Towards Real-Time Frequency Modulated Accelerometer Bias Calibration

This work provides experimental demonstration of a bias calibration method for frequency modulated (FM) accelerometers, that, in the future, could be implemented in a real-time fashion. The foundation for this effort is theoretical work that found connections between bias shift and previously underutilized components of the noise spectrum of a relatively novel class of FM accelerometers. Information on the design, simulation, fabrication, and testing of these devices is provided. Based on experiments, it is found that while secondary components of the noise spectrum can be used to track bias drift, effectively monitoring shifts in the constituent oscillator’s frequency provides a better means to observe in-run bias drift. After removing long-term drift, the remaining stationary noise processes can be filtered using autoregressive methods. Within the limits of the experimental setup, the filtered response provides a Gaussian estimate of local gravity with a bias instability of 180 ng at 1,000 sec.