Downloaded from https://www.everyspec.com AMOCP 7WIN and In order to meet the test requirements, values of N and c must be chosen so that Producer Risk: a Consumer Risk: P P1(1 0 ) a 1 -a For "lI-hot" items, wheo time is not in- Pa(m,) - J (3-9) volved, the item Is tested for performance without considering test time per as, Because c and N ame integers, it usually is not possible to find a 1-".mple plan that satisfies 3-6.1 1-SAMPLE Eq. 3-9 exactly. Presumably this has been realized before contractual obligations were If the lot size is large in relation to sample incurred.
Since the R 0 , R ,'a, and P are size, the binomial distribution can be used somewhat arbitrary anyway, a reasonable set to generate the OC curve of a l-umple plan. of parameters is chosen which Is close to the Accept or reject decisions are made by testing original plan. N items for time T. Table 3-5 can be used to determine the The lot is accepted if the number of fail- l sample plan (N and c) for various values ures is less than or equal to c, the acceptance of the discrimination ratio 7 (see Eq. 3-5), number. If R is the true a-•eliability for the test mission, the probability of acceptance t IA P.
(a point on the OC curve) is - (N%• ,jix- (37) of'e, 0, and -r, Table 3-5 lists c and a pus'am- b') eter D, The sample size N is and various sets of a and P. For a given set where N = the number of items tested where [x] is the largest Integer ,4 x. Table c - the maximum allowable number of 3-5 is an extension of Table 2C-5 in Ref. 8. tfallures (accept with c, reject with c + I) Example Nos. 32 and 23 Illustrate the proiaiire. R the true a-reliability Table 3-6 presents I-sample plans that R= I - R approximate the OC requirements for corm- mon sets of Re, R I, ', and P.
The s-expected If R is small enough, say R 0.1 and N is sample sizes for "equivalent" sequential plans large enough, say N > 10, the Polion ap- are also shown (Ref. 1). proximation can be used to obtain P, from Tables 3-5 and 3-6 have some differences P(A) E i Poif (vi NAR) (3-8) (usually minor) in N and c for the same s.c hi numinal-TVample test specification, Table 3-5 Is based on the Poison distribution, and where the plans are derived so that a is guaranteed and 3 Is no more than specified. Table 3-6 p' NA is based on the binomial distribution, and the criterion used was to meet both the a i - See par. 2-3.1 for a discussion of this dlstribu- and A requirements as nearly a possible.
If tion, [text con.tinue. on post 3-24f