Reliability-Oriented Strategies for Multichip Module Based Mission Critical Industry Applications
The availability is defined as the portion of time the system remains operational to serve its purpose. In mission critical applications (MCA), the availability of power converters are determinant to ensure continue productivity and avoid financial losses. Multichip Modules (MCM) are widely adopted in such applications due to the high power density and reduced price; however, the high number of dies inside a compact package results in critical thermal deviations among them. Moreover, uneven power flow, inhomogeneous cooling and accumulated degradation, potentially result in thermal deviation among modules, thereby increasing the temperature differences and resulting in extra temperature in specific subset of devices. High temperatures influences multiple failure mechanisms in power modules, especially in highly dynamic load profiles. Therefore, the higher failure probability of the hottest dies drastically reduces the reliability of mission critical power converters. Therefore, this work investigate reliability-oriented solutions for the design and thermal management of MCM-based power converters applied in mission critical applications. The first contribution, is the integration of a die-level thermal and probabilistic analysis on the design for reliability (DFR) procedure, whereby the temperature and failure probability of each die are taken into account during the reliability modeling. It is demonstrated that the dielevel analysis can obtain more realistic system-level reliability of MCM-based power converters. Thereafter, three novel die-level thermal balancing strategies, based on a modified MCM - with more gate-emitter connections - are proposed and investigated. It is proven that the temperatures inside the MCM can be overcame, and the maximum temperate reduced in up to 8 %.