An Ultra-low Power Voice Interface Design for MEMS Microphones Sensor

This paper develops an ultra-low power single-chip voice interface consisting of a programmable gain amplifier (PGA) and 12-bit asynchronous successive-approximation register analog-to-digital converter for MEMS microphone sensor. The PGA's current can be scaled down from 100uA to 10uA and the measured SNR in 94dBSPL(14mV) is 74dB and 67dB, respectively. It can also set a programmable gain of 40/34/31/28dB for specific demands of microphone. Total Harmonic Distortion (THD) is measured at different output amplitudes, the design exhibits lower than 0.25% THD+N with 94 dBSPL(14mV) at 1kHz. The SAR ADC operates with an 8-kHz sampling rate and consumes only 400nW from a 1.2V VDD. The measured signal-to-noise and distortion ratio (SNDR) is 67.46 dB and spurious-free dynamic range (SFDR) is 87.97 dB. Its sampling rate can be easily scaled from 1M-S/s to 1S/s with a linear power scaling feature. The proposed circuit realized in 180nm CMOS process demonstrates a successful voice data processing (speech/voice recognition, presentation) with the MEMS sensor and an off-chip platform.