The lower-resistance MOSFET fails before the product starts
Fictional engineering case. All case measurements and outcomes are constructed for this series; no company or reader is being quoted.
Last week’s answer: six loaded contacts changed the conditions
The cold resistance measurements did not support an obvious single bad connection. They also could not explain the full thermal behaviour of the assembled connector.
Using 8 mΩ as a representative complete-path resistance, the initial dissipation in each 5 A path is:
P = I²R = 5² × 0.008 = 0.20 W
Across six paths, that is 1.2 W within the defined measurement boundaries. It includes termination and contact losses, not solely heating at the mating interface. It is an estimate at the stated resistance, not a prediction of temperature rise: resistance changes with temperature and the heat must leave through the surrounding assembly.
The fictional team instrumented the centre and edge positions, representative crimps and the air beside the connector. It tested one loaded contact, then increasing numbers of adjacent contacts, keeping local ambient and current per loaded contact controlled. Temperature rose as the populated group grew, with the centre consistently hottest.
It repeated the experiment with a newly prepared harness using documented crimp settings. The pattern remained. Voltage-drop measurements across the warm paths did not reveal a termination with disproportionate loss. These results supported mutual heating and the installation’s thermal conditions as the main explanation.
Supplier loading data, reviewed against the actual six-contact arrangement and confirmed by the enclosure tests, did not support the original selection at the product’s local ambient and acceptance limit. The team moved to a connector configuration supported by the appropriate loading data and retested it in the closed enclosure. Crimp quality remained part of production control.
The error lay in applying a single-contact rating to a different operating arrangement. No universal derating factor would have established whether this particular assembly passed.
This week’s challenge: a cooler device that fails at power-on
A 48 V industrial unit uses a high-side N-channel MOSFET and hot-swap controller to charge 4,700 µF of downstream capacitance. The downstream load is disabled during charging. The capacitor has been discharged to approximately 0 V before each test, including warm restarts. For the simplified calculation, assume a constant 48 V input, negligible capacitor leakage and an approximately constant charging current of 1.0 A.
The approved replacement MOSFET has the same package and voltage rating as the original. Its lower on-resistance reduces conduction loss after startup. Cold starts generally pass, but some warm restart tests end with the replacement failing short circuit.
The measured startup is approximately 225 ms. Drain-source voltage begins close to 48 V and falls as the output capacitor charges. The controller holds current near 1 A during most of that interval. The controller modulates gate-source voltage to regulate charging current rather than immediately driving the MOSFET fully on.
The mounting base is already at 75°C before a failing restart. Available datasheet safe-operating-area curves describe single pulses with the mounting base held at 25°C. The prior operating history, initial junction temperature and time between restarts have not yet been accounted for. The substitution assessment compared voltage rating, current rating, package and on-resistance. It contains no assessment of the startup trajectory or hot linear-mode capability.
Initial captures show no obvious drain overvoltage. The team has not yet established whether repetitive heating, localised die heating or another mechanism causes the failure.
Which rating governs these 225 milliseconds?
Calculate the ideal charge time and approximate energy dissipated in the MOSFET. Then explain why neither the headline current rating nor a single energy figure establishes that startup is safe.
Which voltage, current, timing and temperature evidence would you obtain before approving either a new MOSFET or a revised current limit?
The answer follows on 29 October, alongside a manufacturing-change decision for engineering and purchasing. Send your reasoning through the Troubleshooter page.
