Skip to Main Content
Vector fitting is a rational approximation technique, which is frequently used to calculate accurate macromodels of electrical and electronical structures. The robustness of the technique is obtained by combining the use of a weighted iterative least squares scheme and a well-chosen partial fraction basis. It was discussed in that numerical problems may occur if poles of higher-order multiplicities are required to approximate a frequency response. This paper shows that the orthonormal vector fitting technique  solves this problem in a fundamental way.