Model-based retrieval of phase-related quantities for two-port microwave devices

Abstract

In this thesis a new method is proposed for the retrieval of phase-related quantities from magnitude-only measurements on two-port microwave systems. The method is applicable when a microwave system is manufactured in relatively large quantities. A reduced-order model is developed using the prototype of such a system. Such a reduced-order model will include a number of quantities or "coefficients" (other than frequency) whose value is determined using a full vector data set on the prototype. Establishing the form required for the reduce-order model acts as the additional constraints that are needed to provide a unique relationship between the transfer function magnitude and its phase. When later production models are tested using magnitude-only measurements, optimization is used (with the above-mentioned "coefficients" as the optimization variables, for which only small adjustments are needed from one device to another) to match the measured and modeled magnitude responses. The retrieved "coefficients" are then used to predict the phase response using the model. The use of the model-based retrieval technique is demonstrated through its application to three different types of device.