Gaining insights in analog design via geometric programming

Abstract

This paper presents a systematic approach for gaining design insights in a variety of analog design problems. Unlike designers* intuition, this method provides both quantitative and qualitative insights. The method relies on formulating a design problem as a geometric program (GP). These nonlinear, constrained optimization problem can be transformed to convex optimization problems, and then solved (globally) very efficiently. Designers can analyze quantitatively trade-offs among different specifications as well as construct trade-off curves very quickly owing to the efficiency of the method.