本页尚未翻译,目前显示英文。
DMR voice calls
What the voice side of the DMR stack does above the modem (dmr-modem.md covers the air interface and the modem
contract). Code: core/dmr/call (the call stack), core/dmr/dmrPrivacy.c (basic privacy), core/radio/radio_sm.c
(the radio core: channel, contacts, settings), core/codeplug (contacts, settings, the DMR ID database). Protocol
references are ETSI TS 102 361-1 and -2.
1. Who we hear
A voice call on our timeslot and color code starts from a voice LC header or by late entry (section 4). It is played
when it passes the filters, and shown either way. Digital monitor (hear all, below) also takes calls on any color code
and either timeslot (dmr_listen_filter_t.any_cc, .both_ts; the modem gets no color code filter and, through a
repeater, no slot lock). The first call heard keeps the voice until it ends: bursts on the other slot meanwhile are
dropped. Turning it on or off changes only the modem's filters: the call being heard and the reply window stay, and
a call heard as not ours plays at once. The view gives the call's color code and slot (rx_cc, rx_ts; the slot 0
when unknown, the code from the first burst that carries one, never burst A) beside the channel's (cc, ts), which
PTT always uses.
| Call | Played when |
|---|---|
| Group (FLCO 0) | its talkgroup is the talk target's, all call (0xFFFFFF), in the channel's RX group list (listen list), the reply target's during the reply window, or hear all, "digital monitor" in the UI (the hear-all latch: the MONITOR key's press toggles it on a DMR channel, its release does nothing (#19); Quick 5; radio_req_dmr_hear_all; until the channel changes, DMR stops (a mode change), or BACK on Home; the end of a call doesn't end it. A zone or VFO scan started with it on keeps it on at every stop and latches it on the channel the scan ends on; turning it off during the scan ends it for the scan, radio-features.md 2). On a trunked channel it follows every talkgroup's grant instead (dmr-trunking.md, Grants): the call plays on its own slot and the system's color code, not any |
| Private (FLCO 3) | addressed to our ID (the channel's DMR ID override, else the radio's), or all call, or hear all |
| Our own ID as source | not while we transmit or for DMR_CALL_OWN_ECHO_MS (700 ms) after our call ended or was cut short (the repeater's echo of it); later it is judged as any other call, so a parrot's playback of our call plays (#294). A terminator from our ID ends only the call with its LC, never another station's |
| Encrypted | only with a privacy key, and never after a PI header (section 5) |
A call that doesn't pass is shown as busy ("not played") while it lasts, and counts for polite channel access.
A voice LC header or terminator whose RS(12,9) check fails is not taken. With the channel's Accept invalid CRC
(dmr_listen_filter_t.accept_bad_crc, open-firmware-cps.md 7.3.3) it is taken when the same
9 LC octets failed among the last four failures (core/fec/rx_seen.h): a system that sends every LC that way repeats
it, and a burst damaged on the air rarely decodes the same twice. The call stack counts those (rx_lc_taken).
The caller is shown by name when one is known: the channel's talk target and listen list contacts first, then every codeplug contact with that ID (group and private IDs are separate name spaces), then, for a private ID, the DMR ID database text (callsign and name, cut to the 16-character name field). The radio core keeps the last four answers (including "no name"), so flash is read once per new ID.
The emergency service option (bit 7) marks the call (emergency in the view, the call screen's warning color). With
CONFIG_DMR_EMERGENCY a played emergency call also raises the emergency alert and tone, and the radio's own
emergency calls carry the option (dmr-emergency.md).
2. Who we talk to
PTT sends to, in order:
- a one-shot target: the row with the focus in the talk target picker or a list (
radio_req_dmr_focus, handed over by the UI as the focus moves). PTT talks to it once; after the call the radio follows it (plays its calls and sends to it on PTT) for the reply window, then goes back to the channel's target. A channel change, Reply cancel, a new manual target, or the focus moving to the talk target itself end it (dmr.one_shot,dmr.focus_id). - the reply target while the reply window runs: after a call we played, its talkgroup (group call) or its caller
(private call to us); after our own call, its target again. The window is the codeplug's group or private call
hang time (rounded up to whole seconds, at most 30 s), or 10 s when the codeplug has none; the reply time setting
(
reply_s, 0 to 30 s, Settings > Talk & listen) replaces the codeplug's for both. BACK on Home ends it. - a talk target chosen for now: a typed ID or talkgroup, a codeplug contact (with its timeslot override), or "call back", which takes the last call heard (its caller for a private call, its talkgroup for a group call). It lasts until the channel changes or it is cleared.
- the channel's TX contact; with none, the first talkgroup of its RX group list (private calls skipped); with neither,
talkgroup 9.
codeplug_channel_talk_contact()is the one place this is decided.
The call is unit to unit (FLCO 3) for a private target, else group (FLCO 0).
For now changes (Quick settings, P7): radio_req_dmr_timeslot and radio_req_dmr_color_code set the timeslot
(temp bit 0x20; it overrides a contact's timeslot) and the color code (0x40) without touching the channel;
radio_req_mode listens in FM or DMR on the same frequencies (0x80). They are refused while transmitting or busy, and
BACK on Home (radio_req_clear_temporary) or a channel change undoes them. Save to channel writes them into the
channel. In the VFO the same requests change the VFO's own settings instead (section 2.1).
TX time-out: the channel's codeplug TOT (15 s units) when it has one, else the radio's (180 s by default). The transmission ends at the next superframe boundary with its terminator, and PTT shows "timed out" until released.
Stalled modem: PTT release and the time-out both wait for a superframe boundary, which only comes while the modem
sends bursts. When the modem takes no burst for DMR_CALL_TX_STALL_MS (1.5 s; a burst counts when it is queued or its
slot starts), the stack aborts the transmission and returns to receive with DMR_TX_ERR_STALLED (dmr.tx_error=stalled
on the debug link). With PTT still held the radio shows "Modem refused to send" until release. A modem without slot
timing (a modem that stops answering) otherwise kept the radio keyed until the battery came out
(gotchas 3.12). Through a repeater, before the first burst's slot, the wait is 300 ms longer
(DMR_CALL_TX_ALIGN_GRACE_MS): the modem's own aligned-start timeout (1.5 s and a slot) must come first, so its
refusal wakes the repeater again (dmr-modem.md 2.7). With an encoder slow to start, the headers are
the last bursts queued, so without the margin the stall cut comes first.
2.1 The DMR VFO
With CONFIG_VFO_DMR the VFO has a mode, FM or DMR, as a setting of its own, next to its frequency, bandwidth and
tones. In DMR it works like a DMR channel that has no codeplug record behind it:
| What | In the DMR VFO |
|---|---|
| Frequency | The VFO's. RX and TX are the same (simplex, direct mode). A VFO A record written by a CPS with a TX offset keeps it, and then the VFO is a repeater-mode channel by the channel rule (RX and TX differ) |
| Color code | 0 to 15, default 1 |
| Timeslot | 1 or 2, default 1. A contact's timeslot override is not applied: the VFO's slot is the slot |
| Own ID | The radio's, as on a channel without an ID of its own |
| Talk target | A typed ID or talkgroup, or a codeplug contact, private or group. Default: talkgroup 9, as a channel with no contact and no RX group list |
| Heard | The talk target's group, all call, private calls to our ID, the reply window, and everything with digital monitor. The VFO has no RX group list, so no other talkgroup is played (it is shown as busy), as on a DMR channel whose list is empty |
| Power, TOT, privacy key | The radio's, as on a channel (the VFO's power level, the radio's TOT, the radio-wide privacy key) |
Not in the VFO: talker alias (a channel's setting), roaming, the "hear all by default" setting, saving to a channel.
These are the VFO's own settings, so they are not temporary changes: no dot on Home, BACK on Home doesn't undo them,
they stay while the radio is on a channel, and they are kept across power cycles with the VFO's frequency
(settings-store.md 3). Squelch has no VFO value: changed in the VFO it becomes the radio's default squelch at once,
so it is not temporary either. The requests are the ones a channel uses and the radio core gives them the VFO meaning:
radio_req_mode (needs both CONFIG_VFO_FM and CONFIG_VFO_DMR; on a channel the mode is no longer a quick menu change, Copy to VFO (2.2) is the way in), radio_req_dmr_color_code,
radio_req_dmr_timeslot, and radio_req_dmr_target / radio_req_dmr_contact for the talk target. A typed or
chosen target is the VFO's from then on (dmr.target_manual stays 0); "call back" is for now as on a channel and
radio_req_dmr_target_clear returns to the VFO's target. All are refused while transmitting.
A mode change retunes the same frequency. FM to DMR starts the call stack and sets the board to DMR; DMR to FM stops
it, and the VFO's FM bandwidth and tones come back. No sub-audio squelch is set while in DMR, whatever tone the VFO
holds. A build with only CONFIG_VFO_DMR has a DMR VFO and no mode change; one with only CONFIG_VFO_FM is as
before.
A split VFO's TX frequency is set by radio_req_vfo_tx (RREQ_VFO_TX, the quick tile TX freq, information-architecture.md 5): any frequency in a transmit band, 0 for simplex, in the VFO only, not while transmitting or in hotspot mode. The receiver isn't retuned. It is saved with the VFO's settings (vfo_tx_hz).
2.2 Copy to VFO
From a channel (FM or DMR), Copy to VFO (radio_req_vfo_copy, RREQ_VFO_COPY) makes the channel's settings the VFO's
own and opens the VFO. It is in the channel's quick menu (tile 4, "To VFO"), on the assignable key function "Copy to
VFO" (KEYFN_COPY_VFO, short or long press) and on the debug link (RADIO op 32). It needs CONFIG_VFO_FM or
CONFIG_VFO_DMR.
What is copied is what is in effect now, so a temporary change on the channel (bandwidth, tone, mode, color code, timeslot, power, a talk target chosen for now) goes along. The channel itself is unchanged: leaving the VFO brings back the channel as it was, without the temporary changes, as any channel change does.
| Channel | VFO |
|---|---|
| RX and TX frequency | Both, so a repeater offset stays (the VFO is then a split VFO; a new RX frequency, typed or stepped, keeps the offset while RX stays in its transmit band and TX lands there too, board_offset_carries; a move to another band, or TX off the band, makes the VFO simplex). Direct-mode channels (DMO) copy RX only |
| Mode, bandwidth | Both |
| Power | The level in effect; a channel on "default" stays on the radio's default |
| CTCSS / DCS, RX and TX (FM) | Both. A DMR channel has none: the VFO keeps its own for its FM mode |
| Color code, timeslot (DMR) | Both. The slot is the one in use: a timeslot chosen for now, else the talk contact's override, else the channel's |
| Talk target (DMR) | The ID and the group / private flag of what the channel would send now: a target chosen for now, else its contact, else the first talkgroup of its RX group list, else talkgroup 9 (the channel default). A one-shot or reply target is not copied |
| FM channel into a VFO that has DMR values | Color code, timeslot and talk target stay as the VFO had them |
Not copied, because the VFO has nothing to hold them: the RX group list, the privacy key, roaming, the data and emergency settings, the talker alias, TOT and the squelch level (a radio setting, not the VFO's). They apply to the VFO as the radio-wide values.
The VFO's values are saved as its own settings (settings-store.md 3) and the VFO's temporary marks are cleared.
Nothing changes, and the UI says why (radio_vfo_copy_check), when:
| Reason | Toast | Debug link status |
|---|---|---|
| No VFO in the build | "VFO: not in this build" | UNAVAILABLE |
A DMR channel without CONFIG_VFO_DMR, or an FM channel without CONFIG_VFO_FM |
"VFO: no DMR in this build" / "VFO: no FM in this build" | REFUSED |
| RX or TX frequency outside the board's bands | "VFO: frequency outside its bands" | REFUSED |
| Already in the VFO | "Already in VFO" | UNAVAILABLE |
The radio core also ignores the request while transmitting and in hotspot mode. A scan in progress ends first (the channel it stopped on is the one copied).
3. Talker alias
Receive: header (FLCO 4) and blocks 1-3 (FLCO 5-7) in the embedded LC, in any order, all four formats. 7-bit text
starts at the header's first data bit; ISO 8-bit, UTF-8 and UTF-16 (big endian) at its second. Characters outside
printable ASCII show as one ? each (a UTF-8 multi-byte sequence or a UTF-16 surrogate pair is one character). The
announced length counts characters, but some senders count UTF-8 bytes or overstate it, so an alias also counts as
complete at the first zero character (padding) or once all four parts are in. A different header (format or length)
starts a new alias; a new talker resets it.
Transmit (CONFIG_DMR_TALKER_ALIAS): the radio name (the codeplug callsign, 8 characters) in ISO 8-bit, when the
channel's talker alias setting asks for text on its timeslot. From the second superframe, every other superframe
carries one alias LC instead of the voice LC (LC, header, LC, block 1, LC, header, ...), so late entry still finds the
voice LC within two superframes. The modem takes the new embedded LC at the next burst A it sets up; the stack hands
it over before queueing that burst A.
3.1 GPS Info
CONFIG_DMR_GPS (= DMR). The GPS Info LC (TS 102 361-2 7.1.1, FLCO 8, FID 0) rides in the embedded LC like a talker
alias block. After the FLCO and FID octets: 4 reserved bits, the position error (3 bits: below 2 m, 20 m, 200 m,
2 km, 20 km, 200 km, more, 7 unknown), the longitude (25 bits, two's complement, 360 / 2^25 degree steps, east
positive) and the latitude (24 bits, two's complement, 180 / 2^24 degree steps, north positive). One step is about
1.2 m. dmr_lc.c (dmr_gps_encode, dmr_gps_decode) packs and unpacks it; a manufacturer FID or 0, 0 (no fix) is
not a position.
Receive: dmr_call decodes every GPS Info LC on the call into dmr_caller_t.gps / gps_valid (counter rx_gps);
a new talker clears it, and last_heard keeps it after the call. The radio core adds the distance and bearing from
our own position (the position service, position.md) in radio_dmr_call_t, the receive screen shows "12.3 km NE"
(or the coordinates when we have no position of our own), Recent keeps the distance with the entry, and in
CONFIG_APRS_RX builds the caller goes into the APRS station list (named by its alias's callsign, else its DMR ID).
Transmit, when either asks for it:
| Setting | Where | Scope |
|---|---|---|
| "Send in DMR calls" | Settings > Location (position_cfg_t.dmr_gps, off by default, P7) |
every DMR call |
| The channel's talker alias field for its timeslot: APRS or Both | the codeplug (open-firmware-cps.md 7.3.3, offset 0x30; channel_t.alias_tx CODEPLUG_ALIAS_TS1_GPS / _TS2_GPS), so a CPS sets it |
calls on that channel; not the VFO |
At
PTT the radio core takes the position in use (position_now: GPS on boards with one, the host's NMEA, else the fixed
position; position ambiguity applied) and hands it to the stack (dmr_call_set_tx_gps); no position, no GPS Info. The GPS Info
LC is one more entry at the end of the talker alias rotation in the odd superframes: LC, header, LC, block 1, LC,
block 2, LC, GPS, ... with an alias, or LC, GPS, LC, GPS without (counter tx_gps_lcs). The position error says 20 m
for GPS and host fixes, 200 m for a typed position, or the ambiguity's blur. It needs no APRS: the setting and the
position are the position service's, built with CONFIG_DMR_GPS. A radio without a GPS sends its fixed position.
The hotspot already passes FLCO 4-8 embedded LCs through to the host (hotspot.c).
4. Late entry
A call heard without its header is acquired from the embedded LC once two consecutive voice bursts were seen (a lone burst is noise) and the LC differs from the previous call's (the modem's stale LC, gotcha 1.4), or the same LC came in two superframes. Playback starts at the burst that completed the LC. A new talker on a running call (a repeater passing overs without a header) switches the caller from the embedded LC. The host tests cover joining at burst C, the lone burst, the stale LC needing confirmation, a fresh LC taken at once, and a talker change.
A call lost for a moment, with no terminator heard, ends and has to be joined again. Where the modem can prove the same call is back, that takes one superframe instead of two, and it counts as one call (dmr-modem.md 2.4).
5. Privacy
Basic privacy (CONFIG_DMR_PRIVACY): Motorola BP (key 1-255) and Anytone 16-bit BP (dmrPrivacy.c). The keystream
is XORed onto the 49 AMBE+2 parameter bits of each frame by superframe position: on TX as the burst is queued (the
encoder's bits as they are), on RX before the voice sink (after the FEC decode, core/dmr/ambe/ambeFec.c). With a key
loaded our calls carry FID 0x10 and the privacy service option (0x40), and received calls with the privacy service
option (or PI) are descrambled. Without a key such bursts are never played; they count as rx_enc_dropped.
The same key covers packet data where its scheme defines a data form (Motorola BP encrypts it, Anytone BP blocks data TX): dmr-data.md 13.
The key is set over the debug link (RADIO op 19, radio_req_dmr_privacy). The radio core saves it in settings slot 1
and loads the active slot at boot (settings-store.md 3.1); the open-firmware codeplug has no key storage, so qdmr can't
set it.
PI header (data type 0, CRC-CCITT masked 0x6969): heard between the voice LC header and burst A, it marks the call encrypted and its first two octets are kept for display (algorithm and key ID in the common layout; not interpreted). A PI header heard just before a late entry is attached to that call. A PI header means an MI-based scheme the static keystream can't undo, so such a call is never played, key or not. We never send PI headers.
The HR_C6000 reports the EMB PI bit unreliably for voice, so its driver clears it (modem-hrc6000.md); on that modem
privacy is recognized from the service options and PI headers only.
6. Encoding the microphone
The vocoder encodes each 20 ms of microphone audio (core/dmr/ambe/ambeVoice.c, in the audio task). Before that the
samples are scaled by the mic gain (Settings > Talk & listen, 2 dB a step). The codec itself can also clean up the
input. These are generic encode options in hal_vocoder_t; the board maps them to its codec, and core/ never sees
codec bits.
| Option | Effect | Default | Setting |
|---|---|---|---|
VOCODER_OPT_NOISE_REDUCTION |
Suppresses steady background noise, also between words. It also lowers a steady tone held into the mic | on | Settings > Talk & listen > Mic noise reduction |
VOCODER_OPT_TONE_FRAMES |
A steady tone or tone pair (DTMF) is sent as a tone frame and comes out at its frequency on the other side. Speech is not affected | on | none |
VOCODER_OPT_AGC |
Input level control: quiet speech raised (about 4x), loud speech lowered. The codec's own parameters; not yet checked on air | off | Settings > Talk & listen > Mic AGC |
options()says which options the board's codec takes. A board without them leavesoptions,set_optionsandencode_flagsNULL. Then the settings are shown dimmed ("Not on this radio") and nothing changes.ambeVoiceSetEncodeOptionsmay be called from any task. The audio callback hands the options to the vocoder before the next frame it encodes, so every codec call of a frame uses the same options. The board keeps them acrossreset(), so a new transmission starts with the options in force.- The options are read at start-up from the settings and on every change (
ui_svc_apply). With AGC on, the mic gain is step 5, the board's levels (FM too on a radio whose board sets the gain), and Mic gain is dimmed, showing "AGC". - Voice is simplex: while a transmission runs, nothing is decoded. A call still being heard goes on as silence, and
the decoder starts fresh once the transmission ends; a call that begins meanwhile resets nothing until then. A
platform may keep the encoder's and the decoder's state in one block (
hal_vocoder_t): a decode during the transmission would cost it its encoder. - Each board's codec mapping is on its
codec-<id>.mdpage ("Encode flags"). Boards with an AMBE+2 host sharedrivers/vocoder/ambe2_flags.h.
6.1 A tone in place of the microphone
ambeVoiceTxTone (ambeVoice.h, radio task) hands the audio callback a tone sequence of up to SIG_AIR_STEPS steps (SIG_STEPS and a guard of silence, SIG_AIR_GUARD_MS)
(radio-features.md 4.3). From the next block of a transmission (the first, if it has not started)
the encoder gets the generated tone instead of the microphone, after the mic gain. While it plays the encode options are
VOCODER_OPT_TONE_FRAMES set and VOCODER_OPT_NOISE_REDUCTION cleared, and the options in force come back after it.
- The encoder waits for room in its queue (
AMBE_VOICE_QUEUE) before each block of the tone, instead of letting the queue overflow, so a tone longer than the headers' time is not cut. final: the microphone stays muted after the tone until the transmission ends.dmr_voice_source_t.busytells the call stack the tone's frames are not all pulled yet. After PTT is let go the stack then goes on at each superframe boundary, for at mostDMR_CALL_TX_TAIL_HOLDSsuperframes, before it ends the over with the terminator. A time-out ends it at once.ambeVoiceTxToneCanceldrops a request, and the source'senddoes it too: a request for a transmission that never went on air is not played in the next one.busyis false while no transmission runs.- The tone's amplitude is fixed (
TONE_AMP_Q15). It is not checked on air, nor how other makes of radio play a tone frame.
7. Observing it
Debug link (debuglink.md): the core view's mode and vfo, the dmr view's dmr.cc, dmr.ts, dmr.target_id,
dmr.target_private, dmr.target_manual, dmr.privacy, dmr.alias_tx, dmr.gps_tx_channel (the channel's own position setting); RADIO op 31 sets the mode, color code and timeslot (the VFO's
own in the VFO), op 15 the talk target and op 32 copies the channel to the VFO (2.2); the dmr_call view's
emergency, pi_header, alias_complete, fid, svc, pi_alg, pi_key; the counters rx_private_calls,
rx_emergency_calls, rx_encrypted_calls, rx_pi_headers, rx_alias_complete, tx_alias_lcs, rx_gps,
tx_gps_lcs; GPS Info as call.gps_valid gps_lat gps_lon gps_err gps_dist_m gps_bearing (and last.). HIL tests:
tests/hil/test_dmr_call_features.py (test_gps_info: the bench sends a call with a GPS block, a bench_tx
voice --gps LAT,LON[,ERR] job), tests/hil/test_dmr_tx.py (test_dmr_gps_tx: our GPS Info decoded by the
bench modem's listener).
7.1 Receive level
A DMR receiver decodes 4FSK well only while the demodulated level at its modem sits in a window: too high and the outer symbol threshold is crossed (corrected bits on every frame, even far above sensitivity), too low and sync is lost. That level is the receiver's audio gain times the transmitter's deviation, and transmitters differ by 10-20 % in deviation, so the gain a board sets has to sit in the middle of the window, not at its best point for an exact transmitter. A carrier offset shifts the same level: a gain with no headroom clips on one side of the channel first, and a fixed receive offset moves the whole window off the channel. Transmitters a kHz or two off frequency are common, so a board keeps its window centered on the channel. Where a board's window was measured is on its hardware page (feature-matrix.md for the radios).
| Tool | What it does |
|---|---|
tools/dmr_rx_sweep.py |
Steps a chip register (debug link CHIP_REG), or one bit field of it (--field HI:LO), through values while the SDR bench sends speech, over bench levels, deviations and carrier offsets (--detune makes the offset by tuning the radio away instead); ranks the values by their worst cell. About three minutes a grid |
test_dmr_audio_rx_strong |
The bench's simplex speech at -75 and -55 dBm at the radio: 95 % of the frames and at most 0.2 corrected bits a frame at each (a level the bench can't reach is left out) |
test_dmr_audio_rx_deviation |
The bench's speech at 0.85 and 1.15 of the standard deviation, at -100 dBm at the radio: 95 % of the frames and at most 0.8 corrected bits a frame, both ways |
test_dmr_audio_rx_offset_wide |
The bench's speech with its carrier 2.5 kHz below and above the channel, at -100 dBm at the radio: 90 % of the frames and at most 2 corrected bits a frame, both ways |
Measure near sensitivity (-90 to -105 dBm at the radio) for the level window: there a fraction of a dB shows. Measure strong simplex signals too (-75 to -45 dBm): a simplex transmitter keys one 30 ms burst every 60 ms, so the receiver's AGC sees the carrier come and go with every burst, and one that settles too slowly overloads the start of each burst. That costs a bit or more a frame however strong the signal, and only on simplex: a repeater's continuous carrier hides it. A clean receiver corrects almost nothing at those levels. Use speech or another varied recording, never AMBE silence: its errors land in bits with no FEC. Let the radio drop the last call before the next one starts (about 2 s): a call that begins while it still follows the last one's slot timing can be decoded at that timing to its end.
8. Open points
- The codeplug's hang time unit is not confirmed. The interop spec lists the group and private hang bytes
(general settings 0x18, 0x19) only as vendor layout that the open-firmware layout does not use
(
open-firmware-cps.md7.3.2), and neither the reference notes nor the vendor guide give a unit. Assumed: 500 ms steps (codeplug.c,GEN_HANG_UNIT_MS), so a stored 6 is 3 s. Confirm with the vendor CPS (set a known hang time, read the byte) on the bench; the reply time setting (rs_reply_s) overrides it meanwhile. - DMR ID database plain format: the 4-byte BCD ID is read as a little-endian word (the spec doesn't say).
- A codeplug TOT of 0 means "off" in the vendor layout; here it falls back to the radio's TOT.
- Emergency alarm and emergency call TX are in
dmr-emergency.md. No enhanced privacy. GPS Info is host-tested (dmrCallTest) and on air both ways against the SDR bench (docs/testing/on-air-coverage.md). Not done: GPS Info in the hotspot's own transmissions (it relays the host's), a separate Settings > Location toggle for it (it lives on the APRS page).