In salety-critical applications, detecting fabrication delects is of utmost importance, even if they do not impact significantly the performance. Defect-oriented test approaches are thus necessary, but their validation is cumbersome. Indeed, defect simulation is unavoidable but computationally demanding. For complex Analog and Mixed-Signal (AMS) circuits and systems, the number of defect candidates may be very large. If the evaluation of each defect candidate requires a complex transient simulation, exhaustive simulation is simply intractable. Sound statistical approaches to estimate defect coverage have been proposed, but one of the main shortcomings of these approaches is that of experimental validation. On one hand, it is almost impossible to get access to delect statistics of commercíal parts since this data is a very sensitive in terms of company image. On the other hand, it is al so impossible to manulacture (and test) a sulficient amount of circuits to get reliable statístics in an academic environmenl. Europractice integration services usually give access to around 50 parts, very far of the production level necessary to estimate a defectivity rate in the arder of tens 01 ppm.
In order to tackle this validation issue, this project propases to adapt the framework of statistical assessment of defect coverage to the study of radiation-induced Single-Event Transient (SET) sensitivity in complex Analog and Mixed-Signal circuits.