A method to estimate task-specific uncertainty for coordinate measuring machine (CMM) measurements

Coordinate Measuring Machines (CMM) has gained a wide acceptance in the manufacturing industry for the measurement of geometric tolerances. With the tightening of tolerances in the manufactured parts to achieve greater performance and reliability, the uncertainty in a measurement has attained a global focus. What the industry does not have and struggling with, is a direct method to compute an uncertainty statement (or at a minimum an upper-bound estimate) for all measurements made in a CMM. Several National and International standards have provided performance tests using calibrated/un-calibrated artifacts for periodic verification of CMM performance. The flip side of these tests is that they only represent a small subset of the measurements made by the CMM and could very well over-estimate or under-estimate measurement uncertainty. In this study the American B89.4 volumetric performance test and the German VDI/VDE test are compared to determine the effectiveness of these tests to provide an estimate of measurement uncertainty. The result of such comparison has highlighted the need for a method capable providing estimates of the CMM geometric errors to be used in a Virtual CMM for task specific uncertainty computation. As a consequence a modified performance test using a ball-step gage measured at 12 locations in the CMM volume is proposed for estimating CMM parametric errors. An application of the proposed method for estimating measurement uncertainty for circular features is also demonstrated.