RCD Won't Reset: Troubleshooting Earth Leakage

What causes an RCD that won't reset? An RCD that trips again the instant you flip it back on — or a toggle that simply refuses to hold in the ON position — is reporting a residual current at or above its trip threshold IΔn, commonly 30 mA, that is still present on the circuit; it is not a random nuisance event. The device is doing exactly what IEC 61008 designed it to do: it will not stay latched closed while an imbalance between line and neutral current is flowing to earth, whether from a failed appliance, wet cable insulation, or a wiring fault in the fixed installation. This article works through the diagnostic sequence in the order an electrician should actually run it: isolating the load, checking for cumulative leakage, ruling out wiring faults, testing the earth path, and identifying when the RCD itself — not the installation — has failed.

Reset vs. Trip: What the Toggle Is Actually Telling You

Inside the device, the contacts are held closed by a spring-loaded latch. The sense winding on the toroidal core only matters at the moment of imbalance — once a fault current above IΔn is detected, the trip coil releases the latch and the spring throws the contacts open. Reset re-engages that latch mechanically. If the imbalance is still present the instant the contacts close, the sense winding sees it immediately and the trip coil fires again — often within one or two AC cycles, so fast it can feel like the switch "won't go up" at all.

Two distinct symptoms get lumped under "won't reset," and they point to different faults. First: the toggle goes up cleanly, holds for a fraction of a second to a few seconds, then trips — a live leakage path exists somewhere downstream. Second: the toggle physically won't stay up even with every downstream circuit breaker switched off — the fault is either in the fixed wiring between the RCD and the first downstream device, in the incoming supply itself, or the RCD's latch is mechanically seized. Getting this distinction right before you start pulling breakers saves a lot of wasted time.

Step One: Isolate the Load Before You Blame the RCD

Switch off every downstream MCB or fused way fed by the RCD, then attempt to reset it with nothing connected. If it holds cleanly, the fault is downstream of the RCD — in one of the load circuits, not in the device or the supply-side wiring. If it still won't reset with every downstream way open, the fault sits between the RCD terminals and the first point of isolation: fixed cabling, the neutral bar, or the incoming tails.

Once it holds with everything off, reconnect one circuit at a time and reset between each. The circuit that trips it is your fault circuit. Don't skip this step by guessing based on "what usually causes it" — a kitchen ring circuit and a workshop machine circuit fail for completely different reasons, and testing in sequence turns a 20-minute guessing exercise into a two-minute isolation.

Cumulative Leakage: The Most Common Reason an RCD Won't Reset

Every switch-mode power supply, EMC filter, and long run of cable contributes a small standing leakage current to earth — typically well under a milliamp per device individually. Put twenty variable-speed drives, PCs, and LED drivers on one 30 mA RCD and the vector sum of all that standing leakage can climb into single-digit milliamps before a single fault exists. Add one marginal fault — a slightly damp gland, a nicked cable insulation — and the total crosses IΔn. The RCD isn't malfunctioning; it's summing correctly.

What we see in the field: boards that worked fine for years start tripping or refusing to reset after a handful of new electronic loads get added, with nobody touching the wiring at all. The fix in that case is rarely "replace the RCD" — it's splitting the board across two or more RCDs so the standing leakage budget per device drops back under threshold, or moving to Type A/Type F devices sized with enough margin for the connected electronic load.

On TT installations, this relationship is why a marginal earth fault can produce exactly the "won't hold" symptom rather than a clean trip-and-stay-off: if R_A is measured close to the limit for the fitted IΔn, the loop has almost no spare margin, and any additional leakage — cumulative or fault-related — pushes it over instantly on every reset attempt. Measuring R_A with a proper earth loop or electrode resistance tester is part of a full diagnosis, not an optional extra, whenever the board is on a TT earthing system.

Wiring Faults That Prevent an RCD From Resetting Even With Loads Off

If the RCD won't reset with every downstream way open, look at the fixed wiring itself. A neutral-earth cross-connection — neutral bonded to earth downstream of the RCD, sometimes from a miswired socket or an old ELCB-era bonding link left in place — creates a permanent imbalance path that has nothing to do with connected load. Shared or "borrowed" neutrals across two circuits on multi-way distribution boards are another common cause: current returns via a neutral that belongs to a different way than the RCD is monitoring, and the device sees an apparent imbalance that no amount of load-shedding will fix.

A tell: if the RCD trips or won't reset even with the main incomer isolated and the neutral disconnected at the bar, the fault current path is inside the cable run itself, not through any load. That points to insulation breakdown in the cable — moisture ingress, rodent damage, or a nicked sheath against a metal conduit — rather than anything a socket outlet or appliance is doing.

When the RCD Itself Has Failed Mechanically

RCDs are electromechanical devices, and the latch does wear. Corrosion in damp switchrooms, dust ingress on the reset mechanism, or simply thousands of trip cycles over a device's service life can seize the latch so it won't hold even with zero current flowing through it — supply disconnected, load disconnected, everything open. That's a bench test, not a site test: isolate the RCD completely from both supply and load, then attempt reset. If it still won't stay closed, the device has failed and needs replacing; no amount of wiring investigation will fix a seized latch.

Test-button behavior narrows this further. A device that trips cleanly on the test button but then won't reset afterward usually has a healthy trip coil and sense winding — the fault is external. A device that won't even trip on the test button, or that clatters without a clean snap action, has an internal fault in the trip mechanism itself and should not be relied on for personal protection regardless of the wiring diagnosis. This is also the point at which a proper instrument test — measuring actual trip time at IΔn, not just the go/no-go test button — separates a borderline device from a genuinely failed one; see our guide on how to test an RCD for the ramp-test procedure.

Step-by-Step Diagnostic Procedure

Run through this in order rather than jumping to conclusions:

1. Note the exact behavior — instant re-trip on reset versus won't hold at all with everything off. This alone tells you whether to look downstream or upstream.
2. Switch off every downstream MCB/fused way, then attempt reset with nothing connected.
3. If it holds: reconnect circuits one at a time, resetting between each, until you find the offending way.
4. On the offending circuit, unplug individual appliances — a single failed appliance (commonly a washing machine heater element, an outdoor pump, or an aging extension lead) accounts for a large share of nuisance trips.
5. If it won't hold even with everything downstream open: check for neutral-earth faults and shared neutrals in the fixed wiring before assuming the RCD has failed.
6. On TT systems, measure earth electrode resistance R_A and confirm it against the R_A × IΔn ≤ 50 V limit for the fitted device.
7. Bench-isolate the RCD completely (supply and load both disconnected) and attempt reset — if it still won't latch, replace the device.

This same load-isolation logic is what separates a "won't reset" fault from ordinary repeated tripping once the circuit is live again — the two symptoms often share a root cause but call for slightly different next steps, and it's worth reading both if the board has a history of intermittent trips as well as a hard "won't reset."

Conclusion

An RCD that won't reset is a symptom, not a diagnosis. Work through load isolation first, check for cumulative standing leakage on boards carrying electronic loads, rule out neutral-earth faults and shared neutrals in the fixed wiring, confirm earth electrode resistance on TT systems, and only replace the device once a full bench isolation test shows the latch itself has failed. Panel builders and maintenance electricians who keep residual current devices and RCBOs on hand in the right sensitivities and types replace a confirmed failure in minutes rather than waiting on a parts order. For the broader selection and sizing background behind everything covered here, see the RCD protection guide, and for the distinction between RCD, RCCB, MCB and RCBO terminology used throughout this diagnosis, see our guide on MCB vs RCBO vs RCD vs RCCB differences.