Non-Monte Carlo Simulation and Sensitivity of Linear(ized) Analog Circuits under Parameter Variations*

In this paper, a non-Monte Carlo approach is presented for the simulation and sensitivity analysis of linear(ized) analog circuits under parameter variations. With uncertain parameters represented as intervals, we present a formulation of circuit equations as a set of interval linear equations. An elegant and efficient algorithm is developed based on some interval-mathematic results derived in this paper. In contrast to Monte Carlo simulation, the proposed approach is rapid—with the computational cost comparable to one nominal circuit simulation—and correct—the computed bounds always contain, and are usually close to, the actual bounds. Further, sensitivity under parameter variations can be computed from the response bounds with minor computation cost. The algorithms have been implemented into SPICE3F5. Experimental results are promising.