Department of Defense MIL-HDBK-816 Notice 1 — MIL-HDBK-816 Notice 1 Radiation Hardness Assurance Guidelines Manual PDF

Downloaded from https://www.everyspec.com MIL-HDBK-816 NOTICE 1 APPENDIX B FIGURE 11. Normal probability plot of actual data that exhibited systematic effects. 60.3.1.1.1 Seeking the causes of inhomogeneity in the example. A study of the plot shows that the ICs must have come from two distinct batches (one batch being ICs R1, R3, R5 and the other being R2, R4). In addition, the response of the circuits varies systematically according to which side of the die the circuit was found (one side being circuits C1 and C2 and the other being C3 and C4). Thus, any circuit belongs to one of four groups as illustrated on figure 12.

This figure is the plot of figure 11 without confidence lines but with the points “colored” to show how each group corresponds to a unique set of attributes. Systematic effects are present, this is an inhomogeneous lot and sample size corrections (see 70.3 and DNA documents referenced in 2.2.1) indicate an effective sample size of six which is not in statistical disagreement with the fact that the data separates into four distinct groups. Any treatment of this data should use an effective sample size between four and six and not estimate probabilities greater than about 84 percent.

If the system survivability were to be calculated, the specific buy of parts (how many batches) would be relevant information. Often it is not possible to find a set of attributes which fit some obvious clumping of points. Nevertheless, it would remain very likely that systematics are a serious perturbation and sample size corrections would be appropriate as well as estimates of a maximum reliable survival probability which could be estimated from that effective sample size. SUPERSEDES PAGE 47 OF MIL-HDBK-816 47