Optimization and comparision of fixed-point implementations for first and second-order digital blocks

The problem of optimizing the organization of fixed-point implementations for first- and second-order digital filters is discussed. The direct and canonical forms are analyzed in detail and compared with each other with regard to the input quantization and roundoff noise effects. A summary of the computation results is plotted in some diagrams depicting on the z plane the zones of greater convenience of either solution.