Bias Linearity and Stablility vs. Wide Tolerances vs. Process Capability



Looking for assistance with MSA calculations. The current excel form we are using rejects if the gage is too stable. Does any one have a form that they are willing to share? Or any help would be appreciated.

(Must complete all 3 calculations)


Fully vaccinated are you?
Have you tried any of the MSA forms in the pdf files directory?


We need more info about what is being rejected. Is the variation > 30% of tolerance or process variation?
I'm going to make a guess that you're using an average and range evaluation for GR&R. It may include a calculation for Upper Control Limit on range and be set to "reject" a system that has any range values outside the control limit.(See pp 20 & pp 46 of MSA manual re: range chart in control.)
This happens when most repeated measurements of parts are the same value, resulting in most range values being zero.
(Part1-2,2,2 Part2-1,1,1 Part3-1,1,1 etc)
This kind of matches with your statement that the gage is too stable.
What it may actually mean is that the gage doesn't have enough resolution for the parts being measured.
To AGS-Why don't you send me a copy of your Excel sheet with the data so I can see what's going on?
mike b

David Drue Stauffer

Hi guys, The MSA calculations in the book work just fine while you are establishing your process controls. When, however you do your gage studies after your process meets and exceeds 1.33 Cp, your acceptance level of your GR&R will go up. Since capability can be reduced to a portion of the tolerance being used ie 1.33 ~ 75% of the tolerance, as you improve your process and reduce the amount of variability in the process moving to 1.67 or even 2.0 Cp, you are using less and less of the tolerance. This will result in the GR&R value to increase as a larger percentage of the variability now is attributable to measurement error. This tells us that the gage is insufficient to control the process. This is a "false indicator". The next response would be to spend greater dollars for a better system which is actually unnecessary because your process is well under control. This is the point where your procedures need a disclaimer that signifies that futher expenditure on gaging becomes less than economical and the capability index indicates that the system is more than capable to control the process.
Use graphical tools to illustrate this control and include them in your records.

Dave S.

Atul Khandekar

It is true that if the tolerances are wide or as the process capability improves, the R&R deteriorates and the 'Study' shows that the instrument in question is not good enough to measure the part in question and hence not good enough to control the already capable process!

However, this alone, IMHO, should not mean that the instrument should be discarded as inadequate or erroneous.

I think we must look at various ways of interpreting the MSA calculation results. viz.;

1. The 10-30 rule based on study variation.
2. Interpretation on the basis of comparison with total tolerance.
3. Interpretation on the basis of comparison with known process variation.
4. Percentage contributions of Part Variation and instrument error.
5. Signal-to-Noise ratio / Distinct data categories etc.
6. Customer requirements.

Acceptance of the instrument can be based on any combination of these.

Disclaimers and graphics would be one way, but one should also discuss such matters with customers ( should I add, auditors?) to decide acceptability.



Date: Wed, 31 May 2000 13:34:59 -0400

I have been requested by our QS9000 registrar to begin performing Gage Stablility and Linearity Studies. I have performed Gage R&R studies but never linearity or stability studies. Can anyone give me some detailed information on what type of data collection is needed to perform these studies and how to perform these studies? I have software (Gagetrak 4.0) for analyzing the data but it doesn't really specify exactly how and what type of data is needed.

Thank you.

Date: Mon, 5 Jun 2000 07:10:35 -0400

Gage Linearity studies are performed by calibrating a series of points over the entire operating range of the M&TE (measuring and test equipment). You can then determine if the bias error is linear, or not, and whether the M&TE is in calibration across its entire range, or not.

Gage Stability studies involve an overtime comparison of the bias results. The simplest construct is to calibrate the same points each time. Then over time you can see if there are trends, etc. When you have enough points, you can calculate a control chart and determine if your M&TE is in control or not, statistically. Gage stability studies assist in determining calibration intervals, since you can see historically how long the piece of M&TE remained in calibration and predict when it will go out.

Date: Fri, 09 Jun 2000 09:08:32 -0400

So does this mean I can use actual calibration data from gage block measurements I've performed for past calibrations for both linearity and stability? I' just confused on what kind of actual data I need to use.

Date: Sun, 11 Jun 2000 19:17:35 -0400

Yes. Stability requires over time data, so past data is fine. For linearity, you typically use current data, and past data could be used as another stability check.
Top Bottom