Quality Scan: Data Redundancy is a Good Thing
Mobile measurement systems are now mainstream tools used in inspection and manufacturing. Portable CMMs such as articulating arms, laser trackers, laser radars, photogrammetric systems, and a variety of scanners are finding their rightful place on the factory floor. Quality inspectors, engineers, and manufacturing personnel are reviewing data output from these systems as part of the everyday product-development lifecycle.
How do you derive quality from a portable 3-D inspection report—not the quality of the part being inspected, but of the data being presented? How do you know the 3-D data are quality data? How do you know if the data are reliable? What evidence satisfies your requirement that 3-D coordinate data be "good 3-D data"?
Redundancy. Therein lies the answer and the outstanding feature of portable metrology equipment.These CMMs have the ability to acquire a plethora of data practically as fast as a single point. Constructing a circle from 100 points shows true circle form when compared to a circle created by three points. Toda'’s rapid data acquisition capabilities bring the days of the minimum data set to an end.
Look at a single point. Why acquire one point when 50 or 100 or 200 can be acquired in a few additional seconds? The data set is averaged to a single point, and the root mean square (rms) or standard deviation is reported. This value gives an indication of the quality or stability of that point. If the rms is high, then the data gatherer may have moved the target off location while data were being acquired, or the point was vibrated by some means (forklift travel, generator vibration, riveting or machining at other end of the tool/part). In either case, the point needs to be acquired again until a low rms is achieved, yielding a quality point.
The same approach holds true for geometric fits. Hundreds of points can be acquired and calculated to determine the feature of interest. The resulting fit can be displayed in a variety of ways:
- Numerically with maximum and minimum values,
- A scatter plot of the data to the average geometric fit, or
- A whisker plot shown on a graphic rendition of the geometric entity and all the data points.
In this manner, any outlying points can easily be identified and evaluated for inclusion-in or removal-from the fit. Data redundancy also allows other geometric quality aspects to be identified, such as hole elongation and planar warping.
Transformations and alignments should also be riddled with redundancy. A seven-parameter least-squares transformation requires a minimum of three points, but provides no information about the quality of their geometric relationship. Scattering or surrounding your measured object or your measurement working volume with numerous reference points for the transformation can show warping, deformations, or systematic shifts in the actual point features. For example, if a detail containing one or more reference points is not properly secured, the results of the transformation will show a systematic shift in that point or points with respect to the configuration of the remaining points. This condition indicates a potential problem with the quality of the data set due to a rigging problem.
Drift checks are often omitted from data reports. These checks are used to determine if the relationship between the measuring instrument and the item measured has changed. Should the value exceed some tolerance, then the measuring instrument needs to be re-referenced back into the working coordinate system prior to any further data acquisition. Periodic readings on a point or point set (again, redundancy) throughout the measurement task permit a quality check on the stability of the setup,which influences the quality of the data set.
Redundancy in data acquisition gives us the means to report statistically generated quality indicators of the acquired data set. In other words, don't be afraid of data, because redundancy yields an indication of the quality of the data set. Gather plenty of data and use it to your advantage. Let data tell on itself. Let data assure you that the information you have gathered or reviewed is a quality data set. Then it's time to move onto the quality of the item measured.
This approach to data acquisition will make quality control, part mating, assembly, and construction of large and small parts easier and more accurate. The ball is in your court, and it's your turn to take a precision shot at your company's data acquisition goals.
This article was first published in the March 2008 edition of Manufacturing Engineering magazine.