hfmodemDocumentation

HF telegraphy reference tables: ITA2, CCIR 476 / SITOR, NAVTEX

Sources (primary, fetched and extracted with pdftotext -layout for exact transcription, not OCR/vision-only reads):

Cross-checks against independent, non-ITU sources:

No unresolved disagreements were found between ITU-R M.476 and M.625 on the 7-unit code tables (they are byte-for-byte identical where both define a signal). One version difference, not a disagreement, is flagged in §3.

1. ITA2 (ITU-T S.1)

1.1 Coding rules

1.2 Full table (combo_no, letters, figure_intl, bits12345)

ITA2_TABLE = [
    # (combination_no, letter_case, figure_case_ITA2, bits "12345" A/Z)
    (1,  "A", "-",              "ZZAAA"),
    (2,  "B", "?",              "ZAAZZ"),
    (3,  "C", ":",              "AZZZA"),
    (4,  "D", "WRU",            "ZAAZA"),   # who-are-you / answerback, printed as an enquiry glyph
    (5,  "E", "3",              "ZAAAA"),
    (6,  "F", None,             "ZAZZA"),   # unassigned internationally
    (7,  "G", None,             "AZAZZ"),   # unassigned internationally
    (8,  "H", None,             "AAZAZ"),   # unassigned internationally
    (9,  "I", "8",              "AZZAA"),
    (10, "J", "BEL",            "ZZAZA"),   # audible signal
    (11, "K", "(",              "ZZZZA"),
    (12, "L", ")",              "AZAAZ"),
    (13, "M", ".",              "AAZZZ"),
    (14, "N", ",",              "AAZZA"),
    (15, "O", "9",              "AAAZZ"),
    (16, "P", "0",              "AZZAZ"),
    (17, "Q", "1",              "ZZZAZ"),
    (18, "R", "4",              "AZAZA"),
    (19, "S", "'",              "ZAZAA"),
    (20, "T", "5",              "AAAAZ"),
    (21, "U", "7",              "ZZZAA"),
    (22, "V", "=",              "AZZZZ"),
    (23, "W", "2",              "ZZAAZ"),
    (24, "X", "/",              "ZAZZZ"),
    (25, "Y", "6",              "ZAZAZ"),
    (26, "Z", "+",              "ZAAAZ"),
    (27, "CR", None,            "AAAZA"),   # carriage return, either case
    (28, "LF", None,            "AZAAA"),   # line feed, either case
    (29, "LTRS", None,          "ZZZZZ"),   # letter shift
    (30, "FIGS", None,          "ZZAZZ"),   # figure shift
    (31, "SPACE", None,         "AAZAA"),
    (32, "NUL", None,           "AAAAA"),   # all-space / unperforated tape / blank
]

1.3 US-TTY figure-case variant

Confirmed against dl-fldigi/src/cw_rtty/rtty.cxx (letters[32] / figures[32] arrays — the figures table is explicitly the "U.S. version"). Exactly 8 combinations differ from the ITA2 figure case above:

USTTY_FIGURE_OVERRIDES = {
    # combo_no: (ITA2_intl_figure, USTTY_figure)
    4:  ("WRU", "$"),
    6:  (None,  "!"),
    7:  (None,  "&"),
    8:  (None,  "#"),
    10: ("BEL", "'"),   # apostrophe moves here...
    19: ("'",   "BEL"), # ...and BEL moves here
    22: ("=",   ";"),
    26: ("+",   '"'),
}

All other letters/figures (including A–C, E, I, K–P, R, T, U, W, X, Y and all control characters) are identical between ITA2 and US-TTY.

2. CCIR 476 / ITU-R M.476 & M.625 seven-unit code

2.1 Coding rules

2.2 Table 1 — the 32 traffic combinations

Verified via pdftotext -layout text-layer extraction of both M.476-5 and M.625-4 — byte-for-byte identical between the two Recommendations for every row.

# (combo_no, letter_case, figure_case, ITA2_bits12345, seven_unit_BY, seven_unit_binary)
# seven_unit_binary uses B=0, Y=1, bit position 1 transmitted first (leftmost char).
CCIR476_TABLE = [
    (1,  "A",    "-",       "ZZAAA", "BBBYYYB", "0001110"),
    (2,  "B",    "?",       "ZAAZZ", "YBYYBBB", "1011000"),
    (3,  "C",    ":",       "AZZZA", "BYBBBYY", "0100011"),
    (4,  "D",    "WRU",     "ZAAZA", "BBYYBYB", "0011010"),
    (5,  "E",    "3",       "ZAAAA", "YBBYBYB", "1001010"),
    (6,  "F",    None,      "ZAZZA", "BBYBBYY", "0010011"),
    (7,  "G",    None,      "AZAZZ", "BYBYBBY", "0101001"),
    (8,  "H",    None,      "AAZAZ", "BYYBYBB", "0110100"),
    (9,  "I",    "8",       "AZZAA", "BYBBYYB", "0100110"),
    (10, "J",    "BEL",     "ZZAZA", "BBBYBYY", "0001011"),
    (11, "K",    "(",       "ZZZZA", "YBBBBYY", "1000011"),
    (12, "L",    ")",       "AZAAZ", "BYBYYBB", "0101100"),
    (13, "M",    ".",       "AAZZZ", "BYYBBBY", "0110001"),
    (14, "N",    ",",       "AAZZA", "BYYBBYB", "0110010"),
    (15, "O",    "9",       "AAAZZ", "BYYYBBB", "0111000"),
    (16, "P",    "0",       "AZZAZ", "BYBBYBY", "0100101"),
    (17, "Q",    "1",       "ZZZAZ", "YBBBYBY", "1000101"),
    (18, "R",    "4",       "AZAZA", "BYBYBYB", "0101010"),
    (19, "S",    "'",       "ZAZAA", "BBYBYYB", "0010110"),
    (20, "T",    "5",       "AAAAZ", "YYBYBBB", "1101000"),
    (21, "U",    "7",       "ZZZAA", "YBBBYYB", "1000110"),
    (22, "V",    "=",       "AZZZZ", "YYBBBBY", "1100001"),
    (23, "W",    "2",       "ZZAAZ", "BBBYYBY", "0001101"),
    (24, "X",    "/",       "ZAZZZ", "YBYBBBY", "1010001"),
    (25, "Y",    "6",       "ZAZAZ", "BBYBYBY", "0010101"),
    (26, "Z",    "+",       "ZAAAZ", "BBYYYBB", "0011100"),
    (27, "CR",   None,      "AAAZA", "YYYBBBB", "1110000"),
    (28, "LF",   None,      "AZAAA", "YYBBYBB", "1100100"),
    (29, "LTRS", None,      "ZZZZZ", "YBYBBYB", "1010010"),
    (30, "FIGS", None,      "ZZAZZ", "YBBYBBY", "1001001"),
    (31, "SPACE",None,      "AAZAA", "YYBBBYB", "1100010"),
    (32, "NUL",  None,      "AAAAA", "YBYBYBB", "1010100"),
]

All 32 rows independently verified to have exactly 4×B/3×Y; all 32 seven-unit codewords are distinct.

2.3 Table 2 — service information signals

# Mode A (ARQ) control/idle/repeat signals, and Mode B (FEC) phasing/idle signals.
# Same B=0/Y=1, bit-1-first convention as Table 1.
SERVICE_SIGNALS = {
    "CS1":              "BYBYYBB",  # mode A only
    "CS2":              "YBYBYBB",  # mode A only
    "CS3":              "BYYBBYB",  # mode A only
    "CS4":              "BYBYBBY",  # mode A only — M.625 addition, NOT in M.476
    "CS5":              "BYYBYBB",  # mode A only — M.625 addition, NOT in M.476
    "idle_beta":        "BBYYBBY",  # both modes A and B, same role in both
    "idle_alpha":       "BBBBYYY",  # mode A idle-alpha == mode B phasing signal 1 / idle alpha
    "signal_repetition":"YBBYYBB",  # mode A RQ == mode B phasing signal 2
}

CS4 and CS5 exist only in M.625 (added for the automatic MMSI-identification handshake — "call signal correctly received during phasing" / "...during rephasing"). M.476 (the legacy standard) defines only CS1–CS3 plus idle-α, idle-β and RQ.

2.4 Dual-meaning combinations (flagged per request)

Checking Table 2's 8 patterns against Table 1's 32 shows 5 of the 8 service signals reuse a traffic codeword, disambiguated only by protocol state (control-signal slot vs. information-block slot in the synchronous ARQ cycle), and 3 are spare codewords (of the 35 total, 32−35=3 left over) not used for traffic at all:

DUAL_MEANING = {
    "CS1": {"also_means": "combination 12 (L / ')')"},
    "CS2": {"also_means": "combination 32 (NUL / blank)"},
    "CS3": {"also_means": "combination 14 (N / ',')"},
    "CS4": {"also_means": "combination 7  (G, unassigned figure)"},
    "CS5": {"also_means": "combination 8  (H, unassigned figure)"},
    "idle_alpha":         {"also_means": "one of the 3 spare (non-traffic) codewords; "
                                          "in mode B this SAME pattern is also Phasing signal 1"},
    "signal_repetition":  {"also_means": "spare codeword; in mode B this SAME pattern is Phasing signal 2"},
    "idle_beta":          {"also_means": "spare codeword; same role (filler) in both mode A and mode B"},
}

RQ's pattern is phasing signal 2, and it is idle-α's pattern that doubles as phasing signal 1 (M.625 Table 2 literally labels that Mode B cell "Phasing signal 1, Idle signal α").

2.5 Cross-check against Wikipedia

Wikipedia's "CCIR 476" article gives the same 8 service-signal patterns as plain binary, e.g. CS1 = 1100101. These are not written in ITU's bit-position-1-first convention — they are the *reverse* of the ITU string with B=1/Y=0 instead of B=0/Y=1. Reversing the ITU string BYBYYBB→(B=1,Y=0)→1010011→reverse→1100101 reproduces Wikipedia's value exactly, and this holds for all 8 signals (CS1–CS5, idle-α, idle-β, RQ). This is a documentation-convention difference, not a technical disagreement — flagged per the instructions, but resolved: both sources describe the identical on-air signal.

3. SITOR mode B (FEC)

Per M.625-4 §4 (current) and M.476-5 §3.2 (legacy, consistent except where noted):

4. SITOR mode A (ARQ) timing

A synchronous system (M.625-4 §3, identical timing to M.476-5 §3.1): the information sending station (ISS) transmits a block of 3 characters (3 × 7 units = 210 ms), then pauses 240 ms while listening; the information receiving station (IRS) replies with one 70 ms control signal, then pauses 380 ms — both sides thus share a 450 ms basic timing cycle, master-clocked by whichever station initiated the call (the "master" station; the callee is "slave" for the life of the circuit regardless of which side later becomes ISS/IRS). CS1 and CS2 alternately acknowledge odd/even-numbered blocks; CS3 requests a change of traffic direction; a mutilated block draws a "signal repetition" (RQ) triplet; a constant equipment delay t_E (kept as short as practicable, occasionally padded 20–40 ms to let a nearby receiver recover from transmitter desensitisation) separates the end of a received signal from the start of the reply at the phase-locked slave station.

5. NAVTEX (ITU-R M.540-2 + IMO NAVTEX Manual 2023)

5.1 Carrier frequencies

5.2 Message format

ZCZC B1 B2 B3 B4 <CR><LF> <message text> NNNN <CR><LF><LF>

5.3 B1 — transmitter identification character

From the current NAVTEX Manual (Table 2): B1 is allocated by the IMO NAVTEX Coordinating Panel, one letter per transmitter, A through X (24 letters; Y and Z are not allocated as B1 values in the current scheme). Each letter has a fixed 10-minute slot every 4 hours, staggered by 10 minutes per letter:

# B1 letter -> six UTC transmission start times (HHMM), repeating every 4 h
NAVTEX_B1_START_TIMES = {
    "A": ["0000","0400","0800","1200","1600","2000"],
    "B": ["0010","0410","0810","1210","1610","2010"],
    "C": ["0020","0420","0820","1220","1620","2020"],
    "D": ["0030","0430","0830","1230","1630","2030"],
    "E": ["0040","0440","0840","1240","1640","2040"],
    "F": ["0050","0450","0850","1250","1650","2050"],
    "G": ["0100","0500","0900","1300","1700","2100"],
    "H": ["0110","0510","0910","1310","1710","2110"],
    "I": ["0120","0520","0920","1320","1720","2120"],
    "J": ["0130","0530","0930","1330","1730","2130"],
    "K": ["0140","0540","0940","1340","1740","2140"],
    "L": ["0150","0550","0950","1350","1750","2150"],
    "M": ["0200","0600","1000","1400","1800","2200"],
    "N": ["0210","0610","1010","1410","1810","2210"],
    "O": ["0220","0620","1020","1420","1820","2220"],
    "P": ["0230","0630","1030","1430","1830","2230"],
    "Q": ["0240","0640","1040","1440","1840","2240"],
    "R": ["0250","0650","1050","1450","1850","2250"],
    "S": ["0300","0700","1100","1500","1900","2300"],
    "T": ["0310","0710","1110","1510","1910","2310"],
    "U": ["0320","0720","1120","1520","1920","2320"],
    "V": ["0330","0730","1130","1530","1930","2330"],
    "W": ["0340","0740","1140","1540","1940","2340"],
    "X": ["0350","0750","1150","1550","1950","2350"],
}

Note: M.540-2 itself only says "B1 is a letter (A–Z) identifying the transmitter coverage area" and delegates the actual assignment scheme to IMO/the NAVTEX Manual — this timing table is from the 2023 NAVTEX Manual, not from the ITU-R Recommendation.

5.4 B2 — subject indicator character

From the current NAVTEX Manual (Table 1):

NAVTEX_B2_SUBJECTS = {
    "A": "Navigational warnings",                                    # mandatory, cannot be rejected
    "B": "Meteorological warnings",                                  # mandatory, cannot be rejected
    "C": "Ice reports",
    "D": "Search and rescue info, piracy, tsunami, other phenomena",  # mandatory, cannot be rejected; triggers alarm
    "E": "Meteorological forecasts",
    "F": "Pilot and VTS service messages",
    "G": "AIS service messages (non-navigational aid)",
    "H": "LORAN messages",
    "I": "Currently not used",
    "J": "GNSS messages regarding PRN status",
    "K": "Other electronic navigational aid system messages",
    "L": "Other navigational warnings — additional to B2 = A",        # mandatory, cannot be rejected
    "M": "Currently not used", "N": "Currently not used",
    "O": "Currently not used", "P": "Currently not used",
    "Q": "Currently not used", "R": "Currently not used",
    "S": "Currently not used", "T": "Currently not used",
    "U": "Currently not used",
    "V": "Special services allocation by IMO NAVTEX Coordinating Panel",
    "W": "Special services allocation by IMO NAVTEX Coordinating Panel",
    "X": "Special services allocation by IMO NAVTEX Coordinating Panel",
    "Y": "Special services allocation by IMO NAVTEX Coordinating Panel",
    "Z": "No message on hand",
}

M.540-2 itself does not enumerate this table (Annex II §2.1: "B2 characters ...are defined by IMO... combination numbers 1-26" of Table 1 of Rec. 476 and 625, i.e. just the ITA2 letters A–Z used as tags) — the semantic mapping above is IMO's, from the current NAVTEX Manual, and has evolved since 1990 (e.g. G = AIS, J = GNSS/PRN status did not exist when M.540-2 was last revised). If implementing against an older document, be aware this table is the operative one now, not something derivable from M.540-2 alone.