Study On The Readout Technique For High-Performance MEMS Capacitive Accelerometer

As the development of the Internet of Things(IoT)has entered a new stage,the era of the internet of Everything(IoE)has officially arrived.In this new era,sensors and systems for the IoE are required to have high integration and high power efficiency to achieve low-cost mass production characteristics and portable characteristics.In order to realize the miniaturization of electronic products,the semiconductor process of manufacturing integrated circuits and the micromechanical process of manufacturing sensors scale down constantly.However,the effect of scaling-down is not conductive to the low power and high performance design.This is because the scaling-down of the MEMS process means small signal output and large parasitic effects,which require high-performance of the post-stage interface circuit to deal with.This in turn increases the power consumption of the interface circuit and leads to the low power efficiency of sensor system.In addition,the scaling-down of the CMOS process results in the degradation of analog circuit performance,which further increase the challenge.In this paper,the influence of process scaling-down effect on the key performance of the readout circuits for MEMS capacitive sensors are analyzed in detail and the solutions are given.The main research contents and innovative achievements of this dissertation are as follows.1.The readout technique named"Oversampling Successive Approximation-Correlated Level Shift"(OSA-CLS)is proposed to improve the linear output swing and the readout accuracy of the MEMS capacitive sensors.The main works are described as follows.Firstly,the defects and limitations of OSA technique and CLS technique are analyzed.Due to the effect of the“holding error”,the conventional OSA technique only increase the effective gain of amplifier to the square of the original gain,and the ability of gain-enhanced is sensitive to the output swing of the operational amplifier.Although the conventional CLS technique can extend the linear output swing of operational amplifier,its gain-enhanced ability is weak and cannot deal with the offset voltage and 1/f noise of operational amplifier.Based on the above defects of the two techniques,the concept of OSA-CLS technique is proposed,which combines the advantages of OSA technique and CLS technique,and overcomes the disadvantages of each technique.A capacitor-to-voltage converter based on OSA-CLS technique is implemented.Finally,the readout circuit based on OSA-CLS technique is simulated at transistor level by 0.18-μm CMOS process.The post-simulation results show that OSA-CLS technique can reduce the gain error of the amplifier to the third-order.The gain error is inversely proportional to the third power of the amplifier’s gain,i.e.(1/A_V)~3.And it can achieve rail-to-rail full linear output swing.2.The readout technique named“Nested-Oversampling Successive Approximation”(Nested-OSA)is proposed to improve the readout accuracy of the MEMS capacitive sensors.The main works are described as follows.Firstly,the main shortcoming of OSA technique named“holding error”effect which is analyzed in detail.Secondly,the concept of Nested-OSA technique is proposed.By using an active error compensation network,the“holding error”effect can be suppressed and the output accuracy of the readout circuit is greatly improved.A capacitor-to-voltage converter based on Nested-OSA technique is implemented.Finally,the readout circuit based on Nested-OSA technique is simulated at transistor level by 0.18-μm CMOS process.The post-simulation results show that the Nested-OSA technique can reduce the gain error of the amplifier to the fourth order.The gain error is inversely proportional to the fourth order of the amplifier’s gain,i.e.(1/A_V)~4.The offset voltage of the amplifier is suppressed to the two order of loop gain.The offset voltage is inversely proportional to the square of the loop gain,i.e.(1/βA_V)~3.3.The readout circuit based on the common-mode charge balance control technique is proposed for the MEMS capacitive accelerometer.The main works are described as follows.Firstly,the advantages and disadvantages of conventional charge balance control readout circuit are analyzed in detail.Then,a readout circuit based on common-mode charge balance control is presented,and its working principle is described.Then the linearity,residual electrostatic force and the electro-mechanical modulation effect of the readout circuit are analyzed.Finally,the readout circuit based on the common-mode charge balance control technique is simulated in transistor level by 0.18-μm CMOS process.The simulation results show that compared with the traditional charge balance control readout circuit,this readout circuit only needs two mechanical sensing to achieve the linear ratio-metric output of full-differential structure.The readout structure has better resistance to the electro-mechanical modulation effect and EMI.Besides,it is also insensitive to the parasitic capacitance.4.A new technique named“Reset Noise Sampling Feedforwad”is proposed to improve the noise performance of MEMS capacitive sensors.The main works are described as follows.Firstly,the noise of MEMS capacitive sensor readout circuit is analyzed in detail,including discrete Type-A noise caused by parasitic capacitor and continuous Type-B noise generated by operational amplifier,and the relationship between of them is analyzed.Then,the concept of RNSF technique is proposed and a capacitor-voltage converter based on RNSF technique is realized.Thirdly,the noise reduction performance of RNSF technique is analyzed in detail.After that,the transistor-level simulation of readout circuit based on RNSF technique is verified.Finally,the readout circuit based on RNSF technique is manufactured using a 0.18-μm CMOS process,and its noise reduction capability is verified by using the RNSF technique to turn on or off the control circuit.The experimental results show that RNSF technique can reduce the noise floor of MEMS capacitive sensors to 10dB,the noise density is 290μg/(?) and power consumption is 200μW.In summary,the main innovation work of this paper is that the new techniques such as OSA-CLS technique,Nested-OSA technique,common-mode charge balance control technique and RNSF technique are respectively proposed to solve the three important performance deterioration of MEMS capacitive sensors:accuracy deterioration,linearity deterioration and noise deterioration.And these techniques have been proved that it can effectively reduce the gain error,reduce the non-linearity error and improve the noise performance of readout circuit in MEMS capacitive sensor system.