IMPROVED SPECTRAL PERFORMANCE OF RANDOM PWM SCHEMES WITH WEIGHTED SWITCHING DECISION

The authors present a statistical approach to the analysis of random p ulsewidth modulation (RPWM) methods which generate PWM signals by comp aring a reference modulating function with random numbers or signals, Such an approach is used to analyze and optimize a new weighted RPWM ( WRPWM) method and also a standard trapezoidal RPWM method, The WRPWM s cheme combines the advantages of the nondeterministic and deterministi c PWM methods by adding some deterministic nature into the nondetermin istic RPWM method. Both theoretical and experimental results confirm t hat the optimized WRPWM method has better spectral performance over th e standard RPWM schemes and has attractive features such as the potent ial of third-harmonic boosting, reduced acoustic noise, and minimal lo w-order harmonics under an overmodulation condition for electronic dri ve applications.