Design of a Novel Current Starved VCO via Constrained Geometric Programming

Presently the design of the optimal analog and mixed signal (AMS) circuits with lesser design cycle time is a great challenge for the designers. This paper describes the optimization of the current starved voltage controlled oscillator (CSVCO) circuit. The objective functions and constraints of the CSVCO circuit are in the form of posynomial functions of the design variables. The convex optimization and geometric programming method can well express the posynomial functions. In this work a novel current starved voltage controlled oscillator is designed by using geometric programming. Geometric programming is an efficient optimization technique in which the problem can be formulated as a convex optimization problem to obtain the global optimal solution for the given constraints with lesser design cycle time. The centre frequency of the VCO before and after optimization is 1.012GHz and 1.0000457GHz respectively. After optimization the frequency deviation of the circuit is reduced to .00457% from 1.2%. With this frequency precision the area is also minimized by the geometric programming method.