LARGE SIGNAL MODELING OF GAAS FETS FOR THE SIMULATION OF NONLINEAR MICROWAVE CIRCUITS

Abstract

The use of gallium arsenide(GaAs) FETs in microwave monolithic integrated circuits (MMICs) has increased steadily in recent years due to the demand for the mobile and personal communication systems. However, the development of MMICs is heavily dependent on the availability of efficient and reliable CAD tools. Large signal modeling of GaAs FETs is a key step in the circuit design. Experience indicates that the current models for GaAs FETs still could not satisfy the stringent demands for the nonlinear circuit design of MMICs. In this dissertation, a comprehensive study of a large signal model of GaAs FETs for the simulation of nonlinear microwave circuits is carried out. The study involves an experimental data-based approach, a bias-dependent behavior investigation of the source and gate resistances of GaAs MESFETs, and a new empirical GaAs FET ' model. The experimental data-based approach has yielded a charge and current conservative model for GaAs FETs. The bias-dependent behavior investigation has revealed for the first time the bias-dependent properties of the source and gate resistances of the GaAs FET at all bias conditions. The results have provided a deeper insight into the operation of the device which in turn will help to model the device more accurately. An empirical model that has the benefits of simplicity and easy extraction, and at the same time has a complete and accurate performance description of GaAs MESFETs has been presented in the dissertation. Experimental techniques for parameter extraction are discussed. A detailed model performance evaluation which includes the DC characteristics, S-parameters and harmonics is carried out. The performance evaluation shows that the predicted values of our model have very good agreement with the measured data. An MMIC amplifier has also been implemented in the dissertation and a comparison is made between the simulated results using the model and the measured performance to verify the accuracy of the model.