Fast algorithms for finding steady-state solutions of switching power converters

Abstract

Analysis of the steady-state waveforms of power converter circuits is important for practical design and implementation as it provides crucial information about the exact circuit operation and the extents of device stresses. The information derived from steady-state analysis is useful for power supply designers in choosing suitable components and enhancing reliability. In this thesis, a detailed study of the computational algorithms for finding the steady-state waveforms of power converter circuits is presented. Time-domain methods and frequency-domain methods are studied. In studying frequency-domain methods, emphasis has been put on a special kind of frequency-domain method which involves representation of waveforms in terms of wavelet functions and polynomial functions. New algorithms have been developed and are presented in this thesis. The benefits and drawbacks of time-domain methods and the specific frequency-domain methods are compared and studied. A hybrid approach, which takes advantage of both time-domain and frequency-domain methods, has been proposed. This new hybrid approach is found to be a very efficient computational approach for solving steady-state solutions of piecewise switched circuits such as power converter circuits.