Module 7: Validation, Reliability & Test
Validation taxonomy & KPIs, HALT/Weibull/bathtub, FMEA root cause, HIL & DFT.
-
1conceptHow do characterization, verification, and validation differ in a reliability program?
-
They are distinct activities with different questions, not synonyms for the final test.
-
Correct. Characterization maps how the thing actually behaves and where its margins are; verification asks did we build it to spec; validation asks did we build the right thing for the user's need.
-
The natural order is characterize, then verify against spec, then validate against need, not the reverse.
-
Verification (against spec) is an internal engineering activity; validation is the one most tied to the user's need, so the split described is wrong.
-
-
2conceptWhat is the difference between HALT and HASS/burn-in?
-
This reverses the two: HASS screens production, HALT finds margins.
-
Correct. HALT (Highly Accelerated Life Test) deliberately overstresses prototypes to discover operating/destruct margins and weak links; HASS and burn-in apply a calibrated sub-destruct stress to screen production units for early-life (infant-mortality) defects.
-
They differ in purpose, not just temperature: margin discovery versus production screening.
-
HASS applies real accelerated stress (it is not stress-free), so calling it non-stress functional testing is wrong.
-
-
3conceptPlotting field failures over time gives the classic bathtub curve. What do its three regions represent?
-
The bathtub axes are failure rate versus time-in-service, not calibration/storage phases.
-
Correct. The bathtub curve shows a falling infant-mortality region (early defects screened by burn-in), a flat useful-life region of low random failures, and a rising wear-out region as parts age out.
-
The whole point of the bathtub is that the rate is not constant: it falls, flattens, then rises.
-
Those are defect sources, not the three time-based regions of the failure-rate curve.
-
-
4calcIn an FMEA, a failure mode is rated Severity 8, Occurrence 4, Detection 5. What is its Risk Priority Number (RPN)?
-
17 is the sum S+O+D; RPN is the product, not the sum.
-
Correct. RPN = S × O × D = 8 × 4 × 5 = 160, the prioritization score for ranking which failure modes to attack first.
-
100 = 5 × 4 × 5 would use the wrong severity value; severity is 8 here.
-
40 = 8 × 5 drops the Occurrence factor; all three terms multiply.
-
-
5scenarioA populated PCB needs production test that can reach internal nets and verify solder joints on a fine-pitch BGA whose pins are hidden under the package. Which design-for-test method addresses the hidden BGA balls?
-
A bed-of-nails ICT needs physical access to a node; the BGA balls are under the package, so probes cannot reach them.
-
Correct. Boundary scan (JTAG/IEEE 1149.1) uses cells at each device pin to drive and observe the hidden BGA connections through the scan chain, testing interconnect without physical probe access.
-
Flying probes also require surface access; they cannot touch balls trapped under a BGA package.
-
Functional test exercises the whole board but does not isolate or confirm individual hidden solder joints the way boundary scan does.
-