The “ Virtual CMM ” a software tool for uncertainty evaluation – practical application in an accredited calibration lab

As a practicable method to evaluate measurement uncertainties for almost any measurement task on CMM PTB has developed the “VIRTUAL CMM” concept (VCMM). This method is based on Monte-Carlo simulations of the error behavior of a real CMM. Meanwhile the VCMM concept has been integrated in the software of two German CMM manufacturers and within the German Calibration Service (DKD) four laboratories are accredited to calibrate workpieces using this technique. FEINMESS as one of the accredited labs is presenting its practical experiences in this paper. Introduction A complete measurement result should consist of the measured value and its uncertainty. The statement of a reliable task specific uncertainty in coordinate metrology is very challenging. This is due to the characteristic of a CMM as a multi-purpose measuring instrument, the large number of uncertainty contributors and their complex kind of propagation. Measurement tasks that include a number of different geometrical features and datums can generally not be described by classical uncertainty budgets. A way to solve this problem is to use Monte-Carlo based simulation tools to calculate task specific uncertainties as described in [3] or [5]. Together with project partners from academia and industry PTB has integrated the simulation method in commercial software: The “Virtual CMM”. Today, accredited calibration labs in Germany use this technique to perform calibrations of prismatic parts as a essential link in the traceability chain down to the shop floor. The role of the Virtual CMM in the traceability chain One of the main tasks of PTB as the National Metrology Institute of Germany is to ensure the comparability and reliability of measurement results by making measurements traceable to the SI units. In the field of 3D metrology for prismatic parts, the traceability can be achieved by a calibration chain as outlined in Fig. 1. Starting from the meter definition and its realization by wavelength standards, it propagates traceability to industrial measurements through a chain set up by 2D-artifacts, calibrated CMMs and calibrated workpieces. The most complex chain link is the CMM, which has to take the step from simple 2D artifacts to complex workpieces, which can incorporate a number of complex features with all their task specific contributors. The Virtual CMM simulation software can generate valid uncertainties for all of the measured or derived features and therefore the process of measurement and simulation can be regarded as a traceable calibration. After calibration, these workpieces can be used for experimental uncertainty determination according to ISO 15530 [1]. In order to supply industry with calibrated workpieces PTB established a calibration service under the aegis of the DKD (German Calibration Service) in cooperation with competent calibration laboratories. These laboratories are now accredited according to ISO 17025 to calibrate arbitrary prismatic parts as reference workpieces, traceable in the strict sense. As a result, industry now can be supplied with calibrated workpieces in order to establish traceability of measuring equipment on the shop floor level. Fig. 1: Calibration chain for CMM Basic principle of the Virtual CMM (VCMM) Basis of the VCMM method is the emulation of the measuring process by statistic simulation. These techniques are well established in many domains of nature and engineering sciences. The model of the measuring process and the uncertainty sources affecting it are, analogous to the classical uncertainty budget, the base of the uncertainty evaluation. However the problem is not solved analytically, but by means of a "virtual experiment". The CMM simulator is now integrated in two German CMM software packages (CALYPSO by Zeiss and QUINDOS by Messtechnik Wetzlar). Fig. 2 illustrates the operation of the VCMM. The real CMM probes the workpiece at specified points and records a set of point coordinates. The evaluation software of the CMM calculates the actual result from the set of coordinates recorded. From the recorded set of points further sets are generated by the VCMM by generating systematic and random deviations which are added to the nominal coordinates. The evaluation software of the CMM calculates results for these additional sets (typically around 100200) of coordinates and obtains a representative sample of potential measurement results. From these samples, a simple statistical routine is calculating an interval that includes approximately 95% of all results for each measurand. This interval is interpreted as the extended measurement uncertainty U (k=2) with a coverage of 95%. The actual measured value and the related uncertainty are stated together in the measurement report of the CMM software. Essential for the performance of the VCMM is the provision of input parameters to describe all relevant contributions to the error behavior of the CMM. Currently the following contributors are considered: CMM / Environment Probing process Workpiece • systematic deviations of the slideways • uncertainties due to calibration • thermal deformation of the slideways • thermal expansion of the scales • drift effects • direction-dependent behavior • uncertainties of the stylus calibration • uncertainties when using a probing system with several styli • thermal expansion of the workpiece • roughness of workpiece surfaces Statistics Real CMM Report CMM software