Push-button vs trip-time RCD testing — what is the difference?

Almost every building in New Zealand has someone who believes the RCDs are tested because a maintenance person presses the buttons twice a year.

That belief is the most dangerous piece of misplaced confidence in workplace electrical safety, and this page exists to take it apart.

The two tests do different jobs

Both have a place. The mistake is treating the first as a substitute for the second.

What the push button actually does

The test button creates a small artificial imbalance inside the device, which the RCD detects and responds to by tripping.

That tells you three genuinely useful things:

• The device is in circuit and energised
• The detection and tripping mechanism operates
• The device has not seized, which mechanical devices sitting untouched for years genuinely do

That last one is not trivial. An RCD that has not moved in six years can stick. Pressing the button exercises it, and there is a real argument that the mechanical exercise is as valuable as the test.

What the button does not tell you:

• How long the device took to trip
• Whether that time is within the regulatory limits
• Whether the trip time has been getting worse
• Whether the device is the correct type for the loads on the circuit
• Whether it will still respond to a real earth fault, as opposed to the internal test circuit

The button answers "did it trip". The regulations care about "how fast".

What trip-time testing measures

A trip-time test uses a calibrated instrument to inject a known residual current and measure the disconnection time.

Typically that means measuring at the rated residual current, and again at a multiple of it, and recording the actual times in milliseconds. Those numbers get compared against the limits that flow from regulation 24 of the Electricity (Safety) Regulations 2010 — broadly, interruption within 300 milliseconds at rated residual current, or within 40 milliseconds at five times rated.

The output is a number per device. That is the difference. A number can be recorded, compared against a limit, and — most usefully — compared against last year's number for the same device.

The failure mode nobody sees

Here is the scenario that makes this worth a page.

An RCD in a workshop board has been in service for a decade. Someone presses the button every six months and it trips, so it goes in the log as tested and working. Nobody has ever measured it.

Its trip time has drifted well beyond the regulatory limit. The button test still trips it, because the internal test circuit produces a clean, generous fault that the device can respond to eventually. The log says "OK" every six months for a decade.

Then somebody puts a hand on a faulty grinder in a damp yard.

The RCD trips. It just takes too long, and the whole point of a 30 mA device with a defined disconnection time is that the time is what keeps the person alive.

Nothing in that story involves negligence by anyone. The maintenance person did what they were asked. The log is honest. The failure is in the belief that a button press is a test.

Who does which test, and how often

Push-button testing should be done by the people who use the site, not by a contractor on a schedule. It takes seconds, needs no tools, and its value is entirely in frequency.

• Fixed RCDs in a benign environment: six-monthly is a common baseline, and quarterly is better.
• Fixed RCDs in wet, outdoor or industrial areas: more often.
• Portable RCDs on construction sites: commonly daily, or before each use.

Two practical points. First, tell people what to expect — an RCD that does not trip on the button is a fault to report, not something to press harder. Second, warn them what goes off. Pressing an RCD that feeds a server, a chiller or a process line at 11am on a Tuesday teaches everyone the wrong lesson about testing.

Trip-time testing is periodic, needs a calibrated instrument, and needs someone who can interpret the result. Intervals depend on environment and criticality — see how often RCDs should be tested.

What changed in the cited standard

This is where the argument stops being ours and becomes the regulator's.

The Electricity (Safety) Amendment Regulations 2025, in force from 13 November 2025, cite AS/NZS 3000:2018 with a set of New Zealand modifications. One of those modifications is to clause 8.3.10, which deals with RCD testing during verification.

Two changes matter here.

The words "either by the operation of the integral test device, or" were deleted. In other words, the option of verifying by the test button alone was removed from that clause.

A list of what function testing must include was inserted, covering trip time, confirmation that the RCD's functions operate as intended, confirmation that its operation provides protection against electric shock, and — for residual sinusoidal alternating current and residual pulsating direct current — the presence of DC leakage and constant DC where applicable.

That last element — DC leakage and constant DC — is not academic. It is there because modern electronic loads inject DC components that can blind an older Type AC device entirely. See RCD types and the 2026 changes.

What a real RCD test record looks like

If someone hands you an RCD test record, this is what separates a real one from a decorative one. It should show, per device:

• Device identifier and the circuits it protects — surprisingly often unknown
• Device type (AC, A, F, B) and rated residual current
• Measured trip time in milliseconds, at each test current used
• The limit it was measured against
• Pass or fail, derived from the numbers rather than instead of them
• Date, tester and instrument, with the instrument's calibration status
• Last year's figures alongside, so the trend is visible

If the record says "RCD test: pass" with no numbers, you have a receipt, not a result. Ask for the readings — any competent tester will have them, because the instrument recorded them whether or not anyone printed them.

Related: when are RCDs required in New Zealand and why an RCD keeps tripping.