If the TNCS is low enough Zs, then it is permissible to feed the SWA submain from a non RCD supply,  so long as the downstream circuits have RCDs at the flat end, the assumption is that any cable damages will operate the MCB when the earthed armour is driven into the live cores, and the end users are protected by the load end RCDs.

Similar logic relaxes the routing of the SWA cable - again the thinking is that unlike non-armoured cables,  damage that exposes a live core but does not trip anything is much less likely to occur.

Another approach is to use an RCD at the source end with a time delay to supply the submain - this is common in outdoor works, TT installations and agriculture especially, where a nested hierarchy of fast low threshold RCDs are fed by higher rated slower ones.

The simplest of course is to ask yourself when the RCDs actually last tripped, and if that is rare enough not to be a major issue, and to say it is acceptable for perhaps a few trips a year - and maybe an odd emergency light may be an alternative mitigation to being left in the dark.

Mike

Thanks Mike

Will remove the main 100amp switch and replace with main RCD. Feeding 6 MCB. 

As for the supply to the granny flat, using a double pole MCB adjacent to the house consumer unit. Would I split the supply in a Henley block then run one supply to the main consumer unit MCB from the Henley block and another from the Henley block to the adjacent garage MCB from the Henley block. This is fed from 100amp double pole MCB at meter. ?

You certainly could. Joking aside, before getting all worried about it how often do your RCDs trip anyway ?

In my place it would be a lot less than once a year, except for testing, excluding the one to the workshop which is a bit of an anomaly as it is where I test things that may be faulty or experimental. MCB trips are almost never - probably less than one per decade - there are far more power cuts.
Mike.

You certainly could. Joking aside, before getting all worried about it how often do your RCDs trip anyway ?

In my place it would be a lot less than once a year, except for testing, excluding the one to the workshop which is a bit of an anomaly as it is where I test things that may be faulty or experimental. MCB trips are almost never - probably less than one per decade - there are far more power cuts.
Mike.