GR&R using 10 Parts with different Specification Limits

[shaunap]

A subcontractor uses a shadowgraph optical projector for measuring the width and length of 5 different products. Each part has different spec limits for width and length.
They have run a GR&R study using the 5 products with a measurement of width and length and called this the 10 parts. They have repeated the measurements 3 times with 3 operators.
Results are 0.76% GR&R and 183 ndc.

To me this is incorrect as there is variation within operators and when repeating measurements on each part. The results are too good and totally inaccurate.

Should I get them to repeat the GR&R 5 times, once for each product but just using the same spec limit (eg, width) for each?

 

[Ron Rompen]

Re: GR&R using 10 parts with different spec limits

That DOES sound awfully good for a shadowgraph. Not sure if this is completely correct, but you could use the same data, but with the TIGHTEST tolerance (this is assuming that the dimensions are similar, and of equal 'ease' to measure).

 

[shaunap]

Re: GR&R using 10 parts with different spec limits

Thanks for the quick reply. Spec limits are as follows

1​

2​

3​

4​

5​

6​

7​

8​

9​

10​

PW (SOT143)​

PW (SOD323)​

PW (SOT5L)​

PW (3L DFN)​

PW (8L SOIK)​

PL (SOT143)​

PL (SOD323)​

PL (SOT5L)​

PL (3L DFN)​

PL (8L SOIK)​

2.79 - 2.87mm​

1.60 - 1.90mm​

2.75 - 3.05mm​

6.4 - 6.6mm​

5.27 - 5.42mm​

1.22 - 1.32mm​

1.15 - 1.45mm​

1.45 - 1.75mm​

3.9 - 4.1mm​

5 - 5.35mm​