Oversampling and Coarse Quantization for Signals

CSIC Building (#406), Seminar Room 4122.
Directions: home.cscamm.umd.edu/directions

Latest Trends in Noise-Shaping
from the Silicon Side

Dr. Bob Adams

Analog Devices

Abstract:   Five years ago the world of noise-shaping converters was quite simple and consisted mainly of single-bit quantizers embedded in discrete-time analog loops with a filter order ranging from 2 to 8. Since then there have been significant new developments such as multi-bit quantization using so-called "Error Mismatch Shaping" techniques as well as a move towards hybrid continuous-time/discrete-time loop filters. Another trend is driven by the need to produce high-efficiency audio amplifiers using a variety of switching schemes, some of which are closely related to 1-bit delta-sigma modulators that that have been modified to minimize the number of switching transitions per unit time. Finally, I will discuss a novel approach to noise-shaping that views the problem as a direct attempt to minimize the error between a filtered version of the input signal and an identically-filtered version of a coarsely-quantized signal.

[LECTURE SLIDES]

Other presentations:
SLIDES ON: A Signal-processing Interpretation of the Riemann Zeta Function