Accurate Multl-polnt Padce Approxlmatlon of Mlicrowave Structures using Orthonormal Polynomlal Basls Functions

Rational least-squares techniques are commonly used to build compact macromodels of passive microwave components. This paper describes a technique which calculates rational least-squares fitting models by matching S-parameter frequency data-samples and higher-order frequency derivatives (or moments) using orthonormal polynomial basis functions to improve the numerical accuracy. Some considerations are given about the optimal choice ofpolynomial basisfunctions.