Minimize system failure rate considering variations of electronic components lifetime data

Temperature and electrical stresses are important factors in affecting the lifetime of microelectronic devices. For a printed circuit board consisting of thousands of microelectronic devices, device temperature and electrical stresses are usually neither constant nor evenly distributed. Temperature and electrical stresses often vary in different subcircuities due to the difference of power consumption and electrical derating. The variations of the thermal and electrical stresses create uncertainties in estimating reliability metrics such as the system failure rate and mean time between failures. This paper proposes an optimal design procedure to minimize the system failure rate by reducing the variations of device temperature and electrical stresses. The system failure rate is treated as a stochastic number. The distribution of the failure rate is approximated by the normal distribution based on the central limit theorem. The objective is to select the best devices from multiple component choices such that the system failure rate is minimized while the cost budget and the six-sigma criteria are still satisfied. This is a non-linear integer-programming problem and the genetic algorithm is used to search for the optimal solution. Finally, an analog printed circuit board is used to illustrate the optimization procedure.