| With the continuous innovation of MEMS technology,MEMS inertial devices are widely used in various fields for their low power consumption and light weight,especially in the field of autonomous driving and robotics.As the internal MEMS inertial devices consists of temperature sensitive material silicon,changes in ambient temperature can affect its performance and directly lead to a degradation in measurement accuracy.Solving temperature drift errors become the biggest challenge in improving the measurement accuracy of MEMS inertial devices.The common methods used today can be divided into two main categories:hardware compensation and software compensation methods.Hardware compensation is generally carried out by studying the internal structure and material properties of the sensor and then changing the internal structure of the sensor,replacing the heat resistant material or by setting up a thermostatic circuit inside the sensor to compensate for temperature changes effects.The disadvantage is that it is difficult to fundamentally reduce the mechanical deformation of inertial devices due to temperature change.Moreover,the readout circuit performance is also susceptible to fabrication process variation and temperature changes effects.Software compensation is based on the temperature output error model,through the fitting algorithm to get the inertial device temperature model of various compensation factors,common fitting models are:polynomial fitting,neural network fitting,etc.The disadvantage of polynomial models is that they require a relatively high order and a large number of coefficients to achieve better compensation,making the amount of operations very large;the disadvantage of neural network fitting is that it requires a large number of training samples and is difficult to integrate into the chip.To address the above problems,this thesis proposes a real-time temperature compensation method,which is based on a new function compensation model.The first-order Fourier function is chosen as the temperature compensation model to address the difficulty of implementing and integrating complex function models.Firstly,the model does not require a large number of coefficients and has a low dependence on the initial value.Secondly,the hardware implementation is simple as the CORDIC algorithm is used to calculate the trigonometric function values.Therefore,the model is easy to implement and highly stable,and is suitable for real-time compensation of inertial devices.Finally,the temperature curve data is tested for real-time compensation,and the test results show that the absolute value of the temperature drift of the output measured at 0g is reduced from 0.18 V to 0.0863 m V,and the maximum deviation is reduced from0.0544 m V to 0.2008 m V before and after compensation respectively.This demonstrates the validity and feasibility of the temperature compensation model and of practical application. |
