CMU MEMS Laboratory Publication Abstract


in M.S. Thesis, May 1998, Carnegie Mellon University, Pittsburgh, PA.
Design, Simulation, and Implementation of Two Novel Micromechanical Vibratory-Rate Gyroscopes
M. S. Kranz
Microelectromechanical systems (MEMS) technology holds the promise of making complex systems that integrate micromechanical and microelectronic components. These complex systems will be extremely small in size and inexpensive when batch fabricated. A particular class of complex systems that is being explored is that of inertial sensors. Systems such as accelerometers and gyroscopes can be implemented in a small package at a small cost, and provide high performance while being rugged and operable in harsh environments. In particular, a MEMS gyroscope system would be orders of magnitude less expensive and take up less space than the macro-systems available.

A number of individuals and groups have worked on micromechanical vibratory-rate gyroscopes with varying designs, and implemented in different fabrication processes. Bernstein demonstrated a tuning fork gyroscope in which two masses oscillate laterally. An external rotation induces oscillation out of the plane of the device. Boeing is now developing a commercial version of this design. Juneau demonstrated a gyroscope with a rotating mass, where deflections induced by rotations about two axes were also measured out of the plane of the device. Clark described a gyroscope with a single mass oscillating laterally, and with lateral rotation induced deflections sensed capacitively. A gyroscope based on a vibrating ring was demonstrated by Putty. The work presented in this thesis focuses on two new designs that reduce some of the numerous sources of measurement error in gyroscopes.

An initial consideration in MEMS design is the fabrication process that will be used. The main process used in this work is a combination of a CMOS process offered by an outside vendor, and a set of post-processing steps done here at Carnegie Mellon University. This process provides a composite structural material that allows for novel designs, a straightforward integration of electronics with the microstructure, and a scalable process. However, the composite structure leads to problems with curvature resulting from gradients in residual stresses. In addition, the post-CMOS micromachining steps are not mature and are a subject of parallel research.

The two gyroscopes being explored are in the class known as vibratory-rate gyroscopes. This class of devices must be mechanically excited in resonance to measure the rotational rate of the device. A configuration of mass is forced to oscillate in a specific mode. When the device experiences an external rotation, the velocity of the mass results in a Coriolis force acting on the mass in the frame of reference of the device. This force acts in a direction orthogonal to the velocity of the mass. The resulting oscillatory motion is then sensed in some fashion and a rotational rate signal derived from it.

There are two primary ways of configuring these devices. In one configuration, resonance matching, the excited mode is a resonant mode and a large displacement is developed due to the Q-factor of the structure. The sensed mode is designed to have nearly the same resonant frequency as the excited mode. Then, when the Coriolis force excites that mode, another gain of Q is seen. The second configuration, resonance separation, uses two modes that are separated in resonant frequency. The sensed mode is then a controlled mode that operates similarly to an accelerometer and measures the Coriolis acceleration.

The two designs in this thesis attempt to reduce sources of measurement errors seen in micromechanical gyroscopes. One device, the three-fold symmetric gyro (3FSG) operates using a matched resonant frequency configuration. This device uses a completely symmetric structure to ensure that the two oscillation modes are matched through process variations and environmental conditions. The second device, the elastically-gimbaled gyroscope (EGG) can be designed using either system configuration, but has only been implemented in the matched resonant frequency configuration. This device places one mass, constrained to motion in only one direction, within a second mass that is constrained to move only in the orthogonal direction. Completely separating oscillation modes reduces errors due to mechanical crosstalk.

Simulation of the gyroscope systems, and MEMS in general, is important in the design and verification of these devices. Mechanical simulation can and is done using a finite element package. However, that strategy will not simulate the interactions of the mechanical elements with electrostatic elements, electronics, and controls. The design and modeling of these gyroscopes contributed to the motivation for and the development of a multidomain MEMS simulation methodology, NODAS. A hierarchical multi-domain nodal simulation methodology is being developed here at CMU, and the two gyroscope devices have been important testbeds for the development, testing, and verification of the simulation tools. The devices have also motivated modeling of components for the tools. Beams with effective mass and plate masses that include global inertial effects due to rotation, were important components in the simulation of these gyroscopes.

Two generations of both gyroscopes have been implemented and fabricated at the point of this writing. The 3FSG has been fabricated in both a polysilicon process and the CMOS-MEMS process, but only the CMOS-MEMS version will be discussed here. The EGG has only been fabricated in CMOS-MEMS, as the structure cannot be operated usefully if fabricated in a polysilicon process. Simulations of the mechanical structures have been performed using finite element analysis in ABAQUS, and multidomain nodal simulations have been performed in Saber. Due to delays in the CMOS-MEMS processing, devices have only now been released and only initial testing has been performed for this report. Future directions for design, simulation, and test are outlined in the conclusions.
© 1998 Carnegie Mellon University, Department of Electrical and Computer Engineering.
Full paper (PDF) (opens in new window).

This page was generated in 0.026043 seconds at 05:13:15 pm EST on 23 Nov 2017.

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