Zero Power, Tunable Resonant Microphone with Nanowatt Classifier for Wake-Up Sensing

This paper shows a high sensitivity (12.6 V/Pa), tunable, and zero-power microphone for use in a near zero power detection system. The microphone uses energy from incoming sound signal to generate enough voltage for detection without any amplifying electronics. It takes advantage of mechanical resonance to achieve high sensitivity and pre-filter a desired frequency band. A spiral structure results in a low resonance frequency (25.2 Hz) for detecting most man-made machines such as generators, cars, and trucks. The microphone output is processed by a near-zero classifier subthreshold CMOS chip. The over-all system power consumption is under 6 nW. We show examples of electric generator detection at 10-meters, and truck detection at 0.5-meter in a rural environment, illustrative of zero-power wakeup sensors for long-lifetime IoT devices. The sensing also demonstrates the ability to distinguish different targets owing to different signal spectrums.