional effects of the component structure that are not considered during the design process. The concept of electromagnetic integration is aimed at the utilization and modification of these dimensional effects. This leads to an integrated structure that fulfills multiple electromagnetic functions with the potential for improved power density, efficiency and reliability. In this dissertation, a family of electromagnetically integrated passives is presented. The related electromagnetic modeling and design approach of these complex electromagnetic structures is presented through a case study of L-L-C-T structures. The development of a sufficiently accurate, yet simplified electromagnetic model for design purposes is presented for the case study. With the electromagnetic model as basis, a comprehensive electromagnetic loss model is created. The electromagnetic design and loss models are combined into a design evaluation program. The graphical output of this design evaluation program allows for rapid selection of improved designs based on external cost criteria. This led to numerous insights into the relationships between the design variables. Through modification of the program, some fundamental limits of the integration approach are addressed. A case study design for a 1MHz, 500W dc-dc converter was considered to evaluate the design program. To assess the accuracy of the electromagnetic modeling, three L-L-C-T prototypes are constructed and experimentally tested. The construction process presented improved power density by 80% over previous processes. The electromagnetic component parameters for three prototypes were within 10% of the required design values, while the electromagnetic loss model estimations were within measurement error. The design evaluation program was enlisted in the design of two of these prototypes. This resulted in a 100% further improvement in power density (480W/in3 or 29.3W/cm3) compared to the original prototype without a loss in efficiency.

Numerical modeling of electromagnetic fields is becoming an important topic in such diverse areas as microwave and RF engineering, antenna design, bio-electromagnetics, and electromagnetic compatibility and interference (EMC/EMI). Among several techniques, time-domain schemes are of particular interest, due to their high flexibility and ease of implementation. This thesis is focused on the Transmission Line Matrix (TLM) method, based on a space and time discrete formulation of Maxwell's equations. The objective of this thesis is to develop, implement and test a number of techniques aimed to the enhancement of the accuracy of the method without increasing the computational load. The link between the electromagnetic theory and the TLM updating equations is first investigated, creating a solid background for the implementation of hybrid schemes characterized by better accuracy. The problem of coarseness error is in particular addressed. Two methods are proposed and analyzed. In the first approach the knowledge of the relationship between field equations and TLM equations is exploited to incorporate the static field behavior in the vicinity of singularities into the three-dimensional TLM mesh. Secondly, the field distribution around a corner is represented in terms of an equivalent circuit derived from a quasi-static approximation of the Green's functions for an infinite conductive wedge. As a result, relatively coarse TLM meshes, in combination with hybrid schemes, can be used to obtain highly accurate results, within the dispersion error margin, across a wide frequency range. By taking advantage of these techniques it is possible to incorporate more information of the structure under study into the TLM solution, thus creating an accurate and efficient CAD tool. / Graduate

M.A. / Emergence of informal settlements on vacant lands in urban areas results from intuition and political decisions, and sometimes cultural background. So any available vacant land is used for settlement development and dwelling sites. An informal settlement named after Elias Motsoaledi, a political activist of the African National Congress (ANC), was developed in the south-western part of the city of Johannesburg in 1993. Motsoaledi informal settlement was established on a vacant piece of land where there are overhead power lines. Motsoaledi community members that have their dwellings in close proximity and even underneath overhead power lines are exposed to an environment that is surrounded by electrical magnetic fields (EMFs). There are no reported cases of biological and health consequences that have been caused by EMFs as yet. This study aims to focus on the health aspect of persons who have their dwelling in the vicinity of the overhead power lines in Motsoaledi. The high levels of EMFs around Motsoaledi is further excelerated by the zinc structures that are conductive to electricity and may result in electrocution for dwellers that are in the direct vicinity of overhead power lines. Longer exposure to athermal conditions may result in bio-effects and lead to changes in human health. Furthermore, effects from EMFs are possible electrocution, often causing immediate death of persons and destruction of property. Due to the lack of proper planning of informal settlements, such settlements become vulnerable to potential environmental hazards such as the destruction of structures. This study hopes to identify the plight of a particular informal settlement, namely, Motsoaledi as an isolated case to reflect dire conditions of most of the informal settlements around the country.

M.Ing. (Electrical And Electronic Engineering) / The metals manufacturing industry is the link between the mining industry, where the ore is mined, and the manufacturing industries, where metals are utilized to create a wide range of products. Melting processes.are used in the manufacturing of metals from the ore, of which induction melting is one of the possible melting processes. In this study the improvement of induction melting of metals is investigated. A new topology for an induction melting furnace is proposed, and this topology is thoroughly investigated. The induction melting process is influenced by certain user-controlled variables as well as the electromagnetic layout of the furnace. In this study, the influence of these variables on the performance of the furnace are investigated through finite element simulations and insight is also gained into the influence of the electromagnetic layout of the system. An experimental induction melting furnace is described in this study, investigated and compared to the predictions from the simulations. Improvements from structural, thermal and measuring viewpoints are also investigated and implemented.

M.Ing. / In power electronics most applications are custom designed. Even though similar topologies are used, each application is designed to fit specific requirements. Presently there is a move towards standard modules that can be connected together to perform the desired power conversion, in much the same way as IC’s have been used for a few decades now. It is important to ensure that the modules can work together without performance degradation. The close proximity of active, passive and logic devices in high power applications dramatically increases the risk of EMI between the various components. This document describes the design and characterisation of a planar conducted RF-EMI power line filter to be used between modules and between modules and power supplies. The filter consists of two sets of transmission lines in parallel. The first is a high permittivity material with nickel conductors, referred to as the attenuator. The second is a set of copper conductors that are placed on the outside of the attenuator. The filter must be able to remove conducted differential mode noise from power line to the module and from the module tot the power line. The filter dimensions are comparable to the wavelengths that have to be attenuated. This allows for a possible combination of RF filtering techniques and power line filtering techniques. The design of the filter and the electromagnetic effects that govern its behaviour is explained. The transfer characteristic of the filter is made possible by the choice of materials and their relation to each other. Each material is discussed and evaluated. Characterisation methods are described and results presented. The influence the materials have on the performance of the filter and the methods and problems of construction of the filter are discussed. The construction of the filter and the commercial processes available are discussed. A prototype filter was built to demonstrate the feasibility of the construction processes. The performance of the prototype filter was then measured. The various test set-ups for different applications are explained. Small signal tests in 50 W systems were used to allow for comparisons with other filter types. The prototype has minimal insertion loss in the pass band and an average slope of attenuation of 40 dB/decade beyond the corner frequency. The corner frequency is at 1 MHz. Transmission line theory is used to develop a distributed element model for the filter using the ABCD-matrix representation of transmission lines. From this matrix an equivalent P-model can be calculated. Applicable assumptions are made to simplify the equations and values for a lumped element P-model is calculated. Both models indicate the importance that the material parameters have on the performance of the simulation results. Good correlation between the measured performance and simulated performance is established. The material parameters are sensitive to temperature. A one-dimensional thermal model is presented to estimate the operating temperature of the filter. It is determined that the attenuation of the filter is primarily determined by the attenuator. Based on the findings, new attenuator designs are made in an attempt to improve the performance of the filter. The small signal measurements of the various designs and the test results under load conditions are compared. A power test is performed. The filter performance varies as the voltage rating is increased. Finally, a temperature characterisation is done. The filter is absorptive and heat is dissipated in both the conductors and the dielectric material. The filter performance is sensitive to the operating temperature. A summary of the technology and the evaluation of this type of filter are presented in the conclusion.

D.Ing. / The necessity of frequency selective interconnects in future integrated power electronic systems are highlighted. A brief historical background illustrates that such interconnects have been utilized successfully in other fields of electrical engineering, although based on different high frequency effects, and configurations. Based on high frequency characteristics of typical interconnects, it is hypothesized that the required frequency selectivity could be obtained through utilization of the skin- and proximity effect, and low conductivity materials, to increase high frequency resistance significantly. Finite element simulation results for a large number of interconnects, and reference structures, are presented in an effort to identify relevant parameters and mechanisms. A hybrid lumped / distributed parameter impedance model is proposed. Parametric analysis is conducted to determine limitations and constraints of the proposed technique. Frequency selective damping of turn-off related power electronic switch and interconnect inductance resonance, is investigated as a possible application of such higher resistance. A simplified analytical model is proposed, and utilized to calculate turn-off waveforms and percentages of damping. An approximation of maximum damping possible is presented. Utilization of enhanced high frequency resistance to realize interconnect based low pass filters for medium power integrated power electronic modules, is investigated as a second application. Based on typical parameter influence, a number of structures are evaluated with finite element simulations. An analytical, lossy transmission line model is developed. Parametric analysis for a chosen structure is conducted, followed by discussion of maximum attenuation, and relative effectiveness. As a third application, the above concepts are applied to 1.5kA nominal current interconnects. A number of structures are evaluated. Application of consecutive impedance mismatches to increase attenuation is investigated. Current and voltage capacity constraints are discussed. Experimental verification of the presented concepts in general, are presented. Technical difficulties and limitations are identified. An objective oriented discussion completes the thesis, with the conclusion that the original hypothesis has been validated.

M.Ing. (Electrical & Electronic Engineering Science) / This study investigates the mitigation of conducted electromagnetic interference (EMI) in a flyback DC-DC converter. Without the use of filters, the maximum mitigation of EMI possible without significantly decreasing converter operating efficiency is investigated. The following parameters are found to influence EMI: · Switching speed: Decreasing switching speed (increasing rise and fall times of the MOSFET) effectively reduces both common mode (CM) and differential mode (DM) EMI above a certain frequency. Series gate resistors up to a certain value were found to not increase power dissipated in the MOSFET. Series gate resistors greater than this value, further reduce CM and DM EMI at the cost of larger amounts of power being dissipated in the MOSFET. · Leakage inductance and inter-winding capacitance: The dominant component of the flyback coupled inductor in terms of EMI generation is the inter-winding capacitance. Increasing inter-winding capacitance increases both CM and DM EMI. Reducing inter-winding capacitance increases leakage inductance. Increasing leakage inductance however, results in reduced converter efficiency. Coupled inductor design is therefore a compromise between leakage inductance and inter-winding capacitance. · Layout inductance: Reducing layout inductance in certain parts of the circuit is an effective method for reducing DM EMI. This is shown to also decrease CM EMI but not to the same extent as DM EMI. · Snubbing: Snubbing is shown to effectively reduce both CM and DM EMI by reducing the magnitude of the voltage overshoot and ringing on the drain of the MOSFET. Snubbing however reduces converter efficiency. This study gives important guidelines to the engineering trade-offs in reducing EMI versus efficiency in a flyback converter.