How an RCD Detects Earth Leakage: Toroidal Core Principle
How does an RCD detect earth leakage? A residual current device sums the instantaneous current in the live and neutral conductors inside a toroidal core transformer, and per IEC 61008 it must trip once that vector sum — the residual current — reaches its rated operating value IΔn, commonly 30 mA for personal protection. Any current returning to source by a path other than the neutral, through faulty insulation, moisture, or a person's body, unbalances the two conductors and induces a signal in a sense winding that releases the trip mechanism, typically within one AC cycle. This article covers the ring-core transformer itself, the sense winding and trip relay, what "balanced" actually means in a healthy circuit, why a shared neutral defeats the detection, and how detection differs across AC, A, F and B device types.
The Toroidal Core: Summing Current, Not Measuring It
An RCD does not measure current in either conductor directly. Both the line and neutral conductors of the protected circuit pass through the same ring-shaped (toroidal) transformer core, wound in opposite senses relative to each other. In a healthy circuit every ampere that flows out through line returns through neutral, so the two magnetic fields the conductors generate in the core cancel. Net flux is zero, and no voltage appears across any winding sensing that core.
The moment current takes a third path back to source — through earth, a person, or a faulty cable insulation — the two conductor currents stop matching. Line carries more than neutral returns. That imbalance leaves residual flux in the core, and residual flux is the only thing the device is built to notice. It has no opinion on load current, power factor, or harmonics; it watches one number: the difference between what went out and what came back through the same two wires.
The Sense Winding and Trip Relay
A third winding, wrapped around the same core, sits between the two main conductors. Residual flux in the core induces a small AC voltage in this sense winding — microvolts to low millivolts at threshold current, scaled by the number of turns. That signal feeds a sensitive polarized relay (an electromagnetic release) or, in electronic RCDs, an amplifier stage ahead of the same type of release. Once the signal exceeds the relay's pickup threshold, it releases a spring-loaded latch that snaps the main contacts open. The mechanical opening is independent of the electronics from that point on — the spring does the work, not a motor or solenoid holding contacts apart.
Electromechanical RCCBs derive their trip energy entirely from the fault current itself; no external supply is needed to open the contacts, and this is why they keep working through a supply-side outage that itself causes leakage. Electronic RCDs add an amplifier ahead of the same relay to raise sensitivity and enable extra functions (auto-reclose, remote trip signaling), but they typically still fail-safe to trip on loss of control power, per the relevant product standard.
What Counts as "Balanced": The Vector Sum in a Healthy Circuit
In a single-phase circuit, balance means line current in equals neutral current out, instant for instant, not just averaged over a cycle. In a three- or four-pole device (3P or 4P, for three-phase loads), the core has one turn per phase conductor plus neutral, and the balance condition is the vector sum of all phase and neutral currents equal to zero. A three-phase motor with no earth fault balances across its three phases even if the phases individually carry different magnitudes during a transient — what matters is that everything leaving eventually returns through one of the monitored conductors.
Formula: Residual Current Trip Condition — Source: IEC 61008-1, Clause 3 (definitions)
IΔ = |Σ I(line, neutral)| ; device trips when IΔ ≥ IΔn
| Symbol | Description | Unit |
|---|---|---|
| IΔ | Residual (vector sum) current sensed by the toroidal core | mA |
| Σ I(line, neutral) | Sum of instantaneous currents in all monitored conductors (line(s) + neutral) | A |
| IΔn | Rated residual operating current — the device's trip threshold | mA |
What we see in the field: a board with several final circuits each leaking a small, harmless amount through cable capacitance or filter networks can push a single upstream 30 mA RCD close to nuisance-trip territory even with zero actual faults, because the core sums leakage across every downstream circuit it protects, not per-circuit.
Why a Shared Neutral Defeats Detection
The core can only sum what passes through it. If a neutral conductor downstream of the RCD is shared with another circuit that is not routed through the same core — a common wiring mistake in older multiwire branch circuits — some of the current returning from the protected line conductor takes a path outside the core entirely. The device reads that as residual current and can trip on normal load, or in the opposite failure mode, a genuine earth fault current returns partly through the borrowed neutral and never reaches the threshold at all.
The same logic is why the neutral must not be earthed anywhere downstream of an RCD (only upstream, at the main bonding point). An unintended neutral-earth connection past the device gives leakage current a second return route that partially bypasses the core, undermining the vector-sum principle the whole detection method depends on.
Detection Limits by Waveform: AC, A, F and B
The physics above assumes a residual current that actually induces flux the core and sense winding can pick up — and that assumption breaks down for certain waveforms. A Type AC device reliably detects pure sinusoidal AC residual current only. Modern electronic loads (variable speed drives, switch-mode power supplies, LED drivers) can produce a pulsating DC or smooth DC residual component under certain fault conditions, and a pure AC-sensing core can be desensitized or partially blinded by a DC component riding on the fault current — the core saturates asymmetrically and stops responding linearly to the AC component it is built to catch.
Type A devices add detection for pulsating DC residual current superimposed on AC, covering most modern electronics. Type F extends this to mixed-frequency residual currents from single-phase variable frequency drives. Type B is built to detect smooth, non-pulsating DC residual current as well — the only type of construction that reliably clears earth faults from three-phase VFDs, EV chargers, and transformerless PV inverters, where a genuine DC fault current would otherwise saturate a lesser core and prevent tripping altogether. Selecting between these types is a construction question, not a sensitivity question — see RCD types AC, A, F and B explained for the full breakdown, and Type A vs Type B for VFD, EV and solar loads when the downstream equipment is a known DC leakage source.
From Trip Signal to Open Contacts: Why This Matters for Sensitivity Choice
Once the sense winding output crosses the relay's pickup point, everything downstream is mechanical and fast — full disconnection at rated IΔn happens in well under the limits set by IEC 61008 for personal-protection devices. But the pickup threshold itself, IΔn, is a design choice with real consequences: set it too low (10 mA on a circuit with normal cumulative leakage from filters and long cable runs) and nuisance tripping follows; set it too high (300 mA "for fire protection only" on a socket circuit) and it will not protect against electric shock at all. The detection principle is identical across the range — only the threshold and the waveform the core can see change. RCD sensitivity from 10 mA to 300 mA covers how to match IΔn to the application, and how to select sensitivity, type and pole count together walks through the full specification checklist.
This depends heavily on where the device sits in the distribution board. On TT earthing systems, where a high earth-loop impedance means overcurrent devices alone often cannot clear an earth fault fast enough, the RCD's flux-summing detection is the only mechanism guaranteed to clear the fault regardless of loop impedance — see RCDs across TT, TN and IT earthing systems for how the same core-and-relay principle behaves differently depending on the earthing arrangement it protects.
For the broader distinction between an RCD that only senses residual current and one that also carries overcurrent protection in the same housing, see what an RCD is and how it works for the basics, and MCB vs RCBO vs RCD vs RCCB differences for how the detection principle described here fits into the wider family of protective devices. Both are covered in Stoklink's RCD protection guide, and the full range of residual current devices and RCBOs ships with datasheets specifying IΔn, type, and pole configuration.
Frequently Asked Questions
Does an RCD measure the actual amount of current flowing to earth?
No. It measures the difference between the current leaving on the line conductor and the current returning on the neutral conductor. If that difference reaches earth through a fault, insulation breakdown, or a person, the physical path the current took afterward is irrelevant to the core — only the imbalance matters.
Why does an RCD need a neutral conductor through the core, not just the line?
Without the neutral passing through the same core, there is nothing to compare the line current against. The device needs both conductors present to compute the vector difference; monitoring line current alone would just be an overcurrent device, not a residual current device.
Can an RCD detect a fault on the neutral conductor itself?
Yes, if the fault causes current to leave the circuit's monitored conductors and return by another path. A broken neutral with no other current path typically does not trip an RCD by itself, since no imbalance is created — that fault is usually caught by other means, such as loss of supply.
Why do RCDs trip on switch-on for some electronic loads even with no fault?
Inrush current through EMI filter capacitors to earth is a brief, genuine residual current, not a false trip. It is smaller in magnitude and duration than a fault but still registers on the core; equipment with heavy filtering (VFDs, some IT loads) is a known contributor to this.
Does the toroidal core detection principle differ between an RCCB and an RCBO?
No. Both use the same ring-core, sense-winding, and trip-relay arrangement for residual current detection. An RCBO simply adds a separate thermal-magnetic overcurrent mechanism acting on the same set of contacts, alongside the RCD stage.
What standard defines how the residual operating current IΔn is tested?
IEC 61008-1 for RCCBs and IEC 61009-1 for RCBOs define the rated residual operating current and the required trip time at IΔn and at 5 x IΔn; IEC 62423 adds the waveform classification (Type A, F, B) tested during that same trip-time verification.
Conclusion
The toroidal core principle is the entire basis of residual current protection: sum the current through every conductor of a circuit, and trip when that sum stops equaling zero. Everything else — sensitivity class, waveform type, selectivity timing, RCBO versus RCCB construction — is a variation on how that one signal is thresholded and acted on, not a different detection method. Specifying the right device means matching IΔn and type to the load's actual leakage behavior, not just picking a number off a shelf.
