PhD defence of Yi Qing Xiao – Advanced Macromodeling in the Loewner Framework

Abstract

Circuit modeling has become a key step in the design of circuit systems given the advantage of being able to evaluate a design before dedicating any physical resource towards it. Macromodeling is an important technique in circuit modeling where the frequency domain data such as Y or S-parameters describing the port-to-port relations of the system are used to generate a black-box model for simulating the system. Macromodeling naturally encourages and synergizes with modular system design leading up to a system wide simulation through combination of individual models, giving great importance to the circuit model property of passivity which guarantees stability to combination of sub-models if the sub-models were passive. Passivity enforcement thus became necessary to perturb non-passive models to become passive, and this thesis present new flexible ways of error control for the enforcement process related to gramian based error control. Another direction of interest for macromodels is the parametric extension to the modeling process where variability with respect to one or more physical property of the system can be included in the macromodels, allowing adjustment of the model with respect to key parameters without having to generated fresh port data repeatedly. This thesis presents two novel parametric macromodeling methods in the Loewner Matrix framework.