Randomized modulation of power converters via Markov chains

Randomized modulation of switching in power converters holds promise for reducing filtering requirements and reducing acoustic noise in motor drive applications. This paper is devoted to issues in analysis and synthesis of randomized modulation schemes based on finite Markov chains. The main advantage of this novel type of randomized modulation is the availability of an explicit control of time-domain performance, in addition to the possibility of shaping the power spectra of signals of interest. We focus on the power spectra of the switching functions that govern converter operation, and on the power spectra of certain associated waveforms, numerical (Monte Carlo) and experimental verifications for our power spectral formulas are presented. We also formulate representative narrow- and wide-band synthesis problems in randomized modulation, and solve them numerically. Our results suggest that randomized modulation is very effective in satisfying narrow-band constraints, but has limited effectiveness in meeting wide-band signal power constraints.