Hi!
I'm a novice engineer QA and I have some problem to solve.


Lot size = 238 (151 - 280 range according AQL)
Ma = 0.4, Mi = 1.0
Normal sampling plans, GIL=II.

What is my sample size??? (according wich kind of defect I have to concidering it? Ma or Mi?? Because according with Ma, sample size is 32, but according Mi is 50.

What to do if 100 each will reamain at the end of shift or at the model change??? What size of the sample will be???

I hope that you help me.

Bizonio,

I'm not sure there is an official answer for your question.

Statistically speaking, it is probably best to draw separate samples testing for majors vs minors, especially if the two are in some way correlated. From a practical perspective, it would almost certainly work well enough to use the same peices to look for both majors and minors.

If the table suggest two different sample sizes, I would draw enough for the larger sample (50 in your example). From that group, randomly test some for the major defects (32 pieces in your case) and then test all the pieces for minors. What you DON'T want to do is make the sample size the same for both while keeping the original AC & RE numbers, because that will throw off the statistics.

If you end up with an odd-sized lot at the end of a shift or production run, I would just look up in the tables the appropriate sample size for that lot.


Tim F

Thank you for the quick answer!


If you end up with an odd-sized lot at the end of a shift or production run, I would just look up in the tables the appropriate sample size for that lot.
Tim F

I also thought so but I am not sure whether I can act into this way because in the case of change of the lot size the arrows are pointing me to the same size of the sample like in case of the lot about value 238. So in this way for example for lot equal 100 I would have test 50 pieces which means so 50%. The same amount for 70 lot pieces, 60, etc. Must I act this way? Is there some method of the selection of the size sample proportionally to the lot size?

bizonio,

There is no other way within Z1.4 (MIL-STD-105E) to choose sample sizes (unless you want to go to a different inspection level for the small end-of-run lots, which sounds like too much trouble). Especially when the lot sizes are small and the AQL is low, you may end up testing the entire lot.

From a statistics perspective, the size of the lot makes little difference - it is the "acceptable" level of defects and the confidence that mostly determines the sample size. Unless you want to realy go into the statistics of sampling AND then convince your boss and/or your customers that you have a valid solution with a smaller sample, it is probably best to stick with a known standard and just do 100% inspection when that is what the table says.


Tim F

about material function, only need sample checking, but for apperance, if possible, need PRD liner worker take 100% sorting/inspection. it apply to start/end of one shift,or modeld change in line.