What Are the Differences?

Comparison between the MSA manual and VDA Volume 5 What Are the Differences? MSA is short for Measurement Systems Analysis. This document was first published by the Automotive Industry Action Group (AIAG) in 1990. A comparison between the original MSA document and the recently published VDA Volume 5 Capability of Measurement Processes shows: The difference is in the details. Over the years, many company guidelines have been created based on the MSA manual whose 4 th edition is available by now. These guidelines are often also referred to as MSA, but their contents differ from the original. Ford, GM, Mercedes, Bosch, etc. created some most relevant company guidelines of the automotive industry and, within these automotive groups, these guidelines are valid worldwide. However, they contain some issues showing major differences from the contents of the MSA manual created by the AIAG. All of the company guidelines mentioned above firstly evaluate the measuring system by means of the and values. In general, these values have to be greater than 1,33. For this purpose, a calibrated reference part with a known nominal quantity value is measured about 20 to 25 times. The resulting standard deviation and the difference between the mean x and the nominal g quantity value of the reference part help to calculate and. In company guidelines, the tolerance of the characteristic to be measured always serves as a reference value. By contrast, the MSA manual does not use any or values and rather uses the variation of the parts inspected as a reference value. The tolerance is also mentioned as possible reference value but it is only one out of four options in addition to the variation of the process and the preliminary process capability. By comparing VDA Volume 5 to the MSA manual, the statistical values of the company guidelines are also considered. VDA Volume 5 is based on the ISO/WD 22514-7 Measurement Process Capability standard that is yet to be published in 2011. It can be assumed that the capability of measurement processes will have to be based on this standard in audits according to DIN EN ISO 9001 in the future, even outside the automotive industry. In the beginning was a definition Even when it comes to terms and definitions, the MSA manual has set standards for lack of international standardization. The first edition of the International vocabulary of metrology (VIM) was already published in 1989, but only the edition released in 1994 defined terms like measuring system, measurement process and measurement uncertainty. Even the Guide to the expression of uncertainty in measurement (GUM) was not published until 1995. Indeed, the MSA manual used some of the terms of the international standardization in its last four editions, but it kept basic terms due to their wide usage, which is quite comprehensible. A typical example is the term measuring system. Figure 1 shows that the term measuring system according to the MSA manual does not comply with the definition of the VIM. A measuring system according to the MSA manual corresponds to a measurement process as defined by the VIM. However, the VIM also contains the term measuring system. As per VIM, a measuring system is a subordinate element of a measurement process combining the typical influence components that

can be mainly affected by the manufacturers of the measuring instrument. These two separate evaluations of the measuring system and the measurement process are also contained in company guidelines by calculating the and the values. If they fall below the required limit of 1,33, it is very likely that the measuring instrument will not be suitable in practice (e.g. in production) when all the existing influence components are included. This differentiation has proved to be most reasonable in practice. Influence Factors MSA 4 Company Guidelines Measuement System Measurement System Measurement Process VDA 5 or ISO/CD 22514-7 Source of Information Resolution / Data Category ndc 5 %RE 5% TOL %RE 5% and RE ure 2 3 Gage Display Uncertainty of a reference *) U 5% TOL UCAL ucal 2 Calibration Certificate Repeatability of a reference to be small Cg 1,33 u s Type-1 study (T1) x - xm Bias t-test Cgk 1,33 ubi Type-1 study (T1) 3 Linearity t-test %LIN 5% TOL u max u T1 on 3 masters Repeatability on an object EV (ANOVA) u EV (ANOVA) Type-2 or type-3 study Reproducibility of an object AV (ANOVA) u AV (ANOVA) Type-2 or type-3 study Uncertainty object Repeat measurements at the TOL a u OBJ or same position 3 3 Temperature *) a ut 3 Stability Quality Control Chart a ustab 3 /Quality Control Chart Other *) *) No precise details or is not considered for %GRR. EVR LIN EVO AV u REST { Bi i} Where a is from: drawings experience estimations trials similar measuring processes long term observations etc. Figure 1: Influence factors Procedure The procedure used in order to evaluate the applied measuring system or measurement process is basically very similar in the MSA manual, the company guidelines and VDA Volume 5. All inspections are conducted under real conditions, measured quantity values are analyzed graphically and numerically, statistical values are calculated and then compared to specified limits. This evaluation decides on the capability of the measuring system or measurement process. The MSA manual and VDA Volume 5 mainly differ in the calculation of statistical values and in the number of observed influence quantities that (might) affect the measurement.

Capability index Limit values Graphical evaluation 2 2 EV + AV %GRR 100% RF MSA 4 VDA 5 or ISO/CD 22514-7 where RF total variation TV, process variation σ, P p, P pk or tolerance TOL used mainly in company guidelines %GRR 10% capable 10 < %GRR < 30 partly capable 30 %GRR not capable GRR Gage variation 2 U TOL MS QMS 100% where TOL tolerance n 2 MS MP å i i 1 U bzw. U 2 u i 1, 2, 3,... u i Standard uncertainty of the i-th influence factor Measurement result y x ± U MP %Q MP 30% capable - U MP x + U MP 2 U TOL MP QMP 100% Remarks: RF TOL L U Incorrect decisions may be caused by measurement values near the specification limits (U or L). TOL L U Measurement result y must lie within the tolerance TOL (s. DIN EN ISO 14253. Figure 2: GRR-value versus Expanded Measurement Uncertainty U mp Evaluation according to the MSA manual The MSA manual assesses capability by comparing the calculated Gage Repeatability & Reproducibility value (%GR&R, Figure 2) to the specified limit. Prior to its calculation, the MSA manual only observes whether the systematic measurement error (or the linearity, if available) is sufficiently small. Moreover, the number of data categories ndc is evaluated. The number must be greater than 5. This requirement is similar in its purpose compared to the company guidelines and VDA Volume 5 where the resolution must be lower than 5 % of the specification (see resolution). If the %GR&R value exceeds the specified limit, the value itself will not indicate why the limit is violated. In order to find the reason for this exceeding, intermediate results must be consulted and, if necessary, further inspections are needed. This is a major disadvantage of the evaluation in accordance with the MSA manual. Evaluation according to VDA Volume 5 VDA Volume 5 evaluates each component of the measurement process affecting the measurement uncertainty separately. The standard uncertainty is calculated for each influence component. This uncertainty provides the basis for the calculation of the expanded measurement uncertainty and the capability ratios of the measuring system and the measurement process. Analogous to the statistical values and used in company guidelines, the measuring system can also be evaluated individually. The single influence components are (Figure 1):

Resolution According to company guidelines and VDA Volume 5, the first step is to evaluate whether the resolution is lower than 5 % of the specification corresponding to the characteristic to be tested. If the resolution does not meet this requirement, no further inspection will be conducted because the measuring instrument would not reflect reality sufficiently. However, even if the 5 % requirement is met, the measuring instrument might always display the same measured quantity value in repeated measurements. In this case, VDA Volume 5 applies the standard uncertainty from the resolution of the measuring instrument. Uncertainty from reference parts The MSA manual hardly contains anything about this problem. Some company guidelines indicate that the uncertainty from the reference part (measurement standard, calibrated working measurement standard, etc.) must be lower than 5 % of the specification. VDA Volume 5 considers this uncertainty explicitly in calculating the expanded measurement uncertainty. This is particularly reasonable for the acceptance of measuring systems because the used working measurement standards often show a high uncertainty. If the measuring points slightly differ in repeated measurements, the variation becomes higher. However, this rise is not caused by the measuring instrument. Without considering the uncertainty of the measurement standard, a higher variation always leads to discussions between the customer and the manufacturer of the measuring instrument. This can be avoided by evaluating the uncertainty of the reference part individually. Repeatability and systematic measurement error In order to evaluate these uncertainty components, repeated measurements are taken on a calibrated reference part with a known nominal dimension. Then the measured quantity values are evaluated. The MSA manual demands a low variation (i.e. standard deviation). However, how low? There is not any specific information available. A t test is conducted regarding the systematic measurement error. If zero lies beyond the calculated confidence interval of 95 %, the systematic measurement error is too high. Company guidelines calculate the and values (see above) from the measured quantity values. If both values exceed 1,33, the measuring system is regarded as capable or the variation and the systematic measurement error are considered adequate. VDA Volume 5 determines the respective standard uncertainties and for both influence components separately. Including the other influence components (resolution and uncertainty from reference part), the expanded measurement uncertainty of the measuring system and the capability ratio are calculated (Figure 2). The capability is then compared to the recommended limit of 15 %. If the ratio meets the requirement, the capability of the measuring system is established. Repeatability on test part and reproducibility of operators In order to evaluate the repeatability and reproducibility, three operators take two repeated measurements on ten test parts that are evenly spread over the entire specification zone (other combinations are possible). The resulting measured quantity values are evaluated. The MSA manual, most of the company guidelines and VDA Volume 5 use the method of ANOVA (Analysis of Variance) for this purpose and determine the variation components EV (equipment variation) and AV (appraiser variation). According to the MSA manual and company guidelines, these components provide the basis for calculating the %GR&R value.

VDA Volume 5 classifies those two components separately as the standard uncertainties and. Form deviation of test part Analogous to the uncertainty from the reference part (measurement standard or working measurement standard), the form deviation may have a major impact on the uncertainty of the measurement process. This evaluation is very important. The 1 st and the 2 nd edition of the MSA manual recommended measuring each test part in three different places of measurement in order to evaluate these influences. However, the 3 rd and 4 th edition do not refer to this issue anymore. There is a huge effort required, of course. In order to avoid this effort, the MSA manual and company guidelines propose to take all the repeated measurements at the same place of measurement. This is often not feasible in practice (typical example: automated measurement processes). VDA Volume 5 evaluates the form deviation based on the standard uncertainty component, which may be determined by means of several different calculation methods. Temperature Neither the MSA manual nor company guidelines deal with the subject of uncertainty from temperature to the full extent. Both, manual and guidelines, assume a constant temperature of the test part, measuring system and the environment. VDA Volume 5 evaluates the influences from temperature by means of the standard uncertainty. The document offers several calculation methods. Stability The evaluation of a measuring system or measurement process at a certain time forms the basis for the decision whether its capability can be established or not. However, the question is if this condition of the measuring system or the measurement process stays the same for the entire period of application. Does the measuring system or the measurement process remain capable or does it change significantly? In order to monitor its condition, repeated measurements shall be taken on a calibrated reference part at regular intervals. The results are documented in a quality control chart. If the new results violate the specified action limits, the capability of the measuring system or measurement process must be re-evaluated. This procedure is recommended in the same way by all documents. Expanded measurement uncertainty and capability ratio Only VDA Volume 5 uses these statistical values. Both, the expanded measurement uncertainty of the measurement process and the capability ratio, are calculated from all standard uncertainty components that were determined before (Figure 2). A measurement process requires a capability ratio lower than 30 %. If the capability ratio meets this requirement, the capability of the measurement process is established. Which system is the best? The advantages of the MSA manual are its high international recognition and the versatile application of the procedures it describes. The detailed observation of the influence components affecting the measurement process and their impacts on the expanded measurement uncertainty argue for VDA Volume 5. In addition, this document is based on an ISO standard, which might lead to a greater recognition of this approach. Time will tell which procedure will become more important in the future.

History of the Evaluation of Measurement Procedures The first edition of the MSA manual was published in 1990. No more than 13 years later, the VDA Volume 5 was released. Almost at the same time, the 4 th edition of the MSA manual and the 2 nd edition of the VDA Volume 5 were published in 2010. The very fact that the MSA manual has been available for more than 20 years and that only now the first international standards about this topic are created shows that the MSA manual has been the ultimate benchmark for decades. Since 1995, with the automotive industry pushing the issue, all suppliers have had to be certified according to the QS-9000 standard (corresponds to ISO/TS 16949 today). This is why the MSA manual gained recognition in the 1990s. The certification requires capability analyses for the applied measuring systems. How to make these evaluations and which limits shall be used, the MSA manual tells. Both, the QS-9000 and the ISO/TS 16949 standard, refer to this manual for further details. In the past, the German automotive groups demanded a certification according to VDA 6.x (corresponds to ISO/TS 16949 today) from their suppliers. This volume says that suitable procedures must be used in order to evaluate the applied measurement procedures. Since the VDA Volume 5 was not published until 2003, company guidelines based on the MSA manual had been consulted about this kind of inspection before. As of 2003, many suppliers had to make capability analyses according to the MSA manual and the VDA Volume 5 because the German automotive groups included VDA Volume 5 in their customer-supplier rating as another applicable document. Sources Author A.I.A.G. Chrysler Corp., Ford Motor Co., General Motors Corp.: Measurement Systems Analysis, Reference Manual, 4th edition, Michigan, USA, 2010 DIN ISO/IEC Guide 99:2007: International vocabulary of metrology (VIM). Beuth Verlag, Berlin, 2010 ISO/WD 22514-7: Capability and performance Part 7: Capability of Measurement Processes. Geneva, 2008 VDA Volume 5 Capability of Measurement Processes. 2 nd edition, VDA, Berlin 2010 ISO/TS 16949:2009-06 Vornorm: Qualitätsmanagementsysteme - Besondere Anforderungen bei Anwendungen von ISO 9001:2008 für die Serien- und Ersatzteil-Produktion in der Automobilindustrie. Beuth Verlag, Berlin, 2009 ISO/IEC Guide 98-3 (2008): Guide to the expression of uncertainty in measurement (GUM:1995). International Organization for Standardization, Geneva, 2008 Neukirch, C.; Dietrich, E.: Measuring System and Measurement Process Are Two Different Things. QZ 4 (2011) 56, S. 16-20 Dietrich, E.: MSA What s new? QZ 1 (2011) 56, S. 39-41 Dr.-Ing. Edgar Dietrich, born in 1951, author of numerous specialist books about statistics and test procedures. CEO of the Q-DAS GmbH, Weinheim, since 1990.