Downloaded from https://www.everyspec.com MIL-HDBK-816 NOTICE 1 TABLE I. Military standard RHA detailed specifications. Letter designator * Ionization effects ** Neutrons *** Hardness assurance levels / No RHA No RHA M 3 X 10 3 rad(SI) 2 X 10 12 D 1 X 10 4 rad(SI) 2 X 10 12 P L 3 X 10 4 rad(SI) 5 X 10 4 rad(SI) 2 X 10 12 2 X 10 12 R 1 X 10 5 rad(SI) 2 X 10 12 F 3 X 10 5 rad(SI) 2 X 10 12 G 5 X 10 5 rad(SI) 2 X 10 12 * For microcircuits, the letter designator for RHA parts replaces the slash in the part number; e.g., 38510R291A in place of 38510/291A. For discrete devices it is a suffix, e.g., JANTXVR. ** At a dose rate of 200 rads(SI)/sec plus or minus 100 rads(SI)/sec.
*** 1 MeV silicon displacement damage equivalent neutron fluence. Test to be conducted only when specified in the purchase order or contract. H 1 X 10 6 rad(SI) 2 X 10 12 4.3.1.1 Advantages and disadvantages of attribute tests. An advantage of the attribute tests is that the data collection and analysis are simple and relatively inexpensive. A further important advantage is that the test is distribution free (i.e., no assumptions are required about probability distributions). However, a major drawback is that the tests require inordinately large sample sizes if significant survival probability requirements are imposed.
For example, to check with 90 percent confidence that a lot has at least 99 percent survivable parts, i.e., that the sample size series number is 1 percent or less, it is necessary to test a sample of at least 231 parts with no failures. It is interesting to note that attribute tests can be applied to attributes that can assume more than just the two values of pass or fail. Acceptance criteria for lots can also be based on a mix of attributes that may be desired (e.g., with confidence C the probability of a given attribute must be between 49.9 percent and 50.1 percent). 4.3.1.2 The n/c test.
The most common type of attribute test is the so called n/c test where a sample of n parts is tested and if more than c of the parts fail the test then the lot is rejected. The significant test result for each part is the attribute of passing or failing and hence the term attribute test. The usual failure criteria for the parts are: (a) functional failure or (b) deterioration of any critical parameter past an acceptable limit. 4.3.1.3 Multiple sampling. Some attribute tests allow the drawing of extra samples if lots fail on the first try (multiple sampling plans).
A common multiple sampling plan used in the military standard system, for example, is a test of 11 parts where the lot is passed if there are no failures. Then, if there is exactly one failure, an additional seven parts may be sampled and tested with no further failures allowed. If there is a total of two or more failures, the original lot is rejected. 4.3.2 Confidence and probability. The results of the military standard attribute lot acceptance test are phrased in terms of a confidence C which is the chance of rejecting a lot where there is less than a probability P that any part in the lot can pass the test conditions. 4.3.2.1 Usual confidence and probability.
In the military standard system, the significance of an n/c attribute test is usually given in terms of 90 percent confidence and a quantity called the Sample Size Series Tests which may be found, for example, in MIL-PRF-38535, Appendix D, the general specification for microcircuits. Thus, for example, the sample size series associated with an 11/0 sample size/accept number test may be found in the table to be 20 percent. An 11/0 sample size/accept number test will therefore provide 90 percent confidence that a lot with 20 percent defective parts or greater will be rejected. In terms of the probability P mentioned in 4.3.2, the sample size series number equals 1 minus P.
SUPERSEDES PAGE 10 OF MIL-HDBK-816 10