Field Failure of 24 V DIN-Rail Power Supplies in Hot Enclosures
A fleet of 500 DIN-rail 24 V, 5 A switch-mode power supplies is installed in car-park ventilation control panels across Hong Kong. After nine months, 38 units (7.6%) have failed: the output sags to about 12 V under load, the unit hiccups, then the internal fuse opens. All failures are in panels where the measured internal air temperature is 52–55 °C in summer. The datasheet gives 5 A up to 50 °C, derating linearly to 3 A at 70 °C. The output electrolytic capacitor is rated 105 °C, 1,800 µF, with 1.8 A ripple rating at 105 °C/100 kHz; bench measurement on a failed unit shows 2.4 A ripple at 100 kHz. Production test is a two-minute functional check at 25 °C; there is no burn-in or thermal cycling. Failure analysis reports capacitor ESR at 12 times the initial value, but the supplier disputes this. The client wants a containment plan within 48 hours and a decision on whether to replace all 500 units. A new production batch is due to ship in three weeks.
1. Set out the immediate containment actions and the evidence you would require before deciding on fleet replacement. Justify the priority order.
Read the worked answer
Immediate containment: (1) identify all panels with internal temperature above 50 °C and load above 3 A; (2) reduce load to 3 A or add temporary external ventilation where possible; (3) replace failed units and fit a temporary 24 V UPS or bypass on critical ventilation panels; (4) quarantine new stock and sample 10% for a four-hour burn-in at 55 °C with ripple measurement. Priority is life-safety ventilation and preventing repeated fuse failures, not blanket replacement. Before fleet replacement, require: failure distribution by site temperature and load; a repeatable thermal and ripple test on returned units; capacitor core temperature or ESR trend; a supplier 8D report with lot traceability; and a cost/risk comparison of replace-all versus targeted replacement. A 7.6% failure rate in nine months is far too high for a five-year design life, so plan for replacement, but do not commit until the mechanism is confirmed. Communicate a 48-hour interim plan with dates and uncertainties.
2. Propose a root-cause verification plan and the design and production changes, including derating and test coverage.
Read the worked answer
The evidence points to output capacitor overstress. The 2.4 A ripple is 133% of the 1.8 A rating, so I²R heating raises the core temperature well above the 52–55 °C ambient; every 10 °C rise roughly halves electrolytic life. The 5 A load at 52 °C also exceeds the linear derating (about 4.8 A at 52 °C), so the unit is marginal on two counts. Verify by measuring capacitor can temperature and ripple current on a loaded unit at 55 °C, and by comparing ESR before and after a 500-hour life test. Design changes: use two capacitors or a lower-ESR part with at least 2.5 A ripple rating at 105 °C, derate to 80% at maximum ambient, and add output LC filtering. Production changes: 100% burn-in at 55 °C for two hours at full load with ripple and ESR screen; lot sampling life test; and a datalogged thermal test on the first article. Do not accept a supplier claim without the 8D evidence.
3. How would you communicate the uncertainty and the replacement decision to the client and supplier?
Read the worked answer
Separate confirmed facts from hypotheses. Tell the client: 38 failures are confirmed; the likely mechanism is capacitor overstress at high temperature and ripple; the failure rate is unacceptable, but a blanket replacement before root-cause confirmation could waste money and still leave the design fault. Give a dated decision gate: containment now, root-cause report within 10 working days, replacement recommendation within 15 days. Tell the supplier: the ripple measurement and temperature data are objective; we require lot traceability, a corrective-action plan, and a written warranty position for the installed fleet. Avoid attributing blame until the 8D is reviewed. Put all advice in writing, state the residual risk for critical ventilation panels, and ask the client to accept the interim risk in writing. If the supplier disputes the data, propose a joint test at an independent laboratory.