A novel method based upon nonlinear optimization for analog filter design with mask constraints

This paper considers the optimization of analog filters which are not covered by classical approaches like Butterworth, Chebyshev, or Cauer approximations. Therefore, a novel and highly efficient method for analog filter design using nonlinear optimization is presented. This approach enables high flexibility and easy handling as compared to existing approximation techniques. By describing particular design specifications as constraints, a solution can be obtained by using suitable optimization software. Apart from the straightforward approximation of filter transfer functions, this approach is extensible to consider coefficient uncertainties of component variations. Furthermore, it can be used to synthesize filters comprised of only real poles. Examples are also provided to validate the benefits of the resulting design method.