A 1.8 V 828 μW 80 dB digital MEMS microphone

Abstract

A single-package digital MEMS Capacitive Microphone (MCM) system is presented. The system consists of a MCM, which is wire-bonded with its readout interface (RI). The MCM sensor is fabricated using a combination of surface and bulk micromachining, employing diaphragm-stiffening to achieve piston-like diaphragm-movement and attaining required sensitivity with a smaller diaphragm-area. The RI is designed in 0.35 lm CMOS and it consists of a preamplifier (PAMP), a sigma-delta modulator (SDM), integrated biasing and digital control, converting the MCM capacitive variations into a single-bit over-sampled digital bitstream. The PAMP employs a two-terminal bootstrapped source-follower buffer to make the readout insensitive to the MCM parasitics, subsequently feeding a third-order single-loop singlebit modulator running at 2.5 MHz. The electrical measurements of the standalone RI demonstrate 55 dB A-weighted @ 1 Pa SNDR at the analog PAMP output and 80 dB A-weighted dynamic-range at the digital output, which corresponds to a conversion range from 40 to 120 dB SPL. The SNDR for acoustic measurements is 33 dB A-weighted @ 1 Pa, limited by the higher MCM thermal noise floor and reduced sensitivity (-53 dB V @ 1 Pa). The frequency characterization of the system for the complete audio-band demonstrates the effect of the system package towards higher frequencies ([9 kHz), giving rise to Helmholtz resonance, and reduction in sensitivity for low-frequencies (\400 Hz) because of acoustic short-circuiting inside the MCM due to flow-by slots. The complete system consumes 460 lA of total current for a 1.8 V single-supply. The total system dimensions are 4.5 9 2 mm 2 (excluding the package), demonstrating the viability of a low-area, low-power and high dynamic-range implementation of digital MCM.

A microphone readout interface with 74-dB SNDR

“…A common way to implement such a readout interface is using a front-end preamplifier followed by a discrete-time (DT) sigma-delta modulator (SDM) [6,7]. An alternative approach that is possibly more power-efficient is a continuous-time (CT) SDM that directly senses the microphone's output, which has been used in the previously proposed electret microphone readout interfaces [1,2].…”

“…Some of previously proposed microphone readout circuits can achieve high SNR (more than 75 or even 80 dB), but the harmonic distortion is relatively high, which limits their SNDR to around 60 dB, corresponding to an ENOB of about 10 [1,2,7]. Although many audio applications could allow higher distortion compared to more strict requirement on low noise, high-performance audio systems still prefer low distortion and therefore high SNDR [8,12].…”

“…The microphone is simply modeled as a voltage source is series with a capacitor. A large resistor is needed to provide the CT SDM with a bias voltage, which can be implemented as a pair of anti-parrallel diode [6,7]. The resistor is not included in this design signal voltage enters the readout interface can be expressed as:…”

The influence of perforation on electrostatic and damping forces in thick SOI MEMS structures

“…A large variety of perforated micro-electromechanical (MEMS) devices have been reported over the years. These include sensors and actuators from all fields of MEMS, such as MEMS microphones [1,2], accelerometers [3,4], rate sensors [5][6][7][8][9], resonators [10,11], optical devices [12], energy harvesting devices [13] and so on. Perforations are designed into MEMS structures to reduce damping forces [14].…”