Abstract: In the semiconductor manufacturing industry, silicon wafer transport robots undertake complex tasks such as precise positioning and rapid handling, which places stringent requirements on their controllers. Therefore, systematic research on the control modules of silicon wafer transport robots is of great significance. This paper takes the PMAC motion controller from Delta Tau Data Systems (USA) as an example to introduce the application of the PMAC board in the controller of a silicon wafer transport robot. Using the company's PMAC2-PC104 control board as the core, and combining electrical and pneumatic circuits, a polar coordinate type controller for a silicon wafer transport robot was designed and established, realizing the basic operational motion of the silicon wafer transport robot. Keywords: wafer-handling robot; PMAC motion controller; robot controller Abstract: In the semiconductor manufacturing industry, wafer-handling robots can execute complex tasks such as precise localization, fast transporting, and so on. To achieve better performance of wafer-handling robots, it is essential to carry out system research on the control units of wafer-handling robots. This paper takes the PMAC movement controller from American Delta Tau Data Systems Corporation as an example, introduces the PMAC board to robot controller application, and designs and builds an R-theta (polar coordinate) wafer-handling robot controller based on PMAC2-PC104. Key words: wafer-handling robot, PMAC movement controller, robot controller Note: This work was supported by the National High Technology Research and Development Program of China (863 Program) (2002AA421230) . 1. Introduction Wafer-handling robots are important transmission and positioning equipment in the semiconductor integrated circuit (IC) manufacturing industry. Their working speed, positioning accuracy, and reliability directly affect the production efficiency and manufacturing quality of silicon wafers. Silicon wafer transport robots achieve rapid transfer of silicon wafers within a limited space, which places high demands on their motion characteristics, responsiveness, accuracy, stability, and reliability. Therefore, these operational characteristics must be fully considered when designing the silicon wafer transport robot controller. A Programmable Multi-Axis Controller (PMAC) is a widely used automation control device in industrial control. It can be configured on a common PC platform, providing an open, modular servo motion control system. It also facilitates computer-based servo control algorithm research and real-time monitoring, making multi-axis linkage parameter matching design more convenient. Silicon wafer transport robots typically consist of 3-5 degrees of freedom robotic arms, while PMAC boards can provide independent motion control for up to 8 axes. Therefore, the silicon wafer transport robot controller can fully utilize a PMAC board as its core to implement its functions. 2. Introduction to PMAC Motion Controllers PMAC stands for Programmable Multi-Axis Motion Controller. It typically uses a high-speed DSP (Digital Signal Processor) as its core, employing a single microprocessor to implement a servo system for multiple motors. Developed by Delta Tau Data Systems, the PMAC open-architecture motion controller is one of the world's most powerful motion controllers. Based on Motorola's DSP56K series digital signal processor (DSP), the PMAC motion controller is available as a card-type product or a standalone controller module supporting multiple bus slots (ISA, PCI, VME, PC104). A single controller can operate 8-32 axes simultaneously and in parallel. Each axis controlled by the PMAC motion controller is completely independent; a single card can control 8 single axes from 8 different machines, 8 axes from the same machine, or any combination thereof. In terms of servo data processing capabilities, axis characteristics, and input signal bandwidth, the PMAC controller, due to its specialized modular structure, achieves a serial processing speed for encoder inputs 10 to 15 times faster than most controllers. Furthermore, it can receive low-interpolated 5-bit parallel data from high-resolution encoders, resulting in an effective input bandwidth of 320MHz. The PMAC motion controller boasts exceptional flexibility, adaptable to a wide variety of commonly used bus structures, motor types, feedback components, and instruction/data structures, allowing for easy hardware upgrades. The PMAC controller enables the same control program to run on all buses, while also allowing for different combinations of motors and feedback components on each axis. Its advanced modular design provides an open architecture, allowing for the selection of appropriate options and accessories based on different application systems. When a system requires multi-axis linkage and interpolation control, a single PMAC board can be used in various automated equipment such as robots, CNC machine tools, coordinate measuring machines, laser processing, engraving machines, printing, and packaging. In this silicon wafer transport robot controller, we selected the PMAC2-PC104 control board. 3. Overall Design Scheme of Silicon Wafer Transfer Robot Controller The silicon wafer transfer robot is a polar coordinate robot with three degrees of freedom. The joints are driven by belts. According to the above design requirements and ideas, the performance indicators of the silicon wafer transfer robot control system can be summarized as follows: (1) Three-axis linkage, semi-closed-loop control mode; (2) Rapid positioning; (3) Circular interpolation function; (4) Serial communication with the host computer; (5) Offline operation function; (6) I/O switch control function; (7) Compensation function; (8) User-friendly human-machine interface; (9) Open control mode; (10) Two modes: linkage and jogging. The design is based on openness, economy, practicality and reliability. 4. Specific Design and Implementation of Silicon Wafer Transfer Robot Controller 4.1 Selection of Motor Drive Mode The control signal output by the computer is insufficient to drive the motor or other actuators. This signal must be amplified by an amplifier to drive the motor and other actuators. Drivers are generally classified into five categories: speed-mode drivers, torque-mode drivers, direct PWM digital-mode drivers, sinusoidal input-mode drivers, and pulse-plus-direction drivers. In computer control systems, the choice of driver mode significantly impacts system performance. The silicon wafer transport robot is a precision point-to-point motion system, controlled by point-to-point control. Considering the high required positioning accuracy and relatively insignificant speed requirements, position control mode was chosen for this silicon wafer transport robot controller. 4.2 PMAC Control of Servo Motors The silicon wafer transport robot requires point-to-point position transfer, and its control method is position control. PMAC uses a pulse-plus-direction mode for position control. This silicon wafer transport robot controller uses the PMAC2-PC104 series multi-axis controller, capable of controlling pulse-plus-direction input servo motor drivers. These drivers can operate in both open-loop mode (actually by introducing the pulse train into their own encoder counter through an internal subroutine to establish a pseudo-closed loop) and closed-loop mode (actually by connecting feedback from external devices to the PMAC2-PC104 to establish a closed loop). The PMAC2-PC104 uses a fully digital pulse frequency control (PFM) circuit to establish its pulse and direction control signals. This circuit repeatedly adds the latest command frequency value to an accumulator. The frequency of the output pulse train is proportional to the command value, and there are no issues such as offset or waveform distortion associated with analog pulse generators. 4.3 Pneumatic Control Circuit of the Silicon Wafer Transfer Robot The flipping action of the end effector of the silicon wafer transfer robot is achieved by a swing cylinder, and the wafer gripping action is achieved by the vacuum adsorption effect of a vacuum generator. Therefore, a pneumatic control circuit is used in the control system of the silicon wafer transfer robot. This pneumatic circuit requires two output ports for solenoid valve control and two input ports to receive signals from the magnetic position switches of the swing cylinder. Since the number of required I/O ports is small, the flag bits on the PMAC2-PC104 can be directly converted into general-purpose I/O ports, where the control of the output ports is controlled by the value of the M variable corresponding to that port. The pneumatic circuit and its electrical control diagram for the silicon wafer transfer robot are shown in Figures 4-1 and 4-2. As shown in Figure 4-1, when the system starts and the air source is connected, the swing cylinder 1 moves to its left limit position, at which point the magnetic switch 1 is engaged; the vacuum generator 2 does not generate negative pressure at this time and does not produce an adsorption effect. When YA1 is energized, the solenoid valve 6 switches to the left position, and the swing cylinder moves to its right limit position, at which point the magnetic switch A2 is engaged. When YA2 is energized, the two-position two-way solenoid valve 7 switches to the left position, air flows into the air inlet of the vacuum generator, and negative pressure is generated in the vacuum hole, at which point an adsorption effect is produced. The end effector of the silicon wafer transfer robot uses this working mechanism to realize the grasping and releasing of silicon wafers, while the flipping action is driven by the swing cylinder. In Figure 4-1, the speed of the swing cylinder is controlled by a one-way throttle valve connected to the two air holes using an exhaust throttling method to ensure the smooth operation of the piston. Intake throttling is generally not used because it results in a small intake flow, slow pressure rise in the intake chamber, rapid exhaust, and very low exhaust pressure. The piston movement relies primarily on the expansion of compressed air, making it difficult to control the cylinder speed to a stable level. Intake throttling is typically only used for single-acting cylinders, clamping cylinders, and low-friction cylinders. As shown in Figure 4-2, the internal output ports 33 and 34 of PMAC's ACC1 (JMACH1) are open-collector (OC) outputs. The two solid-state relays (SSR1 and SSR2) controlling the solenoid valve coils require a +5V voltage to conduct. Therefore, 3.3K pull-up resistors are used at ports 33 and 34 to limit current and prevent excessive current from damaging components. As shown in Figure 4-2, when M114 is set to 1, the OC gate is on, the voltage at 33 in ACC1 (JMACH1) of PMAC-PC104 is 0, the control terminal of the solid-state relay is cut off, and the output terminal is also cut off. No current flows through the coil YA1 of the solenoid valve, and the solenoid valve does not operate. When M114 is set to 0, the OC gate is off, the voltage at 33 is 1, the control terminal of the solid-state relay is on, and the output terminal is also on. Current flows through the coil YA1 of the solenoid valve, and the solenoid valve operates. When A1 is energized, the input value of port 20 (HOME4 flag) is 0, and the corresponding M420 is assigned a value of 0. When A1 is de-energized, the input value of port 20 (HOME4 flag) is 1, and the corresponding M420 is assigned a value of 1. PMAC can determine the current extreme position of the swing cylinder by querying the state of M420 to obtain the state of the magnetic switches A1 and A2. [align=center] Figure 4-1 Pneumatic Circuit Diagram of Silicon Wafer Transfer Robot[/align] [align=center] Figure 4-2 Pneumatic Circuit Control Diagram of Silicon Wafer Transfer Robot[/align] 4.4 Hardware Debugging of Silicon Wafer Transfer Robot Controller Before the silicon wafer transfer robot performs its first movement, the hardware of the control system must be debugged. Using a control platform composed of PMAC and a servo drive system, the stability, speed, and accuracy of the system are adjusted by setting the parameters in the PMAC and servo systems. The hardware debugging of the silicon wafer transfer robot system mainly includes: setting the parameters of the PMAC board, adjusting the PID parameters of the motor servo system, and adjusting the feedforward parameters. 4.4.1 Setting the Parameters of the PMAC Board The relevant parameters of the PMAC board must be set in advance to work under a given system (motor, encoder). The setting process can be done using online commands. The following uses motor #1 as an example to introduce several important I variables that must be set. I100: Motor enable parameter. I100=0, the motor is not enabled and will not run; I100=1, the motor is enabled. I102: Instruction output address. Inform PMAC2 of the command output location for motor 1. Using PFM, the output must be written to the C instruction register of the correct axis interface circuit. For motor 1, I102 = $C004. I116: Maximum permissible programmable speed, which can be adjusted by % as the speed limit. I117: Maximum permissible programmable acceleration, which can be adjusted by % as the acceleration limit. I119: Maximum permissible JOG acceleration. Can be replaced by TA (I120) and TS (I) and TM(). When using the I119 variable, I120 and I121 are always 0. I122: Manual maximum speed. I125: Flag and mode variables, determining where PMAC knows to find its limit and homing flag inputs. If limit switches are connected to +LIM1 or -LIM1, or if the +LIM1 or -LIM1 limit pins are grounded, I125 must be set to 49125 ($C000); if limit switches are not used and the limit pins are not grounded, I125 should be set to $2C000. Incorrect settings will prevent the motor from operating. I129: DAC fine-tuning. This parameter adjusts zero drift and can be manually increased or decreased for optimal results. Automatic adjustment is also possible. I169: Output command DAC limit. This parameter defines the maximum output value from the control loop. If the calculated value is larger than this limit, the output will be limited by it. If this limit is violated for a period of time, the servo error will begin to increase, and the integral circuit in the PID loop will shut down due to overload protection. Therefore, this parameter value should be carefully selected. I910: Encoder and timer decoding control; I910=7 for quadruple frequency control. I916: Output mode selection. 4.4.2 PID Parameter Adjustment for the Motor Servo System In the PMAC board, the proportional gain variable is I130, providing system rigidity. The integral gain variable is I133, used to eliminate steady-state error. The derivative gain variable is I131, used to provide system damping to ensure system stability. Several other servo control I variables can reduce the trajectory error of the servo system: motor speed feedforward gain I132, which can reduce the trajectory error caused by I131, increase system damping, and improve system dynamic performance; motor acceleration feedforward gain I135, which can reduce tracking error caused by inertial hysteresis; and motor friction feedforward gain I168, mainly used to help overcome errors caused by friction. These parameters also play an important role in the PID parameter adjustment process. The PID parameter settings of the PMAC2-PC104 multi-axis controller are executed by the PMACTUNINGPRO software. The PMACTUNINGPRO program has a good debugging tool. It can collect data and draw curves such as DAC-TIME, POSITION-TIME, ACCELARATION-TIME and FOLLOW ERROR-TIME as needed. These curves are used for analysis and adjustment. The following is an example of the PID debugging process of motor #3. The specific adjustment method is based on the above principles. The steps are as follows: (1) Determine the proportional gain P; When determining the proportional gain P, first remove the integral term and derivative term of the PID. Generally, Ti=0 and Td=0 are set so that the PID is a pure proportional control. Set the input to 60%-70% of the maximum value allowed by the system. Gradually increase the proportional gain P from 0 until the system oscillates. Then, reverse the process and gradually decrease the proportional gain P from this point until the system oscillation disappears. Record the proportional gain P at this point and set the proportional gain P of the PID to 60%-70% of the current value. The proportional gain P debugging is completed. (2) Determine the integral time constant Ti; after the proportional gain P is determined, set a large initial value for the integral time constant Ti, and then gradually decrease Ti until the system oscillates. Then, reverse the process and gradually increase Ti until the system oscillation disappears. Record Ti at this time, and set the integral time constant Ti of the PID to 150%-180% of the current value. The integral time constant Ti is adjusted. (3) Determine the integral time constant Td; the integral time constant Td generally does not need to be set and can be 0. If it needs to be set, the method is the same as determining P and Ti, and 30% of the value when there is no oscillation is taken. After adjustment, the step response curve of motor #3 is shown in Figure 4-3. [align=center] Figure 4-3 Step response curve of motor #3 after adjustment[/align] 4.4.3 Adjustment of feedforward parameters The adjustment of the speed feedforward gain Kvff and the acceleration feedforward gain Kaff needs to be carried out after the basic PID parameters are adjusted. The adjustment of the basic parameters can be carried out according to the step response characteristics of the servo axis, while the adjustment of the feedforward control parameters is carried out according to the parabolic response characteristics. Manual adjustment of PID parameters can be performed based on the FOLLOW ERROR-TIME curve. Figure 4-5 shows the adjusted parabolic motion following error curve of the motor. As can be seen from the figure, the error has been greatly reduced, meeting the usage requirements. [align=center] Figure 4-4 Adjusted Parabolic Motion Following Error Curve of Motor #3[/align] 5. Conclusion After debugging, the silicon wafer transfer robot controller has basically achieved control of the silicon wafer transfer robot body, enabling it to perform single-axis single-action and multi-axis linkage movements. It has effectively solved the control problem of the servo motor in the silicon wafer transfer robot drive system and effectively resolved the coupling motion problem of the rotation and lifting mechanisms.