This is video is for electronics assemblers who are required to assess field reliability. It includes determining whether or not sampling can prove that there are zero defects.
Phil Zarrow: Dr. Lasky, engineers and managers that are interested in determining defect rates will certainly find one of your most recent stories from The Adventures of Patty and the Professor very helpful. What happens in the story when Patty is asked to develop a sampling plan?
Dr. Ronald C. Lasky: She was asked by her former boss to develop a zero-defect sampling plan. He wanted a plan in which you could out and achieve, specifically requested a small number of samples, and guarantee that the defect rate was zero, with 95% confidence. This is one of the things that a lot of people don't appreciate. You really can’t' do that with a small sample size.
Patty was really polite, and she went to work and did some math calculations. She developed a solution that would tell him what type of defect rate he could guarantee with any given sample size.
Phil Zarrow: What was the result of her analysis?
Dr. Ronald C. Lasky: Her analysis showed that if you wanted to guarantee a certain defect rate with 95% confidence, and say it would be like 1 in a 100, 1%, if you wanted to guarantee 1 in a 100, you needed 300 samples in the field, and none of them to fail, to show that your defect rate was 1 in a 100, or better.
It's a really good rule of thumb, if you want to guarantee a defect rate of 1 in a 1000, or better, you need 3,000 samples. 1 in a 1 million, or better, you need 3 million samples, all with no fail. The sad news is, if you want to show 0 defects, that would be 3 times infinity. It's just not possible.
Phil Zarrow: Good luck with that. Where could we find this story?
Dr. Ronald C. Lasky: This is one of my most recent blogs, and it's at www.indium.com/blog. For those who are a little math inclined, the derivation that Patty used to get the solution is actually there also.
Phil Zarrow: Dr. Lasky, thank you.
Dr. Ronald C. Lasky: Thank you.
Keywords: indium, Indium Corporation, Dr. Ron Lasky, firstname.lastname@example.org, Phil Zarrow, email@example.com, defect rates, reliability, sampling plan