Research On High-precision Digital Microphone Readout Technology

With the development and maturity of semiconductor and MEMS technology,consumer electronic products are developing in a trend of miniaturization and intelligence,and the requirements for sound pickup units are getting higher and higher.Digital MEMS microphones are gradually replacing ECM because of their high integration,low cost,strong anti-interference and high stability.This thesis follows the development trend,systematic research on digital MEMS microphone readout technology,two key parameters affecting microphone's accuracy are emphasized.Digital MEMS microphone readout technology is mainly embodied in three aspects: 1.Provide high bias voltage for MCM sensor;2.Pickup and amplify the sensor output signal;3.Implement analog-to-digital conversion to realize digital output.A high-precision digital output MEMS microphone readout chip has bean designed to verify the research content,and elucidate the considerations in the design.The main achievements of this thesis are as follows:1.The non-linear index is emphasis taken into consideration in microphone readout technology,abandoning the way in which traditional readout circuits only focus on noise.With the improvement of MCM sensitivity,the non-linear of the circuit becomes more and more serious,and the influence of harmonic distortion on sound quality even as the same as that of noise,The Pre-Amplifier uses a high linearity structure to stabilize the voltage between three terminals of input MOSFET through a high negative feedback coefficient.The harmonic distortion of 5m V sinusoidal input is 0.001572%.2.The bandgap reference adopts a low-noise structure.The equivalent output noise near1 k Hz is /Hz248.1n V.The Pre-Amplifier has a trade-off between noise and linearity.The equivalent input noise after trade-off is (?) near 1k Hz.3.The charge pump adopts an improved four-phase clock structure,effectively prevents countercurrent,the peak value of output voltage ripple is reduced,and effectively reduces the noise generated by the MCM sensor due to unstable bias voltage.4.The adopting of 4-order Sigma-delta ADC with Oversampling and noise shaping actually improved SQNR.Full feedforward structure is adopted to reduce the influence of circuit nonlinearity on signals.An internal feedback node is used to optimize zeros of NTF(Z),which further reduces in-band quantization noise.5.The DC gain of OTA in the integrator is as high as 94 d B,and the unity gain bandwidth is as high as 109 MHz,which effectively reduces the influence of non-ideal factors on Sigma-delta ADC.The readout circuit is designed base on Dongbu Hitek 0.18?m BCD process,Cadence IC platform and Matlab are used simulation and verification,and than design for layout.The overall circuit SINAD is 103.2d B,which is equivalent to ENOB up to 16.85 Bit,meeting the requirements of high precision design.