Search the Elsmar Cove!
**Search ALL of Elsmar.com** with DuckDuckGo Especially for content not in the forum
Such as files in the Cove "Members" Directory

Gage R&R confusion on a part that has little variation

I

isoqn

#1
I am trying to do a Gage R&R on a part that has little variation, according to the calculations from MSA, the gage I'm using is not adequate. My confusion is that the appraisers measurements are with 0.001 of each trial and each other. So how could the gage not be adequate? :bonk:
 

Ninja

Looking for Reality
Trusted
#2
Re: Gage R&R confusion

If I use a ruler with markings every inch, and use it to measure something to the nearest inch...there will be very little variation from run to run or operator to operator.

If my parts vary by +/- 1mm...I will get a GRR result near zero...not because my measurement is so great, but because the tool is not adequate to measure my variation. My test would show near zero variation even though there is some variation.

Insufficient resolution of the gage can give a "false pass" for GRR, which is one of the reasons why the MSA guidance exists regarding the gages.

You see values within 0.001". How much do your parts really vary? Is your gage bottomed out at 0.001" and can't see the rest of the variation?
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#3
What are your spec limits? Can you post your data? The AIAG R&R method can be very confusing and misleading....
 

David-D

Involved In Discussions
#4
Many R&R techniques compare the variation of the gage to the variation of the process. For highly capale processes, the ratio may be "poor' by the boiler plate metric (<10%, etc) but the practical impact when using the gage for acceptance (actual variance or std dev) is minimal as total variation relative to the tolerance limits is small and the process is extremely capable. You dont give up much even if youghten the tolerance limits. On the other hand, it will cause you issues if you use the gage for process control (SPC) as the gage variation will mask process shifts/changes.

On the other hand, you should also be careful of how you select your samples as they nee to accurately reflect the population. If they dont, then comparing the gage variation (std dev) to outside assessments of the variation (stdor dev) of the process may be a better technique.

David
 
I

isoqn

#5
The specification limit is +/- 0.005", I thought about the comment of the ruler and decided to remeasure the parts (using the same gage) and the record the data out to 0.0005", this worked and the gage passes along with the NDC.
I still baffles me that this gage was not passing when the parts were measuring in the middle of the spec and the appraiser's measurements were so close. To me, the gage R & R is a required tool in order to ensure the gage is capable of measuring a certain feature, and the measurements can be repeated, which i believe the data i orignally had showed that.
 

Miner

Forum Moderator
Staff member
Admin
#6
Having adequate gage resolution is a prerequisite for any MSA such as R&R. Inadequate resolution causes chunky data which can bias the R&R results. This bias can distort the R&R results to look better OR worse that they actually are, even for consecutive studies.
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#7
isoqn - miner is correct. We can provide a more specific answer to your specific situation if you post both sets of results (the one with low resolution and the one with high resolution)
 
Top Bottom