MODEL ORDER REDUCTION FOR PEEC MODELING BASED ON MOMENT MATCHING

Abstract

Accurate and efiective system-level modeling has become necessary to address electromagnetic compatibility (EMC) issues in modern circuit and system design. Model order reduction (MOR) techniques provide a feasibility to approximate complex circuit models with compact reduced-order models. In this paper, an efiective MOR technique entitled multi-point moment matching (MMM) is implemented for the partial element equivalent circuit (PEEC) modeling. Moment information at multiple frequency points is used in this method in order to accurately estimate a given system over an entire frequency range of interest, and for each frequency an enhanced asymptotic waveform evaluation (AWE) is applied to obtain a reduced-order model by constructing a pole-residue representation of the original transfer function. The improvements of conventional AWE in aspects of both moment computation and moment matching can avoid ill-conditioned moment matrices and unstable dominant poles. The complex frequency hopping (CFH) technique is employed to select the multiple expansion points by using a newly developed upward- search algorithm. Numerical simulations of coupled microstrip lines in both frequency and time domain indicate the efiectiveness of the proposed method.