DC on AC supply

Weirdbeard:

What I have wondered with regards this situation is if an installation has equipment that can apparently disable commonly found RCDs, would this disablement also apply to other installations that share a supply, for example with looped supplies? 

This is a question I got elsewhere a year ago, mainly about worries with solar PV installations and EV chargers affecting other customers on the network by their possible dc, and so a need to have type-B RCDs as default regardless of whether the installation included such devices. The answer: practically, no.    [edit: note that - in view of the thread's topic - this is about  disablement by dc currents, not e.g. by high frequencies]


Put more generally, it's a question of whether dc(residual)-causing apparatus could affect RCDs that have that apparatus [edit, for clarity: load->apparatus] on their supply side instead of their load side. Various complicated situations can be found where it could matter, but they involve multiple faults or bizarre situations to the extent that it's not worth considering compared to plenty of other risks (like simple RCD failure, or a pure-dc-causing fault on the load side).  And generally only a small proportion of the original dc current would go through the 'victim' RCD.


The main points include that: a healthy circuit on the load side of an RCD has plenty of insulation resistance against dc; a sudden fault even with dc voltage present on the ac supply would trip a type-A RCD except in remarkable conditions; even a gradual fault (condensation) would in most cases pass much more ac than dc due to the voltages involved; and the low resistance of supply cables and transformers prevents significant dc voltage forming between supply conductors.

The above is a bit biased toward the country where the question arose, which has TN* supply so most currents pass through metal unless e.g. to a body or non-bonded electrode.  For a situation possible in the UK, with a chance of causing a problem, consider a TT installation where pure dc is passing from live conductors to earth (we'll ignore how it arises ... and note that it would have to be an awful lot more than just a few mA), in a supply system with rather high neutral-earthing resistance so that significant voltage could arise from neutral to earth. Then consider an RCD on another circuit (or in another TT installation) in which a N-PE fault happens at a time of low load (so the N-PE ac voltage doesn't cause a trip even with e.g. 6 mA of dc flowing through that RCD and the fault). And then a pulsating dc fault in the same direction also happens in that circuit, such as someone poking about in an electronic device; the RCD could have been made insensitive by the first two faults (dc current into the earth, and N-PE fault in the RCD's circuit) so it fails to detect the third fault at the right level of current.  This would require some rather extreme values of current and earth resistance, besides the combination of faults. Plenty of other hypothetical situations can be given, involving external earth contacts instead of a TT system, or funny arrangements with inductors, 3-phase-connected resistors that cancel the ac but add the dc, etc, etc.  But in view of the probabilities, reasonable cost-benefit (plenty of guesswork there, of course!) and so on, my conclusion - particularly for the TN* situation and the low level of expected dc injections - was to ignore the risk.  I think there are more convincing reasons than this one to have type-B RCDs (considering the range of modern appliances) but even those reasons have trouble justifying the cost in present regulations. 


 

Weirdbeard:

What I have wondered with regards this situation is if an installation has equipment that can apparently disable commonly found RCDs, would this disablement also apply to other installations that share a supply, for example with looped supplies? 

This is a question I got elsewhere a year ago, mainly about worries with solar PV installations and EV chargers affecting other customers on the network by their possible dc, and so a need to have type-B RCDs as default regardless of whether the installation included such devices. The answer: practically, no.    [edit: note that - in view of the thread's topic - this is about  disablement by dc currents, not e.g. by high frequencies]


Put more generally, it's a question of whether dc(residual)-causing apparatus could affect RCDs that have that apparatus [edit, for clarity: load->apparatus] on their supply side instead of their load side. Various complicated situations can be found where it could matter, but they involve multiple faults or bizarre situations to the extent that it's not worth considering compared to plenty of other risks (like simple RCD failure, or a pure-dc-causing fault on the load side).  And generally only a small proportion of the original dc current would go through the 'victim' RCD.


The main points include that: a healthy circuit on the load side of an RCD has plenty of insulation resistance against dc; a sudden fault even with dc voltage present on the ac supply would trip a type-A RCD except in remarkable conditions; even a gradual fault (condensation) would in most cases pass much more ac than dc due to the voltages involved; and the low resistance of supply cables and transformers prevents significant dc voltage forming between supply conductors.

The above is a bit biased toward the country where the question arose, which has TN* supply so most currents pass through metal unless e.g. to a body or non-bonded electrode.  For a situation possible in the UK, with a chance of causing a problem, consider a TT installation where pure dc is passing from live conductors to earth (we'll ignore how it arises ... and note that it would have to be an awful lot more than just a few mA), in a supply system with rather high neutral-earthing resistance so that significant voltage could arise from neutral to earth. Then consider an RCD on another circuit (or in another TT installation) in which a N-PE fault happens at a time of low load (so the N-PE ac voltage doesn't cause a trip even with e.g. 6 mA of dc flowing through that RCD and the fault). And then a pulsating dc fault in the same direction also happens in that circuit, such as someone poking about in an electronic device; the RCD could have been made insensitive by the first two faults (dc current into the earth, and N-PE fault in the RCD's circuit) so it fails to detect the third fault at the right level of current.  This would require some rather extreme values of current and earth resistance, besides the combination of faults. Plenty of other hypothetical situations can be given, involving external earth contacts instead of a TT system, or funny arrangements with inductors, 3-phase-connected resistors that cancel the ac but add the dc, etc, etc.  But in view of the probabilities, reasonable cost-benefit (plenty of guesswork there, of course!) and so on, my conclusion - particularly for the TN* situation and the low level of expected dc injections - was to ignore the risk.  I think there are more convincing reasons than this one to have type-B RCDs (considering the range of modern appliances) but even those reasons have trouble justifying the cost in present regulations.