Gage Stability, Linearity and Bias

George Kasdorf

unread,

Nov 21, 1997, 3:00:00 AM11/21/97

to

Can anyone help out with what "acceptance" levels are for gage Stability,
Linearity and Bias studies?

For GR&R the rough rule is 1-10% great, 11-20 okay, 21-30 okay with
approval, etc.

We can calculate the above values, but do not know what the results mean.

Any books available that discuss this?

Thanks,
George

Rejb15kc

unread,

Nov 24, 1997, 3:00:00 AM11/24/97

to

>Can anyone help out with what "acceptance" levels are for gage Stability,
Linearity and Bias studies?
>

For Bias we use 1/4 of the smallest increment the gage can measure
For Linearity, we require the slope be less than .05
For Stability, the Automotive industry standard is that the stability, when
plotted on an X-bar & R chart, mus never go out of control, If it does then the
gage is not stable.

For GR&R the rough rule is 1-10% great, 11-20 okay, 21-30 okay with approval,
etc.
>We can calculate the above values, but do not know what the results mean.

If the percentages are calculated as a percent of Tolerance then they mean that
the amount of variation caused by the act of measuring with that gage is the
calculated percent o the tolerance. So if you have a Tolerance of +/- .005"
with a 20 % gage R&R
that means that 20% of your tolerance is already used up by measurement induced
variation leaving only 80% of the tolerance available for all other sources of
variation including parts. In this case 80% of +/- .005 is +/- .004 !

If the percentages are calculated as a percent of total variation, rather than
tolerance, they are the percent of all variation of your process which is due
to measuring. You would need to know that if you want to reduce the process
variation.

D .Jansen

unread,

Nov 27, 1997, 3:00:00 AM11/27/97

to

Part of the QS-9000 is a manual called Measurement System Analysis.
In this book you will find a lot of usefull information regarding this
subject

regards,

Dirk Jansen

George Kasdorf <kas...@compuserve.com> schreef in artikel
<655iui$5...@examiner.concentric.net>...

George Kasdorf

unread,

Nov 27, 1997, 3:00:00 AM11/27/97

to

D .Jansen wrote in message <01bcfb7d$4e0a1a60$9b1b56c3@zwc00241>...

>Part of the QS-9000 is a manual called Measurement System Analysis.
>In this book you will find a lot of usefull information regarding this
>subject
>
>regards,

Yes it certainly does have a lot of useful information. BUT is does not list
what "acceptable" levels are, or if it does I haven't found the correct
page.

So will continue to look for a reference work on this.

George Kasdorf

unread,

Nov 27, 1997, 3:00:00 AM11/27/97

to

Rejb15kc wrote in message <19971124013...@ladder02.news.aol.com>...

>>Can anyone help out with what "acceptance" levels are for gage Stability,
>Linearity and Bias studies?
>>

>For Bias we use 1/4 of the smallest increment the gage can measure
>For Linearity, we require the slope be less than .05
>For Stability, the Automotive industry standard is that the stability, when
>plotted on an X-bar & R chart, mus never go out of control, If it does then
the
>gage is not stable.
>

We compared these to some of our results and everything looked good. So feel
this will work... but I still think I need to have some evidence of where we
derived these values from, so will continue to hunt down a book on the
subject.

Later,
George

Rejb15kc

unread,

Nov 28, 1997, 3:00:00 AM11/28/97

to

>I still think I need to have some evidence of where we
>derived these values from

The Stability requirement I quote is directly from AIAG book"Measurement system
Analysis". The Bias and Linearity requirements are agreements between us and
the Auto manufacturers. AS for Gage R&R the AIAG book specifically says 10% is
acceptable, up to 30% is acceptable under certain conditions.