Does Gage R&R study makes sense on a CMM?

[gdntnovice]

So this is what I find out after a gage R&R study. Every single feature was poor (high) for reproducibility and repeatability was very good (low). Obviously this is due to the loading and unloading of the part in the fixture, when the parts is not unloaded then repeatability is good but when is unloaded by a different operator then variation are present and minitab interpret this as “reproducibility”. And a little out of topic here but the variation btw is minimal but since is higher than repeatability and part-to-part variation then it appears to be very high and this is why gage r&r is not useful for me but nonetheless mandatory by customer.

What I did then was to redo the gage R&R with one single operator now loading and unloading after every run and of course the results were now very poor repeatability, meaning that it’s not really the difference between operator but the action of loading and unloading the part in the fixture which will never be perfectly placed in the same previous position.

Would you agree then that I should just run a repeatability? Does this makes more sense to you?

Thanks!

[DAN_M]

A Gage R&R Study will show us the repeatability of your measurement method across various operators. Before you bother to collect & compare that data, you need to be 100% sure that your measurement method is sound.

We should look at your fixture & your alignment structure. Specifically:

-What is the alignment structure of this program relative to the part on the fixture?
--If the fixture just holds the part & you go in and make your own alignment off of the part, I would start looking at the program itself to find out why it isn't repeatable.
--If the fixture itself is measured & used as an origin and you then load the part and then measure, then the fixture isn't "good enough" for what you're trying to do OR your program isn't good enough. Is the fixture is dirty? Maybe the fixture isn't holding the parts in a repeatable enough position relative to your tolerances and you need to re-design/tweak in the fixture to get better results? Maybe you need to re-train the operators? Maybe you need to tweak the program?
--If the fixture's location is being controlled by an external alignment, maybe the fixture has been bumped/moved accidentally & the program that controls & exports the external alignment needs to be re-run in order to make the results better on your current program?

Once the above variables are accounted for/mitigated, then a Gage R&R would make sense but right now any Gage R&R activity you did would be wasting your time.

[ThePudds]

To me, this sounds like the fixture may be the issue. Unless I'm reading this wrong. In my opinion, depending on what is being made, the fixture should take out most if not all operator error. I don't know what kind of parts y'all are making, but at my company, we don't rely on operators to do anything. To unreliable. We try to safeguard our fixtures as much as possible.

Like i said, I may have read this completely wrong.

[dwade]

As DAN_M stated, usually the biggest culprit is your fixture (or clamping/locators) or your program (usually your alignments). Here are just a few things off the top of my head.

1. Do your operators load the part in the same manor?
2. Do your operators clamp the part in the same sequence?
3. Does your program loop the alignment (either 3-2-1 or iterative) or do you run it once and say it is good?
4. If you're using an iterative alignment, are your tolerances too lose?
5. If you're using iterative alignment do you have always measure turned on?
6. Part cleanliness (production residue/burrs)?