Research On FPGA-based High Bandwidth Voice Coil Motor Control System

With the rapid development of semiconductors,computers,aerospace and other industries,the machining dimension is getting much smaller,which require higher positioning accuracy of high-end manufacturing equipment.Therefore,the key motion systems of high-end equipment,such as precision positioning system and vibration isolation device which are directly related to system accuracy,are particularly important.The voice coil motor has become a widely used drive in precision positioning and vibration isolation systems due to its low hysteresis,high response,and near-zero stiffness.With the improvement of performance requirements,the thrust nonlinearity caused by the voice coil motor cannot be ignored,which has become an urgent need to study high-bandwidth voice coil motor control systems that can compensate for nonlinearity.This thesis focuses on the permanent magnet force compensated voice coil motor,and studies the hardware design and nonlinear force compensation methods of its control system.A voice coil motor control system model is built to analyze the influence factors of the bandwidth of the current loop.For the high bandwidth requirements of the system,the hardware architectures of different control systems are compared and the FPGA is selected as the main control unit.The nonlinear characteristics of the permanent magnet force compensated voice coil motor are analyzed,and the influence of thrust nonlinearity on the motion accuracy is described.A nonlinear compensation control system scheme based on FPGA neural network is proposed.A linear power amplifier driver circuit is designed for the bandwidth and precision requirements of the driver circuit,and the current sampling circuit design is analyzed in detail.Through reasonable circuit board layout and power device thermal design,the drive circuit achieves its design goals.The digital current loop design is based on FPGA.By properly designing the timing and logic,the current loop bandwidth has been significantly improved.Taking the thrust force and displacement as input,the current is used as a training target,and the basic parameters of the BP neural network are determined using Matlab.Then,based on the analysis of the neural network operation,various modules of the neural network are designed using VerilogHDL.The area-optimized method implements a neural network computing unit and composes a system-on-chip with other IPs.The experimental results show that the neural network can compensate the nonlinear problems of the permanent magnet force compensated voice coil motor within a certain range.