Further improvements in digitizing continuous-time filters

A new method for digitizing a linear continuous-time filter has improved stability characteristics and reduced implementation requirements while retaining the accuracy of the higher order rules recently appearing in the literature. This method approximates the sampled input by a polynomial of degree q fitted to the latest (q+1) samples. The value of q is selected by the designer; values of q up to 14 have been tested successfully. The exact response of the original continuous-time filter to the polynomial input is computed analytically. The method has been formulated using matrices, which lends itself well to machine computation and greatly simplifies the effort needed to obtain numerical results.