Upstream downstream RCD test

Very relevant point. In IT systems, any conductor may become referenced to earth, so on the second fault the RCD must still trip regardless of line/neutral orientation. Residual current sensing is inherently non-directional, although voltage-dependent RCDs may still require correct supply connection for their electronics.

When we use the 0 and 180 test, does this mean it's effectively half wave rectified, or does it just reference the starting direction of the AC waveform?

In this context, the 0° and 180° test is not implying half-wave rectification.

It is simply a phase-angle reference for the AC waveform:

* 0° means the test is initiated with the voltage/current starting in the positive half-cycle.

* 180° means the test is initiated with the waveform starting in the negative half-cycle.

So the purpose is to check whether the equipment under test behaves differently depending on the polarity of the first half-cycle.

If it were half-wave rectified, you would actually be removing one half of the sinewave, producing a DC component — that’s a different phenomenon entirely.

So: it’s about waveform starting direction, not rectification.

Ah, thank you for that clarification about the phasing of the starting point of the 'leakage' current. 

When I pulled apart an old RCD (BG CUR8030; no alpha code), I noticed that internally it has a half wave rectification of the generated differential current transformer (CT) (i.e. it's output) such that only one half cycle of the CT output fed the trip solenoid, no doubt pre-aligned to the magnetisation of the closed trip latch (easier to 'break' that static magnetisation circuit.PDF

This would clearly(?) imply that the reaction to the positive and negative half cycles would be 'different'. Newer devices may use different internal circuits (full wave rectified?)

internally it has a half wave rectification

That doesn't look like a conventional half-wave rectifier circuit to me - if it was I'd expect a diode to be in series with the "load", not in parallel with it. With two of them, both ways round, I wonder it they're more to do with clipping overvoltages and protecting the coil from burning out - although zeners might perhaps be more conventional than IN4004s for that. Hopefully Mike will be along soon to explain...

   - Andy.

It's a (80Amp / 30 mA) RCD.

It took me a while to understand the circuit (too much 'voltage' Thévenin based thinking!). It's first a CT, so needs a deliberate load. Ultimately that load is a pair of 1N4004's. (so ~0.7V type fwd bias)

The trip solenoid has thousands of turns (estimate..) of really fine wire, and just needs break the magnetic loop that's holding the trip closed.

Given the 12Turn step down, It'll be in the low mA range (30mA/12; AC) for the current trip (reminds me of POTS off hook relays which IIRC had 1mA detection.).

That's my understanding, and I'm sticking with it, until a better one gets suggested

Mike is along, and I agree, that is not a normal rectifier - the actuation coil does not see a steady DC, rather something the is almost zero on one half cycle, and quite noticeably negative on the other. Waveforms in microwave ovens are similarly odd.

other phase start

The model. Note if D1 is not a zener then the voltages reached mean the sensing core would saturate horribly - perhaps it does but has already de-latched by then; as it is the integrals are not nice.
  I chose 10mA p-p as being about 3.5mA RMS ,a little over 1/12 of 30mA. The actuator coil resistance is a total guess, but it only alters the amplitude a bit, not the trend we want to understand, that it is not really a rectifier in the normal sense, more of a single sided pulsar.

mapj1 said:

Note if D1 is not a zener then the voltages reached mean the sensing core would saturate horribly

Just to confirm that the diodes are the 1N4004 common place black rectifier diodes.Still have the bits..

There's no indication of normal transformer core saturation just because the diodes are there.

Perhaps you meant that the diodes 'saturate' in forward bias (~1A max withstand peaks up to 30A.). With the 12:1 turns ratio that's a L-N imbalance of 12A 'happily'.

The other option is that the capacitor C1 (Tant bead) is being 'charged up' via D2 (fwd) and then discharges via D1 and the trip solenoid. (the flip-flopping of volts and currents here may have me confused..)

Hmm. been looking again at the micro cct board and interconnection. I could have confused myself about the connections. more soon.

So, I made a mistake. D2 is anti-parallel with D1. Thus the solenoid is fed via C1. Makes more sense,

Though the capacitor reliability may take a slight hit. 99.9% of the time it'll have micro/milli volt signals as the potential 1mA AC (equiv 12mA leakage) keeps tickling the diode forward biases.