When a facility moves from traditional connectors to audio over IP, it gains flexibility but inherits a new anxiety: what if the network drops? A failing analogue cable you hear and replace. An IP packet lost at the wrong moment, though, turns into an on-air hole with no warning. SMPTE ST 2022-7 is the broadcast industry’s answer to that fear: redundancy designed so the listener never hears the failure.
Why redundancy is vital in broadcast
In TV or radio production, the requirement isn’t “rarely down”, it’s “never on air”. A standard data network can reroute a stream after an outage perfectly well — but “after” takes hundreds of milliseconds, sometimes seconds. That’s invisible for a file transfer, catastrophic for a live feed. What was needed was redundancy that switches not in a moment but instantly, and completely transparently. That is exactly the brief ST 2022-7 fulfils.
What is ST 2022-7?
ST 2022-7 is an SMPTE standard describing seamless protection switching. The principle fits in one sentence: the sender transmits two strictly identical copies of the same stream over two separate network paths, and the receiver recombines them in real time to output a single, perfect one. If either path loses packets or fails completely, the other fills the gaps instantly. No renegotiation, no resync: continuity is guaranteed packet by packet.
The standard isn’t audio-only: it protects video streams just as well as the audio streams of the broadcast IP ecosystem. That’s what makes it the common safety net for an entire facility moved onto IP.
How seamless protection works
The mechanism rests on the sequence numbers carried by every packet. The two copies of the stream travel with the same numbering. At the receiver, a recombination engine — often called a hitless merge — reads both arrivals and rebuilds the correct sequence: for each expected packet it takes the first valid copy received, whichever branch it came from, and discards the duplicate. If a packet is missing on one path, it is simply recovered from the other.
For that sorting to work, the receiver holds a small buffer that absorbs the arrival gap between the two paths. The two copies don’t necessarily travel routes of equal length: one may cross an extra switch and arrive a fraction of a millisecond later. The standard sets a tolerance for this skew; beyond it, recombination can no longer guarantee continuity. Hence the importance of designing two paths with comparable transit times.
What the network needs
ST 2022-7 only makes sense if the two paths are genuinely independent. So you build two separate networks, conventionally named “red” and “blue”: two distinct switches, two distinct cabling runs, ideally two distinct power supplies. The idea is that no single failure — a switch rebooting, a cord unplugged, a blown power unit — can cut both branches at once. Duplicating streams through a single switch protects nothing: that’s putting both copies in the same basket.
One thing must not be duplicated: the clock. Both paths must share the same PTP time reference so packets stay coherent. PTP synchronisation is the foundation the whole chain rests on; ST 2022-7 protects the transport, not the timebase, which must stay healthy and common to both branches.
ST 2022-7 and the other standards
Seamless redundancy replaces none of the transport standards: it layers on top of them. You’ll find it above audio over IP as well as video, and it combines naturally with the ecosystem you already know.
| Building block | Role |
|---|---|
| AES67 / ST 2110 | Transport of audio (and video) essences over IP |
| PTP (IEEE 1588) | Common timebase, equipment synchronisation |
| ST 2022-7 | Transport redundancy: two paths, seamless recombination |
In other words, you describe your streams with the transport standards, synchronise them with PTP, and secure them with ST 2022-7. The three layers are complementary. Large Dante environments also have their own dual-network redundancy in the same spirit: two ports, two networks, transparent switchover.
Setting up healthy redundancy
- Physically separate the two paths end to end: independent switches, cables and, where possible, power. Logical redundancy on a single piece of hardware is an illusion.
- Control the transit skew between branches: two routes of comparable length and load stay inside the recombination window.
- Look after PTP: provide a master clock and a backup, because all the coherence rests on a common timebase.
- Test failures for real. Unplug a branch in service and confirm by ear that nothing happens — that’s the only test that counts.
- Document the network plan. A poorly mapped redundant stream becomes unmanageable on incident day; the same goes for your MADI links and SDI islands.
My take
ST 2022-7 is one of those technologies you only notice when it’s missing. Deployed well, it turns a legitimate fear — the network dropping mid-broadcast — into a non-event: a branch fails, nobody hears it, and you fix it calmly. The real work isn’t in the standard, which is elegant and proven, but in the network discipline it demands: two truly separate paths, flawless PTP, and honest failure testing. Skip that discipline and you’ll have redundancy in name only. Respect it, and you’ll sleep well the night before a big live show — which, in this job, is priceless.
