RCBO vs MCB - Garage board

[PeterW]

Quick one ...

 

Got a 40A RCBO feeding a sub main in an integral garage that has 3 circuits - lights, 16A Commando, sockets. 
 

All sockets are ducted / surface mounted and the sparky has been asked  to do a pair of additional sets of sockets and has asked if RCBO or MCB is wanted. (He did not do the original install)
 

His logic is that the main board RCBO covers the wiring to the integral garage and even if you manually trip the sub main RCBOs to test, it takes out the main board garage feed 40A RCBO anyway.
 

Given that ... is it worth adding the new circuits (surface mounted in ducting) using MCB or RCBO, and secondly, any ideas why the main RCBO trips when the test buttons on the sub main RCBOs are used ..? 

[Jeremy Harris]

You shouldn't daisy chain earth leakage devices, be they RCDs or RCBOs, without ensuring that there is an adequate difference in the trip current.  For example, a 100 mA trip current RCD/RCBO upstream of 30 mA RCDs/RCBOs is fine, as there is current discrimination and the downstream ones will trip first.

 

In your case, as you have (most probably) a 30 mA trip current RCBO upstream of the installation, that is providing adequate earth leakage protection for everything downstream.  All that should be fitted down stream are over-current protection devices, fuses or MCBs, not any device that includes residual current/earth leakage fault protection.

 

The exception to this is when a downstream installation is connected as a TT island, with it's own earth electrode, when it needs an RCD/RCBO.

[joth]

Also following as it's likely I'm going to end up in this situation too. (And thanks you @Jeremy Harris for the useful info as always!)
 

 

  On 29/01/2020 at 14:21, Jeremy Harris said:

You shouldn't daisy chain earth leakage devices, be they RCDs or RCBOs, without ensuring that there is an adequate difference in the trip current

Expand  

 

Just to clarify, is this "shouldn't" because it is unsafe or against regs to do so, or just a shouldn't because it will be frustrating and inconvenient to the user for the double-trip reasons given?

 

Also, if you put RCD or RCBOs on all the downstream / sub-main circuits is it then OK to remove the upstream earth leak protection? Or does this then trigger potentially more upgrades need to the cable between the two? (i.e. upgrade to double protection/armoured? Or should that always be there anyway for the feed to a sub-main distribution board)

 

[also to clarify I'll be having a certified person do all the work, just find the conversations go so much easier if I already know what I'm asking for is within regs]

 

[Jeremy Harris]

  On 29/01/2020 at 14:28, joth said:

Just to clarify, is this "shouldn't" because it is unsafe or against regs to do so, or just a shouldn't because it will be frustrating and inconvenient to the user for the double-trip reasons given?

Expand  

 

Really because of the nuisance caused when an upstream RCD/RCBO trips as much as anything else.  There's no added protection from daisy chaining RCDs/RCBOs without discrimination between trip currents.  If there is no discrimination, then upstream RCDs/RCBOs may trip when the downstream ones are tested.

 

  On 29/01/2020 at 14:28, joth said:

Also, if you put RCD or RCBOs on all the downstream / sub-main circuits is it then OK to remove the upstream earth leak protection? Or does this then trigger potentially more upgrades need to the cable between the two? (i.e. upgrade to double protection/armoured? Or should that always be there anyway for the feed to a sub-main distribution board)

 

Expand  

 

It's usually fine.  My garage feed is a run of 6mm² SWA protected at the upstream end with a 40 A DP MCB.  At the garage CU I have an RCD feeding MCBs for the various sub-circuits, so there's adequate residual current protection on every garage circuit, and if there is an earth fault, the RCD in the garage is the one that trips.

 

RCD protection is required at the user end of any circuit, located somewhere upstream of the outlet/lamp/etc, so they are all adequately protected, but there's no real need to RCD protect SWA cable runs leading to a sub-main CU, all that needs is over-current protection usually.  If the cable isn't armoured, then fitting a 100mA RCD/RCBO at the upstream end makes sense and provides the required trip current discrimination.

 

 

[ProDave]

Sounds like another sad case of not actually understanding anything.  His guide says you need an rcd or rcbo in the consumer unit and he does not have the ability to work out you already have that protection.

[PeterW]

I think it’s more that the last sparky fitted RCBOs  everywhere and he is querying whether that’s actually wanted or requested as he can put either RCBO to match or MCB as it’s cheaper. 
 

But thanks all, seems MCB is the way to go. 

[Ed Davies]

  On 29/01/2020 at 14:12, PeterW said:

…and secondly, any ideas why the main RCBO trips when the test buttons on the sub main RCBOs are used ..? 

Expand  

 

I'm puzzled by this. AIUI, the test buttons on many RCDs work by switching in a resistor between line on one side of the coil and neutral on the other to create an imbalance in the coil. AFAICS, this should just look like a normal load to any upstream RCD so shouldn't cause it to trip. Is that wrong or are there other ways the test buttons can be wired that really unbalance the line and neutral. Do normal RCDs in consumer units even have an earth connection? Through the DIN rail clips?

[Jeremy Harris]

Some RCBOs do have an earth lead, so may cause an imbalance that can trip upstream RCDs.  Off the top of my head I know that single pole Crabtree Starbreaker RCBOs have an earth lead, some other makes do, to.

[MJNewton]

  On 29/01/2020 at 14:21, Jeremy Harris said:

You shouldn't daisy chain earth leakage devices, be they RCDs or RCBOs, without ensuring that there is an adequate difference in the trip current.  For example, a 100 mA trip current RCD/RCBO upstream of 30 mA RCDs/RCBOs is fine, as there is current discrimination and the downstream ones will trip first.

Expand  

 

This is not the case. Trip time depends on exactly that - trip time - you cannot with any certainty determine whether the 30mA or 100mA RCD/RCBO will trip first, regardless of where they are located. The only exception to this is if the fault current is between 30mA and 100mA in which case the 100mA RCD/RCBO will of course not trip at all.

 

The only way of being deterministic is with time-delay RCDs/RCBOs which in most situations will undermine the protection that are usually employed to provide.

[Jeremy Harris]

OK, I should have added that the discrimination criteria includes time as well as current, but in general time delay RCDs/RCBOs aren't that common in a domestic installation, are they?

 

I stand by the statement that it isn't good practice to daisy chain RCDs/RCBOs.

[Onoff]

  On 29/01/2020 at 16:07, MJNewton said:

 

This is not the case. Trip time depends on exactly that - trip time - you cannot with any certainty determine whether the 30mA or 100mA RCD/RCBO will trip first, regardless of where they are located. The only exception to this is if the fault current is between 30mA and 100mA in which case the 100mA RCD/RCBO will of course not trip at all.

 

The only way of being deterministic is with time-delay RCDs/RCBOs which in most situations will undermine the protection that are usually employed to provide.

Expand  

 

+1. True selectivity, as it's now called (discrimination being very un pc ?) is achieved by fitting a 100mA time delay ("TD") RCD upstream. Like this:

 

https://www.tlc-direct.co.uk/Products/WYWRMT100slash2.html?

[MJNewton]

  On 29/01/2020 at 16:16, Jeremy Harris said:

OK, I should have added that the discrimination criteria includes time as well as current, but in general time delay RCDs/RCBOs aren't that common in a domestic installation, are they?

Expand  

 

Generally just TT earth systems where the supplier doesn't provide an earth, which can also include outbuildings where the earth isn't exported from the main building.

 

  Quote

I stand by the statement that it isn't good practice to daisy chain RCDs/RCBOs.

Expand  

 

Absolutely agree; I was just pointing out that even daisy chaining a 30mA downstream of an 100mA is still not good as it doesn't provide the discrimination that is often assumed to be the case.

Edited January 29, 2020 by MJNewton

[Markblox]

  On 29/01/2020 at 14:21, Jeremy Harris said:

 

The exception to this is when a downstream installation is connected as a TT island, with it's own earth electrode, when it needs an RCD/RCBO.

Expand  

And remember if on a TT system (Earth Electrode) the RCD must be double pole.

[Ed Davies]

  On 29/01/2020 at 16:07, MJNewton said:

The only exception to this is if the fault current is between 30mA and 100mA in which case the 100mA RCD/RCBO will of course not trip at all.

Expand  

 

I thought the rated trip current of an RCD was the maximum at which it must trip and typically they'll trip at something a bit more than half that current. So if you had a 60 mA fault current the 30 mA one would be more likely to trip first, I imagine, but it's not guaranteed.

[MJNewton]

It is more nuanced than that. The performance characteristics for RCDs are laid down by BS61008 which includes maximum trip times that vary depending on comparison to the rated trip current (In). This is necessary given the increased risk arising from increased exposure time and not just increased current. The base times specified are:

 

  Quote

Earth Fault    Max Trip Time

-----------    -------------

 

  1 x In          300ms

  2 x In          150ms

  5 x In           40ms

 

Expand  

 

Hence, an earth fault of 100mA would require a 30mA RCD (detecting 3.33 x 30mA) to trip within 150ms and a 100mA RCD (detecting 1 x In) to trip within 300ms but there are no *minimum* trip times specified (this is only done for time-delay RCDs) and so it is not possible to determine which would/should trip first. Moreover, for earth faults >150mA a 30mA and 100mA RCD would both be required to trip within the same maximum time (that's not to say they would do so if daisy-chained in the same circuit of course because an early trip from one may limit the exposure time of the other to the current imbalance).

Edited January 30, 2020 by MJNewton

[Onoff]

RCD trip times can and do vary due to the mechanical mechanism. They can get stiff / gum up. It's why they should be regularly excercised by pushing the test button.

[Ed Davies]

Right, so we're all agreed that a 60 mA fault current to earth could trip the upstream 100 mA RCD, leaving the downstream 30 mA one untripped?

[MJNewton]

It is absolutely possible, yes, even if both devices are operating within spec.

 

Edited January 31, 2020 by MJNewton

[Onoff]

  On 30/01/2020 at 23:51, Ed Davies said:

Right, so we're all agreed that a 60 mA fault current to earth could trip the upstream 100 mA RCD, leaving the downstream 30 mA one untripped?

Expand  

 

100%. 

 

The 100mA needs to be time delayed to counter this.

[MJNewton]

  On 31/01/2020 at 00:37, Onoff said:

The 100mA needs to be time delayed to counter this.

Expand  

 

...not wishing to muddy the waters just as we were reaching some clarity but even a time-delayed RCD couldn't necessarilly guarantee the expected outcome as with a 60mA earth fault a 30mA RCD has a maximum trip time of 150ms and yet a time-delayed 100mA RCD has a minimum trip time of 130ms. That still leaves a 20ms window within which the time-delayed 100mA could end up beating the non-delayed 30mA potentially leaving the latter untripped. ? In practice I would expect it to work as intended though as non-delayed trip times are usually significantly shorter than what the standards require.

Edited January 31, 2020 by MJNewton

[Onoff]

  On 31/01/2020 at 00:49, MJNewton said:

 

...not wishing to muddy the waters just as we were reaching some clarity but even a time-delayed RCD couldn't necessarilly guarantee the expected outcome as with a 60mA earth fault a 30mA RCD has a maximum trip time of 150ms and yet a time-delayed 100mA RCD has a minimum trip time of 130ms. That still leaves a 20ms window within which the time-delayed 100mA could end up beating the non-delayed 30mA potentially leaving the latter untripped. ? In practice I would expect it to work as intended though as non-delayed trip times are usually significantly shorter than what the standards require.

Expand  

 

Muddy the waters is probably about right! ? Arguably on a TT you'll have done all you possibly can by fitting a 100mA td rcd upstream (& DP 30mA rcbos downstream). What alternative is there? Imo always try and fit all the RCDs of the same make so you can take it up with the supplier/maker. Makes it easier to get a replacement to swap out.