IRIG-B: a 1950s time code still on board

Born to correlate tapes

In 1951, the US test ranges created the Range Commanders Council to harmonise how they worked, and in March 1952 the Inter-Range Instrumentation Group, IRIG, to standardise their instrumentation. A test was observed from many sites — cameras, tracking stations, telemetry receivers — each recording on its own tape. Comparing those recordings meant knowing exactly when every sample was taken, so a common time code was recorded on a track next to the data. IRIG's telecommunications working group began work on it in October 1956, and the codes were accepted in 1960.

The design shows its origin. Format B sends 100 bits per second: slow enough to ride on a 1 kHz carrier through a voice-grade channel, and to be recorded on an ordinary tape track. The standard has been revised many times since; from 1970 it is called IRIG Standard 200.

A family of codes

IRIG-B is one member of a family that shares the same structure at different speeds.

Format B won because it beats with the second itself: one frame per second is easy to decode, easy to check by eye on an oscilloscope, and precise enough for most uses.

One second, one hundred slots

Each 10 ms slot carries one pulse: 2 ms for a binary 0, 5 ms for a binary 1, 8 ms for a position marker. A marker every ten slots keeps the decoder aligned, and two markers in a row — the last of one frame, the first of the next — announce the new second, which begins on the leading edge of that reference marker.

Within the frame, the time of year is written in binary-coded decimal: seconds, minutes, hours and day of year. Since 2004 the year follows. The control functions are left for the user to define, and the frame ends with the seconds of the day as a plain binary number, from 0 to 86,399, convenient for software.

The full name is a letter and three digits. The first digit gives the modulation: 0 for DC level shift, 1 for a sine-wave carrier modulated in amplitude, 2 for Manchester. The second gives the carrier: 2 means 1 kHz, 0 means none. The third says which fields are present. B002 is therefore unmodulated with the time of year only; B122 carries the same content on a 1 kHz carrier.

How accurate is it?

• Its first promise was modest: about a millisecond, one period of the carrier, on a tape recorder.
• DC level shift, the pulses as voltage levels on a wire: the on-time edge can be resolved to well under a microsecond, tens of nanoseconds with good hardware. The 2004 edition limits jitter to 200 ns from pulse to pulse.
• Amplitude modulation, the pulses as the amplitude of a 1 kHz sine wave in a ratio of 10 to 3: robust on long cables and through transformers, but the decoder has to work from the carrier, and the result is typically good to tens of microseconds.
• The cable adds its own delay, a few nanoseconds per metre: it matters only when microseconds do. Common practice keeps DC level shift runs under about 100 m and modulated runs under about 300 m.

Why it is still on board

Flight-test recorders still accept IRIG-B, and an IRIG 106 Chapter 10 recording notes which source stamped its data: IRIG-B, GPS or an internal clock. Power utilities time-stamp events in their substations with it. And it survives next to GPS and PTP for reasons that have nothing to do with nostalgia.

• It is one-way. A source drives a wire, receivers listen: nothing to negotiate, nothing to configure on a network.
• One generator feeds many receivers through simple isolated inputs.
• It can be checked with an oscilloscope and decoded with a few lines of logic.
• Decades of equipment speak it, on aircraft and in ground stations.

It has limits too. Plain IRIG-B does not say whether it carries UTC or local time, and it has no standard field for leap seconds or for the quality of its source. The extensions written for power networks, IEEE 1344 and later IEEE C37.118, use the control functions to carry a leap-second warning and a time-quality indication. A flight-test installation has to agree on such conventions explicitly, between the generator and every receiver.

And now, no more leap seconds

In 2026, the General Conference on Weights and Measures decided that UTC will become continuous from 20 May 2027: no more leap seconds, the gap between UTC and the rotation of the Earth being allowed to grow to as much as an hour. For IRIG-B, the leap-second warnings of the extended formats will simply fall silent. For a fleet of recorders, it removes one of the rare moments when time codes, GPS receivers and network clocks could disagree by a second.

Using IRIG-B well

1. Agree on the content: UTC or local time, which control functions, whether the year is sent.
2. Choose the modulation by distance and accuracy: DC level shift on short runs when microseconds matter, amplitude modulation for long or noisy runs.
3. Distribute from one source, and measure the cable delay when the accuracy budget is tight.
4. Record the time source with the data, as Chapter 10 does, so that the analysis knows what it can rely on.
5. Look at it: an oscilloscope on the line, and one frame decoded by hand, settle most arguments.