First off, thank you very much for your analysis and responses so far. Dealing with the points raised:
- DrV: We temporarily replaced the connection between loop shield and electrical earth with a 150R resistor purely in order to see if there was any unexpected interaction between the addressable loop +- and the shield, bar the expected capacitive coupling. While the value was too high to keep the alarm panel happy, the 'scope trace of the loop shield does appear to show pure capacitive coupling from + and -, and appear to rule out any connection to + or - anywhere within the loop. The addressable loop may be driven by any single device, one at a time, in the loop. The particular sound of the interference is likely due to the pattern of the alarm panel stepping through each device checking it's ok, and each device responding. The fact that any device may transmit on the loop, in combination with the requirement that a single loop problem (e.g. cable short/break/on-fire) shouldn't stop the panel communicating with all devices on the loop means that the + and - conductors are not continuous on the loop - each device deliberately intercepts them. So, inserting a low value resistor in loop + and - in the panel would unfortunately only affect its transmitting to the first device in each direction, plus them transmitting back. Agree re talking with Cooper Eaton - we're about to do that.
- Bryson: All the fire devices (e.g. senders and sounders) in our system are connected to the control panel via a single wired addressable loop. None of them is wireless. There is a single short-ish spur off the loop, itself connected via a small Cooper box designed for that purpose. I don't think it's meant to broadcast RF. Re the guitar pickups not picking up much above a few kHz, I suspect they're picking up RF, then the downstream amplification (presumably with imperfect HF cut filtering at well above the audio range) is intermodulation-distorting that down into the audible range where we hear it.
- Paul TC & Bryson: The interference is definitely from the fire alarm system addressable loop - we've proved this. That said, good point re intermodulation distortion - I think that's the effect by which we hear the resulting interference picked up by by magnetic pickups through the sound system.
- richardash1981: I've read that part of the manual and, as far as I could tell, these cable requirements have been followed. You wrote "what wiring defect is causing the system to radiate in unexpected ways": Yes... Measurements on the loop shield show it to be continuous (once we'd fixed a single break) and with a single connection to electrical earth in the alarm control panel. The link to the detail shows a 'scope trace of the shield when this connection is broken and replaced with a 150R resistor. The trace looks, to me, like pure capacitive coupling. (The mains wiring does act as an aerial capable of picking up RF - see the spikes for BBC R4 LW (200kHz) plus Radio 5 on MW on the ClubFreq2MHz_2.PNG pic in the details link above. Suspect this is normal behaviour, though.)
DrV & Andrew: As far as we can tell from measurements, the shields at both ends of the loop are joined and connected to electrical earth at a single point in the fire alarm control panel, and the shield is not connected anywhere else.
Replies welcome and I'll keep responding, plus update once we've talked with the manufacturer.