A switch that works ten times will probably work ten thousand times, and that is exactly why endurance testing gets skipped. The test is not there to prove the mechanism moves. It is there to find the point where a cam wears enough that the contacts stop snapping, and the arc starts doing the work the spring used to do.
The verdict is electrical, but the wear that produces it is mechanical. A rocker switch has a pivot, an over-centre spring, a moving contact bridge and two fixed contacts. Every one of those parts changes shape over a few thousand operations, and each change shows up as a different electrical symptom.
What the count actually covers
IEC 60669-1 separates normal operation from making-and-breaking capacity, and the two have very different counts. Normal operation is a switch doing its job at rated current. Making-and-breaking capacity is a switch closing onto a fault-level current, which is a much smaller number of cycles because each one consumes contact material.
| Parameter | Value normally applied to a 10 AX wall switch | Why it matters |
| Rated voltage | 250 V AC | Sets the arc length at break |
| Rated current | 10 A (10 AX = 10 A at 250 V, fluorescent load) | Sets the thermal load in the closed state |
| Normal operations | 10,000 | The headline number in the datasheet |
| Making/breaking capacity cycles | A few hundred, not thousands | Each cycle erodes the contact |
| Test load | Rated current at rated voltage | Resistive load is easier than the real one |
| Contact resistance, after | Tens of milliohms, not hundreds | The first electrical symptom of wear |
| Temperature rise, after | Within the 45 K class limit for the terminal | Wear shows as heat before it shows as failure |
| Dielectric after the run | The full test voltage, one minute | Carbon tracking is the silent failure |
The count is only meaningful with the rate and the duty cycle attached to it. A drive that closes for three seconds and opens for three seconds gives the contact a chance to cool between operations; a drive that runs continuously does not. Two laboratories can run 10,000 operations and report different temperature rises on the same switch design, and neither is wrong.
Rate changes the result, and no standard fixes it for you
Operation rate affects the test in three ways that pull in different directions. A fast rate keeps the contact hot, which softens the silver and accelerates welding. A slow rate lets oxide grow on the contact face between operations, which raises resistance at the start of each cycle. A very fast rate also changes the dynamics of the mechanism: the over-centre spring may not have time to settle, and the bridge bounces more than it would in service.
The practical rule is to pick a rate that a person could produce, and to record it. Somewhere between ten and thirty operations per minute is defensible. Sixty per minute is a stress test, not a type test, and if a customer asks for it the report should say so.
Three failure modes and how to tell them apart
- Welded contacts. The switch stays closed after the drive opens. Cause: too much current for the contact area, or a bounce long enough to melt a bridge. Evidence: a rough, pitted face on both contacts with matching craters.
- Lost snap. The mechanism still switches, but slowly. The contacts part at a lower speed, the arc lasts longer, and the face erodes in one spot instead of spreading. Evidence: a single deep pit near the contact edge, and a cam profile that no longer holds the spring past centre.
- Carbon tracking. The switch still works but the insulation around the contacts has a conductive path. Evidence: a measured resistance from live part to accessible surface that falls after each 1,000 operations, and a brown line on the barrier that will not wipe off.
Only the third one fails the dielectric test. The first fails by inspection. The second passes both and fails in service, which is why the contact resistance trend matters more than the final reading.
Recording a result that can be defended
Measure contact resistance every 1,000 operations, on the same pole, with the same meter and the same four-wire method. Plot it. A switch that starts at 8 mΩ and ends at 12 mΩ is healthy; one that goes 8, 11, 14, 28 mΩ in the last two thousand has already told you how it will fail. Record the ambient temperature, the drive rate, the load type and the mounting, because all four change the number.
The test station itself has to be stable: a switch endurance rig with an adjustable stroke and a force-limited drive, not a motor with a cam cut by hand. If the drive pushes past the switch travel, the mechanism is being tested beyond its design and the result belongs to the rig, not the product. The full range of test equipment on the site index starts from the same principle: the apparatus must not be the variable.
Standards referenced: IEC 60669-1.
