MSA Studies on Fixed Size Gauges

TS16949:2009 requires MSA (Measurement Systems Analysis) studies to be completed on measuring systems depicted on the manufacturing control plan. Variable measuring instruments which are used by different appraisers have both appraiser and repeatability variation which can be evaluated using one of the MSA manual methods for calculating Rand R.

But what of attribute studies for fixed size plug gauges? Have you attempted to carry out studies and did you obtain any meaningful results? My own experience of using the attribute study methods defined in the IATF MSA manual which is used to support TS16949 is that the studies rarely produce any meaningful results.

The problem is that in order to measure any variance you really need both good and bad product in the study sample. If samples are taken from normal production and the process is running in control, you are likely to find that all appraisers confirm that all samples are within tolerance. This doesn’t tell you anything relating to the suitability of the gauging system. The IATF manual suggests making special samples with small dimensional variation to the extremes of tolerance, but this assumes that you have the facilities to do so and maybe impractical?

A far simpler and revealing process for determining the risk associate with the use of fixed size plug gauges is as follows:

1) Tolerance study

Most plug gauges are manufactured to Master Gauge Maker’s tolerances which mean that the go plug has a positive tolerance and the not-go a negative tolerance. Confirm this from the gauge calibration record. This makes it impossible to use the gauge and accept a faulty product. The gauge is guard-banded to ensure only parts within tolerance are accepted. Therefore there is no risk to the customer.

2) Review the process yield (ppm defective)

The risk in using the plug gauge is a very small supplier’s risk, in that we could be rejecting good parts. To decide if this maybe a problem worthy of investigation we only need to look at the process rejects. If we are only rejecting set-up scrap or have no significant rejects then the risk is probably acceptable and a study would not be worthwhile.

3) Appraiser variation

If we do decide to carry out a study, since the gauge is a fixed size there is no equipment variation so we only need to consider appraiser variation. We can do this by using 3 different appraisers to re-appraise the rejects using the gauge. If we have any significant appraiser variation then we need to investigate why. Suggestion is that one or more of the appraisers are not using the gauge correctly.

4) Improve the process

If the process rejects are significant, then the first port of call should be a review of the process capability rather than a more in-depth appraisal of the measurement system. The recommendation here is that time and effort is used to improve the process rather than in quantifying the supplier’s risk of rejecting good parts.

5) Measuring the Supplier’s Risk By Calculation

With any gauge made to master gauge makers tolerances there will be a risk that some good parts are rejected which are near to the specification limits. This risk can be assessed by comparing the difference between the actual gauge size (measured at calibration) and the component drawing tolerance, with the process capability. That is using the normal distribution to calculate what percentage of parts will fall in the size range between the gauge size and the drawing tolerance. It can be quickly realised that for any process running in control at say 1.33 Cpk this risk is miniscule.

5) Measuring the Risk by Practical Means

Alternatively we can measure those parts that have been rejected using a more accurate direct measurement instrument and identify those that were rejected but are actually within tolerance and could have been accepted. The problem with this approach is that we are introducing another level of measurement system variation. To be sure of meaningful results we need an instrument with a resolution of one tenth of the range of measurements we are trying to take and a known R+R. If the difference between the gauge size and the drawing tolerance is .0005mm then a measuring instrument with a resolution of .00005mm and a gauge R+R of <10% would give reasonable results. In any case it’s a tall order.

6) In Conclusion:

MSA studies of fixed size gauges provide little value to the organisation.

Most gauges are guard-banded to prevent acceptance of non-conforming product and this can be confirmed by reference to calibration results.

Only appraiser variation is potentially present.

If the process is running in control at acceptable Cpk the supplier’s risk of rejecting good parts is insignificant. This risk of rejecting good parts can be determined but if there are significant numbers of parts reject by the gauging method, the process capability should be improved as this is the most likely cause.

Please note: TS16949 states that adopting other methods to those defined in customer reference manuals should be approved by the customer. (TS16949:2009 clause 7.6.1)

Leave a Reply