Software and Hardware Automate Cylindrical Coordinate Measurement Machine Control System
- Created: Wednesday, 01 February 2006
The Mechanical Engineering School at the Georgia Institute of Technology needed to reduce the set-up time involved in centering a bearing ring on a rotating spindle in a cylindrical coordinate measurement machine.
The mechanical dimensions of finished bearing rings are measured in the factory using a cylindrical coordinate measurement machine with a manual centering process. The operator places the finished bearing ring on a table mounted to a precision spindle and begins the manual centering process. The operator monitors a digital readout from an LVDT displacement sensor and manually taps the bearing until it is centered and the LVDT reading falls within a preset tolerance. Typically, the manual centering step takes one minute to center a part within the 2.5-micron tolerance window.
At this point, the machine measures the mechanical dimensions of the ring by scanning over the surface of the part with a contact probe. The operator spends 15% of the total measurement cycle time centering the part. Reducing the time required for the centering process also can significantly reduce the cycle time and labor costs of the part measurement process.
By designing an active control system to automate the manual centering process, centering cycle time also can be reduced.
Manual centering is not only used for bearing metrology, but in production processes as well. Therefore, manufacturing engineers could also use an automated centering approach to reduce the cycle time of various production processes for cylindrical parts.
Georgia Tech designed an automated system that consists of a linear slide and a precision spindle. Both motion devices include air bearings to improve precision and smooth motion. The linear slide contains a brushless linear motor and a linear encoder. The spindle uses a brushless motor and a rotary encoder. They used an LVDT displacement sensor as a measurement probe and mounted it to the linear slide. They also mounted a fixed pusher contact to the slide and used it to actuate the bearing ring.
The operation of the system is divided into three separate stages: (1) servo following stage, (2) pushing stage, and (3) modification stage. In the servo following stage, Georgia Tech uses the LVDT measurement probe deviation from null position to command the linear slide velocity. The system captures the probe absolute tip position with respect to rotational spindle position. In addition, the system filters the raw data using a Kalman Filter.
In the pushing stage, experimental data and modeling are used to identify the spindle position where the bearing surface is the greatest distance away from the spindle center. This is used as the target for the bearing position. The system commands the linear slide to the desired target position using a trapezoidal velocity profile.
Finally, in the modification stage, the push stage results are compared to expected values, and adjustments are made before the next servo following and push stages. The overall measures of success include centering time reduction from the current manual centering process and the ability to achieve repeatable centering tolerance.
Georgia Tech currently uses an NI PXI-7350 motion controller from National Instruments (Austin, TX) for velocity control of both the linear and rotary motor. Higher-level control loops were implemented on the PXI controller using NI LabVIEW Real-Time software with parallel LabVIEW timed loops.
The NI LabVIEW Control Design Toolkit was used to design and analyze the higher-level control loops in the system, as well as to design a Kalman Filter for the noisy measurement probe output. The system then models the filtered data with a single-lobe sinewave using least-squares curve fitting.
The NI LabVIEW Simulation Module was used to develop simulations of the various control loops in the system. For example, a motion control loop was modeled using a subsystem for the PID control law used in the motion controller, a transfer function for the motor drive along with a saturation block, and the transfer function representing the motor dynamics. Both position and velocity feedback were included in the model.
Using the LabVIEW control design and simulation tools, Georgia Tech found it easy to design and implement control systems, and experienced ease-of-use when changes or updates to the programs are needed. They can quickly integrate different code components together, including code written in other programming languages. Finally, by having tight integration of the NI software with the hardware, the time required to implement the system can be reduced.