The reliability of the power converter system, an essential energy adapter connecting solar panels and batteries in the satellite, is crucial to an entire satellite. This paper adopts a new fusing failure rate to build a more accurate model of reliability considering PPMS. In particular, the applicability of the new model is demonstrated to not only components following exponential distribution, but also to others following Weibull distribution. Furthermore, for the converter level, the Dynamic Fault Tree and Markov Process (MP) are utilized to model converter's reliability with the help of the state lumping method. In the case study, the reliability modeling of a dual Buck-Boost converter in satellite is conducted, as well as the optimization for redundancy design. The result indicates that the reliability of the converter in the satellite is more accurate and reasonable than that of using traditional methods.