Time-sequenced beacons#
SPEEDBIRD can send short beacons in fixed UTC time slots, in the spirit of the NCDXF/IARU International Beacon
Project. Every station knows when and where every other station will transmit, so anyone can listen and build a
propagation map from what they hear. Code: src/speedbird/core/beacon.py. API: /api/beacon. UI: Control → Beacons.
The beacon is disabled by default. When you enable it, the station transmits automatically, without an operator, as your callsign.
Part 97 compliance (read this first)#
SPEEDBIRD does not decide whether your operation is legal. You do. Points to check (US rules; other countries differ):
- A station that transmits for propagation observation is a beacon (§97.3(a)(9)), and §97.203 applies to it.
- Automatic control of a beacon is allowed only on 28.20–28.30, 50.06–50.08, 144.275–144.300, 222.05–222.06 and 432.300–432.400 MHz, and on 33 cm and shorter wavelengths (§97.203(d)). On the default bands (80–15 m) a control operator has to be at a control point while beacons run.
- At most 100 W PEP output (§97.203(c)). Use the beacon power cap or the radio's
tx_power_max_pct. - No more than one channel per band at the same time from one location (§97.203(c)). The scheduler enforces this across your own radios by skipping the duplicate.
- Beacons must not cause harmful interference and must stop when you are told to stop (§97.203(g)).
- The callsign is in every beacon, and nothing is encrypted.
The Beacons page lists every frequency the configuration will use. It flags each one that lies outside the §97.203(d) automatic-beacon segments and outside the band's automatic-control data sub-band.
Time slots#
- Time is divided into periods of
period_sseconds, counted from the Unix epoch. Periods therefore line up with UTC.period_smust divide 86400. - Each period has
period_s / slot_sslots. - Each station owns one slot per period:
slot: "auto"(the default) usessha256(BASECALL)[:4] mod n_slots. The base callsign is used without the SSID or/P. Any station can work out any other station's slot.- Alternatively, set an explicit slot number.
- With
require_ntpset, a beacon is skipped whentimedatectlreports that the clock is not NTP-synchronised. If the sync state is unknown, the scheduler only shows a warning.
Modes#
| Mode | What | Airtime | Decodes to (nominal) | Slot rule |
|---|---|---|---|---|
call |
SBP CALL burst | ~1.5 s | about −13 dB | ≥ 10 s |
ft8 |
SBP-FT8 | ~12.6 s | about −21 dB | multiple of 15 s |
fst4w120 |
SBP-FST4W-120 | ~109 s | about −32 dB | 120 s slots on even UTC minutes |
auto (default) |
chosen per band by SNR | as for fst4w120 |
How auto chooses a mode for each band:
- It takes the median SNR over the last 24 h at which other stations heard us. Two sources feed this:
- reception reports (
R:CALL@SNRtokens in their beacons); - SBP mesh neighbours' reports. - If there are no reports, it uses how well we hear them: heard beacons, plus SBP rows in NodeDB on that band.
- It maps that SNR to a mode:
- ≥ −10 dB:
call- −10 to −20 dB:ft8- below −20 dB, or no data:fst4w120 - Hysteresis (
hysteresis_db, 3 dB) applies. A faster mode is chosen only whenSNR − hqualifies for it, and a slower mode only whenSNR + hdrops below the current one. - The decision is made once per period and stored in
<state>/beacon/state.json.
A band entry's mode overrides the global setting. Until the SBP code provides send_beacon, every beacon goes out as
one ordinary SBP broadcast. The status and the schedule show this.
Airtime is capped per radio:
- Beacons alone get
max_airtime_pct(default 2 %). - Beacons plus SBP mesh routing together get the mesh cap: 2 % on HF, 5 % on VHF/UHF.
Both caps are measured over max(1 h, period_s). The default 2 h period leaves room for one 109 s FST4W beacon (about
1.5 %).
Band hopping#
With hop: true and a band sequence of N bands, a station whose slot is k uses bands[(p + k + r) % N] during
period p. Here r is the index of our radio, so two radios cover different bands. At any moment different stations
are on different bands, and a listener who stays on one band hears every station once every N periods.
The radio is retuned through the normal modem start (SbpService.start_link) at most once per period, and only when
all of these hold:
- its activity is the SBP modem;
- its control mode is
normal; - the link is idle.
It never retunes in the middle of a session. Without hopping, the beacon goes out on whatever frequency the radio is on.
Payload#
A beacon is one SBP data block:
SBB1 K4MFC-1 EM73tv 40m 7105.0 P10 S5/60 T1763380800 Mft8 R:W1AW@-14 :optional text
The fields are:
| Field | Example | Meaning |
|---|---|---|
| callsign | K4MFC-1 |
|
| grid | EM73tv |
|
| band, kHz | 40m 7105.0 |
|
P |
P10 |
power % |
S |
S5/60 |
slot / number of slots |
T |
T1763380800 |
slot start (Unix seconds) |
M |
Mft8 |
mode |
R: |
R:W1AW@-14 |
up to 3 reception reports |
:text |
:optional text |
up to 40 characters |
What gets recorded#
- Sent and skipped beacons, each with a reason, go to the bus as
beacon.*and to<state>/beacon/log.jsonl. - Heard beacons are stored in
<state>/beacon/heard.db(tablebeacon_heard). - Heard beacons are also fed into NodeDB, with their position taken from the grid, so they appear on Nodes and Topology.
- The API returns a band × UTC-hour best-SNR matrix:
GET /api/beacon/heard.
Mesh HELLO alignment#
align_mesh_hello: true moves SBP mesh HELLOs from their free-running interval into our slot, sent after the longest
beacon the configured modes can produce. The mesh keeps its own idle checks and airtime cap.
API#
| Method | Path | Notes |
|---|---|---|
| GET / PUT | /api/beacon/config |
|
| GET | /api/beacon/status |
|
| GET | /api/beacon/schedule?n=20 |
|
| GET | /api/beacon/heard?band=&since=&call=&limit= |
|
| POST | /api/beacon/test |
body {radio, confirm: true, mode?}. Transmits one beacon now. |