This paper presents a new method to perform efficient first-order symbolic sensitivity analysis of analog circuits by direct differentiation of symbolic expressions stored as elementcoefficient diagrams (ECDs). An ECD is a compact graphical representation of a symbolic transfer function. It is the cancellationfree and per-coeffcient term generation version of determinant decision diagrams (DDDs). The symbolic sensitivity equations obtained from ECDs are stored as a sensitivity-ECDs(SECDs) and can be evaluated extremely fast as it inherits the properties ofECDs. The proposed methodology has been applied to the calculation of sensitivities of four benchmark circuits and it has been demonstrated to be as accurate and more efficient than numerical sensitivity analysis done by SPECTRE.

No abstract

High-performance circuit optimization and synthesis should consider parasitic effects. This paper introduces techniques for parasitic estimation and fast parasitic optimization based on symbolic sensitivity analysis. An effective framework to incorporate parasitic modeling and optimization is presented in order to account for parasitic effects during synthesis. In this paper we primarily focus on using efficient symbolic sensitivity analysis based on element-coefficient diagrams (ECD) to evaluate the dominant parasitic effects so as to eliminate insignificant parasitics. An ECD is the cancellation-free and per-coefficient term generation version of determinant decision diagrams (DDDs). In this paper, parasitic-aware analog circuit synthesis methodology is proposed. The accuracy and efficiency of the parasitic-inclusive optimization have been demonstrated.

We

No abstract