1 in 14 shock protectors silently fail. Most homeowners have no idea.

A residual current device, or RCD, is the protective device that watches the imbalance between line and neutral on a circuit. When a person touches a live wire, the current that flows through their body to earth creates that imbalance. The RCD detects it and opens the circuit before the current is sustained long enough to kill.

In a default NZ switchboard installed to AS/NZS 3000^[2], an RCD covers each lighting and socket final subcircuit. The MCB on the same circuit handles overcurrent. The RCD handles earth-leakage. Neither handles arcing (see arc-fault-risk for that).

For an RCD to do its job, it has to be working when the fault happens. The rest of this page is about what that means and how often it isn't.

The two numbers stamped on every residential RCD: 30 mA and 100 ms. They define what "working" means. IEC 60479-1^[1] models the physiological effect of current passing through the human body and gives the physiology behind both figures.

• 30 mA.The current threshold at which ventricular fibrillation becomes likely with sustained exposure. Below 30 mA the body can usually let go and recover; above it, the heart's rhythm can be disrupted.
• 100 ms.The disconnect time required to keep that exposure short enough that fibrillation does not occur even at currents well above 30 mA. The standard target for a residential RCD.

Where 10 mA is required, not 30 mA

AS/NZS 3000 ^[2] mandates a stricter 10 mA trip threshold (a third of the standard residential 30 mA) on circuits where children or medical equipment are at risk. The 10 mA RCD is the same protection class, set to a lower trip current because the people on the other side of the wire are more vulnerable.

• Child protection.Final sub-circuits intended specifically to protect children from direct contact must be 10 mA, per AS/NZS 3000.
• Patient areas and home medical care.AS/NZS 3003 [9] requires 10 mA Type I RCDs in designated body-protected areas and for home installations that support mains-powered medical equipment used on a patient.
• Higher-risk residential locations (recommended, not mandated).Bathrooms, outdoors, kindergartens, schools, daycares and damp areas are routinely fitted with 10 mA RCDs as a higher-protection step beyond the 30 mA baseline.

The implication for a default residential switchboard is that the 10 mA upgrade is per-circuit hardware. Once installed, it can't be reconfigured. Section 06 covers how Basis handles this differently.

Two datasets anchor the claim. Italian field data, widely quoted in UK and AU electrical-safety literature^[4], gives a 7.1% failure rate among untested in-service RCDs against the spec in section 02, with 2.8% in the regularly tested subgroup. The UK Electrical Safety Council / ERA Technology 2007 study^[5] tested 607 RCDs in housing-authority homes and found roughly 4% failed when push-button or injection tested. Published research across multiple jurisdictions has reported failure rates in the 3% to 15% range, depending on test method, RCD age, and whether the devices were periodically tested or not.

Two caveats apply. The Italian failed sample may have been weighted toward one low-cost manufacturer, which limits generalisability to NZ stock. And the tested-versus-untested comparison carries selection bias: regularly tested fleets replace failed devices when the failure is detected, so the 2.8% understates the lifetime failure rate.

Why they fail without anyone noticing

• Contact welding.Repeated minor trips can micro-weld the internal contacts. The device looks fine, the push-button moves, but the mechanism no longer opens under fault current.
• Debris ingress.Dust, insect bodies, building debris inside the enclosure can prevent the trip mechanism from completing its travel.
• Deliberate bypassing.Six of the failed UK devices had been deliberately shorted out by occupants or installers to stop nuisance tripping. The protection chain is removed and never reinstated.
• Component drift, the 'lazy RCD'.An RCD whose trip time has drifted to 200 or 400 ms still passes the push-button test (the button only proves it disconnects, not how fast). A drift of 100 ms past the 100 ms target is enough to be lethal.

The push-button test is the only verification most homeowners can run. It detects most catastrophic failures (the device doesn't disconnect at all) but it does not detect drift in trip time. AS/NZS 3760 ^[3] recommends a periodic instrument-based trip-time test in workplaces for exactly this reason.

AS/NZS 3000:2018 ^[2] recommends a 6-month interval for residential RCD push-button testing. Manufacturer product labels typically print "test 6-monthly" or "test monthly" on the front of the device. The Safetag NZ consumer guidance ^[7] reflects the prevailing industry position.

• Recommendation, not legal requirement.The AS/NZS 3000 wording on residential test intervals is framed as guidance, not an enforceable homeowner obligation. The Electricity (Safety) Regulations 2010 [6] cover what counts as an 'electrically unsafe RCD' but do not impose a homeowner test schedule.
• What the push-button proves.Only that the RCD will disconnect on a deliberately-injected fault current. It does not prove that it will disconnect within 100 ms. A 'lazy' RCD passes this test and would still kill at 400 ms.
• What homeowners actually do.Industry surveys put quarterly homeowner test rates below 20%. Most homes have never pressed the button.

Practical translation: the test exists, the manufacturers instruct it, the standard recommends it, and almost no one does it. The protection chain is therefore unverified across most of the installed base.

NZ has approximately 1.9 million residential dwellings. AS/NZS 3000 has mandated RCD coverage on residential lighting and sockets since 2007 in NZ, so essentially every home built or wired since carries RCDs. Older stock has partial coverage from retrofits.

The Electricity (Safety) Regulations 2010, reg 24^[6] defines what constitutes an "electrically unsafe RCD" but enforcement is reactive (after an incident or inspection), not proactive (continuous verification). The verification gap is what Basis closes.

Basis constantly self-tests the protection chain on every circuit and flags issues, without homeowner action. Where a default switchboard relies on a 6-month manual push-button that almost no one runs, Basis runs an instrument-grade verification that actually measures trip time, on a continuous schedule.

• Continuous self-test.Every circuit, constantly. Trip time measured against the 100 ms spec, not just disconnection confirmed.
• Silent-failure detection.If an RCD fails or drifts, the failure surfaces in the app the next morning, not at the moment a person needs the protection.
• Selective trip.Default RCDs cover up to 3 circuits each, the minimum AS/NZS 3000 ^[2] allows, so a single earth-leak knocks out the whole group. Per-circuit detection means only the faulty circuit drops. Removing the nuisance trip removes the homeowner's reason to bypass.
• 10 mA or 30 mA, per circuit, in software.Each Basis circuit supports both trip thresholds. Stepping a bathroom or nursery circuit from 30 mA to 10 mA is a configuration change in the Trade App, not a hardware swap. Mandated 10 mA locations are protected without a separate device class.