Research On The MEMS Thermal Wind Sensors Packaging

Wind speed and direction measurement is widely used in traditional meteorological research,transportation,wind power generation,and other scenes.With the rise of 5G applications,it also shows great development potential in intelligent buildings,agricultural planting,and military.These promote the development of wind sensors towards intelligence,miniaturization and low power consumption.Micro-Electro-Mechanical Systems（MEMS）,born out of microelectronic technology,integrates micro sensors,micro actuators,and processing circuits monolithically.It has the advantages of small size,light weight,low power consumption,high reliability,high sensitivity and integration with on-chip circuitry.MEMS thermal wind sensors based on the calorimetric principle has been widely studied since its appearance due to their small size,low power consumption,no movable structure and other advantages.Since researching MEMS thermal wind sensors,our group has accumulated rich experience and achievements in sensor design,interface circuits,packaging,and environmental effects.We have broken through critical nodes to form a complete sensor system solution.However,there is a lack of systematic theoretical guidance for the optimal design,and there is little research on the impact of packaging on sensor performance.To solve the above problems,a comprehensive analytical model is established,a compensation method for benchmark drift induced by packaging asymmetry is proposed,and a high symmetry and low drift sensor system is developed.The research contents and innovations include:（1）A comprehensive analytical model for MEMS thermal wind sensor is established,which reveals the influence of various parameters in the chip design on the sensitivity and linearity,and provides guidance for the performance optimization of the sensors.Previous simplified models assume that the substrate temperature is uniform,and the output is proportional to the sqrt root of wind speed,called quasi-linear characteristic.However,in order to reduce power consumption,a large number of studies focus on low thermal conductivity materials or thin-film structures,which means that the assumption of uniform substrate temperature is not suitable.Therefore,a comprehensive analytical model of MEMS thermal wind sensors is established,the assumption of chip temperature uniform is removed,and the general solution of the sensor output is determined.Furthermore,the effects of sensor heating size,measurement position,substrate thickness and thermal conductivity of substrate on the sensor performance are investigated.The opposite relationship between sensitivity and linearity in sensor design is revealed for the first time,that is,the higher the sensitivity,the worse the linearity in sensor design.（2）The benchmark drift of MEMS thermal wind sensor caused by packaging asymmetry is found.In early stage,the sensors were encapsulated by lab-made ceramic packaging.Experimental results indicates that the benchmark would drift with the change of wind speed,and the greater the wind speed,the smaller the drift.A lumped parameter model of MEMS thermal wind sensor is developed,it is found that the output of the sensor is composed of wind sensitive term（asymmetric heat convection）and benchmark drift term（asymmetric heat conduction）.In the lab-made ceramic packaging,some packaging steps are implemented by manual operation.This will introduce asymmetric factors,resulting in asymmetric heat conduction of the sensor,and then affect the benchmark drift.In addition,the Finite Element Method（FEM）models are developed to further examine the conclusion that the packaging asymmetry leads to the benchmark drift.（3）Aiming at the benchmark drift caused by the asymmetric packaging of MEMS thermal wind sensor,a drift compensation method is proposed,which greatly improves the measurement accuracy.The conventional signal processing procedure using the no-wind output as fixed reference will seriously reduce the measurement accuracy of the sensor with benchmark drift.Therefore,a drift compensation method is proposed:Firstly,the output voltage at zero wind is recorded;Then,the thermal resistance of packaging is solved according to the benchmark drift equation obtained by the lumped parameter model;After that,the thermal resistance of packaging is substituted into the drift function to obtain the relationship between the benchmark and the wind speed and direction;Finally,the drift function is used as the benchmark of signal processing to solve the wind speed and direction information.The experimental results show that in the range of 0～30 m/s and 0～360°,using the conventional signal processing method,the wind direction error of the sensor is±30°and the relative wind speed error is±46%.After compensation,the wind direction error is reduced to±7°and the relative wind speed error is reduced to±8%.The compensation scheme not only greatly reduces the measurement error,but also requires little additional work,and significantly lower the compensation cost.（4）A plastic packaging is proposed to fundamentally cancel the benchmark drift of MEMS thermal wind sensor.Although the compensation can reduce measurement error of the sensor system with lab-made ceramic packaging,the residual error is still intolerable.In order to remove the benchmark drift,three technical paths are proposed:reducing packaging deviation,reducing thermal conductivity of packaging and using constant temperature difference（CTD）heating mode.The technical paths were verified by lumped parameter models and FEM simulation.Finally,drawing on the injection molding process widely used in IC packaging,the plastic packaging suitable for MEMS thermal wind sensor are designed.In addition,according to the results of the comprehensive analytical model,a sensor design based on ceramic substrate is proposed after considering the tradeoffs of power consumption,measurement range,chip size,cost,accuracy,reliability and other factors.The experimental results show that in the full range of 0～30 m/s and of0～360°,the maximum relative wind speed error is±7%,and the wind direction error is less than±7°.Plastic packaging not only solves the problem of benchmark drift,but also greatly reduces the packaging cost,which lays a good foundation for mass production.（5）Aiming at the poor sensitivity in high-speed range of plastic encapsulated MEMS thermal wind sensor,an optimized circuit is proposed.Limited by the heat convection principle of MEMS wind sensors,the sensitivity decreases rapidly with the increase of wind speed,which obviously deviates from the quasi-linear characteristic.A closed-loop Wheatstone bridge（CLWB）measurement circuit is proposed to replace the Wheatstone bridge（WB）.In CLWB,the supply voltage increases with the increase of wind speed,which improves the sensitivity of the sensor in the high wind speed range and makes the output characteristic curve closer to quasi-linearity.The experimental results show that in the wind speed range of 0～33 m/s,using CLWB instead of WB circuit,the exponent term of the wind speed is optimized from 0.424 to 0.544,and the corresponding sensitivity at 33 m/s is increased from 2.31 m V/(m·s^-1)to 3.82 m V/(m·s^-1).With the goal of low drift,high reliability,and high linearity,this dissertation focuses on the theoretical model,circuit,and packaging of MEMS thermal wind sensors.A series of optimization,such as optimizing the analytical model,finding the benchmark drift,proposing the benchmark compensation scheme,designing the injection molding packaging,and developing Closed-Loop Wheatstone Bridge（CLWB）have been presented.