An Iterative Approximation Procedure for Automatic Filter Synthesis

The design of
m
-derived image parameter filters, as well as the synthesis of Butterworth or Chebyshev pass band insertion-loss parameter filters with prescribed frequency-dependent minimum stop band attenuation, normally involves the use of graphical methods to find the necessary number and location of the attenuation poles. These methods are tedious and inaccurate. This paper describes an iterative approximation process, somewhat similar to Remez's second method [3], for the synthesis of low-pass and band-pass filters. The process is very fast and its programmed version (containing numerous options) fits into a 40,000-position computer memory. A brief analysis of the optimality and convergence of the procedure is included. Finally, the actual computer programs are described and some examples given.