No description available.

No description available.

The well-established correlation between the high switching di/dt and dv/dt of power semiconductor devices and electromagnetic interference (EMI) generation leads to a trade-off between power converter losses and emission levels; to reduce emissions, the devices are slowed down, incurring greater losses. This thesis addresses in particular the issue of electromagnetic interference (EMI) generation in IGBT-based variable-speed motor drives. The work presented here begins with the hypothesis that it is the higher-order derivatives of IGBT switching waveforms which are most important in determining EMI generation at the frequencies where direct radiation is the dominant mode of emission. Analysis of idealised waveforms shows that by 'rounding the corners' of switching waveforms, it is possible to reduce their high-frequency spectral content with no increase in transition time. High-bandwidth measurements of IGBT switching behaviour are analysed in detail, enabling the trade-off between EMI generation and losses to be quantified, and the critical aspects of the device behaviour for EMI generation to be identified. As with idealised waveforms, itis found that 'corners' of the measured waveforms are responsible for their high-frequency spectral content. The effect of operating conditions on the trade-off between losses and EMI generation is investigated, showing that both temperature and current have significant influence. Informed by the measurements and analysis presented here, an open-loop gate driving technique is developed, allowing programmed voltage profiles to be applied to the gate of an IGBT. By shaping the device switching waveforms to be more 'rounded', it is shown that their high-frequency spectral content may be suppressed by up to 20 dB in the frequency range of interest, with no increase in switching losses.

Current research and the price of semiconductors are making Matrix Converter topologies more attractive and practically viable for many applications. One of the key benefits claimed for the matrix approach is the possibility of greater power density due to the absence of a DC link. To capitalise on this it is necessary to make the input filter small by having a sufficiently high switching frequency. In order to arrive at optimised solutions in terms of filtering, heatsinking and packaging, it is important to have accurate models to predict power circuit losses as a function of operating point and switching frequency. This aim of this thesis is to further the understanding of Matrix Converter current commutation with a focus on switching losses. The commutation process is analysed in detail and a complete loss model for Matrix Converter circuits is developed. The use of circuit simulation software in the prediction of switching losses is investigated and comparisons are made with experimentally measured results. The loss model is used to compare Matrix Converter losses with a functionally equivalent sinusoidal front end inverter. A review of soft switching techniques for Matrix Converter circuits is given. A new auxiliary resonant soft switching Matrix Converter is presented which overcomes many of the problems of the previously proposed circuits. The design, construction and testing of a 6kW prototype soft switching Matrix Converter is presented to evaluate the feasibility of the topology.