I agree Jon, and a lot of this is based on faulty guesswork as to how these devices operate.

Let us discuss from first principles. Is it possible to get a DC current from an AC supply? This depends on our definition of DC, which is not often clear from RCD manufacturers. My definition is quite simple, DC is a relatively constant voltage or current, which does not vary with time, in other words, a battery. An AC voltage or current is a supply that varies about zero volts in a regular manner in time. Now we get the fuzzy bit, which is an asymmetric AC waveform that is not symmetric about zero and may vary between cycles in some manner. This is the waveform produced by a half-wave rectifier feeding a capacitor, as an example. Current only flows in one direction but is certainly not proper DC as defined above. Next we need to examine the effect of these waveforms on a transformer with two windings connected in opposition. Windings like this give exactly opposite magnetic fields and the net in the core is zero. The result should be that any net magnetic field can be transferred to a third winding by the core, which is then detected as a difference current, and used to make an RCD. Equal DC current in both windings should again cause cancellation of the magnetic field in the core, but any difference will not be transferred to the third winding as transformers need an alternating field to operate. This is the theory.


Now the difficult part, which is how real devices vary from the theory above. It is possible to put two similar windings on a magnetic core in various ways to achieve the theory above. They can be spaced from one another, by a distance as long as they both have the same number of turns, and will work as above with nice waveforms. However, once the waveform has more components at higher frequencies, things get more complex. Magnetic cores are not ideal and have losses, and coils do not couple to them in an ideal way. These two together are usually considered in power engineering as leakage inductance, a measure of how far from ideal a transformer is. It can never be made zero, particularly at low frequencies (50 Hz). It is a measure of how much of the magnetic field fails to couple the two windings and is lost. In an RCD we can wind our a small core with two separate windings or wind them together as a bifilar winding where the two wires are as close to one another as possible on each turn. In RF transformers we also twist the two wires together to improve coupling further and improve the exact balance of the current. As there is a significant voltage between the wires in an RCD, closeness is a problem. The closer the windings the less likely that the waveform will not be transformed as required, or the core cause a problem of coupling. Note a real DC current on L & N should also exactly cancel so the core cannot become saturated.


Next, the DC thing, which seems to focus minds a great deal, I'm not sure why. The question: Can I get a significant DC current through just one winding to saturate the core with a real load? Core saturation prevents normal transformer action and may prevent the difference signal being detected by the sense winding, so prevent operation at least at normal sensitivity. So far I have been unable to come up with any load which can generate such a DC current for more than one cycle, (as may be used in "no trip" tests). Even this is fairly tricky, and certainly nothing like any electronic power supply design. Therefore I consider this DC consideration a rather "RED Herring", and needs a manufacturer to describe how they think this is a problem. I can find no explanation in technical terms anywhere else!


David CEng etc.

I agree Jon, and a lot of this is based on faulty guesswork as to how these devices operate.

Let us discuss from first principles. Is it possible to get a DC current from an AC supply? This depends on our definition of DC, which is not often clear from RCD manufacturers. My definition is quite simple, DC is a relatively constant voltage or current, which does not vary with time, in other words, a battery. An AC voltage or current is a supply that varies about zero volts in a regular manner in time. Now we get the fuzzy bit, which is an asymmetric AC waveform that is not symmetric about zero and may vary between cycles in some manner. This is the waveform produced by a half-wave rectifier feeding a capacitor, as an example. Current only flows in one direction but is certainly not proper DC as defined above. Next we need to examine the effect of these waveforms on a transformer with two windings connected in opposition. Windings like this give exactly opposite magnetic fields and the net in the core is zero. The result should be that any net magnetic field can be transferred to a third winding by the core, which is then detected as a difference current, and used to make an RCD. Equal DC current in both windings should again cause cancellation of the magnetic field in the core, but any difference will not be transferred to the third winding as transformers need an alternating field to operate. This is the theory.


Now the difficult part, which is how real devices vary from the theory above. It is possible to put two similar windings on a magnetic core in various ways to achieve the theory above. They can be spaced from one another, by a distance as long as they both have the same number of turns, and will work as above with nice waveforms. However, once the waveform has more components at higher frequencies, things get more complex. Magnetic cores are not ideal and have losses, and coils do not couple to them in an ideal way. These two together are usually considered in power engineering as leakage inductance, a measure of how far from ideal a transformer is. It can never be made zero, particularly at low frequencies (50 Hz). It is a measure of how much of the magnetic field fails to couple the two windings and is lost. In an RCD we can wind our a small core with two separate windings or wind them together as a bifilar winding where the two wires are as close to one another as possible on each turn. In RF transformers we also twist the two wires together to improve coupling further and improve the exact balance of the current. As there is a significant voltage between the wires in an RCD, closeness is a problem. The closer the windings the less likely that the waveform will not be transformed as required, or the core cause a problem of coupling. Note a real DC current on L & N should also exactly cancel so the core cannot become saturated.


Next, the DC thing, which seems to focus minds a great deal, I'm not sure why. The question: Can I get a significant DC current through just one winding to saturate the core with a real load? Core saturation prevents normal transformer action and may prevent the difference signal being detected by the sense winding, so prevent operation at least at normal sensitivity. So far I have been unable to come up with any load which can generate such a DC current for more than one cycle, (as may be used in "no trip" tests). Even this is fairly tricky, and certainly nothing like any electronic power supply design. Therefore I consider this DC consideration a rather "RED Herring", and needs a manufacturer to describe how they think this is a problem. I can find no explanation in technical terms anywhere else!


David CEng etc.