It seems AMD3 is trying to mitigate tip of the larger problem. The problem of "unidirectional" and "bidirectional" RC devices stems from the way electronics are designed on the secondary winding inside. But the problem is not just the design but the fact that the functionality of modern RCD/RCBO in UK relies on electronics at all. As many of you may already be aware - this type of design was disallowed on mainland continent and most of RCBOs sold nowadays in UK are illegal to use in EU because the electronics will not function without neutral (eg. in the event of PEN fault). The device that's supposed to be part of secondary protection and fire protection will do absolutely nothing if the fault starts by burning through neutral. Just a few years ago just about any RCBO with electronic release would have a flylead connected to MET ensuring the device was still functioning without neutral under fault conditions, but with time even that failsafe was withdrawn from most of the designs.

So - don't add unidirectional and bidirectional malarky to the regs - just ban shoddy electronic designs from UK markets all together before it becomes larger issue. Most of the world already did.

Martin Korczak said:

As many of you may already be aware - this type of design was disallowed on mainland continent and most of RCBOs sold nowadays in UK are illegal to use in EU because the electronics will not function without neutral (eg. in the event of PEN fault). The device that's supposed to be part of secondary protection and fire protection will do absolutely nothing if the fault starts by burning through neutral.

Is this correct? I'm sceptical because the terms in BS EN 61009-1 relating to voltage dependant operation aren't BS national departures.

The overcurrent functions of an RCBO are not dependent upon voltage - they are essentially an MCB in these respects.

The RCD characteristic however is. And you're correct that there's no guarantee this will work if the L-N voltage drops below 0.8 nominal. 

But it would need multiple faults for this to become a hazard. For example a PEN fault is a fault condition so if this was to coincide with a secondary fault that was unable to clear then yes that would be immediately dangerous.

I can see some justification in domestic where the risk of existing faults is sufficiently high - hence the elevated requirements for AP in the first place. But then the advantage of AP by 30mA RCDs [as opposed to e.g. the traditional supplementary bonding approach] is that a problem raises its head and the occupants hopefully ring their electrician about an annoying RCBO.

FEs are a bit of a pain. I've always wondered why they don't put a supercapacitor or something in - but would have its own problems.

Some also argue that we should have upfront PEN protection - I'm not convinced but I think smart meters should report undervoltage to the DNO.

Ultimately protective devices are there for fault conditions not dodgy wiring - which aren't the same thing.

Keir Stitt said:

but I think smart meters should report undervoltage to the DNO.

I think that would be a good idea ... overvoltage too.

The device that's supposed to be part of secondary protection and fire protection will do absolutely nothing if the fault starts by burning through neutral.

I'm not following your thinking here. Protective devices generally only provide protection to downstream conductors, not upstream - if the downstream N burns out, then the device will still have a supply N and can function normally. If the supply N burns out, I agree the device won't trip - but then tripping wouldn't help disconnect the fault - and unless there's a simultaneous fault downstream, there would be no need to disconnect anyway.

   - Andy.

   - 

The device that's supposed to be part of secondary protection and fire protection will do absolutely nothing if the fault starts by burning through neutral.

I'm not following your thinking here. Protective devices generally only provide protection to downstream conductors, not upstream - if the downstream N burns out, then the device will still have a supply N and can function normally. If the supply N burns out, I agree the device won't trip - but then tripping wouldn't help disconnect the fault - and unless there's a simultaneous fault downstream, there would be no need to disconnect anyway.

   - Andy.

   - 

AJJewsbury said:

If the supply N burns out, I agree the device won't trip - but then tripping wouldn't help disconnect the fault - and unless there's a simultaneous fault downstream, there would be no need to disconnect anyway.

I think there's an issue with TT systems if a voltage dependant RCD was providing fault protection.

Personal opinion is that 30mA RCDs/RCBOs are for additional protection and should only be used for TT fault protection where no additional protection is required - I think a TT board should be backed with a type S for fault protection. The regs don't spell it out but I think the clue is in the name *additional* protection - i.e. supplementary to standard fault&overcurrent protection.

S type RCDs are almost universally independent of voltage because of how they work internally although that might change if type S&B combined become a thing.

I certainly think Martin raises interesting points. If you get two electical engineers to discuss PEN faults you'll get four opinions.

I think there's an issue with TT systems if a voltage dependant RCD was providing fault protection.

More generally, a L-PE fault typically causes a reduction in the voltage on L and an increase on PE (a portion of which will be reflected onto N on a TN-C-S system) - so generally it should be expected for the L-N voltage on the supply of any RCD to reduce - as you say often a problem for TT where the RCD element is relied upon for ADS, but also for TN where Zs is too high for the overcurrent devices to open promptly. I had an idea in my head that RCDs were meant to operate down to 50V to allow for such eventualities (below 50V the touch voltage shouldn't be so much of a worry, so lack of disconnection could be tolerated) - but I'm not sure now where I got that idea from..

   - Andy.

AJJewsbury said:

I had an idea in my head that RCDs were meant to operate down to 50V to allow for such eventualities (below 50V the touch voltage shouldn't be so much of a worry, so lack of disconnection could be tolerated) - but I'm not sure now where I got that idea from..

Most will work at much lower voltages - the application notes for the controllers suggest that many would operate at less than 100V. But in the absence of MIs to the contrary the position in EN 61009-1 is 0.85-1.1x rated voltage.

Many of the tests will include impedances so yes there would be some drop from load current in addition. A fault of negligable impedance would operate the MCB function anyway.

We need to know that touch voltages are < 50V. But that's not the same thing as the RCBO needing to operate down at <50V. And 701.415.2 might allow supplementary bonding to be omitted on RCD protected circuits - but only subject to conditions including that the protective equipotential bonding is okay. If the fault is on the DNO's network then the MET does the job of keeping touch currents <50V anyway, supplementary bonding sould only come into play if there's a fault within the installation - and again the MCB function would handle that.

However if someone has heavily loaded submains with too much Vd then it might go very badly very quickly.

if we really do mean it may not work below 0.85 of 230V == 195 volts RMS then in many parts of Europe the mains at the entry point to the building could be perfectly in spec at 207- 253 and in the UK 216V min. (as we seem to never have adopted the proposed -10% limit.) 

Even working well, this leaves very little for in-building drop between entry point and boards at the ends of sub-mains to outbuildings or whatever.

I must admit I have seen, and used a line low of 50V assumption ,and indeed tested and found RCBO units that are quite happy with it (wylex and related if anyone needs a name that works OK)

Equally I have heard of folks who could not get standard RCDs to go on 110V supplies - maybe this is why.

M.