(Note: I am in the UK so the terms 'line' and 'neutral' refer to the two live mains conductors)
I am trying to establish a list of safety issues which I should be aware of if I decide to introduce a resistive path between either of the live mains conductors and earth in a Class I device, or to decide whether I should avoid the idea completely!
The reason for this is that I am working on a circuit which performs a number of diagnostic measurements on the mains supply to be built into a device (earth resistance, correct line/neutral polarity and AC voltage), the circuit will disconnect the supply in case of a fault. One idea for the circuit involves resistive paths between both live conductors and earth (other versions use capacitive sensing but have a number of issues) - see 'Scenario 1' in the image below for a simplified schematic. I can also think of other situations where someone might consider using a resistor between a live conductor and earth - for instance to ensure Y-capacitors in a CM filter are drained ('Scenario 2' below) - so I'm interested in getting to the bottom of the general issue here of allowing a DC conduction path between the live conductors and earth.
The earth in the device is bonded to the chassis and appears on a number of external conductors. I was initially against the idea of using a resistor between live conductors and earth as this compormises galvanic isolation, however as I thought about it more I started to sway towards the opinion that this is possible if done carefully. I have not found any sources which deal specifically with this issue, however indirectly by reading about electrical safety testing and regulations as well as using some common sense I have compiled a list of requirements, which, if met, might make this acceptable:
- The resistance should be sufficiently high to minimise leakage current - most obviously so as not to trip the RCD (the threshold in the UK is 30 mA), however a much more restricting constraint is an insulation test: for instance, if I understand correctly during a PAT test 500V DC is applied between the live and earth connectors, and the measured insulation resistance must not be below 1 MOhm. From this follows that the total resistance of the resistive path must be at least 1 MOhm plus a safety margin (perhaps 2 MOhm?)
- The components in the resistive path must be of a sufficient power rating to handle both dissipation under normal use and dissipation during an insulation test such as above.
- The resistive path should not introduce a risk of voltage spikes on the live conductors reaching the earth conductor, so appropriate transient voltage suppression should be used before the resistive path and/or incorporated into it. In my circuit as drawn in 'Scenario 1', I would use higher-voltage transient suppression between the live conductors and earth, as well as lower-voltage transient suppression between the sense inputs and earth.
- The physical construction of all resistors should be such that clearance and creepage requirements between live conductors and earth are met (e.g. use a through-hole resistor where the distance between leads is greater than 5 mm).
After coming up with this list, I'm still not confident that this is a good idea, partly because I have never seen this done - so the question is have I missed anything, or is there any other reason why resistors between live conductors and earth should be avoided?
Of course I'm aware that insulators always have finite resistance and Y-capacitors have a leakage resistance which in this context is theoretically identical to having very high-valued parallel resistors, but these resistances are much higher than the megaohm range considered here!