Gage R&R Newbie - Everything is Failing!

Hello All,

I am new to Quality Assurance and to Elsmar. Thank you in advance for taking the time to read my thread!

I have only performed a few R&R's and I cant make sense of any of it. I understand that you collect 10 parts over a period of time (for part variation), have 2 or more people test the parts 2 or more times. Once I enter the data my GR&R % is always too high and my NDC is always too low.

Example: we recently performed a GR&R on our CMM. 3 people, 10 parts, 2 times. The GR&R% is at 47.67

Any thoughts or advice would be wonderful!

Re: GR&R Newbie- Everything is Failing!

My first guess is that you don't have enough variation in the parts (as shown by low NDC). Ideally you would want parts spanning the entire range of the tolerance. It's also possible that the measuring equipment doesn't have enough discrimination (needs higher resolution).

Beyond that, it's tough to diagnose with the information you've provided. Can you attach your data?

Re: GR&R Newbie- Everything is Failing!

Thank you Howste!

How can I get more part variation? I've collected parts over a period of time and they seem to measure about the same.

I've attached the CMM report. Not sure if this is what your looking for.

Sorry for the delete

Re: GR&R Newbie- Everything is Failing!

Hello!
It looks like you have quite different readings between operators. Does the part get checked in a fixture. I would make sure everyone loads the part the exact same way(according to instructions), look for good net contact, etc.
We try and run a gage R (1x10) study initially to work out any load problems.
DH

Re: GR&R Newbie- Everything is Failing!

The part gets held onto the surface plate of the CMM by a screw and a clamp.

How can i get more part variation?

Re: GR&R Newbie- Everything is Failing!

You can make parts deliberately near or outside the spec limits to increase the NDC. I think that low NDC is the primary reason for the low R&R value. Is the CMM capable of better resolution than .001? I think that rounding may force a low NDC also?

What is the characteristic being measured?

It also seems that something funny is going on with part number 9. It has more variation than any other part. Is there something on this part that would make it difficult to measure consistently?

Re: GR&R Newbie- Everything is Failing!

GR&R Newbie Welcome,

You have come to the right place, plenty of good smart folks here. I may not have a lot more MSA experience than you, but here is a great GRR Cove Blog to help:
http://elsmar.com/Forums/blog.php?b=130

A key point to remember is that if your TV% or total varaition is small compared to the Tolerance/Spec. Width you may have a problem, not enough part varaition may mean that the gage cannot tell the difference between parts. Make sure the parts you are using are representative of the process, a broad spectrum of variation.

I'm sure there are more experienced posters who can amplify and add..

Regards,
Marty

Re: GR&R Newbie- Everything is Failing!

Howste- Yes the cmm is capable of reading better than .001 the characterist is a circle (4pts) and no i dont believe there is anything unusual with part 9.

Do we measure the same characteristic on all 10 parts?

Optomist1- Thank you so much for the blog!

Re: GR&R Newbie- Everything is Failing!

Measuring different characteristics may introduce other sources of variation. See what happens if you increase the resolution reported to .0001"

How is the hole formed? Is it drilled? Stamped? Do you have the ability to change the dimension?

Re: GR&R Newbie- Everything is Failing!

Hi Breitman,

In case you need it, attached is the following:
1) an excel based GR&R worksheet, I filled in the data from your attachment....although you mentioned a 3 x 3 x 10 MSA I only found 2 operators, 3 trials for 10 parts each
2) my SSBB MSA/GR&R course handout, very good for interpreting the results.

From my brief glance it appears as though the range of your measurements is not in statistical control...a word of caution, as this may indicate the presence of an outlier, the data may require further analysis before analyzing the other parameters. There may be a good expalantion for the an outlier if one is present....this may be what howste is referring to

assuming there is an explanation and you proceed, then you can analyze this from a %Total Variation or from a %Tolerance perspective,,,a good explanation starting on about page 15 in the attached pdf....hope this helps. Use of the course GR&R pdf has been granted.

Marty