Quality Scan: Is Your Gage Good Enough?
How do you know that your inspection device/method is giving you reliable results? Well, you can reference the specification sheet that came with the instrument.The unfortunate truth, however, is that this number often has little direct correlation to the results generated during the measurement of your production workpieces.These specifications often reference "perfect" artifacts, with specific test criteria,which rarely relate directly to your inspection tasks.
Most of my career has been involved with CMMs, so I will focus on these machines, although this same philosophy will similarly apply to almost any gaging systems that I can imagine.
When manufacturers build a CMM, it's run off to an International Standard Organization standard, (ISO-10360).This set of very specific tests has been developed to provide a baseline for comparison of various CMMs. This specification sheet is a good starting point for your evaluation. Remember: these are very specific tests that will show a snapshot of your machine performance under very controlled circumstances. Typically, this will give you a sense of the accuracy and ability to repeat the results on master gages using your CMM at a point in time.
Beware of these spec sheets and tests on artifacts. Experience has shown that two different manufacturers can have the same specification on their respective machines, with totally different results when measuring parts.The reason for this outcome is not always obvious even to experienced observers. Influences that effect the measurement results on your parts include, but are not limited to, the environment, fixturing, methodology, sensor selection, machine speeds, and data densities.
Unless you can provide repeatable and reproducible results, you cannot control your manufacturing process. By far the best solution for evaluating a CMM is to use a combination of the ISO-10360 and a gage repeatability and reproducibility (GR&R) test.This will provide the metrologist with the highest degree of confidence in the measurement data, ensuring consistent results for statistical process control of the manufacturing system, resulting in bottom-line savings.
Although this may sound complicated, it's actually quite simple. The tests involve measuring international reference standards (ISO-10360), which will ensure that the gage is capable of accuracy and repeatability as compared to traceable standards. And the GR&R test will show that the gage can measure your parts on this machine, using software, fixtures, speeds, and sensors necessary to inspect your parts. When the tests are finished, you will have traceable documentation of the gage accuracy and a quantitative comparison of your inspection results as compared to your specific manufacturing requirements.
Gage repeatability and reproducibility methods are described in detail by the AIAG (Automotive Industry Action Group). Most off-the-shelf statistical software packages include GR&R analysis capabilities. There are a couple of relatively simple statistical formulas used to analyze the results of measuring a representative sample of production parts a number of times. Nominally, ten samples of a part are measured three times each in random order, typically using three different operators. This test then breaks the data down into repeatability, reproducibility, repeatability and reproducibility, part variation, and total variation. The typical goal is to use no more than 10% of your manufacturing tolerance as gage variation.
The goal of a reduced gage variation is to ensure that you maintain as much of the available manufacturing tolerance as possible. When there is a high GR&R, you effectively reduce your manufacturing tolerance, thus making it more difficult to manufacture your workpieces. The unpleasant result is increased scrap, which can be costly in terms of material, productivity, and (ultimately) reputation. In addition, you are required to replace or rework tooling before it's necessary, because you are being forced to produce parts where the manufacturing tolerance has been artificially reduced by excessive gaging influences. Finally, the time required to produce the workpieces increases due to the above considerations. Insist on knowing your GR&R before you accept the gage.
Please remember that unless you know how gage variability compares to your manufacturing needs, you will not know if you can control your process. This situation creates a recipe for distrust between manufacturing and quality. Avoid it by insisting on a GR&R study for any inspection process that you will use. Look for results you can trust.
This article was first published in the December 2008 edition of Manufacturing Engineering magazine.