An evolutionary method for analog circuits optimization exploiting Mosfet-C filters

The nature of analog circuits design has lent it perfectly to evolutionary optimization techniques. This is due to components functional and performance dependencies on each other and on a suite of other parameters such as frequency, temperature etc. Traditionally, analog circuits are devised using known structures, which can limit the degree of freedom regarding the specifics of the circuit, along with time to create and test the design. This paper presents an approach that bypasses conventional methods of analog circuit design, by approaching the design technique utilizing an evolutionary algorithmic method, in particular Genetic Algorithms (GA). In a two phase process, the techniques first finds MOSFET characteristics to perform the desired integrated resistance functionality, and in phase 2, a full analog circuit design is performed utilizing integrated components chosen in phase 1. The limitless circuit structure not only allows extra degrees of freedom, but consequently creates a vast search space. This paper not only presents a non-traditional approach to circuit design, but also gives an insight to the power of Genetic Algorithms to solve complex problems by utilizing a simple design. In particular this paper focuses on active low-pass filters.