ChenYuELV

This is the method I use when laying out a 100 V (or 70 V) ceiling-speaker system for PA / BGM and, with the right SPL and intelligibility checks, for voice alarm. Figures below are a planning example, not a certified evac design. Voice alarm still has to be verified to the applicable standard (BS 5839-8, EN 54, NFPA 72, GB 50116, or whatever the contract names) and to the loudspeaker and amplifier data sheets. Coverage and worst-case SPL For a ceiling speaker with a stated dispersion angle, I take the coverage radius at ear height as: radius = (ceiling height − listening height) × tan(dispersion / 2) Listening height is about 1.2 m seated, 1.6 m standing. Grid spacing is then radius × spacing factor. I use 1.41 when I want a square grid with the coverage circles meeting more generously at the diagonal; tighten it if the room is noisy or the speaker is narrow. Speaker count is: count = ceil(length / spacing) × ceil(width / spacing) Worst-case listener distance is the slant range to the edge of the coverage circle: edge distance = sqrt( (ceiling − listening height)^2 + radius^2 ) SPL at listener = sensitivity (dB @ 1 W / 1 m) + 10 log10(tap watts) − 20 log10(edge distance) I then want that SPL to clear ambient noise by a stated margin. 10 dB is common guidance for speech; the contract or the evac standard may ask for something else, and a noisy car park is a different problem from an office. Amplifier connected load = sum(speaker count × tap) + any other 100 V load on that zone required watts = connected load × (1 + headroom) pick the next standard amplifier size (I use a 30 / 60 / 120 / 240 / 360 / 480 / 650 / 1000 W list and edit it to the brand on the job) Headroom of 20–25 % is typical so you can add a few speakers later and are not running the amp at the stop. If even the largest amp on the list is short, split the zone. Cable loss (lumped load at the far end — conservative) Z = V² / P loop R = 2 × length × resistivity / area loss dB = 20 log10( (Z + R) / Z ) Copper at 20 °C is about 0.0175 ohm·mm²/m; use a higher figure in a hot ceiling void. I usually design to 0.5 dB (about 11 % power). 1 dB is about 21 %. If the run fails, upsize the conductor or split the line. Worked example — 100 V, 25 % headroom, 10 dB above ambient, 0.5 dB cable limit, 1.2 m listening height, 120° ceiling speakers, spacing factor 1.41 Open office, 36 m × 18 m, 2.8 m ceiling, 90 dB @ 1 W / 1 m, 3 W tap, 55 dB(A) ambient: - radius 2.77 m, spacing 3.91 m - 10 × 5 = 50 speakers, 150 W connected, 187.5 W with headroom → a 240 W amplifier, 90 W spare - edge distance 3.20 m, SPL 84.7 dB, need ≥ 65 dB, OK The same method on a 60 m × 40 m, 3.2 m high car park at 70 dB(A), 92 dB speakers on a 3 W tap, came out at 13 × 9 = 117 speakers, 351 W connected, 439 W with headroom, so a 480 W amplifier, and the edge SPL just cleared 80 dB. That is a lot of speakers; in practice I would look hard at a higher tap or a horn / projector in the drive lanes before I accepted a 117-way ceiling grid. The cable is what actually failed on that job sketch. A 120 W, 260 m run on 1.5 mm² (Z = 83.3 ohm, loop R = 6.07 ohm) is 0.61 dB, over a 0.5 dB limit. The same run on 2.5 mm² drops to 0.37 dB, or you split the line so each half is shorter. A 54 W office ring 85 m long on 1.5 mm² was only 0.09 dB — the long, heavily loaded run is the one worth checking first. I would be interested in what line-loss limit and what SPL-above-ambient margin other installers here are actually being asked for on BS 5839-8 / EN 54 voice-alarm jobs, versus ordinary BGM.