机电系统动力学论文.doc

Dynamics & control of High Speed High Precision LinearMotion SystemAbstract: This paper presents an improved robust control of a high speed high precision XY-table linear motion system used in semiconductor wire bonding machine. robust control is implemented in both inner velocity loop and outer position loop to achieve high and consistent control performance even in the presence of considerable resonance uncertainties and external disturbances. Auto-tuned feed-forward compensation based on fuzzy logic method is employed to obtain satisfactory dynamic and settling performance. High close loop bandwidth has been yielded, and high acceleration of 20g (1g = 9.81m/s 2), high speed of 0.9m/s can be achieved with the dynamic error of 2 m, settling error of 1.2m and static error of 0.2m.Keywords: Robust control; linear motion; XY-table positioning system; feed forward compensation; fuzzy logic;I. INTRODUCTIONWith the rapid development of semiconductor industry, microelectronics and bioengineering, mechanical systems such as machine tools, microelectronics manufacturing equipment, mechanical manipulators, and automatic inspection machines are intensively implemented with the requirements of robust, high speed, and high-accuracy positioning performance. Linear Permanent Magnet Synchronous Motors (LPMSM) has inherent advantages of direct drive, simple structure, high thrust density, high acceleration/deceleration capability and almost maintenance free 1,2. Therefore it is particularly suitable for linear motion system where high speed & high precision are required.In semiconductor wire-bonding machine application, high-speed operation is often required to yield high productivity, and the precision requirement becomes more and more stringent because less than 30 um bond pad pitch bonding capability is required in the industry. High-speed LPMSM direct-drive XY-table positioning mechanism is therefore implemented in the wire-bonding machine to achieve the required performances. In the XY-table positioning mechanism, Y-table is located on X-table and the bonding mechanical system that composes of bond head, optics, wire-clamps and cable bracket are mounted on the Y-table. The complicated mechanical design results in severe resonances, which vary with different operating conditions and from machine to machine due to manufacturing tolerance. The uncertain resonances cause difficulties in feedback control system design. It is difficult for traditionally designed proportionalintegralderivative (PID) controller, in combination with notch filters, to uniquely deal with a wide range of mass-produced machines with multi-varying resonance modes.Intensive studies on the control of high-speed high precision linear motion systems have been conducted in many literatures. A disturbance observer was developed in 3 and used for table positioning systems 4, 5. An adaptive robust position control design was proposed for linear motor drives 6. Composite nonlinear feedback control is proposed to achieve good tracking performance 7. These designs, which are based on the mathematical modeling of low-frequency plant dynamics, may not effectively take the mechanical resonance uncertainty into consideration. Robust control design including velocity/position dual control loop and feed-forward compensator is presented in 8. However, the acceleration of the XY-table is only 5g, the bandwidth is low and the high frequency resonances are not eliminated enough for close-loop frequency response, which will cause vibrations during settling. An improved robust control design is proposed for a high-speed high precision XY-table positioning mechanism. This improved robust control design is employed in both velocity and position control loop to achieve high bandwidth that can reach 450Hz. Feed- forward compensation based on fuzzy logic control is implemented to obtain satisfactory dynamic and tracking performances. High acceleration of 20g is yielded with such design. Satisfactory performance has been achieved in the implementation of mass production.II. MECHANICAL STRUCTURE AND DYNAMIC MODELINGA. Mechanical StructureFig. 1 (a) shows the mechanical structure of the high-speed XY- table positioning mechanism mounted on a wire-bonding machine. Y table is mounted on top of the X table. A bonding mechanical system including a bonding head (BH) and its assembly accessories, such as bonding capillary, wire clamp and cable bracket are placed on the Y table. Each table is driven by a linear permanent magnet motor. The driving motors are composed of a base-mounted permanent magnet and one or two supporting carbon fiber brackets where the three-phase adjoining start-connection coils are embedded. The coil bracket shown in Fig. 1(b) is sandwiched between 2permanent magnet mounted bases with air-gaps and acts as the moving part of the motor. By feeding appropriate three-phase current to the coils, the interaction between the permanent magnets and the coil bracket will generate the thrust force acting on the coil bracket and drive the table through a screw-tightened connection. X and Y tables are designed with cross-roller way guides to achieve extremely fast acceleration/deceleration. They are synchronized to move in the X and Y directions. X table has the largest moving mass and therefore is designed with 2 set coil bracket and permanent magnet stator, which can generate double moving force to drive the mechanical combination of Y table and bond head. High accuracy optical encoders are attached on XY table with an encoder resolution of 0.05 um to achieve high control accuracy.(a) XY table mechanical assembly(b) Y table coil bracketFigure 1. Mechanical Structure of XY tableB. Dynamic ModelingIn general, the dynamics of the voltage-controllable LPMSM can be described as： （1） （2） （3）where x is the motor position; ,，and are mass of the moving part, viscosity constant, generated force, and system disturbance, respectively; , ,and denote the electrical parameters, i.e., time-varying motor terminal voltage, armature current, armature resistance, and armature inductance, respectively ;is an electricalmechanical energy conversion constant; and is the back electromotive force (EMF) constant.However, the dynamic model can only reflect the low frequency characteristics of the motor. It can not describe the high frequency resonances caused by attached mechanical system. Pseudo-random binary sequence (PRBS) has been employed to get the frequency response of the system by injecting PRBS singles to the current shown in (3) and reading the feedback encoder position signals shown in (1). The velocity frequency response that depicts the relationship between the input current and the velocity derived from the observed position can be obtained as shown in Fig.2, resonances can be observed from 600 Hz. High order model can be fitted based on the frequency response to design the robust controller.Figure 2. Velocity Frequency Response of X TableIII. ROBUST CONTROL DESIGN OF THE SYSTEMA. Control StrategyFig. 3 shows the block diagram of the motion control system. The system consists of a hybrid control structure, which has two digital feedback control loops, inner velocity loop and outer position loop, plus feed forward compensator (FFC). Both velocity and position controllers are designed by employing H robust control. Fuzzy logic method is applied to auto-tune the FFC parameters based on the dynamic and settling performances. These parameters are altering along with the changes of the traveling distance, maximum acceleration and speed. Optimal motion performance can be achieved by employing the fuzzy tuned FFC parameters. The position, velocity, acceleration and jerk reference commands are produced on the basis of position motion profile with minimal time realization. The position error is generated with the comparison of the feed-back encoder position and the command position. The position controller output acts as velocity reference component and adds on the velocity command reference to compare with the encoder velocity.Generated velocity error is sent to velocity controller to produce current reference component. The sum of this component and the FFC output forms the current reference.Figure 3. Control Block DiagramB. Robust Controller DesignRobust control is good at handling the uncertainty by bounding it rather than expressing it in the form of a distribution. The control can achieve results that meet the control system requirements in all cases by applying a bound on the uncertainty. Therefore robust control theory might be stated as a worst-case analysis method rather than a typical case method. It is thus a good choice for the complicated mass production system like XY-table motion mechanism with many uncertainties and high frequency resonances.Figure 4. Control Diagram with Weighting Functions.Figure 5. Standard Problem Control Diagram .Fig.4 shows the feedback control diagram with three weighting functions that are control error, control input and control output respectively. The control structure can be transformed to the standard control problem as shown in Fig.5. Where, w is input signal, u is the control signal, y is measurement output signal and Z is the control output signal that can be defined as （4）The augmented plant P can be derived as （5）And the close-loop transfer function of the control system is shown as （6）Where, S , R and T are the sensitivity, control sensitivity and complementary sensitivity shown below （7） （8） （9）The entropy equation is then be written as （10）The optimal can be derived by minimizing the entropy, and the sub-optimal stabilizing controller will be found to satisfy the weighting equation below （11）IV. SUMMARYThis paper has introduced a hybrid robust control strategy for high speed high precision XY-table positioning system used in wire bonding machine. Both inner velocity loop and outer position loop are controlled by based robust controllers for the sake of consistent and robust control performance. Auto-tuned fuzzy logic feed-forward controller has been implemented to achieve high dynamic and settling performances. And the experiment is needed to demonstrate satisfactory motion performance with the proposed hybrid robust control system later.REFERENCES1 I.Boldea & Syed A. Nasar, “Linear Electric Actuators & Generators”, Cambridge University Press, 19972 M.S.W. Tam and N.C. Cheung, “ A high speed high precision linear drive system for manufacturing automation”, Applied Power Electronics Conference and Exposition, 2001.3 T. Umeno and Y. Hori, “Robust speed control of DC servomotors using modern two degree-of-freedom