Downloaded from https://www.everyspec.com AMCP 70M-1. bar of succeslsful and failed items. This type Multiple-sample (in-sample) plans can test of test is not usually called a life test, be- several samples of items.
Arter every sample, cause the time at which the failures occurred except the last, the plan decides to accept or during the testing period Is not con4idered, reject the lot, or to take another sample, After testling the last sample, the plan decides A variables test is one in which some to accept or reject thf, lot, The analysis characteristic of the test items is measured usually is exact, on a continuous scale, such as amplitude, power output, or life, If the characteristic Sequential-sample (seq'shmple) plans are, is not life, and If each Item is tested for a in effect, m-sample Olans whoso boundaries specified time, the characteristic is measured have been drawn differently.
Most or the at the end of this time. analysts are very approximate and deal with untrunvated plans (sample size -, '). Natur- important, measured thing is the time it takes Ref., 21 ano 22 show exact anialyszs of A lira test is a variables test in which the ally, all real plans are somehow truncated. an item to reach a particular condition, usu- severs, plans and give references to exact ally a "failure", Ihti times at which the fail- anilyses of other plans, urea occur are recorded PFig. 3-2 is a pictorial description of The kind of test which should be used sampling plans.
The illustrations are all for often can be decided by evaluating the ap- attributes plans, but the axes can be relabeled plication of the item and its reliability ob- ror variables plans. The course of the test jective.
If tho reliability objective Is stated can be represented es a path, a series of / in terns of' probability of surviving a fixed points, wherein the cumulative result of test) min•ion, an attributes test ut tihe end of the ing each item is plotted, When the path mission often is used, If the goal is in terms ti.uches or crosses , boundary, the decision of time to failure, a life test usually is appropriate to that boundary is made, Test- Indicated, iug continues until a boundamy is reached, When more than one type of test could be Fig. 3-2(A) is the traditional I-sample plan.
employed' such C'awtors as type of informni- 'The boundary is a rectangle, 'he slash shows tion provided, degree of protection afforded, the dividing line between accept and reject; amount and cost of Inspection, and ease of the point (N,') is on the accept boundary; administration should be considered, Table the point (N, c + I) is on the reject bound- 3-3 summarizts the conventional wisdom on ary. advantages and disadvantages of each type of plan with respect to these considerations. Fig, 3-2(B) Is the traditional 2-sample plai., "nhe point (NI + Nj, c•) is "at. vpt"; the 3-.,1 SINGLE.
OR MULrIPLE.SAMPLE point (NI + N3, cv + 1) is "reject", A tra- rectangles put one on top of the other, 'The The terminology l not standard, and so It accept line is a series of steps; the reject line will be explained here, All samples are pro- is horizontal and even with the top of the sumed to be random from a large lot, last step. Very seldom is r greater than 2 in a widely used plan, and almost never is it PLANS ditional r-sample plan would have r such Single-sample (I-sample) plans test one greater than 3. sample of' items, and then decide to accept j or reject the lot, The analysis usually is Fig. 3-2(C) Ih the traditional sequential- exact, sample plan.
'The shape of the truncation is -i I~~,i •,L