PyCO: A parallel genetic algorithm optimization tool for analog circuits

Analog designers are challenged by increasingly complex device models and lowered signal swing as the CMOS processes scale. At the same time, new trends and emerging technologies pose tighter design constraints. However, the cheap computational resources nowadays enable the use of the mature electrical simulators and device models in simulation-in-a-loop optimization techniques. In this work, we present a flexible circuit optimization tool for analog designs which relies on evolutionary algorithms. Moreover, we employ logistic functions to determine the fittest individuals. The benchmarking test shows that the program is able to dimension an operational transconductance amplifier (OTA) based on the topology netlist in less than 15 minutes, without resorting to any knowledge base, initial guess or simplified models. The achieved performance suggests that the tool may be integrated in an existing design flow with huge benefits.