Downloaded from https://www.everyspec.com AMCP 700-16 Life of a Transistor- when the Temperature Fluctuates Assumptions: 1. The curve of life (appropriate percentile) ve temperature (at constant tem- perature) is known. 2. The severity level can be completely ohiuracterized by a temperature. 3. Linear cumulative damage Is appropriate. 4. No new failure modes, which would decrease the life, are introduced by the temperature changes. (In the theoretical development, this in irrelevant since No.
determines the method of calculation, But when wondering whether or not No, 3 ap- plies, this Is something to consider.) Let the P1 0 life (10% will rail before that time) be given by the life ourve, Filg. '7-2, and the temperature profile be a regularly repeating pattern as shown in Fig. 1t-3, Illustrate& the linear cumulativo damage hypottesi na cuirv, it can be shown (for what it Is worth) that a constatt temperature of 162°C would Table 7.1 can be developed from the P, o life Sive the sa',i.o Pie life.
It is interesting to curve and the temperatute profile, compare thc "% life column" with the "% The domage rate is the reciprocal of Pic life, damage column", etg,, at 3 30'C, 15% of the The fraction of damage has units of 10" life causes 43% of the damagi; while at 50*C, L/hr where L Is the presumed equivalent 40% of the life causes less than 4% of the Pic life. This fraction Is calculated by multi- damage. plying the numbers In the 2 preceding columns. It is from the total of fraction- of-damage column that L is calculated, i.e., From the remarks earlier in this paragraph the total must be unity.
From the P, 0 life it should be remembered that L - 42.9 X TAILS 7.1 PERCENT DAMAGE VS TIMPERATURN TOmp P1o life Pvatlon of Life iCYlWs wet Ratison of Dampie C 10 hr hr/lprlokx % 10.1/hr 10t dIh acual 0 510 10+0 40 0.20 0.00 3.4 100 1,0 4 10 1,0 0,10 4,3 i15 0.50 a ,0 2.0 0.40 17.2 210 0.20 a Is 6.0 0,75 32.2 ISO 0.16 a Is 0.7 1.00 42..? 1 0429 40 100 2,33 2.33 100 ILI,' S•,#t#,' + ''+ '