It may relieve you to know that regardless of the no of intermediate switches, you only ever have to switch one to cover all that would have been covered in a system without intermediates, the number of tests  is not  n! where n is the no. of switches.

Mike

! Factorial notation explained

As others have said above, there are 8 combinations of switch position.

However, for the purposes of insulation resistance, I would assert it's not necessary to do all 8 combinations

mapj1 said:

you only ever have to switch one to cover all that would have been covered in a system without intermediates

Possibly true of the wiring, but is it true if you want to check there isn't a fault in an intermediate switch or either 2-way:

• In any combination where the lamp would be on (check for continuity between "switch supply" and "switch return to light"), all live conductors are effectively "in-circuit", a test of one live conductor to PE tests all live conductors to PE.
  
  
• With a single intermediate switch, you will then need to check 2 positions of the 2-way switches for the "off" state with the intermediate in one position, to check insulation between "Switch supply" to "switch return" - this proves no shorts in:
  • Line to line conductors in wiring throughout
  • insulation resistance in both 2-way switches is adequate C to L1, C to L2 and L1 to L2.
  • Intermediate switch insulation resistance for one position is adequate
• And finally, switch the intermediate switch to the other state, and select an "off" position using the 2-way switches, check insulation between "Switch supply" to "switch return" and this proves adequate insulation resistance for the other combination of 2-way switch.

So, I think 1 "on-state" and three "off-states" will do the trick for a single intermediate switch. Last step repeated for each intermediate switch if there is more than one.

If you are only looking for faults to ground in the wiring to the switch terminals, and not operation under load, then adding any number of intermediate switches only increases the no of tests by one, over and above the no of tests that would have been needed for the chosen style of 2 way circuit without any intermediates , as then all 4 terminals in all intermediate switches on the path are energized in one or another test.

I agree actual switch operation is not tested, but IR does not do that anyway, even single pole.

It is very hard to envisage a credible fault path to earth within a plastic switch body that occurs with the switch crossed that does not occur with it un-crossed or vice versa.

Once under power you should probably still test that all switches operate the lights as expected, unless the switch maker has done that at the factory, but even that only needs to be done in pairs, not all combinations, to exercise all moving contacts. By the time you have half a dozen intermediate switches, this is worth thinking about hard.

Mike

mapj1 said:

It is very hard to envisage a credible fault path to earth within a plastic switch body that occurs with the switch crossed that does not occur with it un-crossed or vice versa.

Switches that are wet inside and appear dry outside ... sadly if you don't catch it at IR test, it makes a lovely black mark up the wall.

The first time I ever encountered this was back in the late 1970s, when my Dad asked my Mum to wipe the kitchen light switches over before refitting them after tiling. Unfortunately, the "wipe over" was done after first soaking the switches in sugar soap !

The fault path could also be L to L ... BS 7671's initial position is that insulation resistance is all live conductors.

Ha! - well with sugar soap in the works all bets are off I think.  In a similar vein  have seen a normal (single pole) switch pass IR and then make  most impressive bang  when turned off due to wet plaster, or it may have been polyfilla,  giving off steam having got trapped somewhere that was not a tested path,,,

However I's pretty sure that it is also the BS7671 view  that L+N to E testing is a perfectly acceptable way to do an insulation test and perhaps more sensible if fixed loads are already in place as then they get tested at the same time. It depends rather, personally I think there are more positive ways to find L-N and L-L faults without hurting anyone, whereas missing earths and faults to earth may be missed for years.

More relevant is the fact that things with very many states, like heating and ventilation  systems in larger buildings are very hard to test in situ and be confident sure that all wiring paths have been covered.

Mike

mapj1 said:

However I's pretty sure that it is also the BS7671 view  that L+N to E testing is a perfectly acceptable way to do an insulation test

Disagree, for initial verification (including additions, alterations and arguably repairs depending on what has been repaired) ... the "go to" for initial verification is still 'between live conductors and between live conductors and the protective conductor connected to the earthing arrangement' (643.3.1) prior to the connection of equipment that might be damaged by such tests (643.3.3).

Only after these tests have been performed, can the 'L+N to E' test be performed.

I suppose it’s not the fault of the standard writers that this is a requirement  that generally never gets done. Anyone who has been at the coal face knows that the erection of an installation rarely allows testing between live conductors as load tends to get connected as the opportunity arises. For example, if tower scaffold is in place to do the first fix lighting , it is likely that second fix of  control systems and luminaires will be simultaneous. Following the prescribed procedure for insulation resistance testing between live conductors in a new installation pressed by commercial imperatives is a logistical nightmare for contractors. It’s far easier to insert  +999Mohms  as a legitimate result to record and we can all pretend that appropriate diligence is in place and shove the matter well below the carpet.

lyledunn said:

It’s far easier to insert  +999Mohms  as a legitimate result to record and we can all pretend that appropriate diligence is in place and shove the matter well below the carpet.

Do I sense irony here? Perhaps it is sufficient to do the IR (and continuity) tests at first fix? The only problem is that conductors could be damaged when the switches or lamp fittings are installed.

lyledunn said:

It’s far easier to insert  +999Mohms  as a legitimate result to record and we can all pretend that appropriate diligence is in place and shove the matter well below the carpet.

Do I sense irony here? Perhaps it is sufficient to do the IR (and continuity) tests at first fix? The only problem is that conductors could be damaged when the switches or lamp fittings are installed.