In this paper, a signal-specific analog-to-digital converter (ADC) with a new structure is proposed, in which the resolution of the ADC is adaptively adjusted by the activity of the input neural signal. The main advantages of the proposed technique for converting sparse and burst-like signals include (1) output data-rate reduction, and (2) power savings in ADC and its succeeding blocks. These benefits are obtained owing to the truncation of bits along in-active part of the signal. The extra blocks for realizing the proposed adaptive-variable resolution technique are fully-digital, which add minimum complexity and design overhead to the ADC. The proposed ADC has a suitable data compression capability at the expense of a tolerable degradation in quality of the reconstructed signal. The simulation results in a 180 nm CMOS technology show power savings of up to 39.5% and a compression ratio of 3.99, as compared to the conventional structure.

Keywords