Frequency-domain adaptive-resolution level-crossing-sampling ADC

In the framework of the large-scale wireless sensor networks involved in the Internet-of-Things (IoT), analog-to-digital converters (ADCs) must target ever increasing levels of power efficiency and amenability to ultra-scaled CMOS technologies. Digitally intensive architectures and smart conversion algorithms are therefore the fuel of future ultra-low power (ULP) designs. The minimization of the output average bitrate is an effective way to maximize the system energy efficiency. Level-crossing-sampling (LC) ADCs are a class of converters that addresses such problem. In their conventional implementation, however, they are mainly impaired by analog blocks (i.e. the high-performance comparators), difficult to be designed in deep nanoscale CMOS. This paper describes a highly-digital frequency-domain implementation of a LC ADC, which replaces the analog comparators with an oscillator-based quantizer and simple digital logic. LC is performed in the digital frequency-domain, where the application of adaptive-resolution algorithms to further enhance power efficiency becomes straightforward. Behavioral modeling simulations demonstrate the appropriateness of the proposed topology by comparing it with the conventional designs and by evaluating the impact of the oscillator-based-quantizer nonidealities on the ADC performance.