Random RCD Tripping and the Demand for a 100 mA Unit
A 25-year-old house has a main distribution board containing a 63 A isolator, a 63 A / 30 mA RCD used as the main switch, and four MCBs: 6 A lighting, 16 A socket outlets, 16 A geyser and 20 A for a 1.5 TR window AC. For several weeks the RCD has tripped at random, most often at night during rain and sometimes a few seconds after the AC compressor starts. An electrician disconnects all appliances and carries out a 500 V insulation resistance test: the whole installation reads 0.8 MΩ; the lighting, geyser and AC circuits each read above 50 MΩ; the circuit feeding an outdoor socket, a 0.5 hp water pump and an outside light reads 0.9 MΩ. A ramp test on the RCD shows it trips at 22 mA. Nothing else in the board has been changed. The owner wants the RCD replaced with a 100 mA unit 'so that it stops troubling us'.
1. Is the RCD defective? Interpret the ramp test and the insulation resistance readings, using the numbers.
Read the worked answer
The ramp test is decisive: a 30 mA RCD must not trip below 15 mA and must trip by 30 mA. Tripping at 22 mA is mid-band and perfectly healthy, so on this evidence the device is working correctly and cannot be condemned.
IR results: 0.9 MΩ on the outdoor circuit is the only value below the usual 1 MΩ minimum and needs repair - but at 230 V that is a leakage of 230/900 000 ≈ 0.26 mA, over a hundred times too small to operate a 30 mA device by itself. The whole-installation figure of 0.8 MΩ is simply the parallel combination dominated by that one circuit, so it adds no new information.
Conclusion: the RCD is not proven faulty. Something intermittent - moisture, appliance filter leakage, or a neutral-earth fault - is producing leakage only under certain conditions, which is why the test done dry and unloaded looks almost normal.
2. Explain why 0.9 MΩ cannot by itself explain the trips, and identify three credible causes you would investigate.
Read the worked answer
0.26 mA against a 30 mA threshold is a gap of roughly 115 times; even several damp sockets would not close it. So look for causes that add up or that bypass the toroid:
- Cumulative appliance leakage. Each item with an EMI filter (AC, refrigerator, washing machine, LED drivers, laptop charger) can leak 0.5–3 mA. Ten such items on one 30 mA RCD can sum past the threshold - matching the pattern of night-time trips when everything is plugged in.
- Neutral-to-earth fault downstream. A N-E short or crossed neutral sends part of the return current outside the RCD toroid, so ordinary load current looks like leakage and the device trips as soon as a heavy load such as the AC starts. Test with the main switch off: IR between neutral and earth bar, then per circuit.
- Moisture-dependent insulation failure. A dry test misses it; the outdoor socket, pump gland or a buried cable may fall to a few kΩ when wet - the classic 'trips when it rains' pattern. Re-test during or just after rain, or wet the area deliberately.
Also check the RCD type: a type AC device can misbehave when electronic loads produce DC leakage; a type A may be required.
3. Give your technical and safety answer to the 100 mA proposal, and the arrangement you would install instead.
Read the worked answer
Refuse, and put the reason in writing. A 30 mA RCD is the accepted additional protection against electric shock for socket circuits; 100 mA is a fault/fire-limiting rating and will not operate before a lethal current has passed through a person for too long. Uprating masks a real defect: the leakage will keep growing as insulation degrades, and the next symptom could be a fire rather than a nuisance trip.
Instead, keep 30 mA protection and divide it. Fit a 30 mA RCBO on each final circuit, or at minimum separate the outdoor/pump circuit and the socket circuit onto their own RCBOs, so one leaky circuit does not black out the house. Repair or replace the wet outdoor socket with an IP-rated enclosure and gland, re-terminate the pump, and confirm no N-E fault exists.
Then re-test: IR on every circuit, RCD trip time at IΔn and at 5IΔn, and a leakage reading per appliance with a leakage clamp. Log the results and explain to the owner that the tripping was the RCD doing its job, not misbehaving.