Estimation of the In-Band Delta-Sigma Noise Power Based on Windowed Data

The power of the quantization noise in a ΔΣ modulator is usually estimated by means of frequency-domain based algorithms. Although coherency is imposed on the input test tone, the finite number of processed samples induces spectral leaking of the wide-band noise, thus affecting the noise power estimate. In this paper, the error in the noise power estimation Induced by the use of the cosine-class windows, commonly employed for reducing such phenomenon, is investigated, and a criterion for choosing the minimum number of samples which bounds the relative leakage error is given.