| dc.description.abstract | Soft errors caused by single energetic particles have an important effect on the reliability of space electronic systems. In combinational logic cells, single event transient (SET) pulse width is a key variable to determine the soft error rate and the pulse width is correlated with PVT (process parameters, voltage and temperature) and design parameters. In order to obtain the statistical distribution of pulse widths in logic cells, the usual methodology requires a large number of SPICE simulations based on sampling the input parameter space. This is time inefficient, especially since the number of parameters is very large for highly scaled technologies.
To reduce the computational effort required to estimate the distribution of single event transient pulse widths, we investigate Design of Experiment (DOE) theory, combined with response surface modeling (RSM), to construct a parameterized mathematical model of single event-induced pulse width in logic gates. A statistical model, which requires a relatively small number of simulations, is built to predict the pulse width distribution. Based on the proposed methodology, we investigate the sensitivity of SET pulse width to multiple parameters. A mathematical model of the pulse width distribution is constructed at only a fraction of the traditional simulation cost. The model accurately describes the probability density function of the pulse width distribution and provides 60X speeds up compared to standard SPICE-based Monte Carlo simulations. | |
