The only way to validate measuring equipment
When replacing old measuring equipment, it is common practice to validate that the old and new devices are measuring the same data and providing the same quality control (QC) results. To achieve this, correlation tests are carried out.
To make the job easier and quicker, it's tempting to test a regularly manufactured part. After all, its specifications are well known. However, this choice of part may lead to an erroneous diagnosis and conclusion, depending on the accuracy of the new measuring equipment.
That's why the best way to check that a measurement tool works within its specifications is to use a calibration artifact, for which the data has been previously validated and is traceable.
The use of a common artifact for both the old and the new device minimizes the variables that can influence correlation tests. These variables, which can induce differences in measurements, include extraction methods, which differ from one technology to another; alignment methods, which are rarely the same; software, which does not process or calculate data in the same way; installations, which generally differ from one technology to another; and the environment, which, if not exactly the same, can greatly influence measurements.
The use of a calibrated, traceable artifact enables operators to validate that both devices are operating within their specifications. So, if the measurements taken on this calibrated artifact give the right values, it's possible to know that both devices are operating correctly.
Scenario
A manufacturing company in the automotive industry wishes to replace its CMM with a 3D scanner. In order to validate the new equipment, a correlation test is carried out between the two instruments, the old and the new. A difference is then observed when comparing the two measurements: the two instruments are not correlated. Why is this? Shouldn't the same measurements be observed on both instruments? What causes this difference? Since we know that the old instrument has always been accurate, should we conclude that the new instrument suffers from accuracy problems?
When looking for a correlation between two types of equipment (i.e., comparing measurements obtained on the same part by two instruments), there are many variables capable of inducing errors in the measurements. These variables include extraction and alignment methods, software calculations, measurement setup and environment.
Extraction methods
Although the same part is measured, the same points are not extracted from one tool to another. This is why a difference is observed in the measurement, due to imperfections in the part geometry. Obviously, when the surface of a plane is probed by taking a point at each of the four corners, the surface defects of the plane are not considered in this method. Conversely, by digitizing this plane, the entire surface is measured and we obtain a measure of its flatness. Therefore, if the surface has a slight curve, the digitized plane may be misaligned with the probed plane. As a result, a difference in measurement will be observed between the two methods.
Alignment methods
Although the same part is measured, two different alignment methods are used. As a result, a slight difference may be observed in the alignment method, which may result in large deviations at the other end of the part, due to leverage. Even if the same alignment method is used, as mentioned above, a difference in the feature extraction method used during alignment may result in misalignment of the part. Positioning values are based on alignment, which must not differ from one instrument to another, either in the construction method or in the way it is measured.
Software processing
Although the same part is measured, the software used is different and processes the data with different algorithms. The consequence is a difference observed in the calculation of a part characteristic, even when the measured data is the same. The more complex the measurement construction, the greater the potential for discrepancies between calculations.
Measurement setup
Although the same part is measured, the measurement setups for the two instruments are not identical. This can result in different measurements for the same part. For example, many dimensions are measured on a CMM. The marble on which the part is placed has excellent flatness (30 microns). The same part is then measured with a 3D scanning system. But the surface on which the part is placed has a different flatness (800 microns). As a result, the part is slightly deformed and twisted when placed on the second marble. Although it's the same part, the two installations result in different measurements due to the different degrees of flatness of the surfaces on which the parts are placed.
Environment
Although the same part is measured, the conditions are not identical. This can lead to different measurements. Of course, if we measure a one-meter aluminum part on a CMM, at an ambient temperature of 20°C, and measure the same part at 25°C, then the temperature difference will result in a length difference of 115 microns at 25°C.
Common artifact
For quality control, it is essential to minimize these different variables, which can generate correlation errors. The best way is to use a common artifact for both instruments, on which measurements have been previously validated and data are traceable.
Artifacts have the distinctive features of being calibrated and traceable. All characteristics have been previously measured and verified in the laboratory, eliminating any doubts or uncertainties about measurements.
Examples of Artifacts
Conclusion
A value commonly obtained with a traditional measuring instrument is not a reference value on which to rely 100 %. This is because equipment is not an artifact. There is always an uncertainty associated with any measuring instrument. This is why the verification, validation or qualification of a measuring instrument cannot be carried out on a part for which the dimensions have not previously been validated.
The only way to certify that a measuring tool works within its specifications is to compare it with an artifact for which the dimensions are calibrated in a recognized laboratory. Only an artifact can correlate measurements between several pieces of equipment, because only an artifact can eliminate all variables that may interfere with the measurement. Thanks to this artifact, there can be no doubt that the equipment is measuring accurately.
If two instruments obtain the same measurement with an artifact, but do not correlate on a specific part, then the difference is not attributable to the instruments. On the contrary, it means that it is necessary to check and examine the measurement processes, in order to obtain the desired measurements.
By Guillaume Bull, Product Manager at Creaform