DC on AC supply

I've not yet found a 'voltage independent' (passive, non-electronic) RCD that doesn't have the permanent-magnet mechanism at its core. This is used in order to operate with low power. The description Mike mentioned is in a long posting in this thread's second page (linking directly to an individual post doesn't seem to work).  RCDs that use the supply as a power source are able to have a more MCB-like mechanism, except that it's a long thin coil across the supply instead of a short thick coil in series.


About Dave's comment: I agree that for a plain ac residual current a bit of superimposed dc won't make much difference to type-A tripping current beyond the level of the dc (approximately), if one considers how typical actual RCDs are made. However, IEC61008's requirement for dc current tolerance is concerned with the case where a 'pulsating dc' fault happens in the same direction as the pre-existing smooth dc. This is the only case tested in that standard with the 6 mA dc.  It's the case where there's a possibility of a few mA of dc making a big difference to the tripping threshold for the pulsating dc. RCDs differ somewhat, but I'd be impressed if you find one where the true-rms tripping value doesn't greatly exceed the rated value (or no trip happen) with e.g. 20 mA dc slowly applied and then a half-wave rectified signal in the same direction.


A generally accepted definition of 'dc' appears beyond what humans can agree on! The issue comes up in many situations outside RCDs, and people take definitions to suit the purpose. In one sense a half-wave rectified waveform is not dc, because it's not a constant value. In another way, if it's a repeating waveform with a non-zero mean then one part of the 'spectrum' of harmonics that you'd need in order to reproduce this waveform must be a constant value ('smooth dc'), added with the fundamental (50 Hz) and any 2nd, 3rd etc harmonics - so the waveform contains a dc component. 

If the waveform described the voltage applied to a transformer or inductor, it would matter a lot: the 'dc component' in the waveform would force the current more and more in one direction, until the current was limited more by the winding resistance than by the inductance. This is, for example, what happens with GIC, when a transmission system at e.g. 400 kV gets a much much smaller dc voltage induced in it (e.g. tens of volts), which pushes enough current through transformers to saturate them, increase their reactive power enormously, and cause power-system disturbances.

When the waveform describes a current - as with the residual current in RCDs - the half-rectified waveform doesn't mean anything special unless there's a saturable magnetic material involved (relevant to type-A RCD with pre-existing dc and a pulsating-dc fault in the same direction, so the core could stay in saturation and give negligible signal on its secondary), or a hard magnetic material that will stay magnetized until pushed in the opposite direction (relevant to type-AC with pulsating-dc, at least if in the 'wrong' direction). 

I've not yet found a 'voltage independent' (passive, non-electronic) RCD that doesn't have the permanent-magnet mechanism at its core. This is used in order to operate with low power. The description Mike mentioned is in a long posting in this thread's second page (linking directly to an individual post doesn't seem to work).  RCDs that use the supply as a power source are able to have a more MCB-like mechanism, except that it's a long thin coil across the supply instead of a short thick coil in series.


About Dave's comment: I agree that for a plain ac residual current a bit of superimposed dc won't make much difference to type-A tripping current beyond the level of the dc (approximately), if one considers how typical actual RCDs are made. However, IEC61008's requirement for dc current tolerance is concerned with the case where a 'pulsating dc' fault happens in the same direction as the pre-existing smooth dc. This is the only case tested in that standard with the 6 mA dc.  It's the case where there's a possibility of a few mA of dc making a big difference to the tripping threshold for the pulsating dc. RCDs differ somewhat, but I'd be impressed if you find one where the true-rms tripping value doesn't greatly exceed the rated value (or no trip happen) with e.g. 20 mA dc slowly applied and then a half-wave rectified signal in the same direction.


A generally accepted definition of 'dc' appears beyond what humans can agree on! The issue comes up in many situations outside RCDs, and people take definitions to suit the purpose. In one sense a half-wave rectified waveform is not dc, because it's not a constant value. In another way, if it's a repeating waveform with a non-zero mean then one part of the 'spectrum' of harmonics that you'd need in order to reproduce this waveform must be a constant value ('smooth dc'), added with the fundamental (50 Hz) and any 2nd, 3rd etc harmonics - so the waveform contains a dc component. 

If the waveform described the voltage applied to a transformer or inductor, it would matter a lot: the 'dc component' in the waveform would force the current more and more in one direction, until the current was limited more by the winding resistance than by the inductance. This is, for example, what happens with GIC, when a transmission system at e.g. 400 kV gets a much much smaller dc voltage induced in it (e.g. tens of volts), which pushes enough current through transformers to saturate them, increase their reactive power enormously, and cause power-system disturbances.

When the waveform describes a current - as with the residual current in RCDs - the half-rectified waveform doesn't mean anything special unless there's a saturable magnetic material involved (relevant to type-A RCD with pre-existing dc and a pulsating-dc fault in the same direction, so the core could stay in saturation and give negligible signal on its secondary), or a hard magnetic material that will stay magnetized until pushed in the opposite direction (relevant to type-AC with pulsating-dc, at least if in the 'wrong' direction).