MSA for a part Automated and part Manual Inspection Machine

[Crimpshrine13]

We had been written up for NC during the last 3rd party audit because the MSA I did for our inspection machine was not done in the way MSA book did. Is it a mandatory to check 50 pieces of samples to get the analysis? My machine has 4 - 6 pins each holding the parts during inspection, and if I must do 50 pieces of samples on each pin, it'd be 200 - 300 times of manual loading and unloading of the parts, times the numbers of characteristics checked (he insisted it had to be done per characteristics), so technically I must do this at least between 800 - 1200 times, which is cumbersome and not very realistic. If I did it with 20 pieces of samples, would it be invalid because it is not the way it's done per MSA book? I've been thinking about this over and over for weeks and I still can't figure out what is the best way to do this. Any recommendations is appreciated.

 

[Golfman25]

Re: MSA for Inspection Machine

As I am sure others will ask, what exactly was the NC?

I always thought is was 10 pcs., 3 reps, 3 operators for the gage stuff.

And then 25 pcs/data points for cpk.

 

[Crimpshrine13]

Re: MSA for Inspection Machine

NC = nonconformity

10-pc, 3 reps, 3 operators is for Gage R&R, but for the attribute studies MSA book shows the sample chart of 50-pc study.

 

[Golfman25]

Re: MSA for Inspection Machine

NC = nonconformity

10-pc, 3 reps, 3 operators is for Gage R&R, but for the attribute studies MSA book shows the sample chart of 50-pc study.

I know NC. But what was the specific write up.

Attribute is essentially go/no go. It sounds like that is what you are doing -- it either fits the pin or it doesn't?

As for the sample size, the MSA book only says "enough" -- a sufficient number of samples to cover the expected operating range. (Pg. 140). Your auditor may be all wet. Good luck.

 

[Crimpshrine13]

Re: MSA for Inspection Machine

He wrote us up because the attribute study was not done according to the MSA book. I calculated the margin of error for false positive and positive false, but it didn't have Kappa calculations. There's no way I can do it according to what MSA book shows for this inspection machine because it is a machine and it cannot have appraiser variation, so all I can do is place the sample parts on the pin, check if the machine decides if it is OK or not OK.

Last year, he reviewed the exact same sheet and he had said nothing, but this year he asked me if the MSA sheet I created was per MSA book method and when I said "no," he wrote us up for that.

 

[Golfman25]

Re: MSA for Inspection Machine

He wrote us up because the attribute study was not done according to the MSA book. I calculated the margin of error for false positive and positive false, but it didn't have Kappa calculations. There's no way I can do it according to what MSA book shows for this inspection machine because it is a machine and it cannot have appraiser variation, so all I can do is place the sample parts on the pin, check if the machine decides if it is OK or not OK.

Last year, he reviewed the exact same sheet and he had said nothing, but this year he asked me if the MSA sheet I created was per MSA book method and when I said "no," he wrote us up for that.

Well you're saying 2 things. First is whether you need a 50 pc. sample and then whether you have a Kappa calculation. Either way, what "shall" in the standard was violated? The standard doesn't refer to the MSA book. I know they have been trying to incorporate the AIAG books into the audits, but the books themselves talk about common sense and guidelines, not absolutes. It certainly does make life miserable.

 

[Crimpshrine13]

Re: MSA for Inspection Machine

The problem was that one of our customer's supplier manual says that MSA needs to be done per MSA book. So, the auditor is pointing to it and says that if it cannot be done by the method stated in MSA book, the other methods can be used, but it needs to be approved by the customer.

He also didn't like the fact that the MSA was not done by the characteristics. Our inspection machines checks multiple characteristics at the same time, so my method of doing MSA was running several NG parts with different types of defects, but he didn't like it and he says that it needs to be done by each characteristic because the control plan shows 4 or 5 different characteristics checked by the machine.

He's also picking on me that I had a typographical error that instead of entering false-positive, I had positive-false data and that concerns him. I told him that was data entry error but he wouldn't listen.

 

[Golfman25]

Re: MSA for Inspection Machine

The problem was that one of our customer's supplier manual says that MSA needs to be done per MSA book. So, the auditor is pointing to it and says that if it cannot be done by the method stated in MSA book, the other methods can be used, but it needs to be approved by the customer.

He also didn't like the fact that the MSA was not done by the characteristics. Our inspection machines checks multiple characteristics at the same time, so my method of doing MSA was running several NG parts with different types of defects, but he didn't like it and he says that it needs to be done by each characteristic because the control plan shows 4 or 5 different characteristics checked by the machine.

He's also picking on me that I had a typographical error that instead of entering false-positive, I had positive-false data and that concerns him. I told him that was data entry error but he wouldn't listen.

Yes, I have had that auditor too. Any you probably have had zero issues with the parts over the years. Your happy, customer's happy. Now this clown screws everything up. We had a guy who required us to do an MSA on "feel" for burrs. It was a real joke.

Well you can either grin and bear it and do the MSA "by the book" or approach your customer and get them to sign off on your current methodology. Unfortunately, many customers will just go into default mode and say "follow the book" rather than decide for themselves. Good luck.