Position error reduction of tool center point in multi-tasking machine tools through compensating influence of geometric deviations identified by ball bar measurements

Abstract

A method to compensate the influence of geometric deviations on tool center point (TCP) for a multi-tasking machine tool is proposed in this paper. Some methods to compensate geometric deviations of a rotary axis in five-axis machining centers have been proposed. However, due to the special topological structure of multi-tasking machine tools, the identification and compensation methods for geometric deviations are different from those of the five-axis machining centers, which have been seldom researched until now. In this paper, the main attention is paid to analyze the eccentricities of the trajectories measured by a ball bar under simultaneous three-axis motions and to reduce the influence of the identified geometric deviations on the position error of TCP by the compensation method. It is divided into two sequential subtasks. At first, the geometric deviations are identified by using the eccentricities of measured trajectories. A simple and practical measuring procedure is proposed to identify geometric deviations of rotary axes existing in a multi-tasking machine tool. For the second step, a method is proposed by modifying the original NC code according to the kinematic chain model of the targeted machine tool to compensate the influence of the existing geometric deviations on TCP. An experiment is conducted on a multi-tasking machine tool with a swivel tool spindle head in the horizontal position. The repeatability of the measured eccentricities based on three experimental results is also investigated to reduce the influence of measuring error on the identified results. As a result, the corresponding values of geometric deviations after the compensation are less than 2.2 arcseconds or 2.4 μm. It is concluded that the influence of geometric deviations on TCP is compensated effectively, and the position error of TCP is reduced significantly.