Lifetime Prediction and Qualification from MTOL, not HTOL! - Joseph Bernstein, Ariel University скачать в хорошем качестве

Lifetime Prediction and Qualification from MTOL, not HTOL! - Joseph Bernstein, Ariel University

Reliability studies of power electronic and advanced deep submicron electronic systems have demonstrated that multiple failure causes occur during normal operation. The most advanced devices exhibit failure after only a few years of operation, exhibiting a random end-of-life behavior that can occur far more frequently than published expectations. Our laboratory is dedicated to understanding reliability physics and performing advanced life testing of multiple, competing mechanisms and their interactions that limit the useful life and results in failures of commercial and military electronic systems. We will show that the current industrial approach of HTOL (High Temperature Operational Life) system is inherently inconsistent with today’s devices and systems. Instead, we can prove that the Multiple Temperature Operational Life (MTOL) approach answers the industries needs for lifetime prediction and reliability qualification. The standard industrial approach to accelerated testing continues to be based on the assumption of a single failure mechanism as canonized in the Mil Handbook 217 approach. In an ideal case, where only one thermal mechanism overwhelms other competing mechanisms and that one mechanism occurs completely randomly, this methodology could be well founded. However today, in our push to faster and more complex system designs, we have found that this methodology falls far short of expectations for reliability qualification. Hence, our assessment must be more sophisticated and consider all the root causes based on physics of failure. Our linear approach to lifetime and reliability calculations is through modeling failure mechanisms as proven by accelerated testing of commercial parts. We analyze failures of the components and compare the mechanisms that we model in the laboratory to be sure that our models accurately reflect the true physics of failure responsible for unreliability in order to better control their effects. We isolate the mechanisms using specific accelerated tests and incorporate these models into a system reliability matrix. We perform a trade-off analysis as part of the design parameters where we can tailor the reliability of a system to meet the performance and reliability specifications before the system is built and in the field. The result is a more accurate and correct reliability assessment compared to the current approach for building in reliability.