A testing service that assesses the performance of form-fitting software used in coordinate measuring machines and related equipment—an often unrecognized and potentially large source of inaccuracy in manufacturing—is now being offered by the National Institute of Standards and Technology.
NIST's new Algorithm Testing and Evaluation Program for Coordinate Measuring Systems, known as ATEP-CMS, identifies data-analysis errors and estimates their contribution to overall measurement uncertainty. Studies, including NIST-conducted, "round robin" comparisons of measurements performed by U.S. companies, found that software-caused errors can be the largest source of inaccuracy.
Such errors can lead to costly mistakes, such as rejecting parts that are within dimensional tolerances or allowing unacceptable parts to pass inspection.
Coordinate measuring machines, vision-based inspection tools, theodolites (an instrument for measuring angles) and photogrammetric systems are among the types of measurement and inspection equipment using analytical software to convert a series of three-dimensional coordinates, or point measurements along the surface of a part, into geometric shapes. The resultant software-generated curves and surfaces are the basis for determining whether part features are within dimensional tolerances.
Industrial use of automated, three-dimensional measuring systems is growing. Coordinate measuring machines are the inspection tool of choice in many manufacturing industries. Some 100,000 are installed in U.S. factories and laboratories, and more than 100 U.S. firms make software for use with these machines.
ATEP-CMS offers two types of testing services. In the first type of test, NIST provides customers with a simulated set of three-dimensional coordinates for analysis with their own software. Results are compared with reference results obtained with NIST's own algorithms. In the second type of test, customers provide data sets and corresponding fit results to NIST. The fit results are compared with those calculated with the NIST algorithms. For both types of services, NIST prepares a Report of Special Test, which explains the results of the comparison and estimates the uncertainty attributable to the customer's data analysis software. The assessment enables customers to quantify measurement uncertainties, check the correctness of programming codes and the suitability of sampling strategies, and identify extreme situations especially prone to measurement errors.
CMS software testing is currently performed by two overseas laboratories, but service is limited to "failing" or "passing" the software under evaluation. "Only ATEP-CMS provides quantified estimates of measurement uncertainty attributable to software-related errors," according to Cathleen Diaz, the computer scientist who directs the new NIST service.
The service now supports least-squares algorithms and seven types of geometry: circle, line, plane, sphere, cylinder, cone and torus. Plans call for expanding the service to include other algorithms and geometries, including sculpted surfaces.
The foundation of the U.S. service is a NIST-developed tool, the Algorithm Testing System, which was field-tested by more than 50 companies. The tool, wrote the product engineering manager for a U.S. maker of industrial controls, "appears to be 'just what the doctor ordered.'"
Fees for the new service are dictated by the type of analysis requested and are set in consultation with prospective customers. For more information, contact Cathleen Diaz, A127 Metrology Building, NIST, Gaithersburg, Md. 20899-0001, (301) 975-2889, fax: (301) 258-9749, e-mail: diaz [at] cme.nist.gov (via Internet).
As a non-regulatory agency of the Commerce Department's Technology Administration, NIST promotes economic growth by working with industry to develop and apply technology, measurements and standards.