Abadeer, Wagdi W. (1986, ASQC) IBM Corp., Essex Junction, VT
The parametric assumption in sampling is usually made when the population is known to have a normal distribution. In many engineering applications one is interested in the population tolerance intervals. These intervals are defined by the tolerance factors, K, such that the probability is ~ (confidence interval) that at least a percentage P of the population distribution will be included between ~ - Ks and x + Ks (two - sided case), where x and s are estimates of the mean and standard deviation computed from a sample of size N.This presentation deals with a new approach to the parametric case with important applications in the area of manufacturing of integrated circuits. These circuits are always designed to operate within certain process limits. These limits can be expressed as (figure 1), where ~ and ~ are the mean and standard deviation, and K~ is a multiplicity factor. For a given sample size, K is always greater than K~ and as N ~ ~, K ~ K~. Also in this case, the tolerance limits x ~ Ks will tend to the design limits (figure 2). Both K and K~ can assume integral as well as nonintegral values.