Master Validation Frequency in production - Absence of leak, thread presence, etc.

[CanuckTester]

Good Afternoon,

I work for a Tier 1 automotive supplier in South Western Ontario. We have end-of-line test/assembly machines to verify a few of the features on the parts(absence of leak, thread presence, backpressure). The settings for these testers are validated once per shift using a 'mastering process'. The master process consists of running several parts with conforming and nonconforming features, and validating whether or not the testers detected all of the features. If the master parts do not pass/fail as required, then we initiate containment on the parts produced prior to the 'mastering procedure'.

My question has to do with the required or recommended frequency of the mastering process. How do I compute the required frequency? I realize that I have not provided data on the actual capability of our process, but I am not so interested in a specific number as I am in the process of arriving at the number. I would really appreciate any suggestions that you may have!

[Miner]

This is normally more of a risk assessment than a true calculation. The interval of mastering should be based on how much material would have to be quarantined in the case of a failure as well as any further processing or possible shipment. A typical approach in calibration is to establish a frequency, monitor the results for out-of-cal occurrences and reduce the interval as appropriate.

If the test provides variable results when mastering, and slowly drifts out of range, you could collect data over time, then perform a regression of results versus time to calculate a time interval to recheck.

[CanuckTester]

Thank you, Miner!

That sounds like a reasonable approach! I think that I will record the number of times that the masters pass/fail on each machine at our current master frequency, and determine the occurence of parts shipped defective for each test (if any) as well as the severity of each defect.

[Statistical Steven]

Originally Posted by CanuckTester

Good Afternoon,

I work for a Tier 1 automotive supplier in South Western Ontario. We have end-of-line test/assembly machines to verify a few of the features on the parts(absence of leak, thread presence, backpressure). The settings for these testers are validated once per shift using a 'mastering process'. The master process consists of running several parts with conforming and nonconforming features, and validating whether or not the testers detected all of the features. If the master parts do not pass/fail as required, then we initiate containment on the parts produced prior to the 'mastering procedure'.

My question has to do with the required or recommended frequency of the mastering process. How do I compute the required frequency? I realize that I have not provided data on the actual capability of our process, but I am not so interested in a specific number as I am in the process of arriving at the number. I would really appreciate any suggestions that you may have!

I hope you are not just adjusting the process every shift. That will lead to overcontrol, and an increase in variability. I would suggest you "sample" once per shift but only make adjustments to the process when a measurement is out of control. Just my

[CanuckTester]

I have found the answer to this question in this forum! I reviewed some of the threads, and found the thread referring to the NCSLI RP-1 Method A3 test for determining calibration intervals. Our mastering process is really a type of calibration of our test equipment, and so we can establish an interval statistically by collecting data on the frequency of pass/fail for each check, and then settig our confidence and reliability targets. This can be done using the Freeware on the IGS website. Essentially, you pick a calibration interval, record pass/fail counts, and use that data to verify your interval.

I have attached the relevant link to the calibration interval discussion.

http://elsmar.com/Forums/showthread....ight=method+a3

Again, a few other things will play into the interval such as containment cost and environmental effects on the test system.

www.igsmax.com