BS 7671 asks for one instrument test on an RCD: an alternating current at the rated residual operating current (IΔn), with the device tripping within 300 ms, or between 130 and 500 ms for an S-type. That has been the requirement of Regulation 643.8 since Amendment 2 was published on 28 March 2022, and Amendment 4:2026 keeps it. The half-times, five-times and 0°/180° tests many of us were trained on are no longer required; they are optional fault-finding tools. This guide covers the test as it stands, why it changed, what the RCD types mean on site, and how to code RCD problems on an EICR.
- One required test: a.c. at 1 × IΔn. Maximum 300 ms for a general non-delay RCD; 130 ms minimum to 500 ms maximum for an S-type.
- The test is the same for Type AC, A, F and B devices. Set the instrument to its a.c. (Type AC) setting whatever the RCD type.
- The ½ ×, 5 × and both-half-cycle tests are optional. The IET lists them as useful for fault finding.
- The test button is a separate functional check under Regulation 643.10 and has its own column on the schedule of test results.
- Type AC RCDs are restricted to fixed equipment with no d.c. components (Regulation 531.3.3). A Type AC where a Type A is needed is a C3 in Electrical Safety First Best Practice Guide 4.
What RCD test does BS 7671 require now?
Regulation 643.8 covers RCDs used for additional protection, and the note to Regulation 643.7.1 covers RCDs used for fault protection. Both say the same thing: effectiveness is deemed verified where the RCD disconnects within the stated time on an alternating current test at IΔn, regardless of RCD type. Regulation 415.1.1 defines the device (IΔn not exceeding 30 mA); 643.8 says how you prove it works.
| Test | Before A2:2022 | A2:2022 onwards, including A4:2026 |
|---|---|---|
| ½ × IΔn, no trip | Routine practice from guidance | Optional |
| 1 × IΔn, general RCD | 300 ms maximum | Required: 300 ms maximum |
| 1 × IΔn, S-type RCD | 130 to 500 ms | Required: 130 to 500 ms |
| 5 × IΔn, 30 mA devices | 40 ms maximum (note to 643.8) | Not required |
| 0° and 180° half-cycles | Both, longer time recorded | Not required |
| Instrument waveform setting | Matched to the RCD type | a.c. setting, whatever the type |
Table 3A of Appendix 3, which used to carry the time/current criteria from the product standards, was deleted by Amendment 2. If your certificates still show a five-times result in the RCD column, that is the wrong figure for the form: the model schedule asks for the disconnection time at IΔn. Our schedule of test results guide goes through the RCD columns.
A2:2022+A3:2024 is withdrawn on 15 October 2026, leaving A4:2026 as the only current edition. Nothing in A4 brings the old three-test routine back, and IET Guidance Note 3 is now in its 10th Edition, aligned to A4.
Why was the five-times test dropped?
Michael Peace's article in the IET's Wiring Matters (issue 91, July 2022) gives the reasoning. A sub-committee of JPEL/64 reviewed the RCD product standards and found the test parameters too complex for anyone on site to know which test applied to which device. Manufacturers already type-test every design, so the committee agreed the field test could be greatly simplified.
There was a practical problem behind it. NICEIC reported in June 2021 that some Type A RCDs of 30 mA or less are only required by BS EN 61008-1 to reach the 40 ms break time at 250 mA, not at 150 mA. Wiring Matters adds that a five-times test on the instrument's Type A setting gives 210 mA of pulsed current, where the product standard uses 350 mA. Healthy devices were failing a test they were never built to pass.
The single a.c. test at IΔn removes that trap. It applies to every type and gives one number to record.
How do you carry out the test, step by step?
- Confirm earth fault loop impedance first. You want a sound earth path before putting a fault current down it. See earth fault loop impedance testing.
- Read the device: IΔn, type, and whether it is S-type. Set the instrument to that IΔn, the 1 × test and the a.c. setting.
- Connect on the load side of the RCD, between line and earth. In practice that is usually a socket-outlet or a convenient point on the circuit.
- Unplug or isolate electronic loads where you can. The IET warns that downstream electronic equipment and the capacitance of long cable runs can affect the result.
- Run the test and record the time in milliseconds.
- Press the integral test button, confirm the device trips, reset it and tick the test button column.
Wiring Matters notes that, strictly, the test belongs at the RCD with the outgoing wiring disconnected, and that this is not usually done. Treat that as the fallback: if a result looks wrong at a socket, isolate, disconnect the outgoing conductors and test at the device terminals before condemning it.
Where one RCD covers several circuits, test it once and enter the same time against each circuit it protects. On an RCBO board, test every RCBO from its own circuit. Our RCBO vs split-load guide covers the layouts.
On a TT system, Table 41.1 gives 0.2 s for some final circuits, yet the test limit is still 300 ms. The IET treats that as acceptable: a real earth fault produces a residual current well above IΔn, and an RCD operates faster as the current rises. BS EN 61008-1 allows a general device 300 ms at IΔn, 150 ms at 2 × IΔn and 40 ms at 5 × IΔn.
A good modern 30 mA device will usually trip in a few tens of milliseconds at IΔn, well inside the limit. Older installations may have BS 4293 devices, a withdrawn standard that specified 200 ms at IΔn, and earlier editions of Guidance Note 3 applied that figure to them.
Which RCD types will you meet, and does the type change the test?
The type describes which residual current waveforms the device can detect. It does not change the required test, but it has to suit the load and it is recorded on the schedule.
| Type | Detects | Where you see it |
|---|---|---|
| AC | Sinusoidal a.c. only | Older boards; fixed loads with no electronics |
| A | a.c. plus pulsating d.c. | The everyday choice for circuits with electronic loads |
| F | As Type A, plus mixed-frequency residual currents | Single-phase inverter-driven appliances and some heat pumps |
| B | As Type F, plus smooth d.c. | Three-phase drives, some EV charging points and PV inverters |
Regulation 531.3.3 says a Type AC RCD shall only be used to serve fixed equipment where it is known that the load current contains no d.c. components. The IET's FAQ on consumer units explains why: residual d.c. from electronic equipment can blind a Type AC device and stop it operating.
Amendment 4:2026 adds a note to 531.3.3 on RCDs in series. A Type AC should not sit upstream of a Type A, F or B; a Type A should not sit upstream of a Type F or B. That matters when you add a Type B for a charging point downstream of an existing Type A main switch: check the manufacturer's data before relying on the combination. A4 also asks designers to consider unidirectional and bidirectional devices (Regulation 530.3.201): unidirectional RCCBs and RCBOs are now marked in/out, line/load or with arrows, which is worth checking wherever PV or battery storage can feed back through the device. Section 722 has its own d.c. protection rules, covered in our EV charger certificate guide.
What does the test button prove?
The button switches a resistor between line and neutral on opposite sides of the sensing coil, which creates an imbalance, so it proves the mechanism moves. It does not time the trip and it does not prove the device operates at IΔn. Regulation 643.10 requires the test facility to be verified, which is why the schedule has a separate tick column next to the disconnection time.
The user notice changed with Amendment 2 as well. Regulation 514.12.2 now asks for the button to be pressed six-monthly, where it used to say quarterly. The IET suggests tying it to the clock changes, which is an easy line to give a customer at handover.
When are the ramp, half-times and five-times tests still worth doing?
Wiring Matters lists the ½ × test, the 5 × test, the type-specific waveform tests, the ramp test and the 0°/180° phase angle tests as additional tests that are not precluded and may be useful for fault finding.
The ramp test earns its place on nuisance-tripping calls. A 30 mA RCD must not trip at half IΔn and must trip at IΔn, so a healthy device ramps out somewhere between about 15 and 30 mA. If it goes at 12 mA with the loads connected, the device is probably fine and the circuit is carrying standing leakage. Unplug loads and repeat, or clamp the leakage. If the pattern points to the wiring, see low insulation resistance troubleshooting.
Why does a healthy RCD fail the test?
The IET's view is blunt: many of the problems met when testing RCDs are down to user error, not faulty devices. Check these before replacing anything.
| What you see | Likely cause | What to do |
|---|---|---|
| Trips as soon as the test starts, or on the instrument's half-rated pre-check | Standing earth leakage from connected loads | Unplug loads, retest; ramp test to confirm |
| Slow or inconsistent times | Electronic loads or long cable capacitance downstream | Disconnect loads; retest at the device with outgoing wiring off |
| No trip at IΔn | Wrong IΔn selected, wrong waveform setting, or a seized device | Check settings, then retest at the device terminals |
| S-type trips in under 130 ms | Device is not time-delayed as labelled, or the wrong device was fitted | Check the marking; selectivity with downstream RCDs is lost |
| Button trips it, instrument does not | Mechanism works but sensitivity has gone | Replace the device |
How do you code RCD problems on an EICR?
Electrical Safety First's Best Practice Guide 4 (Issue 7) gives these classifications for domestic and similar installations.
| Finding | Code |
|---|---|
| Main RCD on a TT system fails to operate on the instrument or its test button | C2 |
| No RCD for mobile equipment reasonably expected to be used outdoors | C2 |
| No RCD for a socket-outlet circuit in a room with a bath or shower | C2 |
| No RCD for a socket-outlet unlikely to supply outdoor equipment | C3 |
| No RCD for domestic lighting circuits | C3 |
| No RCD for cables less than 50 mm deep in a wall, with no other protection | C3 |
| Type AC RCD where a Type A is required | C3 |
The guide has no separate line for a 30 mA additional-protection RCD that is slow or dead. The reasoning most inspectors use is that a device that does not work is the same as no device, so code it as you would its absence, and write the measured time in the observation. More examples are in our EICR codes guide.
How Certio helps
Certio produces the EICR, EIC, Minor Works certificate and PAT records on the BS 7671 model forms and turns them into branded PDFs. Photograph the consumer unit and the circuit schedule drafts itself; point the phone camera at a Megger MFT-X1 and the reading goes straight into the schedule of test results, or dictate it while your hands stay on the probes. Suggested observation wording with the regulation reference is there for you to check and confirm. The judgement stays yours; the typing goes. Starter is free for 7 days, then £4.99 a month.