[ | Next | Previous | Up ]

Re: sampling plan for reliability test

From: Ken Poi
Date: 1/5/2005
Time: 9:50:47 PM

Comments

Dear friend, As indicated in my earlier note, I am faced with the same challenge to come out with a sampling plan for the reliability testing of electronics parts. I was hoping to gather a forum on the determination or validation of statistically valid acceptance sampling? Does this make sense, try out single sampling plan as suggested by Dr Wayne Taylor in his website www.variation.com, determine what type of failure modes you want to detect with your reliability test and then use acceptance sampling to determine the reliability sample size from the protection you want to get from it ie from the AQL ( 95% probability of acceptance of the lot with defectives below this level ) and the LTPD ( 10% chance of acceptance of lot with this defectives level or higher) of the OC curve defining your protection and objectives with the inspection ( test in your case ). From here you can choose a plan from the sampling table indexed with AQLs and LTPDs and you can get an acceptance sample size for the protection you so designed into your Acceptance Sampling plan.

Define Protection: What protection does the ORT single sampling plan n=13 and a=0 provide? Since sampling plans make accept/reject decisions, the AQL describes what a sampling plan will accept. Lots at or below 0.4% defective are routinely accepted. Routinely, in this case, means at least 95% of the time. The LTPD describes what the sampling plan rejects. Lots at or above 16.2% defective are routinely rejected. Routinely, in this case, means at least 90% of the time. Lots between the AQL and LTPD are sometimes accepted and sometimes rejected. Thats how I have merged and interpreted the meaning of acceptance sampling and the idea of reliability sampling into one big plan. Care to expand or refute this idea? Regards Ken


Last changed: November 20, 2007