Specification and Tolerance in Stability Control Chart

[Priss80]

What specification and tolerance should key in control chart for stability study if i use production part to conduct the study? If i use master block, what specification and tolerance should be key?

 

[Steve Prevette]

My assumption is you are trying to figure out the stability of produced product. So for the control chart itself, you are going to measure the parameter of interest on sequentially produced items. Then you make the control chart and see if the control chart is stable. For declaration of stability, you don't need the specification or tolerance. Now, once you understand the production data (and if it is stable or not) you can then compare that to the specification and tolerance to see if the process is producing parts in a predictable manner that will meet the specification and tolerance routinely.

 

[Priss80]

I am talking about the MSA stability, measure the equipment stability. What specification and tolerance to use if I take production part to conduct? What if I use master block to conduct?

 

[Steve Prevette]

Sorry, but you are beyond my knowledge in this. I suspect you will need to provide details as to what you are trying to measure the stability of, or what you mean by a "master block" to get an answer.

 

[Priss80]

I take a molded part from produce in production and measure a dimension let say length5 times repeatedly and record the results in X bar R control chart. The MSA stability study will be repeated every bi-weekly until I collect total of 10 subgroups. This time I should calculate the control limit, now I need to know the specification of the part, I will put the spec of the length 2.5+/-0.5mm in the chart in order to calculate the control limit. Some people I discover they use master block to study the stability of the measurement system. They put the block lenh dimension lets say 2.0+/-0.002mm in order to calculate the control limit. Please help me to understand which method is correct? Why the method is applied?

 

[Priss80]

I take a molded part from produce in production and measure a dimension let say length5 times repeatedly and record the results in X bar R control chart. The MSA stability study will be repeated every bi-weekly until I collect total of 10 subgroups. This time I should calculate the control limit, now I need to know the specification of the part, I will put the spec of the length 2.5+/-0.5mm in the chart in order to calculate the control limit. Some people I discover they use master block to study the stability of the measurement system. They put the block lenh dimension lets say 2.0+/-0.002mm in order to calculate the control limit. The master block is a standard block some in square or round shape that need calibration.Please help me to understand which method is correct? Why the method is applied?

 

[Priss80]

Refer back to my previous post, I also want to know the specification and tolerance put in the control chart to calcualte the control limit is correct?

 

[Steve Prevette]

Refer back to my previous post, I also want to know the specification and tolerance put in the control chart to calcualte the control limit is correct?

That is incorrect. The control limits on a control chart are calculated SOLELY from the data. Specification and Tolerance are not used in calculating the control limits. The UCL is the average of the data plus three standard deviations of the data, the LCL is the average of the data minus three standard deviations of the data.

 

[Bev D]

Expanding on what Steve said:

a measurement system stability study doesn't use the specifications (or tolerances) in anyway. The study is to assess the measurement system, not the conformance of the parts.

The control limits assess the stability of the measurement system. The limits are calculated using standard Xbar, R (or Xbar, S) formulas. do not use the standard deviation of all of the individual values...


There is some utility in plotting the individual values against the spec limits to understand the amount of measurement instability vs the specification. but you must determine the instabtility using the control chart method....

I can post an example if you want...

 

[Priss80]

Yes, please post example for me. THANK YOU.