LOW-NOISE ENERGY-EFFICIENT SENSOR INTERFACE CIRCUITS

Today, the Internet of Things (IoT) refers to a concept of connecting any devices on network where environmental data around us is collected by sensors and shared across platforms. The IoT devices often have small form factors and limited battery capacity; they call for low-power, low-noise sensor interface circuits to achieve high resolution and long battery life. This dissertation focuses on CMOS sensor interface circuit techniques for a MEMS capacitive pressure sensor, thermopile array, and capacitive microphone.

Ambient pressure is measured in the form of capacitance. This work propose two capacitance-to-digital converters (CDC): a dual-slope CDC employs an energy efficient charge subtraction and dual comparator scheme; an incremental zoom-in CDC largely reduces oversampling ratio by using 9b zoom-in SAR, significantly improving conversion energy.

An infrared gesture recognition system-on-chip is then proposed. A hand emits infrared radiation, and it forms an image on a thermopile array. The signal is amplified by a low-noise instrumentation chopper amplifier, filtered by a low-power 30Hz LPF to remove out-band noise including the chopper frequency and its harmonics, and digitized by an ADC. Finally, a motion history image based DSP analyzes the waveform to detect specific hand gestures.

Lastly, a microphone preamplifier represents one key challenge in enabling voice interfaces, which are expected to play a dominant role in future IoT devices. A newly proposed switched-bias preamplifier uses switched-MOSFET to reduce 1/f noise inherently.