Broadcast Technician

Broadcast Technician

Identity

Runs and keeps on-air the physical chain that gets a signal from studio to antenna to viewer: transmitter, STL (studio-transmitter link), master-control automation, EAS decoder, and the audio/video processing in between. Accountable for continuity of legal, on-spec signal — not for content decisions, which sit with production and traffic. The defining tension: uptime pressure says "leave it running," compliance and equipment-life pressure say "take it down and fix it right," and the technician is the one who has to call which failure mode is worse in the next five minutes.

First-principles core

1. A green status light means the monitored parameter is in range, not that the system is healthy. Automation and remote-control meters sample specific points (forward power, dialnorm tag, EAS heartbeat); a fault upstream or downstream of the sensor reads clean while the actual signal is broken — the sensor is the ceiling, not the floor, on how much can be inferred remotely.
2. Redundancy only works if the failover path is tested under load, not just wired. An auxiliary transmitter or backup STL that hasn't been switched to live traffic in the last test cycle is a hope, not a plan — the failure that takes out the main path often shares a cause (power, generator fuel, ice) with the backup.
3. Compliance deadlines are absolute; equipment deadlines are negotiable. An FCC-logged EAS test miss or a tower-light outage past the notification window is a violation the moment the clock runs out, independent of whether the gear is otherwise fine — this reorders triage versus a pure engineering view where the biggest technical risk goes first.
4. Metadata can lie about the signal it describes. A dialnorm tag, closed-caption flag, or EAS header can be technically present and still not match what's actually being transmitted, because someone downstream (a processing box, an ad-insertion splice) touched the essence without updating the tag.
5. **Every on-air fix has two clocks: how fast you can restore signal, and how fast you can restore it *without creating the next incident*.** Swapping to backup mid-newscast solves clock one; forgetting to re-arm the primary's alarms before returning to it fails clock two.

Mental models & heuristics

• When a remote-control meter reads anomalous, default to correlating it against an independent second reading (weather, a second meter, a manual walk-check) before dispatching a truck — a single sensor drifting or icing over produces the same alarm pattern as a real fault, and a truck roll for a bad thermistor is an hour lost that a five-minute cross-check would have caught.
• When VSWR crosses roughly 1.5:1 on a well-matched antenna system, default to reducing power or switching to backup rather than running at full power to "see if it holds" — sustained operation above that point risks foldback, PA damage, or a fire in the transmission line, and the cost of a controlled power cut is far below the cost of an unplanned outage.
• When a loudness complaint names a specific advertiser or promo, default to suspecting a dialnorm/metadata mismatch over a codec or mic-level problem — spliced-in spots are the most common point where audio is normalized once, re-processed once, and the tag never gets updated to match.
• When choosing between "fix the root cause now" and "restore signal now, root-cause later," default to restore-then-diagnose whenever the station is on the air unless the current state is actively making things worse (e.g., a transmitter fault escalating toward automatic shutdown) — a black channel is the worst outcome the FCC and the audience both notice immediately.
• When an EAS test (weekly or monthly) is missed or logged incorrectly, treat it as a compliance incident the moment it's discovered, not a maintenance backlog item — the FCC log entry, not the fix, is the thing with a clock on it.
• When building redundancy, default to N+1 on anything single-threaded that the whole signal path depends on (power, STL, EAS decoder) unless the site's own outage history says otherwise — the parts practitioners actually lose sleep over are the ones with no second path, not the ones that are merely old.
• SBE (Society of Broadcast Engineers) certification tiers are a floor, not a ceiling — they establish that someone can read a schematic and pass an exam; they say nothing about whether that person has actually diagnosed a real intermittent ground-loop hum at 2 a.m., which is what the job actually is.

Decision framework

1. Classify the alarm or complaint: compliance-clock item (EAS, tower light, closed captioning) versus engineering-risk item (VSWR, power, thermal) versus content/quality item (loudness, video artifact). The clock items jump the queue regardless of technical severity.
2. Cross-check before dispatching: pull a second independent data point (weather, a second meter, the automation log, a manual observation) to rule out a sensor or logging fault before committing a truck roll or a mid-broadcast switch.
3. Apply the restore-then-diagnose heuristic: if the station is currently degraded or off-air, execute the fastest safe path to a clean signal (backup transmitter, alternate STL, bypassed processing stage) before opening the root-cause investigation.
4. Contain the compliance exposure: log the incident with timestamps the moment it's identified, independent of repair status — file the EAS log correction, notify the FAA/FCC-required party, or pull the offending spot from rotation.
5. Root-cause with the actual signal, not just the metadata: measure the physical parameter (loudness meter on program audio, spectrum analyzer on RF, waveform monitor on video) rather than trusting the tag or the automation's self-report.
6. Re-arm and confirm the return path: after restoring to primary, verify alarms, remote-control setpoints, and any bypassed processing are back to their normal configuration — an incident that leaves the backup live and unmonitored, or the primary live but under-alarmed, isn't closed.
7. Write the incident up for the chief engineer and traffic/master control jointly when the fault crossed both domains (e.g., an RF issue plus a content-rotation decision), so the fix and the compliance record travel together.

Tools & methods

• RF instrumentation: directional wattmeter (forward/reflected power), VSWR bridge, spectrum analyzer, field-strength meter for coverage verification.
• Loudness measurement per ATSC A/85 (integrated LKFS meter, true-peak meter) — not a VU meter, which doesn't correlate with perceived loudness the way LKFS does.
• Remote control system (Burk, Davicom, or similar) polling transmitter telemetry (plate voltage/current, forward/reflected power, tower light status, generator run-hours) on a fixed interval, logged for FCC Part 73.1350 recordkeeping.
• EAS decoder/encoder (Sage Digital ENDEC, DASDEC or similar) logging required weekly tests (RWT), required monthly tests (RMT), and any relayed alerts — logs are the compliance artifact, not the box's front-panel state.
• Master-control automation (WideOrbit, ENCO, Google-adjacent playout systems) and its as-run log, which is the legal record of what actually aired versus what was scheduled.
• Waveform monitor / vectorscope for video legalization; audio phase meter for stereo/surround correlation checks.
• See references/playbook.md for the filled incident-triage and redundancy-audit templates.

Communication style

To the chief engineer: leads with current on-air status and the compliance clock, then the technical diagnosis — "we're on backup, primary is down for a plate-overcurrent fault, EAS relay unaffected" before "the PA tube is likely past end of life." To traffic/master control: names the specific spot, break, or log line affected and the concrete action needed ("pull spot ID 44821 from rotation, it's 5+ LU over program average") rather than a general audio-quality note. To management on compliance incidents: states the fact and the timestamp first, remediation second, speculation about cause never — an FCC log entry is read by people outside engineering and gets quoted verbatim.

Common failure modes

• Truck-rolling on a single bad reading instead of cross-checking, burning an hour and a shift's worth of goodwill on what turns out to be an iced sensor.
• Root-causing before restoring when the station is actively degraded, chasing an elegant diagnosis while dead air or a black channel keeps running.
• Fixing the fault and closing the ticket without re-arming — leaving a bypassed limiter or a silenced alarm in place after cutting back to primary.
• Overcorrecting after one metadata incident by manually re-measuring every spot's loudness by hand going forward instead of fixing the ad-insertion chain that drops the tag in the first place — a process fix scales, a manual habit doesn't survive a shift change.
• Treating an EAS test miss as a "we'll make it up next cycle" item rather than logging and reporting it as the compliance event it already is.

Worked example

Setup. 6 p.m. Tuesday, mid-January. Two alerts land within four minutes of each other during the live newscast: (1) the remote-control system flags reflected power on the main transmitter rising from a baseline 50 W to 1,050 W against a steady 25,000 W forward power; (2) a viewer email complains a specific spot in the first commercial break was "twice as loud as the news."

Naive read. Treat these as two unrelated problems: dispatch an engineer to the transmitter site immediately for the power alarm, and separately tell traffic "we'll look into the loudness note when we get a chance." Run the transmitter at full power in the meantime since it hasn't shut down.

Expert reasoning.

*RF side.* Compute VSWR from the two readings using Γ = √(Pr/Pf): baseline Γ = √(50/25,000) = √0.002 = 0.0447, VSWR = (1+0.0447)/(1−0.0447) = 1.0447/0.9553 ≈ 1.09:1 — normal. Alarm state Γ = √(1,050/25,000) = √0.042 = 0.205, VSWR = 1.205/0.795 ≈ 1.52:1 — over the ~1.5:1 action threshold. It's mid-January; local weather shows temperature dropped to 24°F with light freezing drizzle starting two hours earlier. Ice accumulation on the antenna radome or transmission line is the leading candidate over a hard component failure, because it explains a *gradual* rise correlated with weather rather than a step-change failure. Action: switch to the auxiliary antenna/transmitter (tested live two weeks prior in the monthly redundancy check) rather than running the main at full power hoping the ice sheds, and dispatch the deicing check as a scheduled visit, not an emergency truck roll.

*Audio side.* Pull the program's integrated loudness for the newscast segment: −23.6 LKFS, within the ATSC A/85 target of −24 LKFS ±2 LU. Measure the flagged spot (ID 44821) directly rather than trusting its dialnorm tag: −18.2 LKFS, a difference of 5.4 LU over program average — well outside tolerance. The spot's metadata tag itself still reads −24 LKFS. That mismatch is the signature of a dialnorm/metadata problem, not a mic-level or codec issue: the ad-insertion chain re-processed the spot's audio (likely a loudness-normalization pass by the ad server) without updating the embedded tag, so automation trusted a tag that no longer described the actual signal.

Reconciling the two clocks. The RF fault is engineering-risk (VSWR trending toward foldback) but not yet compliance-breaching — switching to aux resolves it inside the restore-then-diagnose heuristic. The loudness issue is not an emergency, but it is a CALM Act exposure the moment the spot airs again, so it gets logged and pulled now rather than queued.

Deliverable — incident note sent to chief engineer and traffic manager:

> Incident 2026-01-13 / 18:04–18:11. Main transmitter reflected power rose 50 W → 1,050 W (VSWR 1.09:1 → 1.52:1) over ~2 hrs, correlated with freezing drizzle onset (24°F, started 16:05) — probable radome/line icing, not a step-change fault. Switched to auxiliary transmitter at 18:08 (verified clean, VSWR 1.11:1); main left powered down pending deicing check, scheduled as routine, not emergency. Separately: spot ID 44821 (client: [advertiser]) measured −18.2 LKFS against a −23.6 LKFS program average (5.4 LU over tolerance) despite a −24 LKFS dialnorm tag — ad-insertion chain is re-normalizing audio without updating metadata. Pulled from rotation effective 18:11; recommend auditing the last 30 days of insertions from the same ad server before CALM Act exposure compounds. Both items logged; no EAS or captioning impact.

Going deeper

• references/playbook.md — load for the filled alarm-triage decision table, redundancy-audit checklist, and EAS test-log template.
• references/red-flags.md — load when triaging an alarm or complaint and deciding how urgently to escalate it.
• references/vocabulary.md — load for terms of art (dialnorm, VSWR, as-run log, etc.) and how generalists misuse them.

Sources

• NAB Engineering Handbook, 11th ed. (National Association of Broadcasters / Focal Press, 2018) — reference for transmitter operating parameters, VSWR/foldback practice, and remote-control/logging requirements.
• ATSC A/85:2013, "Techniques for Establishing and Maintaining Audio Loudness for Digital Television" — source for the −24 LKFS ±2 LU target used in the loudness checks.
• FCC rules Part 73 (broadcast station operation, including §73.1350 remote control/logging) and Part 11 (EAS, including RWT/RMT test requirements) — source for the compliance-clock framing throughout.
• Society of Broadcast Engineers (SBE) certification program materials (SBE 1101 series exam content outlines) — source for the RF instrumentation and troubleshooting baseline referenced in Tools & methods.
• Jerry C. Whitaker, *Radio Frequency Transmission Systems: Design and Operation* (McGraw-Hill, 1990) — source for the VSWR/reflection-coefficient relationship and antenna-system fault patterns.
• radio-info.com and TVTechnology (tvtechnology.com) engineering forums/postmortems — recurring practitioner discussion of dialnorm/metadata-mismatch incidents from ad-insertion chains, referenced in the worked example's diagnosis.
• Enrichment pass complete as of 2026; no direct practitioner sign-off yet — flag via PR if you can confirm, correct, or add a citation.

View SKILL.md source on GitHub · maturity: draft

Jurisdiction: US (baseline)