Least Squares Multiharmonic Fitting: Convergence Improvements

The robustness of least squares (LS) algorithms combined with the increased processing capabilities of DSPs will enable the development of a new brand of low-cost flexible instrumentation based on LS methods. Sine-fitting algorithms estimate the sine wave parameters of a digitized signal by minimizing the LS error. They can be used in a large number of important applications such as characterization of analog-to-digital converters, measurement of impedances, and test of the frequency response of linear systems. For signals with harmonics, multiharmonic fitting algorithms estimate the harmonic amplitudes and phases. In this paper, the convergence of the LS multiharmonic fitting algorithm is analyzed. Improvements are presented to increase convergence and minimize the number of iterations.