Notification appliance and voice system testing

A fire alarm system that detects a fire perfectly is useless if nobody hears it. On the third day at Riverlight Tower, Sasha carries a sound level meter from floor to floor while Gerry runs the evacuation message from the fire alarm control center in the lobby. In a corridor on the fifteenth floor the message is loud and clear. Inside a bedroom of suite 1503, with the door closed, it is barely louder than the air conditioner. 'It used to meet the level,' Gerry says, checking last year's record. 'What changed?' This lesson covers the inspection and testing of horns, strobes, speakers and the voice and two-way communication systems.

Inspecting notification appliances

Each horn, strobe, horn-strobe, speaker and bell is inspected for:

• presence and condition: not missing, damaged, painted over or loose on its box;
• visibility: strobes not hidden by new partitions, signs, shelving, or decorations; speakers not covered by ceiling tiles or acoustic panels added later;
• settings: candela and sound settings and speaker taps match the record drawings and the design;
• changes in the space: rooms divided, ceilings lowered, carpet or acoustic treatment added, or a quiet office turned into a noisy workshop, any of which can change how loud or visible the signals are.

Functional testing

NFPA 72 Table 14.4.3.2 requires notification appliances to be tested annually to confirm operation, and audible appliances to have their sound levels measured where the building or space has changed, or as required for acceptance or reacceptance. The technician activates each notification circuit and walks it, confirming that every appliance sounds or flashes, that strobes in the same field of view are synchronized, and that the audible pattern is correct: the temporal-three evacuation pattern, or the coded or voice signal the design calls for.

Walking every appliance is slow, so Gerry's crew uses the walk-test feature of the control unit where it allows signal circuits to be activated in short bursts. Two technicians work together: one at the control unit, one on the floor. Strobes are counted as they flash; horns and speakers are listened for. Any appliance that doesn't operate is traced: is the appliance dead, or is the circuit wired wrongly so that the appliances beyond a point are not getting power?

Measuring sound levels

A horn that sounds isn't necessarily loud enough. The sound level is measured with an A-weighted sound level meter at the locations and heights the standard specifies, and compared with the required level above the ambient sound level of the space. Measurements are made in the spaces' normal condition: with doors closed in sleeping rooms, and with the usual equipment running.

NFPA 72 requires public mode audible signals to be at least 15 dB above the average ambient sound level or 5 dB above the maximum sound level lasting 60 seconds or longer, whichever is greater, measured 5 feet above the floor; private mode signals at least 10 dB above ambient; and the total sound level not to exceed 110 dBA at the minimum hearing distance. In sleeping areas, the signal must be at least 15 dB above ambient, at least 75 dBA, and a low-frequency 520 Hz signal at the pillow.

# Audibility check from field measurements (dBA)
above = 15
sleep_min = 75
rooms = [("corridor 15", 46, 84, False), ("suite 1503 living room", 41, 71, False),
         ("suite 1503 bedroom, door closed", 38, 68, True), ("garbage room 15", 58, 80, False)]
for name, ambient, measured, sleeping in rooms:
    need = ambient + above
    if sleeping:
        need = max(need, sleep_min)
    print(f"{name:34s} ambient {ambient}  need {need}  measured {measured}  {'OK' if measured >= need else 'LOW'}")

corridor 15                        ambient 46  need 61  measured 84  OK
suite 1503 living room             ambient 41  need 56  measured 71  OK
suite 1503 bedroom, door closed    ambient 38  need 75  measured 68  LOW
garbage room 15                    ambient 58  need 73  measured 80  OK

The bedroom of suite 1503 fails. Gerry looks at last year's record: the measurement then was 77 dBA. A visit to the suite's entry hall answers the question. The tenant has replaced the suite's hollow-core bedroom door with a heavy solid-core door, and the hall horn-strobe that served the bedroom is now on the other side of a much better sound barrier. The fix isn't the tenant's door; the suite needs a sounder in the bedroom or a louder setting, and that's a design question for Ravi, so the finding goes in the report as a deficiency with a recommendation.

Strobes and visibility

Strobes are checked to confirm they flash, are synchronized, and are visible from everywhere they are meant to serve. A strobe's candela setting is compared with the drawing; a strobe set lower than designed is a deficiency even if it looks bright, because its coverage was calculated for its rated output. Where a space has changed, such as a large room divided into two, the inspection checks that each new room still has a visible signal, and reports any that don't.

Voice evacuation systems

Buildings like Riverlight Tower use voice evacuation: speakers that deliver recorded messages and live announcements, controlled from the fire alarm control center. Voice systems are tested for more than loudness:

• the pre-recorded messages play correctly, in the right sequence for each zone, with the alert tone and evacuation tone the design calls for;
• live announcements from the microphone reach each zone selected, and override recorded messages as designed;
• speakers are not just loud but intelligible: a message that is loud but garbled by echo or distortion fails;
• the amplifiers are supervised, including the backup amplifier, and switch over on failure;
• each speaker circuit is supervised, and faults are reported.

NFPA 72 covers emergency communication systems in its voice and testing requirements. Intelligibility is checked by listening in each acoustically distinguishable space; where measurements are needed, they use a method such as the speech transmission index, with a target commonly given in the code's annex of an average of at least 0.50 and few points below 0.45.

At Riverlight, Gerry plays each floor's messages, then makes a live announcement on each floor from the control center while Sasha stands in the corridor and in one suite per floor.

Two-way communication

High buildings also have two-way communication for firefighters: firefighter telephones or radio enhancement systems, or both. Firefighter telephone jacks or handsets are tested at each location, confirming a clear two-way conversation with the control center and that the control center shows which location is calling. Stairwell communication for people waiting at an area of refuge, where provided, is tested the same way. Emergency responder radio coverage systems, where installed, are tested by their own specialists, but the fire alarm technician confirms the supervisory signals they send to the fire alarm panel.

# Firefighter telephone test log: every location must give a clear two-way call
locations = [f"stair A floor {n}" for n in range(1, 19)] + [f"stair B floor {n}" for n in range(1, 19)]
results = {loc: "clear" for loc in locations}
results["stair A floor 7"] = "no ring-in at control center"
results["stair B floor 13"] = "one-way only"
bad = {k: v for k, v in results.items() if v != "clear"}
print("locations tested:", len(results))
for k, v in bad.items():
    print("FAIL", k, "-", v)

locations tested: 36
FAIL stair A floor 7 - no ring-in at control center
FAIL stair B floor 13 - one-way only

Voltage at the end of the circuit

FAT-130 showed how voltage drop is calculated when a circuit is designed. The inspection can measure it. With the system on battery and all appliances on the circuit sounding, the voltage is measured at the last appliance. It must stay at or above the lowest voltage the appliances are listed for, 16 volts for typical 24-volt appliances. A circuit that met its design voltage when new can fall short later if appliances were added, wiring was damaged, or the batteries feeding it are weak.

# Voltage at the last appliance, on battery, in alarm
min_V = 16.0
circuits = [("RPS-9 NAC 1", 19.2), ("RPS-9 NAC 2", 17.4), ("RPS-14 NAC 1", 15.6), ("panel NAC 3", 20.1)]
for name, v in circuits:
    print(f"{name:14s} {v:5.1f} V  {'OK' if v >= min_V else 'BELOW 16 V - check load, wiring and batteries'}")

RPS-9 NAC 1     19.2 V  OK
RPS-9 NAC 2     17.4 V  OK
RPS-14 NAC 1    15.6 V  BELOW 16 V - check load, wiring and batteries
panel NAC 3     20.1 V  OK

One circuit on RPS-14 is low. Gerry checks the circuit's load: two strobes were added in a renovated amenity room without a new calculation. The circuit needs to be split or the added strobes moved to another circuit, and the change goes to Ravi.

Amplifiers and backup amplifiers

Voice systems depend on amplifiers, and amplifiers fail. Most voice systems have a backup amplifier that takes over when a primary amplifier fails, and the control unit supervises each amplifier. The test removes or disables a primary amplifier, following the maker's method, and confirms that the trouble is reported, that the backup takes over, and that every speaker circuit the failed amplifier served still plays the message at full level. The amplifiers' battery supply is load-tested like any other, because a voice system in alarm draws far more current than one in standby.

Speaker circuits are supervised for opens, shorts and ground faults, like other circuits, and each fault type is created and confirmed at the control unit. A speaker circuit that has been tapped, with a branch added during a renovation without being wired in and out, may still play the message but leave part of the circuit unsupervised; the inspection finds it when an open on the branch doesn't produce a trouble.

Low-frequency signals and strobes in sleeping areas

Recent editions of the installation standards require a low-frequency audible signal, around 520 Hz, in sleeping areas, because it wakes people more reliably, including people with hearing loss. Sleeping rooms with visual signals for people who are deaf or hard of hearing need high-intensity strobes. The inspection confirms that the appliances installed in sleeping areas are the type the design and the code require, and that replacements over the years, often bought by the building's own staff, haven't swapped in a standard horn or a lower-intensity strobe.

# Check sleeping-area appliances against the design
design = {"suite 1503 bedroom": ("520 Hz sounder", 177), "suite 806 bedroom": ("520 Hz sounder", 177), "suite 1210 bedroom": ("520 Hz sounder", 177)}
found = {"suite 1503 bedroom": ("520 Hz sounder", 177), "suite 806 bedroom": ("standard horn", 177), "suite 1210 bedroom": ("520 Hz sounder", 75)}
for room, (kind, cd) in design.items():
    k2, c2 = found[room]
    ok = (k2 == kind and c2 >= cd)
    print(f"{room:20s} design {kind}, {cd} cd  found {k2}, {c2} cd  {'OK' if ok else 'WRONG TYPE OR SETTING'}")

suite 1503 bedroom   design 520 Hz sounder, 177 cd  found 520 Hz sounder, 177 cd  OK
suite 806 bedroom    design 520 Hz sounder, 177 cd  found standard horn, 177 cd  WRONG TYPE OR SETTING
suite 1210 bedroom   design 520 Hz sounder, 177 cd  found 520 Hz sounder, 75 cd  WRONG TYPE OR SETTING

Recording what was measured

Each audibility measurement, strobe check and voltage reading is recorded with its location, the ambient level, the measured level and the requirement, so that the next inspection can compare. Over several years these records show which spaces have become quieter or louder, which circuits are losing voltage as batteries age, and where renovations have weakened coverage. A report that says only 'audibility OK' throws that history away.