I. Introduction
Spice is not initially designed [1] to handle step-function excitation directly. To avoid discontinuity, SPICE-derived transient simulators substitute inefficient steep ramps for step-function excitations. As the rise time of a steep ramp is reduced to better approximate a step-function, more iterations are required to maintain accuracy and algorithmic stability, thereby consuming substantial computation time. To account for such inefficiency, especially in digitally dominated circuits with rapid turn on/off, either pole fitting methods, such as macromodeling [2] or the abstraction of the simulation environment to switch level [3] can be employed. However, such abstractions impinge on accuracy.