Downloaded from https://www.everyspec.com AMCP 706-198 to step.stressing. There are other cases 1, The rate of doing damage does not where the failure mode changes so drastically depend on the amount of damage already at some level, that it is senseless to continue done. *1 testing above that level. 2. The order in which the severity levels Another advantage of step-stress testing are applied makes no difference. is the elimiitation of "switch.on" problems, such as initial transients and failures due to 3. The total damage is the simple sum or hIgh stress rates. This is becaum the severity Integral of the damage done at each stress level is low at the beginning and the severity level.
increase is gradual. The severity level need not begin at "zero" (iOe,, a level near benign), 4, The rates of doing damage are inde- This can save time and reduce the amount pendent of each other for different severity of cumulative damage at severity levels other levels, than the failure level. Some programs con- corned with investigating cumulative damage S.
The Median' endurance at constant theories may change the severity level only severity level Is unity, once during a test, For example, the Initial part of one test may be at a high severity With regard to Corollary No, 5, the actual level and the remainder at a low severity endurance is I + e, where e is a random vari- level-, a subsequent test reverses the procedure, able:
its statistical properties depend on the Not miuch work of this sort is done in progiamning of the severity levels, on the electronics, but metallic fatigue is a field In probability distributions of the times to which these methods of programming stresses failure at each severity level, and on the per- have received coiisiderablo attention. centage chosen in footnote 2, It usually* is presumed that the calculated life is the Median (or the percentage in footnote 2).
7.8 CUMULATIVE DAMAGE ' The use of a cumulative damage model In order to compare tests (or field ox- does not necessarily mean that the failure perionce) run under different severity-level modes/mechanisms were the same at each programmings, some model of cumulative severity level, although such a case may help damage is necessary, No particular model is the validity of the model. Example No. 32 required, merely some model. In electionics there are very few theories of cumulative iThe Median ( 5.e,, 50th rraontUe) is the conventionsj damage, regardless of the part, but in mech- ffratlon to use. One ooud at eas2ly use tome other per.
dhntile, e, 1% (11% have lives sas than the liven time), anical fatigue, for example, thete are mary The permnitle in the definition and to corollary 5 must models of cumulative damage. Most often CI, or voullrs, 2". 'Coroflaieek 1, 2, 4 are at'ue becaese tits dainq• mrts such a model uses constant,,"tress" test depends only on MtF, not on time (for No. 1) nor on s as its basis. The most common conceptual vetrit v level order (for No, 3) nor on the value of MtP for some othrseverity level (for No. 4), Corollary 3 model, in almost any field, for cumulative i.
true beasuse total damalq D Is one basic issumption, l.e., the rate of doing D f d ( damage is I /MtF where MtF is the Median dr t time to Failure'. l'he MF is for the parti- cular wverity level at which damage is ac- cumulating. There are several corollaries 2 to CorollerU 3 is true doca the nwdian time to (allure, at a this assumption which often are (but эmpro- given ivoktty level it NIP, by hypothesis; the total (median) damage is dam•Nis rate (11MtF) multiplied uy perly) stated as additional assumptions: tienMtF Wics unIty. .. ..