Tolerance design of closed-loop controllers for DC-DC voltage regulators: genetic algorithms and vertex analysis based optimization

In this paper nominal and tolerance design of feedback compensation networks for DC-DC switching converters is presented. Tolerance design techniques, previously presented by the authors and briefly summarized in this paper, are adopted to achieve an optimal design of the compensation network. A genetic algorithm seeks for the set of commercial parameters and tolerance values or the RC components such that the compensation network fulfils design constraints on the acceptability ranges of loop gain cross-over frequency and phase margin. Phase margin and pole-zero matching methods are adopted to generate nominal design solutions using two types of compensation networks. Such solutions are considered as guesses for the tolerance design performed. The population of design solutions evolves towards different optimal final projects depending on the required design preferences. Vertex analysis has been used to check the feasibility of each candidate tolerance region. Moreover, at the end of the design process, the yield value of the optimal solution selected by the genetic algorithm is calculated by means of an interval arithmetic-based procedure. The results presented in this paper, even though concerning the optimal design of the compensation network of buck DC-DC converter, are quite general and show how the techniques used can greatly support designer decisions.