Nonlinear AD converters for integrated silicon optical sensors

State-of-the-art A/D converter research is usually focused on increasing either the resolution or the speed of linear A/D converters. However, in many optical data-acquisition systems, the input transducers used have a dynamic range exceeding five decades and an inaccuracy of around 1%. This favors nonlinear quantizing, which features a constant relative accuracy. Logarithmic readout and A/D conversion can be implemented in an analog signal operation, in a digital signal conversion, or in the A/D converter. The author discusses the advantages of the latter approach, as demonstrated by the performance of two realized types of nonlinear A/D converter. The first converter is based on a floating-point notation and consists of a linear A/D converter preceded by a digitally controlled amplifier or attenuator