FPGA implementation of a low-cost method for tracking the resonance frequency and the quality factor of MEMS sensors

This paper reports the hardware implementation of a low-cost parameter estimation method (LIMBO) and describes its use for tracking the parameters of a resonant MEMS sensor. We show how this method can be used to identify the impulse response (IR) of a sensor with minimal A/D-D/A requirements. Based on this IR, we develop digital estimators capable of extracting the resonance frequency and the quality factor. The full electronic architecture is simulated on a mixed-signal environment and implemented on FPGA-based digital hardware. For a specifically emulated sensor, results show that the resonance frequency can be identified and tracked from 3kHz up to 12kHz with a Normalized Square Error (NSE) of 1.38E-1. The quality factor is estimated and linearized with a correlation coefficient of 0.87731 between Q=2 to Q=60. Those results can be easily scaled to any frequency range by adapting the sampling frequency.