Paul D. Franzon

Algorithms to Design and Implement Self-Calibrating Circuits


Analog and custom digital circuits fail due to process variations and aging effects (especially NBTI and HCI). In this project our goal was to reduce the rate of failures by determining better design . techniques that allow one to properly design in the correct amount of calibration, and to then calibrate rapidly. The design algorithms were largely based on usinga set of modeling techniques, known as Surrogate Modeling, to efficiently map the design space, so that the design variables could be optimized. The calibration algorithm, based on direct optimization, outperforms traditional calibration algorithms. These algorithms were demonstrated on RF and custom digital test circuits designed in collaboration with Raytheon (and pictured above.)

This research was/is funded by: DARPA.

Selected Publications and PhD Theses arising (with links either to a copy of the submitted document or to the version listed in IEEE Xplore):



Last Modified: May 2016