Gauge to Gauge (Gage) Studies

[BrQ]

Re: Gauge to gauge studies

Can anyone tell me why my la's do not match numerically with the bottom or top lines representing them on the graph, and why they are only 2 decimal places when all of my data is 3?

 

[Bev D]

this is simply a rounding/display choice for JMP. you can edit this to display teh LAs at 3 decimal places then the lines will look correct...

 

[BrQ]

Bev,
Called minitab, they had no solution other than to change the limits manually.
I'm a big fan of your postings, keep up the good work!

 

[BrQ]

A shout to the experts here,
Have been asked to help with a tool shop overall R&R on all of their various gauges.
I am going into this task thinking that a minitab driven, Anova study, with %Tol as the qualifier, will be an acceptable R&R.
However, I am torn between the %Tol, or the % study Var being used as the final result.
As this is not a "process control situation" and more of a "does this tool meet it's specification" issue, am I correct in assuming that all that these studies need to encompass on a pass / fail basis would be the %tol in a Gage R&R study?
I know you can lump gages into a "family" situation...i.e. 6" caliper is a 6" caliper...but how much do we have to go into qualifiying this same 6" caliper when it measures one cavity of a certain tool that is 1" in diameter with a +/-.020" tolerance, as opposed to the same tool measuring a 4" cavity with a +/-.020" tolerance?
Does the calibration cover this if it is done effectively to rule out any linearity issues?

Also wanted to know the opinion of the group as to whether or not this would be an acceptable situation for using an X-Bar R method for the R&R. Toughts on why or why not would be appreciated.

As always, your superior guidance and wisdom is much appreciated.


Also,,,how do you add the cool quotes to the bottom of your posts? I have searched the user CP and have not found this..?

 

[Miner]

A shout to the experts here,
Have been asked to help with a tool shop overall R&R on all of their various gauges.
I am going into this task thinking that a minitab driven, Anova study, with %Tol as the qualifier, will be an acceptable R&R.
However, I am torn between the %Tol, or the % study Var being used as the final result.
As this is not a "process control situation" and more of a "does this tool meet it's specification" issue, am I correct in assuming that all that these studies need to encompass on a pass / fail basis would be the %tol in a Gage R&R study?

% Tolerance is the correct metric to use.

I know you can lump gages into a "family" situation...i.e. 6" caliper is a 6" caliper...but how much do we have to go into qualifiying this same 6" caliper when it measures one cavity of a certain tool that is 1" in diameter with a +/-.020" tolerance, as opposed to the same tool measuring a 4" cavity with a +/-.020" tolerance?
Does the calibration cover this if it is done effectively to rule out any linearity issues?

See my blog Creating gauge families for guidance.

Also wanted to know the opinion of the group as to whether or not this would be an acceptable situation for using an X-Bar R method for the R&R. Toughts on why or why not would be appreciated.

If you have software, use the ANOVA method. It provides more data than the Range method. The sole reason for using the Range method is if you must perform the calculations manually.

Also,,,how do you add the cool quotes to the bottom of your posts? I have searched the user CP and have not found this..?

Go to your User CP, then Your Control Panel > Settings & Options > Edit Your Signature

 

[BrQ]

Thanks for your help Miner, always appreciated.
How about a lab that does mold qual. (full layouts of parts, to print values?
I'm thinking %Tol would suffice....would this be correct?

How about if the same lab does capability studies?
I'm thinking the same equip would have to pass % SV.......is this correct?

The question relates to proper application, and I am looking for confirmation of my opinions on this, looking through the AIAG manual, I'm getting a bit of a grey area when looking for defining the line between the 2 and their correct applications other than "Process control" and "product control"
Thanks again for your time,

 

[Miner]

Thanks for your help Miner, always appreciated.
How about a lab that does mold qual. (full layouts of parts, to print values?
I'm thinking %Tol would suffice....would this be correct?

Yes. This is product control.

How about if the same lab does capability studies?
I'm thinking the same equip would have to pass % SV.......is this correct?

Yes.

The question relates to proper application, and I am looking for confirmation of my opinions on this, looking through the AIAG manual, I'm getting a bit of a grey area when looking for defining the line between the 2 and their correct applications other than "Process control" and "product control"
Thanks again for your time,

Any time you must deal with part variation, use %SV. This not only means SPC, but capability studies, designed experiments, hypothesis tests, regression, etc. Basically, any use of statistics applies.

 

[dking2]

Just registered for this very cool forum! You folks deserve a ton of credit for this site . I can't tell you how much I have learned from just reading for the last two days.

I have a slightly unique twist on this two gage/gauge topic. In my semiconductor production line we often collect measurements either across two pieces of automatic equipment or on the same piece of equipment.

I would like to use the production data to determine % P/T and %GRR real time and replace or augment our very costly GRR study. A key point here is that I have more than two tools (many more) but I will never have more than two measurements per unit/piece in production. My data is essentially paired across many different tool-pairs.

My thoughts are to calculate random error (distance perpindicular to unity line in a Youden plot) for each unit. Square it! Then summarize by taking the sqrt of the average random error for each tool-pair. I think this will leave me with the measurement error incorporating any tool offset and repeatability????? Calculation????

With the measurement error I can then summarize the %P/T and %GRR by pair and for the line with little difficulty. I thought I could visualize the paired data on a Youden plot - marker by tool-pair.

Before I started writing JSL I thought it might be good to post to this forum and see if I might have some misinterpretation somewhere... Any thoughts??? Thank you in advance???