Two methods for the reduction of quantization effects in recursive digital filters

Two distinct methods for the reduction of coefficient and product quantization effects in recursive digital filters are described. A degree of freedom is introduced in the design by increasing the approximation order above the minimum. This is then used to increase the allowable margin for coefficient quantization error or to reduce the sensitivity to coefficient quantization. The two methods are used to design several sixth-order low-pass filters. The designs obtained are then compared with the corresponding minimum-order elliptic designs with respect to the effects of coefficient and product quantization. The results show that both methods lead to reduced wordlength for the implementation and also to reduced output quantization noise.