Improvements in Broadband Modeling of Microwave Circuits using Rational Models

Improvements on recent algorithms are used to obtain the rational model from sampled data. The methods described here use frequency transformations to move the sampled data to the normalized interval, thus avoiding extreme values of polynomial coefficients and improving the overall accuracy. Interpolating polynomials are obtained and subsequently used as the reference to obtain the rational approximation. Chebyshev series, least-square techniques and SVD methods are also combined to improve robustness. Finally, some examples to prove the good behavior of these techniques are shown.