CMU MEMS Laboratory Publication Abstract


in Ph.D. Thesis, May 2002, Carnegie Mellon University, Pittsburgh, PA.
Sensing and Control Electronics for Low-Mass Low-Capacitance MEMS Accelerometers
J. Wu
In this work, circuit and system design techniques for sensing and controlling the motion of MEMS structures with ultra-small mass and ultra-small capacitance are investigated and are used to realize low noise integrated CMOS MEMS accelerometers. Structures fabricated by CMOS MEMS surface micromachining have total mass smaller than 10-9 kg and total sensing capacitance smaller than 100 fF. CMOS MEMS accelerometers typically have low sensitivity around 1 mV/g and less than 0.4 fF/g acceleration-induced capacitance change, therefore, noise and other nonidealities must be minimized. There are three sources of noise in MEMS accelerometers: electronic noise from sensor interface circuits; thermal-mechanical Brownian noise due to energy dissipation caused by damping; and quantization noise when analog-to-digital conversion is included. Other nonlinearities include sensor position offset, circuit offset and undesirable charging at the highimpedance sensing nodes.

In the area of sensing circuit design, we introduce a circuit noise model that is validated by experiments and provides insights on design trade-offs. We apply a set of circuit techniques to minimize the circuit noise and suppress other nonidealities, including: a low noise architecture based on chopper stabilized continuous-time voltage sensing; inputreferred noise minimization based on capacitance matching at the sensor/circuit interface; a robust sensing node biasing scheme using periodic reset for charging suppression; and offset cancellation using differential difference amplifier. An integrated CMOS MEMS accelerometer prototype using these techniques achieves 50 μg/rtHz noise floor which is close to the Brownian noise floor, and > 40 dB of sensor offset reduction.

At the system level, force-balanced electromechanical delta-sigma modulation with high-Q micromechanical transducer is investigated to reduce Brownian noise and quantization noise altogether. A single loop architecture is introduced along with the switchedcapacitor circuit implementation of the loop filter. A digital force feedback scheme called complementary pulse density modulation (CPDM) is proposed to realize highly linear offset- insensitive feedback using nonlinear actuators. Simulations show such systems realize high-resolution A/D conversion with 100 dB dynamic range and μg/rtHz quantization plus Brownian noise floor while simultaneously provide robust control to the high-Q micro structure to obtain near optimum closed-loop settling and less than 2 Angstrom proofmass position error.

© 2002 Carnegie Mellon University, Department of Electrical and Computer Engineering.
Full paper (PDF) (opens in new window).

This page was generated in 0.034366 seconds at 10:07:57 am EDT on 26 May 2018.

overview | projects | people | publications | intranet | resources         © 1998-2009  Carnegie Mellon