Imporvment Of MEMS Thermal Wind Sensor Ouput Characteristic

The monitor of wind speed and direction is an essential part of meteorology,which is closely related to people's daily life.With the development of society,the requirements of wind speed and direction measurement has achieved a new level.Recently,microelectronic mechanical system(MEMS)has experienced a rapid development,so a number of breakthrough sensors succeed.Taking the advantage of miniaturization,low maintenance cost and high performance,MEMS wind sensor become research frontier of next generation wind speed and direction measurement,which is promising in market and valued in research.However,in practice,the performance of the MEMS thermal wind sensor is not ideal.For the asymmetry packaging characteristic,past researches lack theoretical analysis so that it is impossible to build a system-level model for simulation and design.For noise,past researches lack analysis and summary of sources and characteristics.And the filter algorithm is not targeted design.So it is necessary to discuss the reason and solution in depth.On basis of previous studies,further investigations are made to the reason and solution of MEMS thermal wind sensor non-ideal output from packaging asymmetry and noise.The main work and innovations of this dissertation are described as follows:(1)An aging test was performed on the sensor,because repeated heating of the sensor may cause changes in the physical properties of the material.Tests show that the electrical parameters of the chip components change less than 1% after aging,and the continuous measurement output is stable.The effect caused by long time working is eliminated by aging test which makes the sensor enter the stable working interval.(2)The model of MEMS thermal wind sensor with asymmetry package is established to qualitatively describe the effect on performance caused by 2-D package deviation,which is the base of system-level design and simulation.Compared to previous 1-D sensor model,2-D distribution of temperature on sensor's surface has been taken into consideration,while effect of packaging asymmetry is included as well,so it is more accurate on output characteristic describe.Furthermore,measurement of wind speed by thermal loss can be applied to decrease the effect of asymmetry package.Under the wind speed of 8 m/s,the experimental results show that the fluctuation of the wind speed decreases from ±15% to ±5%,when measurement method is changed from thermal differential to thermal loss.The maximum absolute error of wind speed measurement decreased from ± 0.34 m/s to ± 0.25 m/s,decreased by 27%,and the maximum relative error decreased from 2.7% to 1.42%.(3)The source of MEMS thermal wind sensor's circuit noise and intrinsic noise is analyzed,and the performance is optimized by hardware and software respectively.On the hardware,a low-noise circuit is designed for circuit noise.The low-frequency noise is reduced by 26%,while the high-frequency noise is reduced by 89.5%.On the software,a Kalman filter algorithm is designed for intrinsic noise.Under the wind speed of 8.0 m/s,compared with the weighted sliding average filter,the experimental results show that the noise is reduced from 19.7 m V to9.7 m V,a reduction of 50.8%,while the response time is 5.76 s dropped to 2.56 s,a reduction of 55.6%.