BROAD BAND BEHAVIORAL MODELING OF ON-CHIP RF INDUCTORS AND TRANSFORMERS

摘要

The broad band behavioral models of on-chip inductors and transformers are extracted from sampled frequency response data. A linearized least square based model parameter extraction algorithm is used to obtain 's'-domain rational function model. Model order is optimized to strike a balance between modeling error and time required for transient simulation of circuits consisting of on-chip inductors and/or transformers. A new cost function is proposed to improve optimal model order selection. Particle swarm optimization (PSO) is applied to maximize the nonconvex fitness function, associated with model order optimization. A modification of PSO algorithm is proposed to speed up optimal model order selection. While modeling on-chip inductors and transformers, the proposed modeling technique extracts highly accurate, yet reduced order models, which are valid over a wide band width and are compatible with SPICE-like simulators.