Dear all,

Good day to you!

I'm not sure if this is the correct place for my question, but if it is, your advice is greatly appreciated! :)

Basically, I am trying to set a spec for a shrink tube process. (The acceptable width of the shrink tube after the heat shrink process). Currently, we are using visual inspection to check that the shrink tube is fully shrunk. However, we are now implementing a Vision Camera for this inspection. This means, we need a "number" to "tell" the Vision System when to accept/reject.

I will be using a statistical formula to calculate a one-sided spec from a pool of data (produce "n" samples and measure the widths). My question is, how do I determine a good sample size? And also how to justify the numbers?

Is AQL suitable for this case? I had some concerns because if I sample the lot produced using AQL, I will not be following the Acc/Rej criteria (assuming Acc 2 Rej 3). I will need to ensure the samples I used for the measurement are all Acceptable (Acc 0 Rej 1).

Please advise. Thank you!

Andrew

## Comments rated to be "almost" Good Answers:

Re: How to determine a good sample size?" by ezatu on 03/01/2017 8:18 AM (score 1)Re: How to Determine a Good Sample Size?" by gutmonarch on 03/01/2017 11:40 AM (score 1)Re: How to Determine a Good Sample Size?" by gutmonarch on 03/01/2017 11:58 AM (score 1)Re: How to Determine a Good Sample Size?" by Del the cat on 03/01/2017 12:24 PM (score 1)Re: How to Determine a Good Sample Size?" by gutmonarch on 03/01/2017 12:30 PM (score 1)