MSA - Using a R&R study for a Tape Measure

[vinici]

Hello, my name is Vinicius, and I'm an quality intern.

I'm doing in my company some MSA studies, and one of them is to qualify our tape measure.



I've done the studies folowing the manual and I am getting the results of the image below:



As it shows, the R - Chart is showing some bad results as it do not have any variation between the measurements of the inspectors, but the R&R values are way below the 10 % that is needed.

The main question is what should I do in this case, since the tape measure have a discrimination of 1 milimeter, so it will be very hard to train the company workers to read half of that. Can the study be approved withouth fixing that chart?

Thank you for your attention! I do apologize for any gramatical error, english is not my native language.

 

[Miner]

The R-chart is showing that the resolution of the tape measure may be inadequate. Low resolution can make the results appear to be better or worse than they truly are. How does a 1mm resolution compare to your available tolerance? To your process variation?

 

[vinici]

The R-chart is showing that the resolution of the tape measure may be inadequate. Low resolution can make the results appear to be better or worse than they truly are. How does a 1mm resolution compare to your available tolerance? To your process variation?

Hi Miner.

The most critical tolerance in the procces is of 6 mm. I know that the tape measure do not answer to the "1 to 10 rule", but it would not worth buying a new instrument with a bigger resolution. The ranges vary between 600 and 1500 mm.

 

[Welshwizard]

Hi Vinici,

With a spec as relatively wide as that, the fact that you can read a scale to a mm should justify not having to run a study. If you have to have the ability to effectively split the mm increment then read on.

The Range Chart is the foundation to the validity of your study. With this in mind and thinking about the fidelity of a tape measure, i cant see the summary statistics but do you think that you could discern to within around 0.030 mm with a tape measure? This is what that chart is trying to tell you.

Happily, there are some further rules we can use to help us, for example at least 3 distinct values within the range limits, again, I cant see from your data but I suspect that is not the case. Ultimately, if you don't have the ability to discern increments of measurements the software under-reports variation and this results in suspiciously small estimates of measurement variation in this case.

The answer? Find a way of estimating the variation with a greater degree of resolution if you can and blind the study so that you minimise the temptation to replicate the results.

Hope this helps

 

[Miner]

Hi Miner.
The most critical tolerance in the process is of 6 mm. I know that the tape measure do not answer to the "1 to 10 rule", but it would not worth buying a new instrument with a bigger resolution. The ranges vary between 600 and 1500 mm.

If this tape measure is only used for inspection, you will be fine as long as the product is in the middle 4 mm. However, if the product is within 1 mm of the specification limit, you may accept bad/reject good product. I recommend a risk assessment on those outcomes.

You did not answer the question about process variation. Are you trying to use SPC on these dimensions?

 

[vinici]

Hi Vinici,

With a spec as relatively wide as that, the fact that you can read a scale to a mm should justify not having to run a study. If you have to have the ability to effectively split the mm increment then read on.

The Range Chart is the foundation to the validity of your study. With this in mind and thinking about the fidelity of a tape measure, i cant see the summary statistics but do you think that you could discern to within around 0.030 mm with a tape measure? This is what that chart is trying to tell you.

Happily, there are some further rules we can use to help us, for example at least 3 distinct values within the range limits, again, I cant see from your data but I suspect that is not the case. Ultimately, if you don't have the ability to discern increments of measurements the software under-reports variation and this results in suspiciously small estimates of measurement variation in this case.

The answer? Find a way of estimating the variation with a greater degree of resolution if you can and blind the study so that you minimise the temptation to replicate the results.

Hope this helps

Thank you for your attention, Welshwizard.

You have appointed some points that I did not payed attention to, like the meaning of the result of the R-Chart.

I will try to train the the workers involved in the measurement to read half of the resolution, but I do not think that it will work. Finding another way to measure will not be worthy, as I said previously, so it will be very hard to fix those problems!

Again, I appreciate your efforts to help me!

 

[vinici]

If this tape measure is only used for inspection, you will be fine as long as the product is in the middle 4 mm. However, if the product is within 1 mm of the specification limit, you may accept bad/reject good product. I recommend a risk assessment on those outcomes.

You did not answer the question about process variation. Are you trying to use SPC on these dimensions?

Hello Miner!

I do not have the data about the process variation, I'm sorry. As soon as i get those, I will give you a reply.

I really think that it isn't necessary to do a study about that too, but rules are rules... The main reason to do this post was to discover a way to justify just a calibration of the instruments, that it's already done regularly.

Thank you for your attention!

 

[Miner]

Are you using SPC or is the gage only used for inspection?

 

[vinici]

Gage, only.

 

[Miner]

Gage, only.

I will take this to mean that you are only using it for inspection. Refer to my comment in post #5.

If you are in the middle 4 mm, don't worry about the resolution. If you are within 1 mm of the specification, I recommend trying to interpolate to a half mm.