A new methodology for constraint-driven layout design of analog circuits

Layout design of analog integrated circuits suffers from a lack of automation due to the multitude of complex design constraints. Most of them are specified and considered manually by expert designers (expert knowledge). We present a new methodology that enables the automatic inclusion of expert knowledge in the form of layout constraints. The resulting comprehensive constraint-driven design approach allows defining and verifying the analog layout constraints by transforming them between the different design domains. In order to verify our methodology, we present a new constraint-driven placement optimization engine. It includes a deterministic decision maker. The decision maker is used to solve effectively the hard instances of the optimization problem that are resulting from complex correlations between the constraints. We have verified our methodology successfully by applying it to the placement of analog circuits in an industrial design environment.

I. Introduction

Contrary to digital layout design, the layout design of analog circuits is often a manual and time-consuming task. This is due to the various stringent and complex design requirements (constraints) that must be considered simultaneously. In analog designs, most of the constraints are currently specified and considered manually by experts (so-called expert knowledge). These constraints are often used implicitly based on the designer's experience. This prevents the effective use of the constraints in design automation.