Variation-Aware Performance Verification of Analog Mixed-Signal Systems

Variations in process, voltage, temperature, and other parameters can have a significant impact on the performances of Analog/Mixed-Signal (AMS) systems. We describe a systematic, automated approach to making these variations visible in system-level simulations using behavioral models. The approach characterizes the dependence of properties on parameters in a compact linear kernel and brings this kernel into behavioral models. It is demonstrated using the example of a charge pump PLL, for which we analyze the impact of process and temperature variations characterized by a commercial circuit simulator with a SystemC AMS simulation.