# Gage Linearity

Six Sigma – iSixSigma Forums Old Forums General Gage Linearity

Viewing 10 posts - 1 through 10 (of 10 total)
• Author
Posts
• #29319

Rouge BB
Member

I am looking for rules-of-thumb for acceptable %’s of gage linearity. Any input?

0
#74774

Mike Carnell
Participant

Rouge,
Calculate the regression line. Use the slope value. (Y= ax + b > use a ; I think that is how AIAG specifies the formula) Lower is better.
I think the AIAG manual has a section on this.
Good luck.

0
#74776

Rouge BB
Member

Mike, thanks for the input. I understand that lower is better, what I am specifically interested in is how low is excellent, vs how low is good enough. Is there a rule of thumb? Should I relate it to a % of the total tolerance?

0
#74777

Martínez
Participant

This metric is more like the % of process than % of tolerance.
Suggest you use the same percentages as for GR&R – 30% unacceptable.
AIAG MSA doesn’t really give acceptance criteria, but this is what I use, and it makes sense.

0
#74778

Rouge BB
Member

Thanks for the input. My initial thinking was using this critera as “acceptance”. I wanted to raise this question to see if anyone out there has a different way of looking at it.
Thanks again.

0
#74779

Khandekar
Participant

Unfortunately there is no Rule of Thumb for accepting linearity. Also you should not look only at the slope of the regression line. For the regression line to be meaningful, goodness of fit also should be high. Again, no rule to decide how high is good enough. For linearity, we normally take parts throughout the range of the instrument. Which tolerance (or process variation) should you consider?
-Atul.

0
#74787

Mike Carnell
Participant

Rouge,
I try to relate anything that has to do with gages to the tolerance if there is one. The error that I get from the gage always eats away at the workable space I have in the tolerance. The percent tolerance is probably a good guideline but I would look at it as a whole. The worse the process behaves the less room I have for error (All 5 types) in the gage. If the process behaves well I can tolerate more gage error. Ultimately I want them all reduced.
That really wasn’t much better was it?

0
#74796

Withheld
Member

MBB,
I agree with your %’s. To add color to your comments, I use 10% or less for critical characteristics and no greater than 30% for the others.

0
#74799

Martínez
Participant

We also created a modified Minitab gage linearity macro that runs a regression analysis on the bias points to test if the slope is significantly different from zero. If the p-value is greater than 0.05, we don’t worry about linearity – although we will check the plot in case the data are indicative of important non-linear trends, such as a quadratic.
I hear rumors that the newest MSA edition from AIAG may incorporate better techniques, such as this, but haven’t yet seen it.

0
#74925

Eoin
Participant

R-squared value of great than 0.95 is a good rule of thumb over the range of values which you expect. Might want to consider intermediate precision as part of your criteria alos. say a relative standard deviation (RSD) of less than 2%.
Best of luck, Eoin

0
Viewing 10 posts - 1 through 10 (of 10 total)

The forum ‘General’ is closed to new topics and replies.