The calculation are based on a binomial distribution, which
assumes that the sample is small compared to the lot size (say the sample is 10% or less of the total lot). If this is not the case, then the sampling plans listed will be
tighter than the alpha & beta that were inputted.
For example, if the spreadsheet says to use Ac=2 for a sample size of 130, but the lot is only 200 total, then you will be much better at accepting good lots and much better at rejecting bad lots than you predicted. If the lot size is 2000, then the predictions should be quite good.
In fact, given the numbers in the example, it would be impossible to reject a good lot for a lot size of 200, since Re=3 and the worst possible good lot would have 1% of 200 = 2 bad pieces. As the lot size increased, the odds of rejecting it would increase toward a maximum of 5%.
The next stage would be to make a new version based on the hypergeometric distribution. This would work better for small lots, where the sample size would likely be a large % of the total lot.
Tim F
P.S. It looks like Statistical Steven was half a step ahead of me in responding about lot size.