A real-time temperature compensation algorithm for a force-rebalanced MEMS capacitive accelerometer based on resonant frequency

This paper presents a real-time temperature compensation algorithm for a force-rebalanced MEMS capacitive accelerometer which relies on the linear relationship between the temperature and its dynamical resonant frequency. A phase-locked loop (PLL), where we inject a drive signal to excite the accelerometer into resonance and track the real-time resonant frequency, is added to the conventional force-rebalanced loop to compensate for the temperature-induced output drifts. Experiments demonstrate that the first-order temperature dependence of the resonant frequency is −0.1308 Hz/°C. The bias offset of the accelerometer has a first-order temperature coefficient of 3.54 mg/°C, which is significantly reduced to 0.05 mg/°C after compensation. Besides the improved bias instability of 1.6 µg after compensation, the long-term drift at the integration time of 100 s is suppressed more than one order of magnitude, from 35 µg to 2.4 µg, in the Allan deviation plot, which reveals the inhibition of the temperature-related performance deterioration.