A nonparametric Bayes approach to decide system burn-in time

Burn-in is the preconditioning of assemblies and the accelerated power-on tests performed on equipment subject to temperature, vibration, voltage, radiation, load, corrosion, and humidity. Burn-in techniques are widely applied to integrated circuits (IC) to enhance the component and system reliability. However, reliability prediction by burn-in at the component level, such as the one using the military (e.g. MIL-STD-280A, 756B, 217E) and the industrial standards (e.g., the JEDEC standards), is usually not consistent with the field observations. Here, we propose system burn-in, which can remove many of the residual defects left from component and subsystem burn-in. A nonparametric model is considered because (1) the system configuration is usually very complicated, (2) the components in the system have different failure mechanisms, and (3) there is no good model for modeling incompatibility among components and subsystems. Since the cost of testing a system is high and, thus, only small samples are available, a Bayesian nonparametric approach is proposed to determine the system burn-in time. A case study using the proposed approach on MCM ASICs shows that our model can be applied in the cases where (1) the tests and the samples are expensive, and (2) the records of previous generation of the products can provide information on the failure rate of the system under investigation.