Research On Self-calibrated Digital Closed-loop Interface ASIC For MEMS Acclerometers

Nowadays,due to the advantages in volume,power consumption and cost,MEMS accelerometers have been widely used in consumer markets.In recent years,the digital electrostatic servo-controlled ∑Δ closed-loop accelerometer has achieved further performance improvement in noise,bandwidth and linearity,which has driven the use of MEMS accelerometer to higher-end application.However,the residue stress induced long-term performance drift problem has been the main limitation when using MEMS accelerometer to high-end industrial and military applications.The most pernicious form of drifts is the long-term bias drift.Purely electric excitation based self-calibration technique could be an effective solution.But current research on electrostatic self-calibration could only achieve the low precision calibration of open-loop MEMS accelerometer.The self-calibration of high-precision digital closed loop accelerometer has not yet been deeply researched.Aimed at above problems,this paper will conduct in-depth research on self-test,selfcalibration technique for the dc bias drift of digital closed-loop accelerometer.The residue stress is inevitably involved during the fabricating and packaging process of MEMS structure.The residue stress will slowly release during the longterm usage,which is the main cause of bias drift of the sensor.The research has found that the stress release will translate to closed-loop servo position drift,cause system imbalance,and induce even-order harmonics.In this paper we established the mathematical model of the harmonic distortion and closed-loop bias drift,revealed the linear relationship between closed-loop bias drift and even-order harmonic distortion,and verified the correctness of proposed theory whereby experimental test.Based on this relationship,through the method that excite the MEMS sensing element via by on-chip generated 1Bit ∑Δ modulated electrostatic force and observe the even order harmonics in the output response,the self-test of dc bias drift has been realized.On this basis,this paper proposed an on-chip self-adaption digital selfcalibration method.Whereby the on-chip synchronous orthogonal demodulation circuit,the background noise is suppressed,the second-order harmonic distortion is extracted.According to the information extracted,a digital self-calibration loop is estabished,which automatically adjusts the value and direction of front-end capacitors array to compensate the drift of sensing capacitors.Thus the on-chip digital calibration of dc bias is realized.Under sef-test excitation,the displacement variation of movable mass is enlarged,which will made the loop stability problem more severe.For this,we made an in-depth research on the displacement modulation effect,estabilised an imporved loop stability model which takes this phenomenon into consideration.Further analysis has found that,the displacement modulation effect of electrostatic force will on one hand introduce a local positive feedback path,and on the other hand cause an expansion in the loop gain,both of which are detrimental to loop stability.At last,the upper and lower boundray of loop coefficients ensuring the loop stability are derived.The experimental test results shows,the improved stablity model is more consistant to the real condition.According to the derived relationship,the coefficients are optimized,the unstable problem under self-test mode has been effectively solved.Based on above research,a digital closed-loop interface ASIC for capacitive MEMS accelerometer is designed which has the unique dc bias self-calibration function.The proposed ASIC is fabricated and taped-out using 0.35μm HV-BCD process.The total chip area is 4.2mm×4.0mm,the total power consumption is37.5m W,in which the digital self-calibration circuit only takes up 1/5 area,and 1/20 power consumption.With the MEMS sensing element provided by cooperating research institute,a series of system tests are carried out.The test results show that,the digital closed-loop accelerometer system has achieved a noise floor of 1μg/√Hz,and a signal bandwidth of 1k Hz,a dynamic range of 140 d B.When enabling the selfcalibration function,the system output dc bias is reduced from 150 mg to 4mg.And under different placing statuts,the self-calibration results are unaffected and able to achieve a residue bias below 4mg.The experimental results show that the proposed harmonic-distortion based bias self-calibration method is effective,and is unaffected by the external acceleration,thus can be carried out on-line whenever needed.