Estimation of single-event transient pulse characteristics for predictive analysis

Abstract

In this paper a methodology to predict single-event transient (SET) pulse characteristics is proposed. Analytical models and technology pre-characterization are used to estimate SET pulse-widths for different standard cells. The model uses graph analysis of the cell netlist to identify similar circuit structures for reduced computational complexity for the characterization of standard cells. The error between the proposed model and simulations is between 3% and 9.3%. Model predictions are also compared with results from heavy-ion experiments for a test chip fabricated at the 65-nm technology node showing excellent agreement. The proposed model will allow designers to model effects of soft errors at the circuit-level during the design phase.