Yes, I'm familiar with that, but it leads me to wonder exactly what you meant when you talked about limiting the shock current to 5mA. As your link shows, AC-2 does, indeed, limit the magnitude of shocks to 5mA for durations of 5 secs or more, but with the 100 msec disconnection time you go on to mention, it allows a shock current of something like 80mA.IEC/TS 60479-1:2005 is the standard describing the effects of AC current, in terms of magnitude and time. Shown here .... We had to be within zone AC-2 ....
I'm a bit confused here. One surely designs either for a particular maximum shock current (you indicated 5mA) OR, equivalently, for a particular touch voltage (assuming some value of shock path impedance) - I therefore find your "and" a bit confusing.... and less than 50V touch voltage. We arrived at a touch voltage of 40V.
I personally feel that hand-to-hand is a far more important consideration since, unless people have bare feet or are wearing deliberately conductive footwear (including socks) paths to earth via the feet will usually be of sufficiently high impedance to not represent a major risk. Hands are much more likely to be 'naked'. If you were working on the basis of a max touch voltage of 40V and an assumed hand-to-hand impedance of 2,600Ω (which, as I said before, assumes fairly dry hands), that would equate to shock current of about 15mA - still fine for AC-2 with 100msec disconnection, but still very different from the 5mA you said was being aimed for. I still think that hand-to-hand shock path impedances can easily be much lower than that assumed figure, in which case the actual shock current would probably be a lot more than 15mA in some cases (although still less than the ~80mA which would be acceptable for AC-2 with 100msec disconnection time).The same standard gives a hand to hand impedance of 2600 Ohm at that voltage, and 6500 Ohm hand to foot impedance. That gives approx 6.25mA, and in our case, the OCPD operated in 100mS.
Returning to your original point, if you work to a max touch voltage of 40V, that obviously means that (assuming a 230V supply and that both conductors are the same length) that the CPC CSA has to be about 4.75 times greater than the CSA of the L conductor [ (230-40)/40 ] - which is more realistic/do-able than the enormous figure I previously mentioned one would get if one designed for a max shock current of 5mA (the figure you mentioned) and and a body impedance of 1kΩ (which I would personally think would be a reasonable 'bad case' figure).
Kind Regards, John