How to Plan a Live Signal Path That Stays on Air

Not every step deserves a backup, and trying to duplicate everything makes a setup heavier and harder to understand. Look at the ranking from your drawing and choose the points where a failure would end the feed. Those usually include the main output from the mixer, the encoder, the power supply and the final link to the broadcaster.

A second route should be genuinely separate. Two cables laid side by side through the same trunking will both be cut by the same trolley wheel. Two encoders plugged into one power strip fail together when the strip trips. When Daniel Whitcombe builds a backup path, he asks one question: what single event would take out both routes? If the answer is anything at all, the routes are not yet independent.

Where a full second route is out of budget, a cold spare is a fair compromise. A spare encoder in a labelled case, already configured and within arm's reach, turns a long outage into a short one.

Monitoring should answer one question: is the picture that leaves the building good? Checking that a device has a green light is not the same thing. A green light means the box is powered, not that the pictures are right. We place a monitor at the end of the chain, fed from the delivered signal rather than from the mixer, so the operator sees what the broadcaster sees.

Alongside the picture, watch the technical readings that warn you before a failure. Audio levels drifting, a bitrate dropping, or a link reporting repeated errors all show up minutes before the picture breaks. A person who is watching these readings can switch to the backup calmly, with no visible interruption.

Assign monitoring to a named person. When everybody is partly watching, nobody notices the small change. One operator with a clear view of the output and the readings is worth more than three people glancing at a screen in between other tasks.

A backup that has never been used is a guess. Include a switch to the second route in every rehearsal, and time it. The aim is that the operator can move the feed across without hesitation, and that the broadcaster sees nothing worse than a brief change in the picture, or nothing at all.

Agree in advance who decides to switch and what triggers it. If the rule is that a dropped link for more than a few seconds means going to backup, then the operator does not need to ask permission while a producer is waiting. Written triggers remove the pause that makes small faults grow into long ones.

After the switch, the work is not over. Someone should find out why the main route failed while the backup carries the feed, then repair it and test it, so that the protection is restored before the next fault arrives. Many outages become serious because a second fault hits while the first one is still unrepaired.

These weeks bring damp ground, early darkness and cold mornings, and each affects a location setup. Connectors that were fine in summer collect moisture, and a cable lying on wet grass can behave very differently from the same cable indoors. Use covered junctions, raise connectors off the ground, and tape or sleeve any joint that cannot be avoided outside.

Cold also shortens the life of batteries and can make some equipment slow to settle after switching on. Power up devices early and let them reach a steady state long before the feed is due. A unit that has only just woken up is more likely to drop a link under load.

Long cable runs deserve a test of their own. Run the full length with a live signal for a good while, not a quick glance, and look for errors rather than a picture that merely appears. A run that passes a one-minute check can still show faults after an hour when the temperature has changed.

After each job, write down what was used, what failed and what was changed. A short log kept in a shared place is among the most useful items in our kit. It shows which cable has failed twice, which encoder setting caused trouble, and which location needs extra protection from the weather.

Write the entries plainly, as if for a colleague who was not there. Note the time, the symptom and the fix, with no attempt to assign blame. A log that people are afraid to fill honestly stops being useful quite quickly.

Review it before the next similar job. Patterns appear that nobody noticed on the day, and the plan for the next feed becomes a little stronger each time. Reliability is rarely one clever decision. It is a long series of small, recorded corrections.