additional RCD or not

[NotHimAgain]

Why are you attempting to draw parallels between cable faults - that in most cases would be due to some form of damage - and RCDs which can 'fail' (and failure needs to be defined) simply because they have been inactive for a long period?

As I said, this is a common method of improving reliability in engineering systems - I am not suggested that you have to use it .

 

[RF Lighting]

I'm just trying to understand your reasoning for using two RCDs in series.

I can fully understand having a backup in place incase your first method of protection fails. This is the very reason RCDs are required by BS7671. It is there as a backup incase your properly designed installation fails to protect the user from electric shock perhaps due to an unforseen circumstance.

The bit I am less sure of is why you feel it is necessary to add a third method of protection into your circuit design.

531.2.9 even advises against the use of two RCDs in series in some circumstances.

Would you install two fire alarms into a building incase the first one failed?

 

[NotHimAgain]

First there are no regulations that call for RCDs in series, but who knows what the future holds, we are using RCDs to safe idiots now (see 4151.1 last sentence ).

There is research that shows that RCDs can 'fail' due to inactivity - this would generally be due to 'stickiness' in the mechanism. Remember, we require them to operate within 40 ms for many applications. RCDs are often left for years without operating.

An RCD is just an mcb with an additional sensor. When an mcb responds to an overload or a short circuit it is either hot or subject to an intense magnet field or both. This would tend to overcome any minor 'stickiness' in the mechanism. When working in RCD mode they are not subject to such disturbance. In the precision relay type - a small pin is pushed against the mcb trip lever - and in the electronic type a solenoid drives the pin in a similar way. If the pin sticks for a moment the RCD will be slow and you may be dead

Now I haven't checked this next bit and it is a long time since I have used it - but it goes something like this:

If we say an RCD has a probability of failure of 0.05, then two similar RCDs in series would have a probability of failure of

0.05 * 0.05 = 0.0025

which is much better.

Now I have probably got that wrong but the idea is correct.