Have you heard of
EnDat yet?

If not click here!

 
 

Gauging Machine Motion-System Accuracy With Optics

Rick Glos
HEIDENHAIN CORPORATION
Schaumburg, IL

Optical scanning techniques measure motion precisely enough to tune feedback in fast servocontrollers.

Motion systems today frequently must deliver accuracies on the order of a micrometer, often while moving at relatively high speeds. Servocontrol and NC machine designers find themselves at the point where traditional test equipment is too large and cumbersome to measure small dimensions or limited in ability to measure complex contours at high speed. These problems crop up in particular for multiaxes motion, as in milling machines and robots that assemble microminiature components.

Equipment commonly used for checking motion accuracy are comparators, interferometers, and double-ball bar systems. One difficulty with ball-bar devices for dynamic measurement of contours is their fixed radius. Because the bars are typically more than 100 mm long, they cannot follow circle tests with smaller radii, and the errors they detect usually trace back to problems with the machine's geometry rather than control circuits. These machine errors are larger than those produced by the control loops and consequently hide them. Measurements taken with the other devices, comparators, or coordinate measuring machines, are taken on completed parts, and do not represent true dynamic measurements. Determining error sources from a completed part for controller tuning requirements, is a questionable procedure at best.

The positioning accuracy of motion-control systems depends on how well controllers can execute linear and nonlinear interpolation programs at high rates. These systems are typically validated by measuring and plotting various motion profiles to see how accurately they follow input commands. Furthermore, because circular motion profiles, in particular, let users easily analyze systems and optimize control parameters, they are the preferred test contour. For higher feed rates and axis accelerations, the additional stresses of masses in motion amplify the effects of geometric error.

New instrumentation has recently been developed to help tune high-speed systems and let machine makers check nonlinear motion accuracy at speeds to 24 m/min. One involves a two-coordinate measuring device consisting of a cross grid and an optical head. The cross grid serves as a 2D optical scale, providing measurements in a plane.

For testing motion profiles, as might be traced out by a milling machine, the cross grid mounts on the machine table and the noncontact optical head sits in the tool spindle, locked against rotation. The cross grid need only be aligned approximately parallel to the table axis during initial setup. For other motion- control configurations, customized fixtures replace the spindle and table to provide the same kind of platform for mounting the head and grid.

An advantage of cross-grid systems is their ability to measure any move within a circular area of 235 mm in diameter, like 2D coordinate measuring machines. Because cross-grid systems can measure these contours with great accuracy at high speed, they are becoming an essential tool for servosystem, machine tool, and numerical controls manufacturers for optimizing motion-control parameters.

For example, say a given servo controller was to move a spindle around a 90° corner. A plot of the spindle movement, using the cross-grid measurement system, reveals the corner area is off by about 10 µm at various points. This kind of error is indicative of a mistuned integrator in the PID control loop. The reason such errors point to the integrator as the culprit is that an integrator is in a lag mode. The time constant it introduces in this controller is too long. Reducing the time constant reduces the error, but going too far runs the risk of creating an unstable control loop.

Similarly, say a circular servocontroller spindle path for a typical electrical discharge machine, measured via the cross-grid, turned out to be somewhat elliptical. Examining plots at different speeds might typically reveal that this kind of error arises from several factors:

  In one case, cyclic errors (oscillations) were visible with a pitch of 6 mm and an amplitude of 4 mm in the X axis. This happened to be the pitch of the ball screw, so the remedy was to reduce whirling vibrations by readjusting the ball screw.
 
  A 6 mm step motion appeared in the second quadrant of the trace. An inspection revealed a flexible hose in the table path. Moving the hose eliminated the collision and the troublesome step.
 
  The Y-axis scale was 2 mm/90 mm 


When checking servo system performance on a large circle, both controller and machine geometry are analyzed and measured. Checks for the controller preferably take place on small circular paths with diameters about 10% of the machine's shortest axis. Acceleration effects such as overshoot are more obvious around small circles. Also, the maximum contouring velocity for making measurements can be as high as the maximum rate for the system's motion controller.

When checking servo system performance, designers need to take special precautions so the accelerations of instrument components as they enter and exit the test paths do not affect circular tests. When testing a circle with a ball-bar device, the measurements takes place along a fixed arc. Because of this, the beginning and ending of the motion may contain some small errors contributed by the accelerations and decelerations. To eliminate the errors, start the test before 0° and end it after 360° .

By contrast, the grid plate encoder can start a circular test at any point. The software typically supplied with the encoder starts at the center of the circle, moves on an incremental straight line in the positive X direction, follows a small arc, then enters the circle tangentially. After the circle has plotted more than 370° , it leaves the arc and returns to the center of the circle. In addition to performing circular tests at low and high feed rates and analyzing acceleration effects, cross-grid systems can check linear interpolation for straight lines at any angle to a machine axis, up to 235 mm long.

Copyright © 2005 Optimized by Cherryoneweb.com Website Optimized by: Cherryoneweb.com

 

© 1999-2004 Automation and Metrology, Inc.  All rights reserved.