This project concerns the comparison of three type F2 software measurement standards, in the form of internet-based software, for testing surface texture software, by NIST, NPL and PTB. A set of reference data was examined by the type F2 standards and three commercial software packages. Calculations of the parameters given in ISO-4287 (1996) were compared and the differences between the software packages were evaluated.
Final Report 2009-07-06
In general the results for R-parameters obtained from the three type F2 standards are in good agreement. The exceptions are the RSm and Rc parameters. The reason for this disagreement is the ambiguous and unstable definitions given within ISO standards. The three software measurement standards performed better than the three commercial packages by giving high precision results and their specifications adhere closely to ISO standards. For commercial packages, the results indicate that software is a primary contributor to variability in the results of surface profile measurement. One commercial software package delivered significantly different results. The variation of the results obtained from these software packages is even greater than the variation caused by the surface inhomogeneity, variation of measurement environment and different data collection methods. Therefore, it is not safe to ignore the calibration of software embedded within a surface instrument. Some particular conclusions are as follows: The current specifications of parameters Ra, Rq, Rsk, Rp, Rt and Rz are clearly defined and stable. The three type F2 software standards are qualified to provide accredited results for those parameters for commercial packages. The specifications of parameters RSm and Rc are ambiguous and unstable. The variation of RSm is significant. The revised specification of RSm proposed by Scott is mathematically stable in this test. The specifications of P-parameters are unambiguous in standard documents. However, there are different understandings of the meaning of P-parameters, which leads to different interpretations. The effect of rounding error is insignificant in the test. The major contributor to the variation is the specification variation. In addition, there are significant variations on the results of W-parameter as well.