The current has to have somewhere to go
A residual current device measures a difference, not a fault. It compares the current leaving on the line conductor with the current returning on the neutral; thirty milliamps of difference is read as current that found another route, through a body or wet ground or a steel frame. The device is only as good as the path a fault current can take.
On a diesel generator that path is usually where the plan says. A local source, an output referenced to the protective conductor, an electrode near the compound, an arrangement written down. Not elegant, but documented.
A battery unit is a different source, and the difference is electrical. The output is generated electronically, and the unit stands on its own wheels. Whether the output neutral is bonded to the protective conductor at the socket is a design decision, and it settles whether a single fault to earth produces any residual current at all. Get it in writing.
One device type does not cover every source
Type AC responds to sinusoidal residual current and nothing else. Type A adds pulsating DC components. Type B exists because smooth DC residual current can saturate the sensing core of an AC or A device and leave it unable to trip at all — the familiar cases are drives, chargers and large rectifiers.
An inverter output is not a rotating machine. It is switched, filtered and regulated; a pure sine with total harmonic distortion below 3 % describes the shape at the socket, not a fault at the terminals. What matters is whether one device type was assessed against one source, and who signed it off.
The weak answer names a type without naming a source. If the reply to your question is "Type A is fine", ask which source, which fault and which test produced that sentence. If nobody can answer, you have located the gap rather than closed it.
What the unit carries, and where that stops
The numbers to plan around are the ones at the socket. A P25 gives 2,500 W continuous and 5,000 W peak for up to 10 s from 2.7 kWh at 29 kg, on 230 V 16 A single phase. A P25-D makes it 5,000 W continuous, 10,000 W peak, 5.4 kWh, roughly 60 kg, on 230 V 20 A with the parallel box. A P25-T reaches 7,500 W continuous and 15,000 W peak from 8.1 kWh at roughly 90 kg, on 400 V three-phase at 10 A.
Above 16 A the protection question gets harder. A 160–200 A class inverter welder on 400 V does not draw a smooth current, and the pattern during an arc start is not what a 30 mA socket device was built for. That is a statement about the load, and the reason three-phase work needs its own protection decision, not an inherited one.
The chain is short and every link counts. Source, protective device, cable, tool. The enclosure rating does not enter into it: IP54 describes resistance to splashing water, not a certification.
Two measurements before the handover
Insulation resistance first, on the equipment, before it is connected. A 500 V test on tool and lead shows whether the fault path already exists. No device type compensates for a wet motor that is leaking today.
Trip time second, at the real source, under the real load. An RCD tester at the socket with the unit running measures what the installation does; the test button measures the device. Two different questions, and only one is asked on site.
The third item is paper, and it is the cheapest. A written statement of how the output neutral is referenced, which device type the manufacturer considers suitable, and the CE documentation under the EMC Directive 2014/30/EU — test report A2409245-C01-R01-V1 on request. For transport there is UN 3480 Class 9 packaging, UN38.3 and the MSDS. It is what an insurer asks for first.
What this is not
This is not a UPS. There is a gap between the mains dropping and the unit being connected; only the C24 is specified with 10 ms or less.
The unit does not replace the earthing of the position and cannot be tied into a building's fixed distribution. It supplies its own sockets, is not explosion proof and is not submersible. Nor does it modernise an old device: a Type AC unit from the generator years is still the same device.
And it does not decide the question for you. We can state what the output delivers and what the enclosure is rated for; the protection concept belongs to whoever signs it.
Send us the load list, the device type printed on the RCD and the earthing arrangement at the position. We will work through the configuration at https://www.hitama.net/en/contact.