A New Analytical Model of SET Latching Probability for Circuits Experiencing Single- or Multiple-Cycle Single-Event Transients

Abstract

As technology scales down, more single-event transients (SETs) are expected to occur in combinational circuits and thus contribute to the increase of soft error rate (SER). We propose a systematic analysis method to precisely model the SET latching probability. Due to the decreased critical charge and shortened pipeline stage, the SET duration time is likely to exceed one clock cycle. In previous work, the SET latching probability is modeled as a function of SET pulse width, setup and hold times, and clock period for single-cycle SETs. Our analytical model does not only include new dependent parameters such as SET injection location and starting time, but also precisely categorizes the SET latching probabilities for different parameter ranges. The probability of latching multiple-cycle SETs is specifically analyzed in this work to address the increasing ratio of SET pulse width over clock period. We further propose a method that exploits the boundaries of those dependent parameters to accelerate the SER estimation. Simulation results show that the proposed analysis method achieves up to 97% average accuracy, which is applicable for both single- and multiple-cycle SETs. Our case studies on ISCAS'85 benchmark circuits confirm our analysis on the impact of SET injection location and starting time on the SET latching probability. By exploiting our analytical model, we achieve up to 78% simulation time reduction on the process of SET latching probability and SER estimation, compared with Monte-Carlo simulation.