Get Answers for CMM Calibration Uncertainty
The year 2000 will be remembered in dimensional metrology history as the year of the accreditation boom. Because of the Jan. 1, 2001, deadline for suppliers to meet the accreditation requirements spelled out in the Third Edition of QS-9000, an accreditation movement spread across the country last year. When QS-9000, Third Edition, was released in 1998, only a handful of calibration laboratories had considered accreditation. By the end of 2000, hundreds of calibration service providers had been, or were in the process of becoming, accredited. In addition, business opportunity hit the accreditation field, and many companies began offering ser-vices to accredit calibration laboratories. In a few short years, the U.S. accreditation business for dimensional calibration went from basically nothing to hundreds of companies across the country.
Proving technical competence sets accreditation apart from registration or certification to quality systems such as QS-9000 and ISO 9000. The biggest challenge in proving technical competence is the correct estimation of measurement uncertainty. Unfortunately, uncertainty is still not generally well understood and few standards exist to help end users. Estimating the uncertainty of dimensional measurements made in temperature-controlled laboratory environments is a challenge, but the difficulty reaches higher levels when measurements are made in the field. The mandate for accredited calibrations applies not only to calibration laboratories, but also to any provider of calibration services, including OEM or third-party calibrations of major instruments at customer sites.
When field service engineers calibrate measurement equipment, whether that equipment is next to a machining center, stamping press, shipping dock or inside a measurement lab, the measurements should come from an accredited source, and the measurement uncertainty should be properly estimated. This applies whether any repairs or adjustments are made or not--any data on a calibration certificate should be accompanied by uncertainty estimates and a proper accreditation number. If not, then the value of the work should be questioned.
Environmental influences are com-monly listed as contributors in an uncertainty analysis for dimensional measurements. Temperature is one of the biggest factors and, for this reason, most dimensional calibration laboratories precisely control temperature. If temperature variation is not controlled or, even worse, unknown, then determining the impact of temperature on any measurement is difficult. This is the real challenge in estimating uncertainty for field-service calibrations--the uncertainty estimation methodology must be able to handle all potential environmental situations at a customer's facility. The uncertainty of calibrations made when the customer has the instrument in a room that is precisely temperature-controlled will differ from an instrument that is located near the bay door of an active shipping dock.
Managing a variety of environmental conditions is a problem that most laboratories do not have to consider in their uncertainty. Calibration laboratories will typically have a fixed uncertainty for each type of measurement. This is normally listed as the "best measurement uncertainty" on the scope of accreditation for the particular measurement. A fixed uncertainty normally works for calibration laboratories because the environment is controlled. The best measurement uncertainty listed on the scope of accreditation is often the actual uncertainty of the calibrations performed. For field service calibration, particularly for dimensional measurements, this is rarely the case. Instead, uncertainty varies from situation to situation, and the best measurement uncertainty only occurs in specific, and sometimes uncommon, situations. This is particularly true for large dimensional measuring instruments, the most common example of which is the coordinate measuring machines (CMMs).
Because of the availability of nationally and internationally recognized performance test standards, CMMs are calibrated and tested using well known and understood techniques. Most CMMs are sold using specifications from one of three available test standards: the international standard, ISO 10360-2:1994; the U.S. standard, ASME B89.4.1-1997; and the German guideline, VDI/VDE 2617. These standards were developed for purposes of commerce, in that they detail specific tests that customers can use to easily compare the accuracy of various machines, and they become part of purchase contracts to ensure clear communication of expectations and deliverables. Though not written as calibration standards, end users have adopted the performance test standards for calibration purposes. They are now the de facto calibration standards for CMMs.
For those new to uncertainty estimation, the mathematics involved is a concern. In general, however, the math is straightforward. The most important challenge is in knowing the measurement process. Uncertainty is all about estimating potential error. The more you know about the sources and magnitudes of errors in the measurement process, the simpler the estimation of uncertainty becomes. The first step in estimating uncertainty is to carefully examine the measurement process. For CMM calibration, the test that is common to all the standards is length measurement using a calibrated artifact or instrument. The one artifact that is common to all the standards is a step gage.
The setup and measurement is simple. Once the proper coordinate system has been established, points are taken on the faces of the steps on the gage. The distances between the points determine the length measurements. The measurement results are then compared to the calibrated values of the step gage.
The first step in estimating the uncertainty of a particular measurement process is to analyze the process and list the various sources of uncertainty. In practice, this is often done as a brainstorming session among those who are most familiar with the measurement process, such as the technicians and engineers who operate and manage the measuring equipment. For a step gage measurement on the CMM, the following questions will typically arise:
- What is the uncertainty of the calibration of the step gage? This should be found on the step gage calibration certificate.
- Did a laboratory accredited by a recognized accreditation body calibrate the step gage? If not, then the calibration values are useless for your accreditation purposes. This can be a major issue, and it is usually checked carefully by accreditation assessors. The more established international accreditation bodies do not recognize many of the new accreditation groups in the United States, as reviews of their processes have not been completed.
- How repeatable is the measure-ment process? A statistical study may be needed, such as a gage repeatability and reproducibility study.
- Does the alignment or workpiece coordinate system vary much? If so, how does it impact the length measurement? In general, this effect is small, but it may be a problem depending on fixturing and choice of datums.
- Is the step gage bending or de-forming? How does the fixturing com-pare to that used when the step gage was calibrated?
- How well do you know the temperature of the step gage or machine scales during measurement? Is the temperature measured directly or is the machine's temperature compensation system used? Either way, what is the uncertainty of the temperature sensors? Who calibrated the sensors, and which organization accredited them? If the temperature isn't measured, then a much bigger problem exists because an estimate of the temperature has to be made. Without a calibrated temperature sensor, this is difficult, if not impossible.
If the CMM is not located in a room that is exactly 20 C (68 F), then how well known is the thermal expansion coefficients of the step gage and the machine scales? These values will always just be estimates, and because of this, additional uncertainty occurs.
The answers to these questions are used to build an uncertainty estimate. Clearly, the answers to the tempera-ture questions vary depending on the environment in which the CMM is located. If the machine is located in a stable, temperature-controlled room, then it is much easier to determine the temperature of the step gage and the machine. If the temperature around the CMM varies quickly, then the uncertainty can increase dramatically. These are reasons that field calibration uncertainty estimation is more difficult than laboratory uncertainty.
Shortened versions of two uncertainty budgets are shown in the table, "Uncertainty Estimates for a One-Meter Length Measurement." The estimates are for the exact same measurement of a one-meter length on a steel step gage, except one is for a more tightly controlled environment. The values shown were chosen as examples; however, the list of uncertainty sources and the general approach has success-fully passed the scrutiny of technical accreditation assessors for CMM calibration of length measurement.
Coordinate measuring machine calibrations must take into account the uncertainty factors caused by environmental conditions. Photo: Mitutoyo America Corp.