Type B time delayed rcd

The hidden question is should it really have tripped or should it have ignored it ?

DC blinding is one thing, and larger RCDs are harder to blind, as the cores are larger area, but with switching electronics there are other possible factors, and RCD specs are not written in a way that defines 'correct' behaviour for all cases.

Very short duration high amplitude spikes may be perfectly safe - that electrostatic 'nip' you get from combing a cat or running on nylon carpet may be a few kV and tens of amps flow though your finger tip, but it is only there for 10 microseconds and produces no significant heat, and no burns or life threatening muscle twitch - should an RCD respond to non- dangerous events that or not ?


A switching supply or inverter may well introduce just such a glitch every time the internal switching transistors reach a particular point in their cycle  of changing states.  That may or may not mean there is a problem that needs a trip.


Of course if the things were properly filtered then the short spikes of kV and microseconds on the dirty side of the filter would be spread out into milliseconds of a few volts and milliamps on the clean side, so the RCD wold then not see it, but the "value engineering" teams have taken out any filtering that is not mandatory smaller lighter cheaper product is always a competetive advantage. Making it work is someone else's problem.


In terms of 50Hz leakage it is perfectly possible to differentiate between capacitance between L and E,  which is leakage through filters and so  on, and benign, and resistance which could be a human, but no RCD I know of does look at the I-V phase shift to do this, and if it did it probably would not meet any of the product standards.

In the example of the no-trip 30mA and the firing 100mA, if it was  DC flowing then I agree that is very bad, and the 30mA is being blinded.  If it was short duration switching spikes, then it is not clear which if any RCD should have tripped. The problem is, rather like the conducted emmissions question of the LED drivers in this thread , without kit that most folk do not have, you can only really 'try another model' and see if it is any better, which can be very expensive and unsatisfactory, a bit like proposing to find the deepest crater on the moon by walking round with a bag over your head...

The hidden question is should it really have tripped or should it have ignored it ?

DC blinding is one thing, and larger RCDs are harder to blind, as the cores are larger area, but with switching electronics there are other possible factors, and RCD specs are not written in a way that defines 'correct' behaviour for all cases.

Very short duration high amplitude spikes may be perfectly safe - that electrostatic 'nip' you get from combing a cat or running on nylon carpet may be a few kV and tens of amps flow though your finger tip, but it is only there for 10 microseconds and produces no significant heat, and no burns or life threatening muscle twitch - should an RCD respond to non- dangerous events that or not ?


A switching supply or inverter may well introduce just such a glitch every time the internal switching transistors reach a particular point in their cycle  of changing states.  That may or may not mean there is a problem that needs a trip.


Of course if the things were properly filtered then the short spikes of kV and microseconds on the dirty side of the filter would be spread out into milliseconds of a few volts and milliamps on the clean side, so the RCD wold then not see it, but the "value engineering" teams have taken out any filtering that is not mandatory smaller lighter cheaper product is always a competetive advantage. Making it work is someone else's problem.


In terms of 50Hz leakage it is perfectly possible to differentiate between capacitance between L and E,  which is leakage through filters and so  on, and benign, and resistance which could be a human, but no RCD I know of does look at the I-V phase shift to do this, and if it did it probably would not meet any of the product standards.

In the example of the no-trip 30mA and the firing 100mA, if it was  DC flowing then I agree that is very bad, and the 30mA is being blinded.  If it was short duration switching spikes, then it is not clear which if any RCD should have tripped. The problem is, rather like the conducted emmissions question of the LED drivers in this thread , without kit that most folk do not have, you can only really 'try another model' and see if it is any better, which can be very expensive and unsatisfactory, a bit like proposing to find the deepest crater on the moon by walking round with a bag over your head...