Solder-Joint Field Failures Under Combined Thermal and Wind Loading
A manufacturer of outdoor LED luminaires has a field failure rate of 2.1% within six months, against a baseline of 0.3%. The failures are open-circuit solder joints at the corner of a 15 mm × 15 mm QFN package on a 1.6 mm FR4 board. The luminaire operates from -20 °C to 70 °C and is mounted on a metal pole that flexes in wind. The board is secured by four screws at its corners; the QFN is 18 mm from the nearest screw. Laboratory thermal cycling (-40 to 125 °C, 1000 cycles) had passed before launch. The proposed corrective action is to add underfill to the QFN only, leaving the board and mounting unchanged. A wind-tunnel test shows the pole's first mode at 4.8 Hz and board strain of 450 µε at the QFN corner during 15 m/s gusts.
1. Distinguish between thermal fatigue and mechanical flexure as the failure mechanism, and state the evidence you would collect.
Read the worked answer
Both mechanisms can crack a corner joint, so the distinction must come from evidence, not from the failure location alone.
Thermal fatigue is driven by the mismatch in coefficient of thermal expansion between the package, solder, and board. Cracks typically initiate at the package corner on the board side, propagate along the diagonal, and show a smooth fatigue surface under magnification. The laboratory thermal cycle passed 1,000 cycles, which makes pure thermal fatigue a weaker explanation unless the field cycle is more severe than the test, for example due to power cycling or a different temperature ramp.
Mechanical flexure is driven by board bending. Cracks tend to follow the bend line rather than the package diagonal, and the failure should correlate with wind events. The 4.8 Hz pole mode and 450 µε at the QFN corner are significant: repeated gusts at that strain can produce high-cycle fatigue. The QFN is 18 mm from the nearest screw, so it sits in a relatively flexible region of the board.
Evidence to collect: dye-and-pry and cross-sections of returned units; fracture-surface microscopy; strain-gauge data from a field-mounted board correlated with wind speed; and a laboratory bend test at 450 µε with and without thermal cycling. A combined test is more informative than either test alone.
2. Evaluate the underfill-only proposal.
Read the worked answer
Underfill can help, but the proposal as written is incomplete and may make things worse.
Underfill reduces the shear strain in the solder joints caused by thermal expansion mismatch by mechanically coupling the package to the board. That is a reasonable thermal-fatigue countermeasure. However, it transfers more stress into the solder pads and the board beneath the package, raising the risk of pad cratering and laminate cracking, especially with a brittle or thin board.
If the dominant field mechanism is board flexure, underfill does not reduce the bending strain at the QFN corner. It may even stiffen the local region and concentrate stress at the underfill fillet or at the nearest screw. Leaving the board mounting unchanged means the 450 µε and the 4.8 Hz pole mode remain.
The proposal also lacks a control: only the QFN is underfilled, so any improvement could be confounded by lot or assembly differences.
A better corrective action would combine underfill with a reduction in board strain: add a mounting point closer to the QFN, stiffen the board, relocate the package nearer a screw, or damp the pole. The choice should follow the mechanism evidence, not be fixed in advance.
3. Propose a validation plan with acceptance criteria.
Read the worked answer
The plan must reproduce the field environment, not just the original thermal qualification.
Test legs:
- Thermal cycling alone, as a control, to confirm the original qualification.
- Combined thermal cycling with simultaneous board bend at 450 µε and 4.8 Hz, representing wind gusts.
- Bend-only at 450 µε for high-cycle fatigue.
- Field trial of at least 500 units for six months on exposed poles, with wind data logged.
Sample size and duration: at least 30 units per leg, run to twice the expected field life or until 50% of samples fail, whichever comes first; use Weibull analysis to estimate the shape and scale parameters.
Instrumentation: strain gauges at the QFN corner, event counters for wind gusts, and continuity monitoring to detect open joints in real time.
Acceptance criteria:
- No open-circuit joint in any leg before twice the field life.
- Predicted failure rate at six months below 0.3%, with the upper confidence bound below 0.5%.
- No pad cratering or laminate cracking in cross-sections.
- Field trial return rate below 0.3% and no wind-correlated failures.
If the combined leg fails while the thermal-only leg passes, the mechanism is flexure-assisted and the underfill-only action should be rejected.