Efficient System-Level Statistical Design of Analog Systems via Variance Minimization

H. Chang and H. Tang (USA)

Keywords

Systemlevel design, Statistical design, Variance minimiza tion, Worstcase design, Analog systems, Symbolic analy sis

Abstract

Due to increasing system complexity, statistical design has to be performed at both system-level and circuit-level. This paper presents an efficient method for system-level statis tical design of analog systems. In particular, the method considers system performances that can be obtained from symbolic analysis, which are then linearized at the nominal design point with first-order Taylor series expansion. As suming that all system-level design parameters are Gaus sian random variables, the system performances are also Gaussian variables. To maximize yield, it is equivalent to minimizing performance variances. The overall statistical design problem is formulated as a nonlinear programming problem and is solved efficiently.

Important Links:



Go Back