Memory less Non Linear System Model

This Paper presents a methodology for the design of analog circuits. Under this methodology, the ports of all circuit blocks are bidirectional. The stability of circuits designed following this methodology can be determined by local, rather than global,analysis. If the circuit blocks follow a certain design discipline, they can be connected in nearly arbitrary configurations while maintaining stability. This methodology may be applied to designing circuits that solve several classes of problems. In particular, it is applied to designing a small circuit that can build a model of a memory less nonlinear system, and to predistort to linearize this system. First, this document shows how this methodology can be used to solve basic con-strained optimization problems in analog circuits. Given a set of constraint equations, it shows how to build a circuit that will solve that set of equations. This circuit is sufficiently small that it can be used for simple analog processing in basic analog circuits; for instance, it may be used to set optimal bias currents for another circuit. Next, this Paper shows how to use this methodology to build a circuit that can be used to determine the parameters of a model of a system through observation of the system’s behavior. This circuit can, for instance, be used to monitor and model the nonlinearity of the power output stage of an amplifier. Given this model, a symmetric, matched circuit can predistort to compensate for the nonlinearity. These circuits are small enough that they can be used as blocks for other analog circuits, particularly in places where digital processing is impossible due to size and cost constraints.