BS7671;2018 Minor works certificates.

Andrew Jewsbury:

I've definitely seen RCDs with longer trip times at 5IΔn than at IΔn, so IΔn is not always the "worst case" either!

In which case, can we have any confidence that a 1x (or 5x) test will give realistic assurance that the RCD will open in the required time when subject to a real earth fault - which could be many amps in magnitude?

Perhaps ... there are a number of problems related to RCD testing, and it perhaps could be said that it's obvious that trying to do an "in-situ" test based on the parameters of what is effectively Table 3A in BS 7671 doesn't always lead to expected results. See below on the L-N test for the "test button".

 

These require RCDs for Additional Protection - in this case this would invoke the 5IΔn test. To add all these requirements together, we have a maximum loop impedance for a 30 mA RCD of 230*.95/0.15 Ω (where additional protection is needed), i.e. 1457 Ω and not 1667 Ω

Humm - not sure of the logic there - additional protection is usually aimed at situations of direct contact (nails through cables, flexes cut by lawnmowers, touching the contacts of lampholders) - where the c.p.c. and hence Zs aren't usually of relevance (likewise any calculations involving Uo or Cmin). In a functioning installation we're only really worried about 150mA going through the victim's body, where it goes after that isn't a concern. Certainly for a successful test we'd need a lowish loop impedance, but that requirement doesn't seem to be obviously written into BS 7671.

 

Given the fact that another recognised means of additional protection is supplementary local equipotential bonding, I don't quite follow your argument (although I'm not dismissing it entirely). I'm not sure anyone would say that someone touching just a line conductor of a circuit protected by a 30 mA RCD would be 100 % safe ... however, I do follow where you are coming from here.

Would you think it acceptable to perform a 5x test by placing the tester's earth probe on the supply N rather than PE? That would avoid problems due to a high Zs but still test the RCD's fundamental operation?

Perhaps not, as the RCD test is designed to be conducted downstream of the RCD, using the associated protective conductor. The RCD is not guaranteed to work with an earth return path by another means (although I'm happy to accept for additional protection, it's a possibility).


Basically, as things stand (with BS 7671 Part 6 and the guidance in GN3 and OSG for testing), the designer should ensure Zs is low enough to accommodate the RCD at 5x for additional protection, and not simply Table 41.5 alone. When we are talking about automatic disconnection in times less than 0.3 s, this does present another issue, and definitely one for further discussion.


We should certainly be looking at the performance of the installaition as part of the Verification, and not a particular protective device.

Andrew Jewsbury:

I've definitely seen RCDs with longer trip times at 5IΔn than at IΔn, so IΔn is not always the "worst case" either!

In which case, can we have any confidence that a 1x (or 5x) test will give realistic assurance that the RCD will open in the required time when subject to a real earth fault - which could be many amps in magnitude?

Perhaps ... there are a number of problems related to RCD testing, and it perhaps could be said that it's obvious that trying to do an "in-situ" test based on the parameters of what is effectively Table 3A in BS 7671 doesn't always lead to expected results. See below on the L-N test for the "test button".

 

These require RCDs for Additional Protection - in this case this would invoke the 5IΔn test. To add all these requirements together, we have a maximum loop impedance for a 30 mA RCD of 230*.95/0.15 Ω (where additional protection is needed), i.e. 1457 Ω and not 1667 Ω

Humm - not sure of the logic there - additional protection is usually aimed at situations of direct contact (nails through cables, flexes cut by lawnmowers, touching the contacts of lampholders) - where the c.p.c. and hence Zs aren't usually of relevance (likewise any calculations involving Uo or Cmin). In a functioning installation we're only really worried about 150mA going through the victim's body, where it goes after that isn't a concern. Certainly for a successful test we'd need a lowish loop impedance, but that requirement doesn't seem to be obviously written into BS 7671.

 

Given the fact that another recognised means of additional protection is supplementary local equipotential bonding, I don't quite follow your argument (although I'm not dismissing it entirely). I'm not sure anyone would say that someone touching just a line conductor of a circuit protected by a 30 mA RCD would be 100 % safe ... however, I do follow where you are coming from here.

Would you think it acceptable to perform a 5x test by placing the tester's earth probe on the supply N rather than PE? That would avoid problems due to a high Zs but still test the RCD's fundamental operation?

Perhaps not, as the RCD test is designed to be conducted downstream of the RCD, using the associated protective conductor. The RCD is not guaranteed to work with an earth return path by another means (although I'm happy to accept for additional protection, it's a possibility).


Basically, as things stand (with BS 7671 Part 6 and the guidance in GN3 and OSG for testing), the designer should ensure Zs is low enough to accommodate the RCD at 5x for additional protection, and not simply Table 41.5 alone. When we are talking about automatic disconnection in times less than 0.3 s, this does present another issue, and definitely one for further discussion.


We should certainly be looking at the performance of the installaition as part of the Verification, and not a particular protective device.