Downloaded from https://www.everyspec.com AO 706-191 where s * number of test stations ( " consumer risk- probability of accept- j a AT* ing a lot with A a X1. Xo A : E Ir, .o and E ,I, A , for various test plans, are given in Table 3-8, T = total-test-time 3-7.2 SEO4AMPL! It is often more convenient to use a normal- ized total-test-time XoT or X, T, rather than Consult pars, 3-S and 3-6 for a general T, itself, discussion of seq-sample tests and references for further reading, Fit. 3-4(B1) shows the The minimum accept time is for rA 0 0, and decision boundary for a typical test. The is (from Eq.
3-29a) equations for the lines are similar to those in par, 3-6 for the binomial parameter XoTnl.t -A/So a aA/bso (3-3U&) (adapted from Ref. 13). Sai ( )n The minimum number or failures to reject is "0 for T 0, and is (from Eq. 3-29b) i ~b, n= X, \- Io;'i *0 The computation of the exact a, P with 4 -(3-27) rPmin " Ahj a 1 /b, (3-31) truncation Is done with special computer b W, 1 - 1 bilI programs. The basic algorithm is simple enough, but the calculations are horrendous for large tests. Ref. 17 shows the exact a, bt ,Јn -fn 7.0 '-2n (XI/X0) 3 for the truncated plins, In practice, the truncation of a plan as calculated from Eqs.
h" a A/b2 (intere.pt) 3-27, 3-28, and 3-29 can be severe without having the actual a, 0 exceed the nominal hiS - a/bi (lntercepL) ones. a -b/bI so "beo/ I The usual difficulties occur with this test s, w b(,1bg slope) plan as with any such plan. If W, and Xo are close together, and a and P are small, the test-time for A near & can be quite long, Ref. 15 discusses some of these difficulties; they r0- thA + &T - hA + SOX (A 0 T) are far from being resolved. i"--ht+ x tT) (3-29n) It is usually convenient to work with ti - h + *T - hA + so X A 0 oT) a I 0. it makes little difference how they - A 5 + u, x ()iT) (3-29b) am chosen if the discrimination ratio is avail- able for adjusting.
After all, we are just pickinC 2 points on the OC curve and they where might as well be convenient ones. Then so and s, can be written in terms of the dis- o l producer ritk; probability of rejectwng criniination ratio. 1The subscripts A and R . Sa lot with A •e, Xo Ion a and h can be dropped. Eqs. 3-27, i , ; ,' " V -,, ;1 ,