Cold Start Discipline For 24 Hour AV Racks

A call that a room is dead is usually not a hardware failure. It is the consequence of the order in which things came back. After a power cut, a breaker trip or a maintenance shut down, the network switch, the matrix, the decoders and the displays all begin booting according to whatever the circuits allow, and the sequence they land in determines whether the room works or presents a black screen with a lit power LED. Designing for that moment is cheaper than the second visit it prevents, and it is mostly a matter of three things: current, order and hot plug behaviour.

Current first. A populated rack does not draw its running load at the instant the breaker closes. Power supplies charge their input capacitors in the first few hundred milliseconds, and the aggregate inrush of a matrix, a PoE switch and a dozen displays can be several times the steady state figure, which is how a circuit that is comfortable all afternoon trips on restoration. PoE makes the curve stranger still, because the switch's load only arrives as endpoints boot and negotiate: eight decoders at 12 W each plus two cameras is a load that appears in stages over the first minute, and if the switch and the decoders are on the same circuit the sequencing has to account for both. Measure the inrush rather than assuming it, and put displays on a delayed stage so that the room comes up in a known order.

Order is where video usually fails. A source that powers up before its sink is connected sees no hot plug detect, and a matrix that comes up before the display may latch the default EDID from the last session or fall back to a safe 1080p timing. When the display finally arrives, nothing renegotiates, and the room sits at the wrong resolution until somebody power cycles the matrix. The fix is unglamorous: fixed EDID per output set to the native timing of the sink, forced hot plug detect, and a documented boot order of network, then sources, then matrix, then displays. On decoders and extenders, the same logic applies at the far end - a decoder that boots into a link that is not yet up should retry rather than settle into a permanent no signal state.

Recovery then becomes policy rather than improvisation. A watchdog that pings the room controller and power cycles the media circuit when the room has been silent for a defined interval will clear most overnight faults, provided the reboot count is capped and someone is alerted on the second event rather than the tenth. A nightly scheduled reboot of unattended rooms is unfashionable and effective, particularly where consumer grade sources are involved. Size the UPS from measured draw, not nameplate sums, and keep the control processor and the PDU on a circuit that the media devices do not share. Then hand over a document that lists the boot order, the breaker map and the recovery steps, because the person who gets the call at 8am is rarely the person who designed the rack. Amtune hybrid matrices, AV over IP decoders and signal converters ship with documented consumption and fixed EDID options for exactly this kind of estate. Commissioning checklists from sales@amtune.com.