As suspected, and with thanks to you all but especially "Sonic" for such a compreshensive trawl of the possibilities, it's all down the the induction loop being picked up by the gear.

Turn off the loop and everything is fine. I'm looking to re-route the loop wire to try to give a "dead" area around the "stage" area, at present I thing the loop goes straight underneath. Also the intermediate stage box is currently directly on top of the loop amplifier, and this cannot be helping, so I'm looking to move things about but I'll have to make some new wires up to give the necessary freedom of movement.

I have not actually found any miswiring of connectors so far, but that's now a work in progress.

For information here's the setup now.

The wiring is using commercial multicore cabling (all cores are twin and earth, with unique idents) with XLR/Jacks at one end and a stagebox at the other.

The "intermediate" box provides 12 feeds to the mixer (1-12) and receives 4 feeds back (A,B,C,D). A and B are main PA L/R, C is now a foldback line fed from the "summed mono out" of the desk, and D is the Induction loop drive from the "Aux 1" feed of the desk. The loop drive uses the line level input of the loop drive amp, I haven't dragged the PA amps out to check them but I'm confident that they work as expected.

The "stage" box may have 16 "ins" and 4 "outs" but only 8 of the "ins" and one of the "outs" are wired up into a multicore that comes out next to the intermdiate box to give 8 XLRs and a single TRS jack respectively. The 8 XLRs go straight into 1-8 on the intermediate, and the jack goes into "C" on the intermediate box to route the foldback signal to the stage box. This is also now working as expected.

So I think this situation, if not fully sorted yet, is at least in a position where we know the issues and can take some steps to improve things.

Thanks again!

John