Search the Elsmar Cove!
**Search ALL of Elsmar.com** with DuckDuckGo Especially for content not in the forum
Such as files in the Cove "Members" Directory

What Do 'They' Want in a Gage R&R? Every gage on every feature on my control plan?

D

Don Winton

#11
With regard to the first post by thomash, part variation should not influence your gage study.
My intention in including the part variation in the ANOVA was to demonstrate the significance (or lack thereof). This is not the same as saying that part variation is a part of the gage R&R study.

Regards,
Don
 
W

wslabey

#12
If you don't use enough parts, part variation will be small and magnify the appraiser/equipment variation. Or if you use parts that don't cover the part tolerance range (i.e., minimal part variation) then the same applies.
 
D

DavidT090616

#13
Re: What Do 'They' Want in a Gage R&R? Every gage on every feature on my control plan

I list part variation as a consideration in the evaluation as a whole. Agreed not a part of classic R&R.

Nothing specific. Just fishing for comments.
Hello Marc,

I too would like feedback regarding the use of multiple part features utilizing the same gage. Is it proper to include different features, each with different nominal values and tolerances into one gage study? Would this not assist in understanding linearity as well? In particular I am looking at a 3D laser scanner.

If anyone could jump into this conversation or provide me with a link to a reputable source I would appreciate it. I have searched for discussions on this topic but am coming up empty. Perhaps wrong search terms?
 

Miner

Forum Moderator
Staff member
Admin
#14
Re: What Do 'They' Want in a Gage R&R? Every gage on every feature on my control plan

Hello Marc,

I too would like feedback regarding the use of multiple part features utilizing the same gage. Is it proper to include different features, each with different nominal values and tolerances into one gage study? Would this not assist in understanding linearity as well? In particular I am looking at a 3D laser scanner.

If anyone could jump into this conversation or provide me with a link to a reputable source I would appreciate it. I have searched for discussions on this topic but am coming up empty. Perhaps wrong search terms?
There is a similar discussion going on in this thread.
 

Ninja

Looking for Reality
Trusted
#15
Re: What Do 'They' Want in a Gage R&R? Every gage on every feature on my control plan

In my experience, it is common to use one gage to measure a number of features.
I have a CMM specifically to measure a broad range of things. I'm not going to buy a CMM for OD and a separate CMM for length...that would be silly.

GRR tests the Test Method...and the test method is different for length and OD
Same gage, two test methods...two GRR's.

The question then becomes whether I need two GRR's for measuring 1-2inches and 5-6 inches...and the famous answer is: Try it.

If I pass GRR for 0.25-0.5", and the same test method passes for 5.5-5.75", darned if I'm going to rerun it for the spans in between...
What I WOULD do is run one with parts spanning the whole range of interest (and a bit outside) and call it a day.

Passing GRR for length and passing for OD does not mean you can trust it for Angles...that's a whole different test method.
 
#16
Re: What Do 'They' Want in a Gage R&R? Every gage on every feature on my control plan

The problem with your question is oft times, people don't stop and assess the situation. A Gage R&R is a statistical tool to describe and categorize variation. Nothing more, nothing less. It is good at it, generally accepted by industry though it has its detractors.

The question on the table is: can you use different features to validate it? The answer, at least to me, is yes, you can.

Let's come up with a scenario that most people on here will at least recognize. Say we are manufacturing ground dowel pins and we want to R&R our 1" micrometers.

So Customer A wants to buy 1/4" dowels and wants a Gage R&R. So we grab 10 1/4 inch dowels and conduct the study and we get a pass. Can we infer that the mics will work on our larger 3/4" dowels that go to customer B? Well, we are more confident in this because we have seen a pass on the 1/4" dowels. But.... maybe a tech was carrying the mics open to about 3/4" and dropped them, damaging the threads in this area. They're looser in and around 3/4".

We would expect that they would be OK as long as we were away from the damaged area. So fine for 1/4" and 1/2", but not so for 3/4".

But remember, our 1/4" only study answered the question "Can this gage successfully repeat when measuring 1/4" dowels?" It did NOT answer the DIFFERENT question of "Can this gage successfully measure dowels throughout its measuring range?"

We could test this as well, and we would do so by introducing more, different sized dowels. We could pick increments of 1/8" and gather dowels throughout the range. (Is a 1/2" dowel different than a 1/4" dowel, or is a 1/2"' dowel a 1/4" dowel that is WAY out of spec? The R&R doesn't know.)

THIS would have a better chance of finding our problem in the 3/4" range. But be careful - we may just gloss over it. And here's why - because the study is statistical, we would have parts all throughout the range. If MOST of them are out of the bad range of our dropped micrometers, it will hide the part in the bad range.

We may have an R&R on 1/4" dowels of 3%. And because there is damage on the 3/4" area, an R&R on JUST 3/4" dowels may be high. 40%, let's say. A study of dowels throughout the range will be MOSTLY like the 3% result with the one 40% result sort of like an outlier. And overall, the study may be say 7%. We'd still pass this if we were looking at the results alone.

So here is the key point: IT IS NEVER GOOD ENOUGH TO JUST CONSIDER THE RESULT OF THE STUDY. If you don't consider ALL the typical output, especially the graphs, and understand what you are trying to study, you will not use the tool effectively.

In my example, looking at the range chart, would show all the reading with a good range with the reading on the 3/4" dowel having a big spike (where the threads on the mic are loose). Now because I have an understanding of the condition of the study, that spike is going to be VERY suspicious to me. And I am going to at the very least want to get more samples in this region to understand the spike. Why would it be so different?

If, however, I was looking at a study of ONLY 3/4" dowels and had 10 points on my R chart with all of them low and ONE spike, that's a different thing. Because I have 9 others that are good, I am likely to chalk up this one outlier to a typo or an operator mis-writing a number. (And, I'm going to look at individual readings on this measurement too. Did ALL the operators see an increase in range here - pointing at the gage. Or did only one of of the 3 show it - pointing at a typo or misread.)

Hopefully this is enough to get you thinking and I don't want to drone on. But it is also important to bring in the customer.

We have two customers. One buying small and one buying large dowels. Now, if you did a good, solid study of parts throughout the micrometers range, that SHOULD be good enough. That doesn't mean it will because many people do not get the concepts. They want to check a box. A micrometer is a standard device. The customer may say "I don't care, do the study" because that is the easiest choice on them. They get to check their box and it doesn't cost them anything.

But you still have to be careful in this. Let's say in your grinding line you put in a special, automatic, inline checker that is 100% checking your 1/4" dowels. You have a good gage R&R on this. Now, you want to use the same machine on your 3/4" dowels, but, you have to swap out the probe. Do you have to do a gage R&R? Maybe. If the 3/4" probe is heavier, maybe that affects things. Maybe it doesn't Maybe you do a gage R&R with your lightest probe (1/8" dowel) and your heaviest probe (1" dowel) and if they are COMPARABLE (that doesn't mean they both pass, that means good stats and good plots) you could very reasonably argue the equipment is OK for everything in between.

So you can see why I dislike the questions that go "What gage R&R is acceptable?" Without understanding the goal of the study, that cannot be answered.

Last point which is VERY important. A Gage R&R in NO WAY tells you if the gage is "right." ONLY that it is repeatable UNDER THE CIRCUMSTANCES OF THE STUDY. Let's go back to our mics. You could spin the mikes open 1/8", press the zero button and then start measuring your 1/4" dowels. Guess what? The readout would be telling you they are 1/8" and it would still pass the Gage R&R because it would REPEAT. (As long as nobody pressed the "zero" button during the study, which is why it actually IS a good idea to periodically zero it during the study....) To know if it is "right" you need to calibrate it - either measure a certified object of known thickness, or check several mics against one another. And now lets talk about "under the circumstances of the study..." You may do your study on dowels throughout the range of the mic and come to the conclusion that a mic is a good gage for dowels within it's measuring range. The conditions of the study are hard steel dowels. You cannot then infer that you could check rubber bands with it. Those are different conditions.
 
Top Bottom