“But the datasheet says 10 A…” — The Spec That Actually Fails First in a Relay Roundup

Scenario: A maintenance light panel for a conveyor line, using Omron MY2 relays (5 A, 250 VAC, AgCdO contacts, 12 VDC coil). The designer picked the MY2 because each lamp draws 0.8 A — well under the 5 A contact rating. But within 18 months, three relays failed: two stuck open, one chattered intermittently.

Number → Mechanism: The MY2 coil is rated 12 VDC ±10%; the panel’s unregulated 12 V supply drifted to 13.0–13.4 V after a rectifier replacement (about +8–11% above nominal). That’s within the coil’s absolute max (usually 110% of rated voltage per IEC 61810-1), but the continuous over-excitation raised coil temperature 8–12 °C above the 70 °C upper limit of the MY series. The temperature rise reduced the mechanical clearance of the armature pivot, causing incomplete stroke and eventual contact bounce. Each bounce under 0.8 A resistive load generated a micro-arc that eroded the AgCdO surface asymmetrically — not enough to weld, but enough to create a high-resistance zone that eventually stopped current flow.

Worked consequence: The failure mode was contact resistance creep, not weld. The relay still “clicked” audibly, but contact resistance rose from ~15 mΩ to >2 Ω. The circuit breaker stayed closed, but the lamps dimmed. Three unscheduled line stops cost ~$4,700 in lost throughput over two years.

↻ When this reverses: If the supply voltage had been regulated within ±5% and the ambient stayed 50,000 operations at 0.8 A — the coil would not be the weak link. This case fails first on coil voltage tolerance, not contact rating.

Scenario: A large warehouse retrofit used eight 480 W LED fixtures driven by electronic drivers. Inrush current per fixture peaked at 110 A for ~200 µs (datasheet typical). The designer chose an Omron G7J-4A (40 A, 250 VAC, AgSnO₂ contacts), reasoning 40 A continuous × 0.8 derating = 32 A, and steady-state load was only 16 A. Within 3 weeks, the G7J-4A welded closed on one pole.

Number → Mechanism: The G7J-4A contact rating is 40 A at 250 VAC for resistive loads per IEC/UL 61810-1. The inrush crest of 110 A (≈275% of rated) lasted 200 µs — too brief to trip a breaker, but enough to initiate a micro-weld junction in the AgSnO₂ material. AgSnO₂ is more weld-resistant than AgCdO, but only if the inrush is contact welding, but the root cause was inrush vs. contact material capability, not the 40 A steady-state rating.

Worked consequence: The fixture bank stayed on 24/7 after the weld. A maintenance electrician saw the G7J still clicking but no open-circuit — they had to replace the relay and add an NTC inrush limiter. Cost of the NTC bank: $210. Cost of one unscheduled night shift: $960.

↻ When this reverses: If the LED drivers had active PFC with inrush inrush vs. contact material tolerance. For any electronic load, measure inrush crest and duration — the 40 A number is a fantasy if the spike exceeds 150% of rated.

Scenario: A bank of 1.2 kW resistive heaters (≈5 A at 240 VAC) switched by an Omron G2R-1 (10 A, 250 VAC, AgCdO) on a PCB mount. The load was purely resistive, no inrush. After 8,000 cycles (about 18 months), the relay failed open — the armature no longer pulled in.

Number → Mechanism: The G2R-1 has a rated mechanical endurance of 10 million operations (no load) but electrical endurance at 10 A resistive is typically ~250,000 cycles (illustrative, based on Omron derating curves). At 5 A resistive, one might expect >500,000 cycles. But the coil drive came from a 24 VDC supply that sagged to 21.4 V during heater on-time due to undersized cabling. The G2R coil is rated 24 VDC ±10% (21.6–26.4 V). At 21.4 V, the coil voltage was below the minimum pick-up voltage (typically about 80% of rated, i.e., 19.2 V) — but only by 0.2 V. The relay did pick up, but the magnetic force was marginal. Over thousands of cycles, the armature seated with less overtravel, causing increased contact bounce and secondary arcing. The arcing eroded the AgCdO asymmetrically, and one contact crater became deep enough that the cantilever spring lost preload — the contact gap widened, and at 21.4 V the coil could no longer close the gap.

Worked consequence: Heater bank lost one phase, causing uneven temperature in a lab oven. Product quality drifted for three days before the fault was found — $6,200 in reprocessing costs.

↻ When this reverses: If the 24 V supply had been stiff (≥22 V under load) and the ambient coil pick-up margin — the contact rating was irrelevant. For DC coil circuits, measure the minimum supply voltage during load on-time; a drop of 10–15% can kill a relay that looks “oversized” on contact rating.