# Need Help with Gage R&R Questions

Six Sigma – iSixSigma › Forums › General Forums › Tools & Templates › Need Help with Gage R&R Questions

This topic contains 10 replies, has 4 voices, and was last updated by DavidTran 1 week, 1 day ago.

- AuthorPosts
- January 30, 2019 at 4:50 am #235823

DavidTranParticipant@DavidTran**Include @DavidTran in your post and this person will**

be notified via email.Dear all,

I am a new member of our isixsigma. I would like to have asking some question relating the Gage RR.

The firstly that, i have chance to do comparing the result of Gage RR( Crossed) and Gage RR ( Expanding). However, with same the raw data. But the result wasn’t difference with 2 ways of generate. Can some one help to explaining why? ( The difference result with repeatability, stdev,reproduct,Number of Distinct Categories,..)

The secondary that, I want to get CPK value based on the Gage RR. But i haven’t seen it after run Gage RR. Can someone help to tell me, whether i can get it with thanks.

I have attached raw data for reference with thanks.

0January 30, 2019 at 4:50 am #235824

DavidTranParticipant@DavidTran**Include @DavidTran in your post and this person will**

be notified via email.January 30, 2019 at 8:54 pm #235849

StrayerParticipant@Straydog**Include @Straydog in your post and this person will**

be notified via email.Regarding your second question, I wouldn’t use the GR&R results in calculating CPK unless it shows consistent skew. For instance, if you find that your gage tends to read high or low you might adjust for that. Otherwise GR&R only tells you about confidence in your measurement accuracy and only says something about confidence in your CPK.

1January 30, 2019 at 9:03 pm #235850

DavidTranParticipant@DavidTran**Include @DavidTran in your post and this person will**

be notified via email.@Straydog : Thank for your answered question.

Based on the guidline of minitab. It look like not easy to telling how is the confidence/ acceptable based on the Gage RR result. Could you have any samples of Gage RR analysis data. Please share me with thanks.

0January 30, 2019 at 10:03 pm #235852

Mike CarnellParticipant@Mike-Carnell**Include @Mike-Carnell in your post and this person will**

be notified via email.January 30, 2019 at 10:18 pm #235854

DavidTranParticipant@DavidTran**Include @DavidTran in your post and this person will**

be notified via email.Dear Mr @mike-carnell,

As i understood that, Minitab will get Gage RR based on the UCL & LCL (Control limit level – 6 sigmal). Correct me please if i am wrong.

However, CPK is just based on the USL & LSL ( Customer limit level – Almost wider than control limit as well).

I want to have run Gage RR based on the customer limit and run CPK as same the limit as well?

Can we do that think together?

Really thanks for your great explaining always.

0January 31, 2019 at 5:22 am #235893

Suresh JayaramParticipant@SJayaram**Include @SJayaram in your post and this person will**

be notified via email.Hi David,

There are different percentages calculated within Gage R&R (based on study variation, historical standard deviation, and customer tolerance limits). If you specify the customer tolerance limits (LSL and/or USL), you will get the measurement error reported as a percentage of the customer tolerance. This is 11.47% based on the data you have shared on your Excel worksheet. This percentage is calculated as the measurement error * 100 / customer tolerance.

Cpk does not consider measurement error in its calculations. Though measurement error will impact your capability numbers but the assumption is that you have an acceptable measurement system before you want to check its process capability. Cpk is just a comparison of your process performance (mean, standard deviation) with respect to the customer tolerance requirements (LSL and USL).

I would recommend the following sequence:

a) First perform your measurement systems analysis to ensure you have an acceptable measurement system

b) Next, check the stability of your process

c) Finally, get into process capability calculations to calculate your capability indices.Hope this helps,

Suresh.

0February 1, 2019 at 12:12 pm #235967

Mike CarnellParticipant@Mike-Carnell**Include @Mike-Carnell in your post and this person will**

be notified via email.@sjayaram You said “Cpk does not consider measurement error in its calculations.” I want to make sure we understand what you are saying. Measurement error is part of the standard deviation calculation. That is exactly what the chart I attached it saying – if your Actual Cp is X what you will see is something less capable the worse the measurement error is.

0February 1, 2019 at 12:32 pm #235968

Mike CarnellParticipant@Mike-Carnell**Include @Mike-Carnell in your post and this person will**

be notified via email.@davidtran I am not exactly sure what you are asking in terms of capability so let me explain what I do with the numbers. If I have a tolerance width of 0.100 and a percent tolerance of 10% that means at each spec limit there is a band of +/- 0.010 that is an unknown region. It is inside a band that may or may not be what the measurement says it is. This called guard banding. What it does is it leaves you operating inside a spec width that is not 0.100. It is 0.080 when I take 0.010 off of either end. If I wanted to know my capability with regards to measurement error then I would calculate it based on the LSL + % tol and USL – % tol.

We went through a MSA question the other day and I said the only MSA number I cared about (after number of distinct categories) was percent tolerance and of course there was that cacophony of esoteric nonsense about the other categories. The percent tolerance reduces the area you get to operate in with confidence. Inside the guard band is unknow. You can do multiple measurements to reduce the std. error and improve your confidence but who wants to test something 4 time just to cut the measurement error. If my capability (Cpk) is less than 1.0 now I worry about how much of the variation is coming from the gage. If I am greater than 1.0 I am capable enough that everything except % tolerance is a “who cares.”

In general the stats people and the old timers aren’t going to like this. Just remember that when we started doing the Allied deployment in 1995 for the most part GR&R was relatively unknow except to people who had read Mario Perez Wilson’s book on MPCpS. For the last 24 years people have ranted that it wasn’t applicable to them. Poor GR&R caused the recession of 2008. You need to stay practical with the application of this.

Just my opinion.

0February 10, 2019 at 11:07 pm #236200

DavidTranParticipant@DavidTran**Include @DavidTran in your post and this person will**

be notified via email.@sjayaram : Thanks for your suggestion.

Do you have any example analysis for Gage R & R for reference. I would like to have see your analysis data.

0February 10, 2019 at 11:12 pm #236201

DavidTranParticipant@DavidTran**Include @DavidTran in your post and this person will**

be notified via email.@mike-carnell : When i worked for Nokia. They just introduce me to have analysis Gage R & R based on the CPK level. It more visible for user. Event now, when i working with Minitab. It look have more information. That why, i got some confusing when i was moving to using Minitab function. Could we have some examples or advicing then it can help me ( new member) have more easier to understood and pratice on minitab with thanks.

0 - AuthorPosts

You must be logged in to reply to this topic.