Analog-to-digital converters (ADCs) are indispensable building blocks of wearable and implantable biomedical data acquisition systems. Ultra-low-power ADCs for biomedical signal sensing have witnessed a dramatically reduced power consumption in recent years, but we have to admit that our biomedical systems need more breakthroughs than just squeezing harder in conventional ways.
As is known to all, many biomedical signals are born with a sparse nature. A large amount of redundant digital samples will be thus generated if we use Nyquist-rate ADCs to convert such signals. Most likely, ADC power savings are not a major concern in a system in which transmission power dominates the overall power consumption. However, if this is not the case, from a signal point of view, new ways of sampling or sensing are necessary to further improve the performance of the whole system.
A new and promising ADC approach for biomedical data acquisition is based on so-called level-crossing (LC) sampling, in which samples are generated only when the input signal crosses the threshold levels, so there is no redundant sample in this case. However, the conventional LC-ADC utilizes power hungry comparators and DACs, which causes the LC-ADC to consume much more power than ultra-low-power Nyquist ADCs (e.g., SAR ADCs). In our new approach (mentioned by Wouter earlier in the weblog), innovations at both system level and circuit level enble us to design a more power-efficient LC-ADC. Power consumption is now in the range of hundreds of nanowatts. We are currently investigating the possiblity to further improve its performance and reliability.