NDC (Number of Distinct Categories) - Long Gage R&R Study in MSA 3rd Edition

Hi
I am also facing a similar problem my MSA Rand R % is less then 10% but my NDC not more then 5
We use a 0.01 vernier caliper
I am also attaching my MSA report if any one can help me find whats going wrong
We make Plastic parts from of injection moulding machines

Your GR&R to the tolerance is very low, but your GR&R to the product variance is very high. The variation of product presented to the study was very minimal (the parts were too similar in size). ndc uses total variance, not tolerance. If the product you measure does not represent the variation you expect to see over time, that is it does not show as much variation as you will expect to see, GR&R to %TV and ndc will be way artificially low.

To expand upon Bob's comments, your gage is perfectly adequate for inspection purposes, but does not appear to be adequate for SPC and possibly for capability studies.

The key question that Bob raised was are these parts representative of the expected process variation. If they are representation, then the gage is not adequate for SPC. If they are not representative, then you should repeat the study using representative parts. You can also substitute actual process variation from a capability study in for the study variation.

I noticed that your gage resolution is borderline inadequate for SPC. Correcting this might help a little, but don't expect major improvements. Appraiser C tends to measure smaller dimensions. Could they be applying greater pressure when measuring?

Hi Mr. Bob

I am new to the concept of MSA could u tell me what do you mean by the part representing the expected process variation.

We take the 10 parts from each hour from one shift on random basis is that what is gone wrong or am i missing basic fundamentals

Hi Mr. Miner

I am new to the concept of MSA could u tell me what do you mean by the part representing the expected process variation.

We take the 10 parts from each hour from one shift on random basis is that what is gone wrong or am i missing basic fundamentals

If you are taking 1 part per hour over a 10 hour period, you should have representative samples. The concern would be if you had used 10 consecutive samples as that would not exhibit a full range of process variation.

Hi Miner,

Thanks for your response
In that case is my Vernier calliper not appropriate it's recently calibrated and is a digital 0.01 lc Vernier
Then should I be using a micrometer

If you have access to a micrometer, I would repeat the R&R study using the micrometer to see whether it helps. Poor resolution can affect the results of an R&R study in an inconsistent manner, one time making the results better, another time worse. In your case, it alone may not be enough to get a good R&R for SPC. You will have to address the operator variation also.

Finally, you may get into a situation where your process variation is so small that there are no gages that are economically feasible that can measure the variation with sufficient precision.

Why most of the MSA forms don't calculate NDC for tolerance method? I can't find answer in the MSA manual.

Peters said:

Why most of the MSA forms don't calculate NDC for tolerance method? I can't find answer in the MSA manual.

Click to expand...

It is not in the manual, the only ndc they consider is based on the process variation.

Remember, the calculations believe that the variation you present in your samples represents the variation in the process over time. They rarely ever do, though - creating significant Gage R&R error. I always recommend using the historical process variation whenever possible for a more accurate result (see bottom of page 121 of AIAG MSA 4th edition). In Minitab, they have a check box for historical standard deviation. Newer versions of GagePack also support this approach.

Most customers have not figured out that if you are using your gage to qualify product, rather than control the process, ndc to process variation does not apply. They rubber stamp ndc requirements. It is sad...and wrong.