Design of Nonlinear Circuits, the Linear Time-Varying Approach

Over the last years the ever-growing demand for higher performance has led to much interest in using nonlinear circuit concepts for electronic circuit design. For this we have to deal with analysis and synthesis of dynamic nonlinear circuits. This thesis proposes to handle the nonlinear design complexity by dividing the design process in two steps. In the first step, a high-level synthesis/analysis step, a circuit topology implementing the wanted (nonlinear) function is found. We conclude that an expansion in basic functions, chosen to fit the nonlinear building blocks used, appears to be the best option for implementing this step. The second step consists of a low-level analysis/synthesis step, in which the quality of the topology is determined. The thesis research has focused on using the linear time-varying (LTV) small-signal model for describing the dynamic behaviour of nonlinear circuits in the context of low-level analysis/synthesis. This model is a generalization of the conventional linear time-invariant small-signal model and allows the use of equivalent stability criterions. Linear eigenvalues and poles are generalized to dynamic eigenvalues, Floquet exponents and Lyapunov exponents. The LTV small-signal model decreases low-level modeling complexity by using knowledge from the high-level step. The linear time-varying approach was applied to various specific nonlinear circuits: a negative feedback amplifier with class-B output stage, a dynamic translinear filter and oscillator and a differential pair used as a limiter. These design examples show that the linear time-varying approach is a good modeling candidate for low-level synthesis/analysis.