Dynamic range and error dissipation of adaptive quantizers

An analysis of backward estimating adaptive quantizers is presented. The algorithms studied are linear in logarithmic step size, which is the case for many quantizers presented in the literature. The dynamic range in terms of signal-to-distortion ratio and the effect of digital errors on the decoding process are investigated. It is shown that an unlimited dynamic range is possible only at the expense of a nonzero digital error residual offset between the analog-to-digital and the digital-to-analog converters.