Real-time tuning of MEMS gyro dynamics

This paper reports real-time tuning of the JPL-Boeing micromachined vibratory rate sensor. The ideal sensor is designed to operate in a degenerate condition in which two modes of vibration have equal resonant frequencies. This condition achieves the best possible signal-to-noise ratio thereby maximizing sensor performance. A frequency split between the two modes, however, is inevitable in actual devices and leads to degraded performance. To modify the sensor dynamics to a desired condition, we have studied the bias potential effect on the sensor dynamics and successfully implemented a real-time tuning process via electrostatic forces to reduce the frequency split to less than 0.1 Hz when the nominal modal frequencies are near 4.4 kHz. A closed-loop identification method is employed for rapid and precise empirical frequency response estimates of the sensor dynamics. An LMI-based parameter estimation scheme produces an excellent fit of the model to the frequency response data and this enables the successful implementation of a steepest descent algorithm. Transformations for decoupling the MIMO sensor dynamics are also motivated and demonstrated.