Repeatability issue in Gage R&R study for inner diameter of a bushing

[mstfbakr]

Since this gage is used solely for inspection, the metric to use is % Tolerance. Ignore %Study Variation, %Process Variation and ndc as they are irrelevant for an inspection-only device.

BevD's comments are right on target. Since the resolution of the gage appears to be adequate, I would start by investigating the possibility of Within-part Variation (WIV).

After you have made a determination of the impact of WIV, update us on your progress and we can advise further.

hi miner,

Please let me know how can i explain to supplier ndc or %Study Variation, %Process Variation and ndc as they are irrelevant.

Why % Tolerance is important for this study.

 

[mstfbakr]

I would add that your repeatability is very poor and so you can make NO conclusions about the reproducibility. although the spreadsheet will calculate a number for you it is essentially meaningless until you get a decent repeatability.

looking at the Range chart operator 3 appears to have less repeatability error than the other two operators.
also the maximum range for both operator 1 and operator 2 is close to your tolerance (,0009)

Since this is a diameter measurement, I'm going to suggest that you investigate the possibility that this is really within piece variation: you may have an oval tube instead of a round one. so the first measurement is of the smaller diameter, then the next one is in a different location around the diameter - a slightly larger dimension...try marking the tube(s) and measuring the same location multiple times...

Hi Bev D,

per to your comment regarding ovality. Do we have to mark 10 pieces with e.g. rubber stamp to ignore roundness problem and measuring parts in each measure at same position.

and how can i explain to supplier that their repatability is poor so making comment is not possible to reproducibility ?

Is type 1 study is not enough to show gage is ok ?

 

[mstfbakr]

What [Bev D] said above. Also, that tolerance seems pretty close and we've not been informed of the material type yet.

hi normnzone material type is 1.6604 ( F1260 ) spec per LN 668 rpund steel bar

 

[Miner]

hi miner,

Please let me know how can i explain to supplier ndc or %Study Variation, %Process Variation and ndc as they are irrelevant.

Why % Tolerance is important for this study.

%Study Variation, %Process Variation and ndc are irrelevant for an inspection-only device because they are only meaningful to assess a gage's suitability for use in SPC.

%Tolerance assesses a gage's suitability for inspection to tolerance.

 

[Bev D]

Hi Bev D,

per to your comment regarding ovality. Do we have to mark 10 pieces with e.g. rubber stamp to ignore roundness problem and measuring parts in each measure at same position.

and how can i explain to supplier that their repatability is poor so making comment is not possible to reproducibility ?

Is type 1 study is not enough to show gage is ok ?

I would mark the parts in two places and measure each of these places twice (or 3 times). this will allow you to see how 'oval' the parts are and determine if your repeatability problem is due to within part variation.

the within part variation may be important when compared to the specification, but if it exists it is confounded with your repeat measures. so you must unconfound them...

as for the reproducibility not be trusted in the presence of large repeatability: the formula for reproducibility is essentially the square root of the total variance - the repeatability variance. If the repeatability is very large the estimate of reproducibility will be very small but probably wrong. In effect it clouds the true reproducibility figure...

 

[bobdoering]

Is everyone measuring the part in the same location? They should. Don't blame within part variation on variation in the gage- it's not the gage's fault. Should not be a part of the gage R&R. That's a process capability question.

At a 9 um tolerance, you should have a gage that can read sub-micron. Also, if the tolerance is that tight, you now need to be using controlled conditions - clean parts, clean gage...darn near white glove measurements. Even finger heat will affect the dimension. That tolerance is asking a lot from your measurement technique. I hope it really needs to be that tight!!

 

[bobdoering]

%Study Variation, %Process Variation and ndc are irrelevant for an inspection-only device because they are only meaningful to assess a gage's suitability for use in SPC.

%Tolerance assesses a gage's suitability for inspection to tolerance.

Write that on the wall!!!!