Tools › Reliability › System Models › Reliability Prediction (217F) › Example

Digital control board of 42 parts: MIL-HDBK-217F part stress prediction in ground benign and airborne inhabited environments

Tool	Reliability Prediction (217F)
Source	U.S. Department of Defense (1995). Reliability Prediction of Electronic Equipment. MIL-HDBK-217F, Notice 2, 28 February 1995. Sections 5.1, 5.2, 5.8 to 5.11, 6.1, 6.3, 9.1, 10.1, 15.1, 16.1, 19.1 and Appendix A.
Level	Professional

Problem

A small digital control board carries an 8-bit CMOS microprocessor, a 64K UVEPROM, a 16K static RAM, six CMOS logic circuits, an operational amplifier, two switching transistors, four switching diodes, a Zener reference, twelve film resistors, eight ceramic and two solid tantalum capacitors, an 8 MHz crystal and a rectangular connector on a four-layer board with 270 wave-soldered plated through holes. Predict the failure rate, the MTBF and the reliability over 1,000 hours by the part stress method in a ground benign environment at 40 °C ambient and in an airborne inhabited cargo environment at 55 °C, and compare with the parts count method.

Inputs to Enter

Microcircuits class B: microprocessor 40 pins, 0.5 W; UVEPROM 28-pin glass-seal DIP, 0.3 W; SRAM 24 pins, 0.3 W; six logic circuits of up to 100 gates, 14 pins, 0.05 W; linear circuit of 30 transistors, 8 pins, 0.1 W; junction-to-case 28 °C/W
Semiconductors JANTX: two switching transistors rated 0.5 W at 0.1 W and 40 percent voltage; four switching diodes at 30 percent voltage, 0.02 W; Zener at 0.2 W; junction-to-case 70 °C/W
Twelve RLR resistors level M at 0.05 W of 0.25 W; eight CKR capacitors 0.1 µF at 5 V of 50 V; two CSR capacitors 10 µF at 5 V of 20 V with 1 Ω/V; crystal 8 MHz; rectangular connector, 0.1 mating cycles per 1,000 h, 1 A in 22-gauge contacts; board with 270 holes, 4 planes
Environment 1 ground benign, ambient 40 °C; environment 2 airborne inhabited cargo, ambient 55 °C; mission 1,000 h

Open the Reliability Prediction (217F) tool (loaded with this example)

Answers

Worked part, the microprocessor in ground benign: T_J = 40 + 28 · 0.5 = 54 °C, π_T = 0.1 exp[−(0.35/8.617×10⁻⁵)(1/327 − 1/298)] = 0.3349, C_2 = 2.8×10⁻⁴ · 40^1.08 = 0.01504, λ_p = (0.14 · 0.3349 + 0.01504 · 0.5) · 1 · 1 = 0.05441 failures per 10⁶ h
Ground benign at 40 °C: λ = 0.2542 failures per 10⁶ h, MTBF = 3,934,135 h, R(1,000 h) = 0.99975; largest contributors the connector (35.8 percent) and the microprocessor (21.4 percent)
Airborne inhabited cargo at 55 °C: λ = 1.690 failures per 10⁶ h, MTBF = 591,596 h, R(1,000 h) = 0.99831, 6.65 times the ground benign rate
From 0 to 100 °C ambient the ground benign failure rate rises from 0.1293 to 1.173 failures per 10⁶ h (factor 9.07); every stress is inside the default derating limits
Parts count method with the Appendix A generic rates for the same parts: 0.2779 failures per 10⁶ h in ground benign and 2.841 in airborne inhabited cargo

Notes

The tool's default inputs are this board; press Submit (select the parts count method for the last result). The board is illustrative and has no printed source. The models are checked by test_reliability_prediction.php (2,933 checks): the handbook's worked examples of Sections 5.13 (0.15; 0.038, 0.031, 0.0015, 0.0069), 6.15 (0.011), 10.2 (0.049 from the printed rounded factors, 0.047 unrounded) and 16.2 (18,893 cycles, 0.4), 1,964 Appendix A cells recomputed from the part stress models at the appendix's stated defaults, and an independent numpy implementation on this board and 20 seeded boards. The two methods differ because the appendix assumes its own default temperatures and stresses and a board of 1,000 plated through holes.