Search

MEMS sensors often suffer from distance mismatch that the proof mass does not lie in the right middle of the two fixed plates. The mathematical formulae presented in this paper show that this asymmetry of MEMS sensor does not influence the linearity of accelerometer as a whole, and only introduces an offset. This is true not only for continuous time closed loop, but also for time divided architecture. For this kind of asymmetric sensors, Matlab/Simulink models are established. Simulation results confirm theoretical formulae. To testify this theory, a readout circuit of time divided architecture is designed and fabricated using 0.35µm HV CMOS process. Test results show that the nonlinearity of the tested accelerometer with the asymmetric sensor is within 0.3%.

Matlab/Simulink tool has been widely used in system modeling of circuit design. However, its application in the discrete-time circuit is not enough mature. Based on the continuous-time circuit simulation results by Matlab/Simulink, an s-z transformation method for the discrete-time accelerometer system design is proposed in this paper. In this method, a switched-capacitor readout circuit for closed-loop micromechanical capacitive accelerometer is designed and fabricated in a 0.35µm 15V CMOS process. The accelerometer operates under ±6.5V supply with a full scale of ±50g and the equivalent noise acceleration (ENA) is tested to be 54µg/√Hz at 60Hz.

Linearity which determines the system precision is one of the main performance parameters of accelerometers especially the wide-range ones. The design parameters of the interface circuit have a great effect on the linearity of the accelerometer system utilizing time-division force feedback (TDFF). In this paper, the nonlinearity resulting from time-division force feedback is discussed and the formulae reflecting the relationship between the system linearity and the design parameters are derived, which can help the designers to adjust timing parameters according to the linearity demand. The theory is verified by Matlab/Simulink numerical simulation. Based on the design formulae presented, a ±30g closed-loop TDFF capacitive micro-accelerometer is implemented in 0.35µm high-voltage CMOS process, which operates under ±6.5V supply and achieves linearity of 99.72%in the self-test mode.

In order to design accelerometers, electromechanical hybrid simulation with the sensor and interface circuit together is required. In this paper a Verilog-A model for the accelerometer sensor is presented, and based on the Verilog-A model presented, an open-loop capacitive accelerometer with a full-scale range of ±2g is presented. The Verilog-A model, characterizing the properties of the sensor, is compared to be right with the Matlab model published before. The accelerometer is tested to achieve the equivalent noise acceleration (ENA) about 7.7µg/√Hz.

A new low-power EEPROM circuit for sensor interface circuit is introduced in this paper. The calibration data and parameter setting data stored in EEPROM can be reloaded automatically by the power-on-reset (POR) technique when the power is on. The module only consumes power during power-on-reset and data changing by status optimization and low-power design. This EEPROM circuit design has been used in a MEMS accelerometer readout circuit verified by 0.35μm CMOS EEPROM process. The results are satisfying in that the module can write and store 25bit data and automatically reloads data within 0.1μs after power-on and then turns to low power mode.

A programmable readout circuit for capacitive accelerometer is presented. The inequality of initial capacitance of sensor's differential capacitor pair is properly compensated. The gain of CSA and the cut-off frequency of LPF are both adjustable. To lower noises, amplifier of CSA is carefully designed. Its equivalent input noise density is 12.7nV/ √ Hz. A rail-to-rail buffer is chosen to enlarge the measurement range. EEPROM instead of pads is exploited, which saves about 33 percent of the layout area.