← All resources

Standards & formats

IRIG 106 Chapter 7: sending packets, Ethernet and video down a PCM link

Telemetry links were built for PCM: fixed frames, fixed rates, and decades of receivers, bit synchronisers and decommutators on every range. Modern installations produce packets — Chapter 10 packets, Ethernet frames, compressed video. Chapter 7 of IRIG 106 carries those packets inside an ordinary PCM stream, so the radio link and the ground equipment stay the same. Here is how it works, and what to watch when you use it.

7 min read·Updated October 2026

A PCM telemetry stream is a fixed sequence of words repeated at a fixed rate. That rigidity is its strength: every range has receivers, bit synchronisers and decommutators that lock on it, and every flight-test engineer knows how to read it. It is also its limit. Video, Ethernet traffic and Chapter 10 packets vary in size and arrive irregularly; forcing them into fixed word slots wastes bandwidth or loses data.

Chapter 7, the packet telemetry downlink, solves this without touching the link. It reserves part of each PCM minor frame for a byte stream of packets: the transmitter, the receiver and the bit synchroniser see an ordinary Chapter 4 stream, and only the equipment at both ends needs to know about packets.

What can be carried

  • Chapter 10 packets — called Chapter 11 packets in the standard since the packet format moved to its own chapter — slightly slimmed down for the link and rebuilt on the ground.
  • Raw Ethernet frames, from the destination address to the frame check sequence, whatever they carry.
  • IPv4 and IPv6 packets.
  • TmNS messages, for installations built around the network chapters of IRIG 106.
  • Application-specific packets, and two utility types: fill packets, bytes of 0xAA that keep the stream flowing when there is nothing to send, and a 12-bit test counter for checking the link.

Two layers of packets

Each source packet is wrapped in an encapsulation packet: a short header, then the packet. The header — two 12-bit words — says what the payload is, whether it is a complete packet or the first, middle or last fragment of one, how long it is in bytes, and whether a CRC-16 closes the packet. A packet of 64 kilobytes or more must be fragmented; any packet may be.

Encapsulation packets follow one another as a continuous byte stream, mixing types freely. That stream is cut into transport packets of fixed length, one per PCM minor frame. A transport packet starts with a byte carrying a stream identifier and a version, then a protected word giving the offset of the first encapsulation packet that begins inside it. A receiver that joins mid-stream, or loses a frame, finds the next packet boundary without reading everything before it.

scroll

Source packets, variable length: CH. 10 / 11 PACKET ETHERNET FRAME IP PACKET TmNS MESSAGE FILL 0xAA WHEN IDLE Encapsulation packets — Golay-protected header + one packet or a fragment, back to back: HDR packet 1 HDR packet 2 HDR packet 3 (fragment) HDR fill HDR packet 4 Transport packets — fixed length, one per minor frame; header = stream ID + offset to first EP: HDR transport packet 1 HDR transport packet 2 HDR transport packet 3 Chapter 4 PCM minor frame — the transport packet may be split between classic words: SYNC SFID PCM words TRANSPORT PACKET PCM words CRC-32 On the ground: Golay-decode, follow the offsets, reassemble — the original packets come back out.
Packet telemetry, layer by layer: source packets wrapped in encapsulation packets, cut into fixed-length transport packets, one per PCM minor frame.

For a Chapter 10 packet, the wrapping removes what the link does not need — the sync pattern, the packet length, most of the trailing fill — and keeps the channel ID, the data length, the sequence number, the relative time counter and the data unchanged. On the ground, the packet is rebuilt with its 0xEB25 sync pattern and a recomputed length, and comes out as an ordinary Chapter 10 packet that any compliant tool can read or record.

Protecting the structure, not the data

A single bit error in a length field can throw away every packet that follows it. Chapter 7 therefore encodes the structure-critical words — the encapsulation and transport packet headers, and the protected part of Chapter 10 packet headers — with the extended binary Golay code: each 12-bit word becomes 24 bits and survives up to three bit errors.

The payload gets no such protection. Errors in it are detected, not corrected: by the optional CRC-16 at the end of an encapsulation packet, by the checksums of Chapter 10 packets, and by a CRC-32 on the PCM minor frame, which the standard recommends. The link margin still decides how much data arrives intact.

Placing transport packets in the minor frame

The transport packet takes a fixed share of every minor frame, written byte by byte, most significant bit first. Editions after 106-15 allow it to be split into several segments interleaved with ordinary PCM words; the frame sync pattern, the subframe counter and the classic parameters keep their places and are not part of the transport packet. The TMATS setup record describes where the segments sit. When there is nothing to send, the encoder fills the transport packets with fill packets, so the PCM stream never stops and the ground never loses lock.

scroll

Design choiceWhat it sets
Transport packet bytes per minor frame × minor frame rateThe packet throughput; the rest of the frame stays available for classic PCM words
Size of the largest packet in the queueThe worst-case delay for the packets behind it
Low-latency encapsulation packetsShort urgent packets placed at the head of the next transport packet, ahead of a long packet being sent
Stream IDSeparates several application streams sharing the same transport packets
Encoder bufferHow large a burst — a video key frame, a burst of Ethernet traffic — can be absorbed without loss
Minor frame CRC-32Whether corrupted frames are detected before their packets are reassembled

Latency, and why low-latency packets exist

Packets go down in order. A long packet — a video frame of several hundred kilobytes — occupies the transport packets for as long as it takes to send, and a short, urgent message queued behind it waits. Chapter 7 lets the encoder insert low-latency encapsulation packets at the start of the next transport packet, ahead of the long packet being transmitted, which then resumes. A low-latency packet must fit entirely within one transport packet.

Buffering matters as much. Cameras and network sources send in bursts far above their average rate; the encoder must hold the largest burst while the fixed-rate link drains it. Sizing that buffer from the average bit rate is the classic way to lose video frames.

On the ground

A Chapter 7 stream decommutates like any PCM stream, up to the minor frame. The ground equipment then extracts the transport packet bytes, Golay-decodes the headers, follows the offsets to rebuild the encapsulation packets, reassembles fragments, and delivers what was sent: Chapter 10 packets to a recorder or an analysis tool, Ethernet frames back onto a network. Test counter packets, if the encoder sends them, measure the link directly before the real data depend on it.

Chapter 7 is a one-way downlink that keeps everything a range already owns: PCM transmitters, receivers, antennas and decommutators. The network chapters of IRIG 106, known as TmNS, go further, with IP radios, two-way traffic and remote control of the airborne units, but they need network-capable radios and ground infrastructure. To send packets through an existing PCM link, Chapter 7 is the standard way.