Compare commits

..
20 Commits
Author SHA1 Message Date
rouggy a621a3ee62 chore: release v0.26.16 2026-08-26 00:25:37 +02:00
rouggy ae4a56dac1 chore(changelog): open 0.26.16 for the TRX audio-source work
v0.26.15 had already shipped, so its entry about setting the source by
hand in ExpertSDR3 goes back exactly as released — it describes the
version operators are running. What replaces it belongs to the version
that carries the change.
2026-08-26 00:25:17 +02:00
rouggy fdc2378191 feat(tci): name the transmit audio source when keying, per the protocol
The specification settles what a night of experiments could only guess at.
TRX takes an optional THIRD argument naming the signal source — tci, mic1,
mic2, micPC, ecoder2 — and TCI 2.0 says it plainly: 'The signal for
transmitting is always taken from the microphone selected in the
ExpertSDR3. If a third-party software connected via TCI wants to transmit
its audio signal, you must specify the third argument - TCI.'

Without it the radio sends no chrono at all, whatever the mode. That is
what the SSB attempts ran into, and what was misread here as 'digital
modes only' — the mode was never the rule, the missing argument was.

So OpsLog says it, following the 'To radio' device: 'tci' when the voice
keyer owns the transmission, nothing at all otherwise, which leaves the
operator's own microphone alone for every other PTT. Nobody has to find
that setting in ExpertSDR3 and set it again for every mode, which is how
it is remembered there.

Two more things from the same document. A chrono with no audio ready is
now answered with silence rather than left unanswered — the vendor calls
that preferable. And the receive stream declares float32 and two channels
instead of trusting the defaults: they are the documented defaults, but a
default is something another program sharing this radio can have changed,
and a stream in an unexpected format is heard as noise rather than as a
mistake.
2026-08-26 00:21:32 +02:00
rouggy 5388f76733 chore: release v0.26.15 2026-08-26 00:14:16 +02:00
rouggy ceb88d2e29 refactor(tci): remove the test bench now that choosing the device is the setup
The TCI section of the audio settings was an investigation: open the
stream, read what arrives, record ten seconds to listen to, key a tone.
It answered every question it was built for — the frame layout, the
sample width, that the radio asks rather than follows a clock, that the
transmit audio source decides — and confirmed 80 W on real hardware.

None of that belongs in front of an operator now. The radio is simply one
of the devices in the two dropdowns, and choosing it IS the configuration:
one control, in the place where the question is already being asked.
Keeping the tick box beside it would have been two switches for one
decision, with the second one where nobody looks.

Gone with it: the stream/record/probe bindings, the audio.tci_rx setting,
and twenty-four translation keys. The plumbing they proved out stays and
now carries the voice keyer.
2026-08-26 00:13:04 +02:00
rouggy ec3e60e47b chore(changelog): open 0.26.15 with the TCI audio work
The RDA and Elecraft SWR entries move with it: both were written after
v0.26.14 was tagged, so they ship in the version that carries them rather
than in the one already on operators' machines.
2026-08-26 00:09:32 +02:00
rouggy 5c6df90526 merge: TCI audio — receive, transmit, and the voice keyer over the CAT link
A SunSDR carries its audio on the same WebSocket as its commands, so
OpsLog can take it directly: no virtual cable, no second sound card, no
Windows mixer between the recording and the air. The radio appears as a
device in both audio lists and can be chosen for either direction.

Everything here was settled on real hardware over one evening, and none of
it was guessable from the documentation:

  - The receive stream answers format=3 for four-byte floats, so the
    sample width is derived from the frame rather than trusted from the
    field.
  - The radio asks for transmit audio only when the transmission is the
    CLIENT'S, and only when its transmit audio source is TCI rather than
    the microphone.
  - The chrono is a REQUEST, not a clock: no payload, carrying the size it
    wants, 47 times a second. Audio goes out in answer to it and never on
    a timer of our own — a timer was the first attempt and all 234 frames
    of it were ignored.

Confirmed: a clean test recording, and 80 W out of a 1 kHz tone.
2026-08-26 00:05:40 +02:00
rouggy deeb654482 feat(tci): the voice keyer can send its messages through the radio
The radio now appears as a device in both audio lists — 'Radio (TCI
network audio)' — so the receive audio and the voice keyer can both take
the CAT link instead of a sound card. No virtual cable, no second card, no
Windows mixer between the recording and the air.

The transmit side reuses the exchange the tone probe established, with a
WAV in place of the sine: the radio asks, we answer with the next slice,
and it sets the pace. The message is converted once, up front, rather than
per frame — a voice message is a few hundred kilobytes, and resampling
inside a callback that has 21 ms to answer would put arithmetic on the
path where a late frame is a gap on the air.

The PTT is untouched by all this: the keyer keys before and unkeys after
exactly as it does with a sound card, so a transmission is bracketed by
the same code whichever way the audio travels. And the playback runs OFF
the CAT goroutine, since everything else about the rig goes through that
one place and a ten-second message would otherwise freeze the frequency
display and the antenna following for ten seconds.

Two refusals rather than silent failure. A radio that has gone away stops
being offered as a device at all, and a radio whose transmit audio source
is still the microphone is caught within a fifth of a second — a voice
keyer that transmits silence is worse than one that says why it will not.
2026-08-25 23:59:09 +02:00
rouggy 8e0865a71f fix(elecraft): read the SWR as three digits, not four
Elecraft's release note settles it: SW; returns the most recent reading in
transmit or TUNE as three digits in tenths of a ratio — SW023 is 2.3:1,
and SW999 is the 99.9:1 it reports instead of infinity.

This asked for four, so every answer failed to parse and the bar stayed
empty for the whole transmission. A tester saw exactly that: no SWR at
all, which reads as an unsupported radio rather than an off-by-one.

SW also comes out of the probe list. It is known now, and asking again
mid-transmission costs a round trip on the link the carrier depends on.
2026-08-25 23:58:20 +02:00
rouggy d4d22eb4b2 fix(tci): the mode was never the rule — ask the radio instead
Transmit over TCI is confirmed on a SunSDR: 80 W out of a 1 kHz tone at
70% of full scale into 80% drive, every request answered.

Which makes the earlier rule wrong. 'Digital modes only' came from a real
observation — SSB silent four times over, DIGU answering at once — but the
mode was a coincidence. ExpertSDR3 has a transmit audio SOURCE, microphone
or TCI, kept per mode, and it was on the microphone in SSB. Refusing SSB
would have blocked the one thing a voice keyer exists for.

So nothing is refused on the strength of the mode. The radio declares what
it wants by asking for audio, 47 times a second when it wants any: key,
wait for one request, and stop within two tenths of a second if none
comes — naming the setting to change rather than theorising about it.

That is better on three counts. It works in SSB when the source is set
right, it cannot be wrong about a mode nobody thought to test (AM, FM,
RTTY), and a misconfiguration costs a quarter-second of carrier instead of
five seconds.
2026-08-25 23:51:28 +02:00
rouggy 592dd08835 fix(tci): send the tone near full scale, and log both levels
The transmit path works: six passes, the radio asked 231 times and was
answered 231 times, none missed, and the tone was there on the panadapter.

What was missing was power on the meter, and the cause was the level. The
tone went out at a quarter of full scale, out of caution, into a radio set
to 15% drive — enough to draw a clean signal and not enough to move a
needle. In a digital mode the radio expects a line level it can drive to
full output; the POWER is its own drive control, so sending quietly only
wastes the range. Now 0.7, short of 1.0 to leave room for the peaks.

The start line carries both numbers — ours and the radio's drive — because
a quiet transmission has two possible causes and one line should settle
which, rather than an evening of guessing. Reading 'drive' off the radio
is the only reason it is parsed at all.
2026-08-25 23:35:08 +02:00
rouggy c6294d9eb3 feat(tci): answer the radio's requests instead of pushing audio at it
Three transmissions on a real SunSDR settled how the transmit side works,
and none of it was guessable from the documentation.

The radio asks for audio only when the transmission is the CLIENT'S: with
the operator keying the microphone it sent 282 receive frames and nothing
else. And it asks only in a DIGITAL mode — keyed from here in SSB it stayed
silent four times over, and answered in DIGU immediately. In SSB the
modulator is wired to the microphone, which is also the honest answer to
'why can I hear myself but not the tone'.

The chrono turns out to be a REQUEST, not a clock. It carries no payload —
the message itself is the ask — and it names the size it wants in the
header: 2048 samples, two channels interleaved, 47 times a second, which
is 1024 pairs at 48 kHz, exactly real time.

So audio goes out in answer to a request and never on a timer of our own.
The timer was the first attempt and the radio ignored all 234 frames of
it. Answering also hands the pacing to the radio: no drift, no buffer to
tune, and the size taken from what it asked for rather than from what we
assumed. The sine keeps its phase across frames, since one restarted every
frame is a click 47 times a second.

A pass in SSB is now refused rather than attempted. It keys the
transmitter, produces nothing and teaches nobody anything — and it is
still a transmission.

The feed mechanism is the one the voice keyer will use: WAV samples in
place of the sine, everything else unchanged.
2026-08-25 23:27:26 +02:00
rouggy 848ce68ec5 feat(tci): key the radio ourselves and push a tone, to see what it wants
A first real transmission settled one question and raised a better one.
With the receive stream open and six seconds of transmit, the radio sent
282 frames of receive audio and NOTHING else: no chrono, no transmit
audio. So the chrono the documentation describes is not offered to a
client that merely happens to be connected while the operator keys the
microphone, and waiting for it to appear is waiting for nothing.

The reading that fits is that the radio asks for audio when the
transmission is the CLIENT'S and takes the microphone when it is the
operator's — which makes the experiment obvious. Key it from here, push a
1 kHz tone, and watch. Chrono frames appearing gives their size and
cadence by measurement instead of by guesswork; no chrono but a tone on
the meter is just as useful, because then the pacing is optional and the
voice keyer can push frames at the rate the stream already runs at.

A tone rather than silence so the answer shows on the power meter and not
only in the log.

It transmits, so: an explicit button inside a warning box, five seconds,
capped at ten, and every path out unkeys — including the panic that has
not happened yet and a socket that dies mid-tone. A transmitter left keyed
by a defect is the one fault here that would reach somebody else's band.

Writes are now serialised too. send() held the lock only long enough to
read the connection, which was enough while every command came from the
poll loop; a stream of audio frames from a second goroutine is not, and
gorilla panics on a concurrent write rather than failing quietly.
2026-08-25 23:18:11 +02:00
rouggy 95e57fb812 chore(tci): mark the transmit passes and count every frame type
The first transmit test came back with a log that said nothing, which is
the one answer that cannot be read: either no transmit frames arrived, or
they arrived and went unlogged.

So each pass is now bounded by a line of its own, and every stream type is
counted without limit. A pass that reports 'receive audio: 240, and
nothing else' is a result — it says the radio sends no chrono unless
something more is asked of it — where a log with no transmit lines was
merely a silence. The forty-frame logging budget is also handed back to
the transmit types on each pass, since it was always spent on receive
audio long before anyone got round to keying.

The start line says whether the receive stream is even open, because a
radio with nothing streaming has no reason to send chrono, and that is the
likeliest reason the first attempt saw nothing.
2026-08-25 23:03:14 +02:00
rouggy 4441b16699 fix(awards): the RDA comparison list no longer jumps back to the top
Two faults, one behind the other.

RDAPanel is nested inside SettingsModal so it can read its state, and it
was written as <RDAPanel />. A nested function is a new component TYPE on
every render: React cannot know it is the same panel, so it unmounted the
tree and mounted a fresh one — and a fresh scroll container starts at the
top. It is called now, like the other panels, which produces the same
elements in place and never disturbs the scroll position.

What it was reacting to should not have reached it either. The PSK
Reporter and grid-cache poll ran every three seconds from wherever you
were in Preferences, re-rendering the whole dialog for a count shown only
on the cluster section. It now runs while that section is open, and the
one-time loads it was sharing an effect with have stayed where they were
rather than refetching on every click in the sidebar.

The visible result: a hundred contacts to correct can be read from top to
bottom.
2026-08-25 22:32:56 +02:00
rouggy f50bbc005c feat(tci): the QSO recorder can take its audio from the radio
Confirmed on a real SunSDR: the stream decodes and the test recording
plays back clean. So it can do the job a virtual audio cable was doing —
this wires it to the QSO recorder, which already accepts a pushed source
(the Icom network audio uses the same door).

The conversion lives here rather than in internal/cat: the radio's job is
to hand over what it sent, not to know that the recorder works in 16 kHz
mono. Three samples are AVERAGED rather than two of them dropped —
decimating by picking every third folds everything above 8 kHz back into
the voice band, and on a receiver that is hiss, which a QSO recording has
plenty of already.

Off by default, and applied the moment it is switched: it replaces a
sound card the operator has already wired up, and an option that needs a
restart to take effect reads as an option that does not work.
2026-08-25 20:14:45 +02:00
rouggy 6a6b7ad6c2 chore(tci): probe binary frames per stream type
The frame log had one budget for the whole session, and the first forty
receive-audio frames spend it in under two seconds. A transmit-chrono
frame — the thing the voice keyer will have to answer, and whose size and
cadence cannot be read off the documentation — only appears once the
operator keys the radio, by which time nothing would have been logged.

Counted per type now, so the first frames of each kind are recorded
whenever they turn up.
2026-08-25 08:29:24 +02:00
rouggy 01a23ccb77 feat(tci): read the radio's declared format, and record a test WAV
The SunSDR announces its own stream at connect —
audio_stream_sample_type:float32 and audio_stream_channels:2 — and both
were being logged as unhandled while the code worked the format out from
frame arithmetic. The declaration is better evidence and arrives before
the first frame; the arithmetic stays as the check on it. The channel
count now drives the mix-down instead of an assumed stereo.

Adds a ten-second test recording, written as a WAV beside the QSO
recordings. Counting frames proves a socket is delivering bytes; it says
nothing about whether those bytes are the receiver's audio, at the right
rate, in the right order. A stream decoded with the width wrong or the
samples misaligned counts exactly as well as a correct one and sounds
like a fan — so the test is a file the operator can play, the same way
the CW decoder was settled on the air rather than on a spectrogram.

The file is written at the rate the RADIO reported, not a constant: a
recording at the wrong rate plays at the wrong speed, which is the one
fault that would be blamed on the decoding.
2026-08-25 08:25:29 +02:00
rouggy 9b8168370f fix(tci): derive the sample width instead of trusting the format number
A real SunSDR answers format=3 where this expected 0 — and 0 was read
from the documentation, which is exactly the kind of detail a memory of a
document gets wrong. The stream was refused outright: 'format=3, expected
float32', zero frames, silence.

Swapping one magic number for another would only move the guess, so the
width is now MEASURED: the header says how many samples the payload
holds, and dividing gives the bytes per sample. Four is float32, two is
16-bit PCM scaled to the same -1..1 the rest of the audio path uses, and
anything else is reported rather than mangled. That stays true whatever
number the format field carries on the next firmware.
2026-08-24 22:36:21 +02:00
rouggy efa711af78 feat(tci): receive audio over the TCI WebSocket (experimental)
A SunSDR already carries its receive audio on the same WebSocket as its
commands, so a virtual audio cable and a second sound card are two pieces
of plumbing an operator installs for no reason. This is the receive half:
what the QSO recorder and the CW decoder need.

The reader now looks at the frame type. It used to ignore it and split
every frame on ';' -- harmless only for as long as no stream was ever
opened, since audio bytes would otherwise have been handed to the command
parser a hundred times a second.

NOTHING HERE IS CONFIRMED ON A RADIO. The header layout comes from the
TCI documentation, and the stream-type numbers are exactly the sort of
detail a document gets right and a memory of it does not -- so the first
forty frames of a session are logged verbatim, and a test bench in
Preferences > Audio reports the sample rate the radio chose, the frames
arriving and the peak level of the last second. 'The stream is open' and
'audio is arriving' are different claims and only the second is worth
anything to whoever tries this first.

Transmit (the voice keyer) is the other half and is deliberately absent:
it has to answer the radio's chrono packets at the right pace, and that
is worth doing once the format is settled on real hardware.
2026-08-24 20:34:06 +02:00
20 changed files with 1394 additions and 324 deletions
+61 -306
View File
@@ -1,322 +1,77 @@
# OpsLog
# Transceiver Control Interface
<a href="https://discord.gg/FYM8yw5pT" target="_blank">
<img src="https://img.shields.io/badge/Discord-Join%20the%20server-5865F2?logo=discord&logoColor=white" alt="Join our Discord" />
</a>
## Introduction
A modern, fast ham-radio logger for Windows — single-strip entry, real-time CAT
for **OmniRig**, native **FlexRadio/SmartSDR**, native **Icom CI-V** (USB **and**
remote-over-internet, replacing RS-BA1) and **TCI** (SunSDR / Expert Electronics),
DX cluster with spot alerts, awards tracking, maps, contest logging, QSL
management and a QSL-card designer. Built with **Wails v2** (Go backend +
React/TypeScript frontend), **pure Go** (no CGO): SQLite for configuration,
optional **shared MySQL** for the logbook so several operators can run one log.
Fully themeable and bilingual (English / French).
TCI (Transceiver Control Interface) is a network interface for control, data transfer and
synchronization between transceiver/receiver, contest loggers, digital mode software, skimmers
and other software, as well as external power amplifiers, bandpass filter units, antenna switches,
radio controllers and other devices.
Developed by **F4BPO**.
TCI was created as a modern alternative to the outdated COM port and audio cable
interfaces, it uses a full duplex web socket protocol that runs on top of a TCP connection and
serves for server-client communications, providing cross-platform connectivity. Transceiver works
as a server, all other software and devices as clients. The server and clients can be inside the
same computer (program-server, hardware log, etc.-clients) and/or in separate physical devices
connected through the local network (classical transceiver, power amplifier, antenna switch, FFT
unit, etc.).
---
The TCI interface contains basic transceiver control commands (analog of CAT system),
receives CW macros from clients and broadcasts them, outputs transceiver IQ stream to clients,
receives spots from skimmers and Internet clusters, receives/outputs audio signal to work in
digital modes.
## Building / developing
The TCI uses an extensible architecture and can be supplemented with new functions and
commands, while keeping the old ones operational. Thus, the TCI interface can be extended and
supplemented to meet the specific needs of any software manufacturer and/or device
manufacturer (receivers, transceivers, power amplifiers, switches, etc.). The presence of a device
identifier allows the manufacturers of transceivers and receivers to switch to the TCI interface
while maintaining the device model designation. The extensibility of the TCI interface allows you
to create an individual set of commands and functions for each device model, while maintaining
the basic command set inherent to all transceivers.
- **Dev:** `wails dev` (Vite hot-reload; Go methods reachable at http://localhost:34115).
- **Build:** `wails build` (use the project's wails v2.11 — `~/go/bin/wails.exe`).
- **Regenerate Go↔TS bindings** after changing exported `App` methods:
`wails generate module`.
- **Release:** `.vscode/release.ps1` (Ctrl+Shift+P → *Tasks: Run Task*
*Release OpsLog*) — bumps the version, pushes source to Gitea, builds the exe
and publishes it to Gitea + GitHub releases.
Our company advocates universal unification of data exchange between devices and
software by creating the TCI interface for this purpose. Modern transceivers and software must
communicate using one protocol - the TCI protocol.
---
## Interface description
## Logging
Any command represents an ASCII string that contains a command name and a list of
arguments corresponding to this command. There are reserved characters that cannot be
included in the command name and command arguments.
- **Single-strip entry:** callsign, RST tx/rx, name/QTH/grid, band/mode,
TX/RX frequency (split), start/end time, comment/note. The contacted entity's
**flag** is shown large next to the RST fields.
- **Callsign lookup** (QRZ.com / HamQTH) with photo, auto-fill of name/QTH/grid
and the QRZ.com tab.
- **Offline DXCC** resolution from `cty.dat` (country, CQ/ITU zones, continent),
with `/MM` `/AM` `/B` (beacon) and call-area (`/8`, `/W6`) handling, plus
ClubLog DXpedition date overrides.
- **Recent QSOs**, **Worked-before** matrix (per band/mode slot), bulk re-resolve
from cty/QRZ/ClubLog, bulk send to QSL services. A live **selection count**, a
**Select all / Unselect all** toggle, and a row limit that keeps the full log
fast **but is lifted while a filter is active** (every match is shown, not just
the first page).
- **Advanced QSO filter builder** (field / operator / value, AND / OR, saved
presets) with filtered- and selected-row **ADIF export**.
- **Find duplicates** (Tools) — groups QSOs by same call + band + mode (optionally
same day / minute) and lets you pick which to delete.
- **ADIF 3.1.7 compliant** import/export: a full field dictionary, 30 promoted
columns, a generic "extra fields" editor and standard/all export modes.
- **Profiles:** every setting is per-profile; each profile can point its logbook
at the local SQLite file or a **shared MySQL** database (multi-operator).
List of reserved characters: «:», «,», «;».
## Maps & antenna
Command structure:
1. Name of the command;
2. Separating character between command name and arguments «:»;
3. Separating character between arguments «,»;
4. End of the command character «;».
- **Main view = two configurable panes** (per profile, Settings → General →
*Main view*): great-circle map, locator (street) map, the cluster grid, the
worked-before grid, recent QSOs, the **FlexRadio controls**, the **Icom
console** or the **Net control** panel.
- **Great-circle map** with short/long-path distance & azimuth, selectable
basemaps (Light / Voyager / Street / Satellite, all key-free and labelled) and
the **antenna beam lobe(s)** drawn from the rotor azimuth.
- **Rotor compass** (azimuthal-equidistant, click-to-turn) driven by
**PstRotator** (UDP), a **4O3A Rotator Genius** (native TCP) or a **microHAM
ARCO** controlled natively — no PstRotator needed — over the **LAN** or **USB**
using its Yaesu GS-232A protocol.
- **Ultrabeam** support (Normal / 180° reverse / Bidirectional): the radiating
direction is shown in green and the **mechanical boom** in grey, on both the
compass and the map, so you never lose track of where the antenna points.
If a command has no arguments, an end of command symbol is placed after the command
name. If the command is invalid, it is ignored. The case of letters does not matter.
## DX Cluster
The ExpertSDR3 program acts as a server, which can have several client connections at
the same time, they will be synchronized with each other by the server. When connecting to the
ExpertSDR3, the client receives the current status of the ExpertSDR3, first sending initialization
commands, then parameters to set the status, such as frequency, modulation, etc.
- Multiple cluster servers with auto-reconnect, a master for commands.
- **Filter sidebar** (callsign search, hide-worked, group duplicates, band /
mode / status / source) shared by the Cluster tab and the Main-view cluster
pane, with a show/hide toggle.
- Per-spot **status** (new / new-band / new-slot / worked), click-to-tune the
rig, and a multi-band **Band Map** (panadapter-style strips). Optionally, all
**digital modes count as one** (DXCC-style) for the new/new-slot colouring and
the worked-before matrix badges (Settings → General).
- **POTA** spots are tagged with their park reference (via `api.pota.app`).
- **Spot alerts:** rules on call / country / band / mode /
spotter, with sound, visual and e-mail notification (Tools → *Alert
management*).
When a parameter change occurs in the ExpertSDR3 (server) program, the server notifies
all connected clients, i.e., clients do not need to poll the server constantly, any change of state
will be sent in time to all clients. If the client sends a new state, the server will set it to itself, as
well as send it to all clients, that is, the server acts as a synchronizer. All clients connected to the
server will be automatically synchronized. This way of work allows to minimize network load,
reducing traffic.
## CAT control
The TCI protocol implements the transmission of receiver IQ stream to clients, which is
necessary for the work of special skimmer software, they automatically find the station and
decode it throughout the band, and it also allows you to record radio signals in the file.
Four native backends (Settings → CAT), each with auto-reconnect and a fast,
non-blocking connect so a powered-off radio never freezes the app:
TCI is also used to transmit audio signals of the receiver to clients and to receive audio
signals from clients, i.e., the client can transmit audio signals to ExpertSDR3 for radio
transmission. The audio stream exchange is designed to work with digital modes, where encoding
and decoding is performed by third-party software, as well as in voice modes, where audio macros
can be broadcasted, which is very much in demand in contest loggers.
- **OmniRig** (Rig 1/2, hot-swap) — works with any OmniRig-supported rig.
- **FlexRadio (SmartSDR)** over the radio's TCP API — real-time slice freq /
mode / split, UDP discovery, and **panadapter spots** (cluster spots pushed to
the Flex display; a click fills the call and **tunes the rig to the right
frequency AND mode**).
- **Icom CI-V** — native, over the radio's **USB** port *or* over the internet
via the radio's **built-in LAN server** (see *Remote Icom* below). No RS-BA1 or
Remote Utility needed.
- **TCI** (WebSocket) — SunSDR / ExpertSDR2 and any TCI-compatible server:
freq / mode / PTT / split, plus optional panorama spots.
Mode is taken from the radio; the digital sub-mode (FT4 vs FT8) is inferred from
the frequency. **Per-band Flex RX/TX antennas** can be configured and are applied
automatically on band change.
### FlexRadio control tab (SmartSDR-style)
Shown only when the CAT backend is a FlexRadio:
- **Transmit:** RF power, tune power, TUNE, MOX, speech processor (NOR/DX/DX+),
VOX (+ level + delay), monitor (+ level), mic gain.
- **Receive (active slice):** RX/TX **antenna** selectors and a **DAX** toggle
(TX audio through DAX, for WSJT-X & co), AGC mode/threshold, audio level,
NB / WNB / NR / ANF, and — on **SmartSDR v4** radios (8000 / Aurora series) —
the extra DSP tools **NRL, NRS, NRF** (with level) plus **RNN** (AI noise
reduction) and **ANFT** (FFT auto-notch), shown automatically when the radio
supports them. **RIT / XIT** with wheel / ± tuning.
- **Antenna tuner (ATU):** tune / bypass / memories.
- **Amplifier:** the amp card follows whichever amplifier is configured, with a
dropdown to pick it when **several amplifiers** are set up (e.g. two SPEs run
in parallel). See *Amplifiers & switches* below.
- **Live meters** over the UDP VITA-49 stream: S-meter (S-units), forward power
(W), SWR, ALC, PA temperature, voltage, plus the amplifier's meters.
### Icom control tab
Shown when the CAT backend is Icom (USB or network). A full RS-BA1-style console:
- **Twin VFO readout** (MAIN / SUB) with the big tabular frequency, mode badge,
band and RIT/ΔTX offset, and a **mode-button row** (SSB / CW / RTTY / PSK /
AM / FM).
- **Spectrum scope + waterfall** (panadapter): ON/OFF, CTR/FIX, double-click to
tune, and **◀ ⊙ ▶** buttons to centre the scope on the current frequency
(±50 kHz) and pan left/right.
- **Live meters** always visible: S-meter (click → fill RST), power in watts, SWR.
- **Receive DSP:** AF / RF gain, squelch, AGC, preamp, attenuator, filter
(FIL1/2/3), NB, NR, ANF and — **on CW only** — the **APF** (audio peak filter).
- **Passband / notch:** Twin PBT (inner / outer), manual notch + position.
- **Transmit:** RF power, MOX, TUNE, **split with an automatic offset**
(+5 kHz on SSB, +1 kHz on CW), and monitor. On **voice modes only**: mic gain,
speech compressor, VOX (+ gain + anti-VOX). Controls that don't apply to the
current mode are hidden automatically.
- **Bands & antenna:** one-touch band buttons and ANT1/ANT2 selection.
- **Clarifiers:** RIT and ΔTX with wheel / ± tuning (Ctrl+←/→ nudges RIT).
- **Power ON / OFF** buttons (manual by design — the app never wakes the rig on
connect).
- **CW keying** can run through the radio's own keyer (see *Keyers* below).
### Remote Icom (over the internet, no RS-BA1)
OpsLog speaks the IC-7610's built-in network protocol directly — it **replaces
both the Icom Remote Utility and RS-BA1**. Enter the radio's IP, the Network
User1 name/password and the CI-V address, and the whole Icom console works over
the LAN/internet: login + token (auto-renewed), CI-V tunnel, receive-side
retransmit for a rock-solid link even with the panadapter streaming, and manual
power ON/OFF. (Audio is out of scope — use the radio in USB + a voice link such
as Mumble.)
## Keyers & audio
- **CW keyer** with macros and F-key macros. The keyer engine is selectable:
**WinKeyer** (K1EL WK1/2/3 over a COM port), **FlexRadio CWX** (the radio's
built-in keyer over the SmartSDR API — type-ahead and backspace, no WinKeyer or
SmartCAT needed), **Icom** (the radio's own keyer over CI-V — no extra hardware,
works over the remote link too) or **TCI**.
- **Digital Voice Keyer** (DVK): record F1F6 voice messages and transmit them.
- **QSO audio recording:** continuous rolling capture; on *Log QSO* the contact
is saved to a per-QSO WAV (`CALL_YYYYMMDD_HHMMSS.wav`); mixes RX + mic.
## Amplifiers & switches
- **Amplifiers** — configure **one or several** amps (Settings → Amplifier is a
list; e.g. two SPEs run in parallel for more power). Each amp's control card
appears on the FlexRadio tab and in **Station Control**, with a dropdown to
choose which one it shows; the bottom status bar carries **one clickable chip
per amp** (green = OPERATE, orange = STANDBY, red = offline). Supported:
- **PowerGenius XL** (4O3A) over direct TCP — operate/standby, fan-mode
selector and fault display.
- **SPE Expert** (1.3K-FA / 1.5K-FA / 2K-FA) over **USB** (virtual COM) or the
**network** (RS232-to-Ethernet bridge) — operate/standby, ON/OFF,
Low / Mid / High output level, an output-power bar and live status (band,
SWR, PA current, temperature, warnings/alarms).
- **ACOM** (500S / 600S / 700S / 1200S / 2020S) over **USB** or the **network**
(RS232-to-Ethernet bridge) — operate/standby/off and live telemetry (forward
& reflected power, SWR, PA temperature, band, fan, faults). Power-ON works
over a serial cable that wires the DTR/RTS lines.
- **Antenna Genius** (4O3A) antenna switch over TCP/GSCP — a docked A/B
antenna-switch widget.
- **Station Control** panel (dockable, drag-to-reorder widgets): the **rotator**,
**Ultrabeam** element control and **relay boards** — WebSwitch 1216H, KMTronic,
**Denkovi** USB (FT245 D2XX bit-bang, 4 or 8 relays) and generic USB-serial
(CH340 / LCUS, A0 protocol) — for station power, antennas and accessories.
- **Relay auto-control** (Settings): switch Station-Control relays automatically
from the rig frequency / band (like PstRotator) — per relay, a frequency window
or a set of bands.
## QSL & awards
- **Awards engine:** built-in + custom award definitions (shared **globally**
across profiles) — DXCC, WAS / WAZ / WAC, WPX, IOTA / POTA / SOTA / WWFF,
**DDFM**, worked/confirmed/validated by band & mode, OR rules and manual
reference assignment, live reference detection on call entry, **reference-list
import** for totals/names, and a **Rescan** that re-pulls the logbook (picks up
fresh LoTW/QRZ confirmations).
- **QSL services:** ClubLog (batched ADIF upload), LoTW, QRZ.com, eQSL — upload
and **confirmation download** (which auto-refreshes the award stats).
- **QSL Card Designer** (see below).
- **E-mail eQSL:** right-click a QSO → *Send eQSL by e-mail* via the configured
SMTP account. (Outlook/Hotmail disable basic-auth SMTP — use Gmail with an app
password, or a Microsoft app password.)
## Contest logging
- **Contest tab:** pick a contest (built-in ADIF `CONTEST_ID` list) and an
exchange (running serial or a fixed exchange). OpsLog auto-fills `CONTEST_ID`
and the sent/received serials (`STX` / `SRX`), enforces a window start/end,
flags dupes and keeps a live scoreboard.
## Statistics
- **Logbook statistics dashboard:** headline tiles (QSOs, unique callsigns, DXCC
entities, continents, % confirmed) plus charts — QSOs **by mode**, a per-band
**CW / phone / data** split, **activity over time** (rolling day / 7-day /
30-day / 12-month views), by operator and by continent. Date-range, per-operator
and per-contest filters, and a **Table** view that mirrors every chart.
## Multi-operator live status (special events)
For a multi-op special-event call on a shared MySQL logbook (e.g. **TM74TFR**),
publishing is **automatic** — no setting to turn on. Each OpsLog instance
heartbeats its current activity (operator call, band, frequency, mode) into a
`live_status` table every ~15 s, and drops back to *off air* automatically 5 min
after the last logged QSO. A small PHP renderer
([`docs/livestatus/tm74-status.php`](docs/livestatus/tm74-status.php)) on your
own web server reads that table and produces a live page/image you can embed on
the station's **QRZ.com** bio (`<img src="…/tm74-status.php?img=1">`). OpsLog
only writes to the DB — it is not a web server.
## Net control
- **Directed-net logging** (Tools → Net): a global roster (`nets.json`) plus an
in-memory active session — check stations in, then log them individually or the
whole net at once (**Log everyone**) using the CAT frequency. **Drag & drop**
between the two lists (roster → on-air starts a QSO, on-air → roster logs it),
and after each log the next on-air station is selected automatically so you can
chain contacts.
## Appearance & language
- **Themes:** four complete themes (Warm light, Warm dark, Graphite dark, High
contrast) plus **Auto** (follows the OS light/dark preference), selectable in
Settings → General. Every panel and every AG-Grid table follows the theme.
- **Bilingual:** full **English / French** UI, with a first-run flag chooser and
a switcher in Settings → General.
## Security
- **Secret vault:** opt-in passphrase encryption of the stored passwords
(AES-GCM + PBKDF2). Encrypted values are portable; a single unlock prompt at
launch decrypts them for the session.
## Integrations (outbound)
- **UDP emitters:** push the current frequency to **PstRotator**, radio info in
**N1MM `RadioInfo`** format, or an **ADIF record on each logged QSO** — so
external tools (rotator control, digital apps, other loggers) stay in sync.
## Other
- **Autostart:** launch external programs (WSJT-X, JTAlert, rotator control…) at
OpsLog startup, skipping any already running.
- **Backup:** optional database + ADIF backup at shutdown.
- **Update check** at startup and every 5 minutes (and on opening Help → About),
with a toast (toggleable), plus a **What's new** dialog that shows the changelog
(English / French) on the first launch after an update — reopenable any time
from the Help menu.
- **Anonymous usage telemetry** (a once-a-day heartbeat: random install ID +
version + OS — no callsign or QSO data; opt-out in Preferences).
---
## QSL Card Designer
Tools → *QSL Card Designer…* turns a few photos into a polished eQSL card:
1. Pick 16 photos (jpeg/png). OpsLog analyzes them offline (detail/luminance
grid) and proposes **3 designs** — callsign in the calmest zone of the best
photo, operator name, CQ/ITU zones + locator line, country flag, the other
photos as bordered inserts, and a per-QSO confirmation box.
2. Pick a proposal and fine-tune it: click an element to select, drag to move,
change font / style preset (gel gold, gel silver, classic white outline,
script, flat) and per-preset knobs in the right panel.
3. Save the template (photos are copied into `data/qsl/templates/<id>/`, so the
originals can move). One template can be the default per profile.
Sending: right-click a QSO → *Send eQSL by e-mail*. The card is rendered with
that QSO's data, rasterized to a ≤ 800 KB JPEG, archived in `data/qsl/outbox/`
and sent through the configured SMTP account to the address found by the
QRZ/HamQTH lookup. On success the QSO is stamped `EQSL_SENT=Y` (ADIF). The
e-mail subject/body templates live in the designer
(`{CALL} {DATE} {BAND} {MODE} {MYCALL}` variables).
Fonts: Archivo Black, Lilita One, Baloo 2, Oswald, Great Vibes, Allura (all
OFL, embedded — licenses in `internal/qslcard/assets/fonts/`); Cooper Black is
offered when MS Office installed it. Flags: flag-icons (MIT), embedded for the
commonly-worked DXCC entities.
---
## Data & storage
- **Config** (settings, profiles, rigs/antennas, cluster nodes, lookup cache,
award lists, QSL templates) always lives in the local SQLite file under
`data/` — instant even when the logbook is on a far-away MySQL.
- **Logbook** (QSOs) lives where the active profile points it: the local SQLite
file or a per-profile shared **MySQL** database.
---
*A French version of this document is available in [README.fr.md](README.fr.md).*
When working in contests, it is important to record all on the air operation, for this purpose
the audio stream from the line-output is sent to all clients. The resulting audio stream can be
recorded to a file or played back with a PC sound card.
+72 -4
View File
@@ -8350,8 +8350,49 @@ type AudioSettings struct {
// ListAudioInputDevices / ListAudioOutputDevices enumerate WASAPI endpoints
// for the device dropdowns.
func (a *App) ListAudioInputDevices() ([]audio.Device, error) { return audio.ListInputDevices() }
func (a *App) ListAudioOutputDevices() ([]audio.Device, error) { return audio.ListOutputDevices() }
// ListAudioInputDevices lists the microphones and line inputs, plus THE RADIO
// when the CAT link carries its receive audio.
//
// Same reasoning as the output list: over TCI there is no sound device for
// Windows to show, so without this the one correct answer to "where does the
// received audio come from" could not be chosen at all.
func (a *App) ListAudioInputDevices() ([]audio.Device, error) {
devs, err := audio.ListInputDevices()
if err != nil {
return devs, err
}
if a.tciAudioAvailable() {
devs = append([]audio.Device{{ID: audio.NetworkDeviceID, Name: "Radio (TCI network audio)"}}, devs...)
}
return devs, nil
}
// tciAudioAvailable says whether the active CAT backend is a radio that streams
// its audio over the CAT link.
func (a *App) tciAudioAvailable() bool {
if a.cat == nil {
return false
}
_, ok := a.cat.TCIAudioState()
return ok
}
// ListAudioOutputDevices lists the sound cards, plus THE RADIO ITSELF when the
// CAT link can carry transmit audio.
//
// Offered only while it is actually available, and named as a radio rather than
// as a protocol: an operator choosing where their voice goes is picking between
// "my sound card" and "the radio", not between WASAPI and TCI.
func (a *App) ListAudioOutputDevices() ([]audio.Device, error) {
devs, err := audio.ListOutputDevices()
if err != nil {
return devs, err
}
if audio.NetworkPlayerReady() {
devs = append([]audio.Device{{ID: audio.NetworkDeviceID, Name: "Radio (TCI network audio)"}}, devs...)
}
return devs, nil
}
// GetAudioSettings returns the stored audio config (preroll defaults to 8s).
func (a *App) GetAudioSettings() (AudioSettings, error) {
@@ -8453,6 +8494,20 @@ func (a *App) SaveAudioSettings(s AudioSettings) error {
return err
}
}
// Choosing the radio as the receive device opens its stream, and choosing
// anything else closes it. Done HERE rather than left to the next restart:
// a device chosen in a dropdown that only takes effect after a relaunch
// reads as a device that does not work.
// The transmit source follows the "To radio" device, and is applied now:
// the radio is told which input to use at the moment it is keyed, so the
// setting has to be right before the next message rather than after the
// next restart.
a.applyTCITXSource()
if s.FromRadio == audio.NetworkDeviceID {
a.startTCIRecording()
} else if a.tciAudioAvailable() {
_ = a.cat.TCIAudioDo(func(t cat.TCIAudioController) error { return t.StopTCIAudio() })
}
// Apply device/preroll/enable changes to the running recorder.
a.startQSORecorderIfEnabled()
// And to a monitor ALREADY RUNNING: the operator is listening while they
@@ -8497,7 +8552,7 @@ func (a *App) startQSORecorderIfEnabled() {
// nothing right to point at. The stream is pushed into the recorder instead
// — same samples, no sound card in the middle, and no virtual cable to set up.
from := cfg.FromRadio
a.qsoRecPushed = a.icomNetAudioActive()
a.qsoRecPushed = a.icomNetAudioActive() || cfg.FromRadio == audio.NetworkDeviceID
if a.qsoRecPushed {
from = audio.PushedSource
}
@@ -15178,6 +15233,12 @@ func (a *App) reloadCAT() {
} else {
a.catSig = sig
}
// Withdraw the radio as an audio output before deciding anything else. The
// TCI case below puts it back; every other backend, and a CAT link turned
// off entirely, leaves it withdrawn — a voice keyer that still lists a radio
// it can no longer reach would play a message to nowhere, and the operator
// hears their own PTT click and assumes it went out.
a.installTCITXPlayer(false)
if !s.Enabled {
a.cat.Stop()
return
@@ -15300,7 +15361,14 @@ func (a *App) reloadCAT() {
a.cat.Start(cat.NewIcomNet(s.IcomNetHost, s.IcomNetUser, s.IcomNetPass, s.IcomAddr, s.DigitalDefault, audioSink))
case "tci":
// Expert Electronics TCI (WebSocket) — SunSDR / ExpertSDR2, or any
// TCI-compatible server.
// TCI-compatible server. The receive audio rides the same socket, so
// the QSO recorder can take it without a virtual cable — see
// app_tci_rec.go. Armed after the backend is up, since it is the
// backend that carries the stream.
defer a.startTCIRecording()
// And the other direction: the voice keyer can send its messages over
// the same link — see app_tci_dvk.go.
defer a.installTCITXPlayer(true)
tb := cat.NewTCI(s.TCIHost, s.TCIPort, s.DigitalDefault, s.TCISpots)
// Clicking one of our spots on the ExpertSDR panorama fills the entry form.
tb.OnSpotClick = func(call string, hz int64) {
+86
View File
@@ -0,0 +1,86 @@
package main
// The voice keyer, through the radio's own link.
//
// Selecting the radio as the "To radio" output makes the voice keyer hand its
// messages to the CAT backend instead of a sound card. Everything around it is
// unchanged — the same PTT before and after, the same gain, the same files —
// which is the point: the audio takes a different road, not a different route.
import (
"fmt"
"hamlog/internal/applog"
"hamlog/internal/audio"
"hamlog/internal/cat"
)
// tciTXPlayer hands one message to the radio.
//
// The controller is fetched on the CAT goroutine and the message is then played
// OFF it. Playing on it would hold that goroutine for the length of the
// message, and everything else about the rig — frequency, mode, PTT state —
// goes through the same place: a ten-second call would freeze the display and
// the antenna following for ten seconds. The TCI backend serialises its own
// writes, so this is safe to call from here.
func (a *App) tciTXPlayer(pcm []byte, rate, ch, bits int, stop <-chan struct{}) error {
if a.cat == nil {
return fmt.Errorf("CAT not initialized")
}
type txPlayer interface {
PlayTXAudio(pcm []byte, rate, ch, bits int, stop <-chan struct{}) error
}
var player txPlayer
err := a.cat.TCIAudioDo(func(t cat.TCIAudioController) error {
p, ok := t.(txPlayer)
if !ok {
return fmt.Errorf("this radio cannot take transmit audio over its CAT link")
}
player = p
return nil
})
if err != nil {
return err
}
return player.PlayTXAudio(pcm, rate, ch, bits, stop)
}
// applyTCITXSource tells the radio where to take its transmit audio from,
// following the "To radio" device.
//
// This is the setting an operator would otherwise have to find in ExpertSDR3
// and set again for every mode, because it is remembered per mode there. It is
// the third argument of TRX, so OpsLog can simply say it each time it keys —
// and a radio told nothing keeps the operator's microphone, which is what
// every PTT that is not a voice message should do.
func (a *App) applyTCITXSource() {
if a.cat == nil {
return
}
cfg, _ := a.GetAudioSettings()
src := ""
if cfg.ToRadio == audio.NetworkDeviceID {
src = "tci"
}
_ = a.cat.TCIAudioDo(func(t cat.TCIAudioController) error {
if s, ok := t.(interface{ SetTXAudioSource(string) }); ok {
s.SetTXAudioSource(src)
}
return nil
})
}
// installTCITXPlayer offers the radio as an audio output, or withdraws it.
//
// Withdrawing matters as much as offering: a radio that has gone away must stop
// being a device the voice keyer will happily "play" to, or a message goes
// nowhere and the operator hears their own PTT click and assumes it worked.
func (a *App) installTCITXPlayer(on bool) {
if !on {
audio.SetNetworkPlayer(nil)
return
}
audio.SetNetworkPlayer(a.tciTXPlayer)
a.applyTCITXSource()
applog.Printf("tci: the radio is available as an audio output — no virtual cable needed for the voice keyer")
}
+98
View File
@@ -0,0 +1,98 @@
package main
// Feeding the QSO recorder from the TCI stream.
//
// The recorder works in 16 kHz mono, which is what its files and its mixing are
// built around; TCI delivers 48 kHz stereo float32. The conversion is the whole
// of this file, and it happens here rather than in internal/cat because the
// radio's job is to hand over what it sent, not to know what the recorder wants.
//
// Confirmed on a SunSDR (ExpertSDR3 1.5): 2048 samples a frame, 8192 bytes,
// four bytes per sample — and a test recording that plays back clean.
import (
"encoding/binary"
"hamlog/internal/applog"
"hamlog/internal/audio"
"hamlog/internal/cat"
)
// tciRecordSink pushes the receive stream into the QSO recorder.
//
// Installed whenever the TCI backend starts, and harmless when nothing is
// recording: PushRX drops what arrives unless a QSO is being captured, so the
// cost while idle is a decimation and a function call.
func (a *App) tciRecordSink(rate int, samples []float32) {
if a.qsoRec == nil || len(samples) == 0 {
return
}
a.qsoRec.PushRX(tciToRecorderPCM(rate, samples))
}
// tciToRecorderPCM converts the stream's mono float samples to the recorder's
// 16-bit PCM at its own rate.
//
// Averaging rather than picking every third sample: dropping samples aliases
// everything above 8 kHz back down into the voice band, and on a receiver that
// is hiss — the one thing a QSO recording has plenty of. A three-tap mean is a
// crude low-pass, but it is a low-pass, and it costs two additions.
func tciToRecorderPCM(rate int, samples []float32) []byte {
if rate <= 0 {
rate = 48000
}
step := rate / audio.RecorderSampleRate
if step < 1 {
step = 1
}
out := make([]byte, 0, (len(samples)/step)*2)
for i := 0; i+step <= len(samples); i += step {
var sum float32
for j := 0; j < step; j++ {
sum += samples[i+j]
}
v := sum / float32(step)
if v > 1 {
v = 1
}
if v < -1 {
v = -1
}
var b [2]byte
binary.LittleEndian.PutUint16(b[:], uint16(int16(v*32767)))
out = append(out, b[0], b[1])
}
return out
}
// startTCIRecording opens the receive stream and routes it to the recorder.
//
// Called when the TCI backend comes up, and only when the operator has asked
// for it by choosing the radio as their receive device: opening a 384 kB/s
// stream on a station that records nothing is work the radio does for nobody.
func (a *App) startTCIRecording() {
if a.cat == nil {
return
}
// One switch, and it is the one an operator is already looking at: the
// "From radio" device. There used to be a tick box here as well, from when
// this was an experiment with no device to choose — two controls for one
// question, and the second was where nobody would look.
cfg, _ := a.GetAudioSettings()
if cfg.FromRadio != audio.NetworkDeviceID {
return
}
err := a.cat.TCIAudioDo(func(t cat.TCIAudioController) error {
if s, ok := t.(interface {
SetTCIAudioSink(func(int, []float32))
}); ok {
s.SetTCIAudioSink(a.tciRecordSink)
}
return t.StartTCIAudio(0, 48000)
})
if err != nil {
applog.Printf("tci: could not open the receive stream for recording: %v", err)
return
}
applog.Printf("tci: recording the receive audio over TCI — no virtual cable needed")
}
+47
View File
@@ -0,0 +1,47 @@
package main
import (
"encoding/binary"
"testing"
"hamlog/internal/audio"
)
// The stream is 48 kHz and the recorder works at 16 — three to one. A
// recording that keeps every sample plays back three times too fast, which is
// the fault that gets blamed on the decoding rather than on the rate.
func TestTheStreamIsResampledToTheRecorderRate(t *testing.T) {
const in = 48000
samples := make([]float32, in/10) // a tenth of a second
pcm := tciToRecorderPCM(in, samples)
want := (audio.RecorderSampleRate / 10) * 2 // 16-bit
if len(pcm) != want {
t.Fatalf("a tenth of a second produced %d bytes, want %d", len(pcm), want)
}
}
// Full scale must arrive as full scale: a conversion that quietly halves the
// level turns a recording into evidence of a fault that is not there.
func TestFullScaleSurvivesTheConversion(t *testing.T) {
samples := make([]float32, 12)
for i := range samples {
samples[i] = 1
}
pcm := tciToRecorderPCM(48000, samples)
if len(pcm) < 2 {
t.Fatal("no samples came out")
}
v := int16(binary.LittleEndian.Uint16(pcm[:2]))
if v < 32000 {
t.Fatalf("full scale came out at %d", v)
}
}
// A rate the recorder already works in is passed through rather than mangled by
// a division that would round to nothing.
func TestAStreamAtTheRecorderRateIsNotDecimated(t *testing.T) {
samples := make([]float32, 160)
if got, want := len(tciToRecorderPCM(audio.RecorderSampleRate, samples)), 160*2; got != want {
t.Fatalf("%d bytes, want %d", got, want)
}
}
+26
View File
@@ -1,4 +1,30 @@
[
{
"version": "0.26.16",
"date": "",
"en": [
"TCI transmit needs no setting up in ExpertSDR3 any more. The protocol lets the audio source be named when keying, so OpsLog does it — the voice keyer works in SSB as well as in the digital modes, nothing has to be set per mode in the radio's own window, and the microphone stays the source for every transmission that is not a voice message."
],
"fr": [
"L'émission TCI ne demande plus aucun réglage dans ExpertSDR3. Le protocole permet de nommer la source audio au moment de passer en émission, et OpsLog le fait — le manipulateur vocal fonctionne en SSB comme en numérique, il n'y a plus rien à régler mode par mode dans la fenêtre de la radio, et le micro reste la source pour toute émission qui n'est pas un message vocal."
]
},
{
"version": "0.26.15",
"date": "",
"en": [
"SunSDR / ExpertSDR3: the radio's audio now travels over the TCI link itself. Pick 'Radio (TCI network audio)' as the From Radio or To Radio device and the QSO recorder and the voice keyer work with no virtual cable, no second sound card and nothing to set up in the Windows mixer.",
"For transmit, ExpertSDR3's own transmit audio source must be set to TCI rather than the microphone — it is remembered per mode, so setting it in SSB does not set it in DIGU. OpsLog says so within a fifth of a second rather than transmitting silence.",
"Awards, RDA district comparison: the list stays where it was put. It was thrown back to the first row every three seconds, which made a long list of contacts to correct impossible to work through.",
"Elecraft console: the SWR bar works. The radio answers SW; with three digits in tenths of a ratio — SW023 is 2.3:1 — and OpsLog was reading four, so every answer was discarded and the bar stayed empty."
],
"fr": [
"SunSDR / ExpertSDR3 : l'audio de la radio passe désormais par la liaison TCI elle-même. Choisis « Radio (TCI network audio) » comme périphérique From Radio ou To Radio et l'enregistreur de QSO comme le manipulateur vocal fonctionnent sans câble virtuel, sans seconde carte son et sans rien à régler dans le mixeur Windows.",
"Pour l'émission, la source audio d'émission d'ExpertSDR3 doit être réglée sur TCI et non sur le micro — elle est mémorisée par mode, donc la régler en SSB ne la règle pas en DIGU. OpsLog le dit en deux dixièmes de seconde au lieu d'émettre du silence.",
"Diplômes, comparaison des districts RDA : la liste reste où on l'a laissée. Elle revenait à la première ligne toutes les trois secondes, ce qui rendait impraticable une longue liste de contacts à corriger.",
"Console Elecraft : la barre de ROS fonctionne. La radio répond à SW; par trois chiffres en dixièmes de rapport — SW023 vaut 2,3:1 — et OpsLog en lisait quatre, si bien que chaque réponse était jetée et la barre restait vide."
]
},
{
"version": "0.26.14",
"date": "",
+34 -4
View File
@@ -2035,6 +2035,13 @@ function SettingsModalImpl({ onClose, onSaved, initialSection, onMainPaneChanged
const [spotTTL, setSpotTTL] = useState(0);
const [spotTTLText, setSpotTTLText] = useState('0');
const [spotMaxText, setSpotMaxText] = useState('1000');
// TCI receive-audio test bench. Polled only while the stream is open: a panel
// that asks the backend twice a second for a stream nobody started is work
// done for nothing.
// Whether the QSO recorder takes its audio from the radio's own stream.
const [gridStat, setGridStat] = useState<any>(null);
const [pskrStatus, setPskrStatus] = useState<any>(null);
const saveBandOpen = async (next: any) => {
@@ -2050,14 +2057,22 @@ function SettingsModalImpl({ onClose, onSaved, initialSection, onMainPaneChanged
try { const n = await GetSpotTTLMinutes(); setSpotTTL(n); setSpotTTLText(String(n)); } catch { /* defaults stand */ }
try { const n = await GetSpotMax(); setSpotMaxText(String(n)); } catch { /* defaults stand */ }
})();
// Poll the feed while the panel is open: a live count is the only thing that
// distinguishes "connected" from "connected and receiving nothing".
}, []);
// Poll the feed only while the section that SHOWS it is open.
//
// A live count is the one thing that separates "connected" from "connected
// and receiving nothing", so it has to be polled — but it was polled from
// everywhere, re-rendering the whole dialog every three seconds whichever
// panel was in front. That is a heartbeat through every list and every form
// in Preferences for a number nobody is looking at.
useEffect(() => {
if (selected !== 'cluster') return;
const t = window.setInterval(async () => {
try { setPskrStatus(await GetPSKReporterStatus()); } catch { /* ignore */ }
try { setGridStat(await GetGridCacheStatus()); } catch { /* ignore */ }
}, 3000);
return () => window.clearInterval(t);
}, []);
}, [selected]);
const [selfSpot, setSelfSpot] = useState({ enabled: false, minutes: SELF_SPOT_MIN_MIN });
const [selfSpotText, setSelfSpotText] = useState(String(SELF_SPOT_MIN_MIN));
const [clusterStatuses, setClusterStatuses] = useState<ClusterServerStatus[]>([]);
@@ -6738,6 +6753,12 @@ function SettingsModalImpl({ onClose, onSaved, initialSection, onMainPaneChanged
<strong>{t('aud.fromRadioShort')}</strong> {t('aud.explainFrom')}{' '}
<strong>{t('aud.toRadioShort')}</strong> {t('aud.explainTo')}
</p>
{/* The radio is one of the devices above when it can carry its own
audio see ListAudioInputDevices. What used to be here was a test
bench: open the stream, record ten seconds, key a tone. It settled
how TCI works and has no business in front of an operator now that
choosing the device is the whole of the setup. */}
<div className="flex items-center gap-3">
<Button
variant={monitorOn ? 'default' : 'outline'}
@@ -7560,7 +7581,16 @@ function SettingsModalImpl({ onClose, onSaved, initialSection, onMainPaneChanged
uscounties: USCountiesPanel,
databases: DatabasesPanel,
autostart: () => <AutostartPanelComponent />,
awards: () => (<div className="space-y-6"><AwardsSelectionPanel profile={activeProfile ?? undefined} /><RDAPanel /></div>),
// RDAPanel is CALLED, not written as <RDAPanel />.
//
// It is nested inside this component, so as an element it would be a new
// component TYPE on every render — React cannot know it is the same panel,
// so it unmounts the old tree and mounts a fresh one. A fresh scroll
// container starts at the top, which is what threw the district comparison
// back to the first row every three seconds. Calling it produces the same
// elements in place, and the scroll position is simply never disturbed.
// (Safe because RDAPanel holds no hooks of its own — see PanelHost.)
awards: () => (<div className="space-y-6"><AwardsSelectionPanel profile={activeProfile ?? undefined} />{RDAPanel()}</div>),
cat: CATPanel,
rotator: RotatorPanel,
winkeyer: WinkeyerPanel,
+1 -1
View File
@@ -1,6 +1,6 @@
// Single source of truth for the app version shown in the UI (header + About).
// Bump this on a release (the release script updates it alongside telemetry.go).
export const APP_VERSION = '0.26.14';
export const APP_VERSION = '0.26.16';
// Author / credits, shown in Help -> About.
export const APP_AUTHOR = 'F4BPO';
+15 -1
View File
@@ -157,7 +157,21 @@ func (m *Manager) Play(deviceID, path string, gainPct int) error {
// instantly, the PTT is released 120 ms later, and NOTHING says why —
// which is exactly what a station heard as "it plays once, then never
// again": the call succeeded, the sound did not.
if err := playPCM(deviceID, pcm, rate, ch, bits, stop); err != nil {
play := func() error { return playPCM(deviceID, pcm, rate, ch, bits, stop) }
if deviceID == NetworkDeviceID {
// Straight to the radio over its own link. Decided HERE rather than
// inside playPCM because there is no Windows endpoint to open: asked
// for one, the system complains about a missing device instead of
// saying the true thing, which is that no radio is connected.
fn := networkPlayer()
play = func() error {
if fn == nil {
return errNoNetworkRadio
}
return fn(pcm, rate, ch, bits, stop)
}
}
if err := play(); err != nil {
LogSink("audio: playback on %q failed: %v", DeviceName(deviceID), err)
}
m.mu.Lock()
+67
View File
@@ -0,0 +1,67 @@
package audio
// Playing a message through the RADIO instead of a sound card.
//
// A SunSDR takes its transmit audio over TCI, on the same socket as the
// commands, so the voice keyer can hand it the message directly: no virtual
// cable, no second sound card, no Windows mixer between the recording and the
// air. To everything above, that radio is simply another output device.
//
// The device it presents itself as is a name rather than a WASAPI endpoint id,
// which is why Play checks for it before opening anything: there is no endpoint
// to open, and asking Windows for one produces a confusing error about a device
// that does not exist rather than the truth, which is that nothing is connected
// to the radio.
import (
"errors"
"sync"
)
// NetworkDeviceID is the id the radio-over-network output carries in the
// settings and in the device lists. A fixed string, not a Windows endpoint id:
// it is chosen by us and must survive a radio being switched off and on.
const NetworkDeviceID = "net:radio"
// NetworkPlayer sends already-decoded PCM to the radio, returning when the
// message has been played or when stop is closed.
//
// It carries the same arguments as the sound-card path so that Play can hand
// over whatever it read, and the radio can decide what converting it needs —
// the sample rate a WAV was recorded at is not the radio's business until the
// moment it has to be resampled.
type NetworkPlayer func(pcm []byte, rate, ch, bits int, stop <-chan struct{}) error
var (
netMu sync.RWMutex
netPlayer NetworkPlayer
)
// SetNetworkPlayer installs (or clears, with nil) the radio's transmit path.
//
// Package-level rather than per-Manager: there is one radio, the CAT backend
// owns it, and a Manager that happened to be built before the radio connected
// would otherwise be permanently unable to reach it.
func SetNetworkPlayer(fn NetworkPlayer) {
netMu.Lock()
netPlayer = fn
netMu.Unlock()
}
// networkPlayer returns the installed player, or nil.
func networkPlayer() NetworkPlayer {
netMu.RLock()
defer netMu.RUnlock()
return netPlayer
}
// NetworkPlayerReady says whether a radio is currently able to take transmit
// audio, so the settings panel can offer the option honestly rather than
// listing a device that would fail when used.
func NetworkPlayerReady() bool { return networkPlayer() != nil }
// errNoNetworkRadio is what a message played to a radio that is not there
// comes back with. Named, because "the device could not be opened" would send
// an operator hunting through Windows sound settings for a device that never
// existed.
var errNoNetworkRadio = errors.New("no radio is connected to take the audio — check the CAT link (the radio output only works with a TCI radio)")
+9
View File
@@ -0,0 +1,9 @@
package audio
// RecorderSampleRate is the rate the QSO recorder works in.
//
// Exported because a source that is NOT a sound card — the TCI receive stream,
// the Icom network audio — has to resample into it, and hard-coding 16000 at
// each of those call sites is how one of them ends up at the wrong speed after
// this constant is ever changed.
const RecorderSampleRate = sampleRate
+1
View File
@@ -14,6 +14,7 @@ import (
// any device regardless of its native mix format.
const (
sampleRate = 16000
channels = 1
bitsPerSample = 16
blockAlign = channels * bitsPerSample / 8 // bytes per frame (=2)
+12 -4
View File
@@ -15,6 +15,10 @@ package cat
// good match on a bad antenna. So the raw answers are LOGGED, for a real
// radio to settle, and until then the panel says the scaling is provisional.
//
// SWR is settled: SW; answers three digits in tenths of a ratio ("SW023;" =
// 2.3:1), from Elecraft's release note. The power meter is still read from the
// bargraph and still provisional.
//
// The same discipline as the Yaesu meters, which were guessed wrong twice and
// only settled when an FTDX10 keyed a carrier at two known power levels.
@@ -309,7 +313,9 @@ func kenwoodAGCValue(name string) int {
// answered and what it said, next to the power SETTING: the meter that tracks a
// known carrier at two different power levels is the power meter, and no amount
// of reading the reference settles that as well as one transmission does.
var kenwoodMeterProbes = []string{"SM;", "SMH;", "BG;", "SW;", "PO;", "TQ;"}
// SW is no longer among them: it is known, read above, and asking again during
// a transmission costs a round trip on the one link the carrier depends on.
var kenwoodMeterProbes = []string{"SM;", "SMH;", "BG;", "PO;", "TQ;"}
// readTXMeters reads the transmit meters.
func (k *Kenwood) readTXMeters() {
@@ -321,10 +327,12 @@ func (k *Kenwood) readTXMeters() {
if v, ok := k.askNum("BG;", "BG", 2); ok {
k.panel.PowerMeter = k.powerPeak.update(kenwoodBargraphPercent(v), now)
}
if v, ok := k.askNum("SW;", "SW", 4); ok {
// SW; — SETTLED, from Elecraft's own release note: three digits, tenths of a
// ratio. "SW023;" is 2.3:1, and "SW999;" is the 99.9:1 it reports instead of
// infinity. This was reading FOUR digits, so every answer failed to parse
// and the bar stayed empty — which is why a tester saw no SWR at all.
if v, ok := k.askNum("SW;", "SW", 3); ok {
k.panel.SWRRaw = v
// Tenths of a ratio, provisionally: 15 → 1.5. Reported as raw as well,
// so the log can correct this without anyone having to trust the bar.
if v > 0 {
k.panel.SWR = float64(k.swrPeak.update(v, now)) / 10
}
+94 -2
View File
@@ -34,6 +34,16 @@ type TCI struct {
OnSpotClick func(callsign string, freqHz int64)
unhandledSeen map[string]bool // log each unknown TCI message type once
// audio holds the receive-audio stream — see tci_audio.go. TCI carries it
// on this same WebSocket, which is what lets a SunSDR record and decode
// without a virtual audio cable in the way.
audio tciAudio
// One writer at a time. send() held the lock only long enough to READ conn,
// which was enough while every command came from the poll loop — a stream of
// audio frames from a second goroutine is not, and gorilla panics on a
// concurrent write rather than corrupting the socket quietly.
wmu sync.Mutex // serialises writes to the socket (text AND binary)
mu sync.Mutex // guards conn + writes + state
conn *websocket.Conn
dialCancel context.CancelFunc // cancels an in-flight Connect dial (Interrupt/Stop)
@@ -56,7 +66,14 @@ type TCI struct {
// txAllowedKnown keeps an OLDER ExpertSDR, or a TCI-compatible program that
// never sends TX_ENABLE at all, from being treated as refusing: without a
// word from the radio we key and let it decide.
txAllowed bool
txAllowed bool
// txSource is the TRX third argument: "tci" while OpsLog has audio to send,
// empty for the operator's microphone. See SetPTT.
txSource string
// drive is the radio's transmit drive, 0-100. Kept because a quiet
// transmission has two possible causes — our level or the radio's — and a
// log that names both settles it in one line instead of an evening.
drive int
txAllowedKnown bool
lastSig string // last logged state signature (log only on change)
@@ -330,9 +347,49 @@ func (t *TCI) SetPTT(on bool) error {
"check the frequency is inside a transmit band and that TX is enabled in ExpertSDR")
}
}
// THE THIRD ARGUMENT NAMES THE AUDIO SOURCE, and it is the whole answer to
// "why does the radio ignore what I send it".
//
// TCI 2.0 §TRX: "The signal for transmitting is always taken from the
// microphone selected in the ExpertSDR3. If a third-party software connected
// via TCI wants to transmit its audio signal, you must specify the third
// argument - TCI." Without it the radio never sends a single chrono frame,
// whatever the mode and whatever is configured in its window — which is
// exactly what a night of experiments showed and misread as "digital modes
// only".
//
// Sent only when a transmission is ours to feed. A plain trx keeps the
// operator's own microphone, which is what every other PTT in OpsLog means.
t.mu.Lock()
src := t.txSource
t.mu.Unlock()
if on && src != "" {
return t.send(fmt.Sprintf("trx:0,true,%s;", src))
}
return t.send(fmt.Sprintf("trx:0,%t;", on))
}
// SetTXAudioSource says where the radio should take its transmit audio from
// while OpsLog keys it: "tci" for the stream this program sends, "" for the
// microphone the operator chose in ExpertSDR3.
//
// Set from the audio settings — it follows the "To radio" device — so keying
// for a voice message and keying for anything else behave differently on
// purpose: only the first one takes the audio away from the microphone.
func (t *TCI) SetTXAudioSource(src string) {
t.mu.Lock()
changed := t.txSource != src
t.txSource = src
t.mu.Unlock()
if changed {
if src == "" {
debugLog.Printf("TCI: transmit audio will come from the radio's own microphone")
} else {
debugLog.Printf("TCI: transmit audio will be taken from %s when OpsLog keys the radio", src)
}
}
}
// send writes a command to the WebSocket (one writer at a time).
func (t *TCI) send(cmd string) error {
t.mu.Lock()
@@ -341,6 +398,8 @@ func (t *TCI) send(cmd string) error {
if c == nil {
return fmt.Errorf("tci: not connected")
}
t.wmu.Lock()
defer t.wmu.Unlock()
_ = c.SetWriteDeadline(time.Now().Add(3 * time.Second))
if err := c.WriteMessage(websocket.TextMessage, []byte(cmd)); err != nil {
debugLog.Printf("TCI: send %q failed: %v", cmd, err)
@@ -354,10 +413,18 @@ func (t *TCI) send(cmd string) error {
// connection closes.
func (t *TCI) reader(conn *websocket.Conn) {
for {
_, data, err := conn.ReadMessage()
mt, data, err := conn.ReadMessage()
if err != nil {
break
}
// TEXT frames are commands, BINARY frames are streams. The type used to
// be ignored and every frame split on ';' — harmless only for as long as
// no stream was ever opened, since audio bytes would then have been fed
// to the command parser a hundred times a second.
if wsMessageIsBinary(mt) {
t.handleBinary(data)
continue
}
// A frame may carry several ";"-terminated commands.
for _, cmd := range strings.Split(string(data), ";") {
t.handle(strings.TrimSpace(cmd))
@@ -393,6 +460,16 @@ func (t *TCI) handle(msg string) {
switch strings.ToLower(name) {
case "device":
t.device = strings.TrimSpace(args)
// The radio ANNOUNCES its audio format at connect —
// "audio_stream_sample_type:float32" and "audio_stream_channels:2" — which
// is better evidence than anything derived from a frame, and it arrives
// before the first frame does. Both were being logged as unhandled.
case "audio_stream_sample_type":
t.audio.declaredType = strings.TrimSpace(args)
case "audio_stream_channels":
if n, err := strconv.Atoi(strings.TrimSpace(args)); err == nil && n > 0 && n <= 8 {
t.audio.declaredChans = n
}
case "ready", "start":
t.ready = true
case "stop":
@@ -421,7 +498,22 @@ func (t *TCI) handle(msg string) {
}
case "trx":
if get(0) == "0" {
was := t.tx
t.tx = get(1) == "true"
// Said out loud, every time. The transmit side of TCI can only be
// written from a log of a real transmission, and the first one came
// back without a single line to say whether the radio had even been
// keyed — which left the interesting question, why no transmit
// frames, indistinguishable from nobody having pressed anything.
if was != t.tx {
t.noteTXTransition(t.tx)
}
}
case "drive":
if get(0) == "0" {
if v, err := strconv.Atoi(get(1)); err == nil {
t.drive = v
}
}
case "tx_enable":
if get(0) == "0" {
+424
View File
@@ -0,0 +1,424 @@
//go:build windows
package cat
// TCI audio — receiving the radio's audio over the same WebSocket that carries
// the commands, so a SunSDR needs no virtual audio cable.
//
// TCI mixes two kinds of frame on one socket: TEXT frames are the commands
// ("trx:0,true;"), BINARY frames are streams. A binary frame is a fixed header
// followed by float32 samples:
//
// uint32 receiver which receiver the stream belongs to
// uint32 sampleRate Hz
// uint32 format 0 = float32
// uint32 codec 0 = uncompressed
// uint32 crc unused in practice
// uint32 length samples in the payload
// uint32 type which stream this is (see tciStream*)
// uint32 reserved[9]
// float32 payload[…] stereo, interleaved
//
// The stream is asked for with "audio_samplerate:" then "audio_start:<rx>;",
// and stopped with "audio_stop:<rx>;".
//
// NOTHING HERE IS CONFIRMED ON A RADIO YET. The layout above is read from the
// TCI documentation, and the stream-type numbers in particular are the sort of
// detail a document gets right and a memory of it does not — so every header is
// logged for the first few seconds of a session, and the numbers the radio
// actually sends will settle it. Same discipline as the Yaesu meters and the
// Flex spot feed: measure on the real thing, then write the constant down.
import (
"encoding/binary"
"fmt"
"math"
"strings"
"sync"
"time"
"github.com/gorilla/websocket"
)
// TCI stream types. RX audio is the one this file consumes; the others are
// named so a log line says what arrived rather than "type 3".
const (
tciStreamIQ = 0
tciStreamRXAudio = 1
tciStreamTXAudio = 2
tciStreamTXChrono = 3
)
// tciHeaderWords is the header length in uint32 words (7 named + 9 reserved).
const tciHeaderWords = 16
// tciHeaderBytes is the same in bytes.
const tciHeaderBytes = tciHeaderWords * 4
// tciAudioProbeMax bounds the header logging. Enough frames to see the shape
// and the rate; few enough that an evening of listening does not fill the log.
const tciAudioProbeMax = 40
// TCIAudioStatus is what the panel polls while testing the stream.
type TCIAudioStatus struct {
Running bool `json:"running"`
SampleRate int `json:"sample_rate"`
Frames int64 `json:"frames"` // binary frames accepted
Samples int64 `json:"samples"` // audio samples decoded
// PeakDB is the loudest sample of the last second, in dBFS: the one number
// that says "audio is really arriving" rather than "a socket is open".
PeakDB float64 `json:"peak_db"`
LastErr string `json:"last_err,omitempty"`
}
// tciAudio is the receive-side state, kept on the backend so it lives exactly
// as long as the connection does.
type tciAudio struct {
mu sync.Mutex
want bool // the host asked for audio
rx int // which receiver
rate int
frames int64
samples int64
peak float64
peakAt time.Time
probeByType map[int]int
// countByType counts EVERY frame per stream type, capped by nothing.
// The probe above stops logging after forty frames of a type; these keep
// counting, so a transmission that produced no transmit frames at all can
// be reported as a fact rather than inferred from an absence of lines.
countByType map[int]int64
lastErr string
// widthLogged keeps the one-line note about the sample width to once a
// session — it is a fact about the radio, not an event.
widthLogged bool
// txMark is the per-type frame count when transmission began, so the census
// at the end reports the pass rather than the whole session.
txMark map[int]int64
// txFeed supplies the next frame of transmit audio when the radio asks for
// one, or is nil when nothing is being sent. Set under this same lock, and
// read on the reader goroutine — the radio's request and our answer are two
// halves of one exchange and must not straddle a race.
txFeed func(samples int) []byte
txSent int64
txShort int64 // requests the feed could not fill (it had run out)
// What the radio SAID about its stream at connect (audio_stream_sample_type,
// audio_stream_channels). Its own declaration, and it arrives before the
// first frame — the frame arithmetic below stays as the check on it rather
// than as the only source.
declaredType string
declaredChans int
// OnSamples receives decoded MONO samples (the two channels averaged) at
// the negotiated rate. Mono because everything downstream — the QSO
// recorder, the CW decoder — works on one channel, and a receiver's two
// channels carry the same audio.
OnSamples func(rate int, samples []float32)
}
// StartTCIAudio asks the radio to stream receiver rx's audio.
func (t *TCI) StartTCIAudio(rx, rate int) error {
if rate <= 0 {
rate = 48000
}
t.audio.mu.Lock()
t.audio.want = true
t.audio.rx = rx
t.audio.rate = rate
t.audio.frames, t.audio.samples, t.audio.peak = 0, 0, 0
t.audio.lastErr = ""
t.audio.mu.Unlock()
// Sample rate first: the radio applies it to the stream it is about to
// open, and asking afterwards restarts the stream on some firmware.
if err := t.send(fmt.Sprintf("audio_samplerate:%d;", rate)); err != nil {
return err
}
// Said rather than assumed. float32 and two channels are the documented
// defaults and what this radio streams, but a default is a thing another
// program can have changed — they share the radio, not just the protocol —
// and a stream arriving in a format the decoder was not expecting is heard
// as noise, not as a mistake.
_ = t.send("audio_stream_sample_type:float32;")
_ = t.send("audio_stream_channels:2;")
return t.send(fmt.Sprintf("audio_start:%d;", rx))
}
// SetTCIAudioSink installs (or removes) the consumer of the decoded samples.
//
// One sink, not a list: today it is a test recording, tomorrow the QSO
// recorder, and two consumers of a live stream would need a policy about which
// one wins that nothing yet has an opinion about.
func (t *TCI) SetTCIAudioSink(fn func(rate int, samples []float32)) {
t.audio.mu.Lock()
t.audio.OnSamples = fn
t.audio.mu.Unlock()
}
// StopTCIAudio closes the stream.
func (t *TCI) StopTCIAudio() error {
t.audio.mu.Lock()
t.audio.want = false
rx := t.audio.rx
t.audio.mu.Unlock()
return t.send(fmt.Sprintf("audio_stop:%d;", rx))
}
// TCIAudioStatus reports what has arrived.
func (t *TCI) TCIAudioStatus() TCIAudioStatus {
t.audio.mu.Lock()
defer t.audio.mu.Unlock()
st := TCIAudioStatus{
Running: t.audio.want,
SampleRate: t.audio.rate,
Frames: t.audio.frames,
Samples: t.audio.samples,
LastErr: t.audio.lastErr,
}
// A peak older than a second is not a level, it is a memory. Reported as
// silence rather than left standing, so a stream that has stopped arriving
// looks stopped.
if time.Since(t.audio.peakAt) < time.Second && t.audio.peak > 0 {
st.PeakDB = 20 * math.Log10(t.audio.peak)
} else {
st.PeakDB = -99
}
return st
}
// handleBinary decodes one binary WebSocket frame.
//
// Called from the reader goroutine. Anything malformed is counted and dropped:
// a stream frame is not worth breaking the command connection over, and the
// command connection is what keeps the radio usable.
func (t *TCI) handleBinary(data []byte) {
if len(data) < tciHeaderBytes {
t.audioErr(fmt.Sprintf("binary frame of %d bytes is shorter than a header", len(data)))
return
}
le := binary.LittleEndian
receiver := int(le.Uint32(data[0:]))
rate := int(le.Uint32(data[4:]))
format := le.Uint32(data[8:])
codec := le.Uint32(data[12:])
length := int(le.Uint32(data[20:]))
stype := int(le.Uint32(data[24:]))
// Counted PER STREAM TYPE, not overall.
//
// A single counter was spent on the first forty receive-audio frames, which
// arrive twenty-four times a second — so a transmit-chrono or transmit-audio
// frame, the two this needs to see before the voice keyer can be written,
// would never have been logged at all. They only appear once the operator
// keys the radio, long after any global budget is gone.
t.audio.mu.Lock()
if t.audio.probeByType == nil {
t.audio.probeByType = map[int]int{}
}
if t.audio.countByType == nil {
t.audio.countByType = map[int]int64{}
}
t.audio.countByType[stype]++
probe := t.audio.probeByType[stype]
if probe < tciAudioProbeMax {
t.audio.probeByType[stype]++
}
t.audio.mu.Unlock()
if probe < tciAudioProbeMax {
debugLog.Printf("TCI: binary frame — rx=%d rate=%d format=%d codec=%d length=%d type=%d payload=%d bytes",
receiver, rate, format, codec, length, stype, len(data)-tciHeaderBytes)
}
if stype == tciStreamTXChrono {
// The radio asking for the next frame of transmit audio. It is empty —
// the whole message IS the request — and it carries the size it wants in
// the header's length field, so the answer is written from what it says
// rather than from what we assumed.
t.serveChrono(rate, length)
return
}
if stype != tciStreamRXAudio {
// IQ and transmit audio. The latter is ours to send, not to receive:
// counted above, and dropped.
return
}
if codec != 0 {
t.audioErr(fmt.Sprintf("stream is codec=%d, and nothing here decodes a compressed stream", codec))
return
}
// The FORMAT number is decided by measurement, not by the number itself.
//
// A real SunSDR answered format=3, where the code expected 0 — and 0 was a
// guess from reading the documentation, which is exactly the kind of detail
// a memory of a document gets wrong. Rather than swap one magic number for
// another, the sample width is derived from what arrived: the header says
// how many samples the payload holds, so the bytes per sample follow from
// dividing. That is true whatever number the format field carries, on this
// firmware and the next.
payload := data[tciHeaderBytes:]
if len(payload) == 0 || length <= 0 {
return
}
width := len(payload) / length
var n int
switch width {
case 4:
n = len(payload) / 4 // float32
case 2:
n = len(payload) / 2 // 16-bit PCM
default:
t.audioErr(fmt.Sprintf("frame carries %d bytes for %d samples (format=%d) — not a width this reads",
len(payload), length, format))
return
}
if n == 0 {
return
}
// Under the lock like the rest of the counters: the reader is the only
// writer today, but a fact about the radio that is read from another
// goroutine has no business being the one field left unguarded.
t.audio.mu.Lock()
first := !t.audio.widthLogged
t.audio.widthLogged = true
t.audio.mu.Unlock()
if first {
debugLog.Printf("TCI: audio is %d bytes per sample at %d Hz (format field says %d)", width, rate, format)
}
// Stereo interleaved → mono. Both channels of a receiver carry the same
// audio, and everything downstream works on one.
// How many channels are interleaved. The radio says so at connect; two is
// the fallback, which is what every SunSDR seen so far streams.
t.audio.mu.Lock()
chans := t.audio.declaredChans
t.audio.mu.Unlock()
if chans <= 0 {
chans = 2
}
mono := make([]float32, 0, n/chans+1)
var peak float64
sample := func(i int) float32 {
if width == 2 {
// 16-bit PCM, scaled to the same -1…1 the rest of the audio path
// works in, so a change of format cannot change what a level means.
return float32(int16(le.Uint16(payload[i*2:]))) / 32768
}
return math.Float32frombits(le.Uint32(payload[i*4:]))
}
for i := 0; i+chans-1 < n; i += chans {
var sum float32
for c := 0; c < chans; c++ {
sum += sample(i + c)
}
v := sum / float32(chans)
if a := math.Abs(float64(v)); a > peak {
peak = a
}
mono = append(mono, v)
}
t.audio.mu.Lock()
t.audio.frames++
t.audio.samples += int64(len(mono))
if rate > 0 {
t.audio.rate = rate
}
if peak > t.audio.peak || time.Since(t.audio.peakAt) > time.Second {
t.audio.peak = peak
t.audio.peakAt = time.Now()
}
cb := t.audio.OnSamples
t.audio.mu.Unlock()
if cb != nil {
cb(rate, mono)
}
}
// audioErr records a decoding complaint, once, so the panel can show it without
// the log filling with the same line at fifty frames a second.
func (t *TCI) audioErr(msg string) {
t.audio.mu.Lock()
first := t.audio.lastErr != msg
t.audio.lastErr = msg
t.audio.mu.Unlock()
if first {
debugLog.Printf("TCI: audio: %s", msg)
}
}
// resumeAudio re-opens the stream after a reconnect, if the host had asked for
// it. A dropped WebSocket takes the audio with it, and an operator who switched
// recording on does not expect to switch it on again.
func (t *TCI) resumeAudio() {
t.audio.mu.Lock()
want, rx, rate := t.audio.want, t.audio.rx, t.audio.rate
t.audio.mu.Unlock()
if !want {
return
}
if err := t.StartTCIAudio(rx, rate); err != nil {
debugLog.Printf("TCI: re-opening the audio stream failed: %v", err)
}
}
// wsMessageIsBinary keeps the type test in one place — the reader used to
// ignore the message type entirely and split every frame on ';', which would
// have fed audio bytes to the command parser the moment a stream was opened.
func wsMessageIsBinary(mt int) bool { return mt == websocket.BinaryMessage }
// noteTXTransition reports what the stream did across a transmission.
//
// The voice keyer needs two numbers the documentation does not give: the size
// and the cadence of the frames the radio expects while transmitting. They can
// only be read off a real transmission — and the first attempt came back with a
// log that said nothing at all, which is ambiguous: either no transmit frames
// arrived, or they arrived and went unlogged.
//
// So the boundaries are marked and every stream type is counted. A pass that
// produces "type 1: 240, and nothing else" is a RESULT — it says the radio
// sends no chrono unless something more is asked of it — where a log with no
// transmit lines in it was merely a silence.
func (t *TCI) noteTXTransition(on bool) {
t.audio.mu.Lock()
if t.audio.countByType == nil {
t.audio.countByType = map[int]int64{}
}
if on {
// Let the transmit types speak again on every pass: forty frames is a
// budget spent long before the operator gets round to keying.
if t.audio.probeByType != nil {
delete(t.audio.probeByType, tciStreamTXAudio)
delete(t.audio.probeByType, tciStreamTXChrono)
}
t.audio.txMark = map[int]int64{}
for k, v := range t.audio.countByType {
t.audio.txMark[k] = v
}
streaming := t.audio.want
t.audio.mu.Unlock()
debugLog.Printf("TCI: TRANSMIT started — watching for transmit-audio (type %d) and chrono (type %d) frames; receive stream is %s",
tciStreamTXAudio, tciStreamTXChrono, map[bool]string{true: "open", false: "CLOSED (tick the TCI recording option, or the radio has no reason to stream)"}[streaming])
return
}
names := map[int]string{
tciStreamIQ: "IQ",
tciStreamRXAudio: "receive audio",
tciStreamTXAudio: "transmit audio",
tciStreamTXChrono: "transmit chrono",
}
var parts []string
for _, k := range []int{tciStreamIQ, tciStreamRXAudio, tciStreamTXAudio, tciStreamTXChrono} {
if n := t.audio.countByType[k] - t.audio.txMark[k]; n > 0 {
parts = append(parts, fmt.Sprintf("%s (type %d): %d", names[k], k, n))
}
}
t.audio.mu.Unlock()
if len(parts) == 0 {
debugLog.Printf("TCI: TRANSMIT ended — NO binary frames of any type arrived during it")
return
}
debugLog.Printf("TCI: TRANSMIT ended — frames during the pass: %s", strings.Join(parts, ", "))
}
+39
View File
@@ -0,0 +1,39 @@
//go:build windows
package cat
import "fmt"
// TCIAudioController is the receive-audio capability of the TCI backend, kept
// as an interface for the same reason as the Flex and Yaesu ones: the host asks
// the manager, and a station running something else gets a clear "this backend
// does not do that" instead of a nil dereference.
type TCIAudioController interface {
StartTCIAudio(rx, rate int) error
StopTCIAudio() error
TCIAudioStatus() TCIAudioStatus
}
// TCIAudioState returns the stream's state, or (zero, false) when the active
// backend is not a TCI radio.
func (m *Manager) TCIAudioState() (TCIAudioStatus, bool) {
m.mu.RLock()
b := m.backend
m.mu.RUnlock()
if tc, ok := b.(TCIAudioController); ok {
return tc.TCIAudioStatus(), true
}
return TCIAudioStatus{}, false
}
// TCIAudioDo dispatches an audio command onto the CAT goroutine, like every
// other backend-specific control.
func (m *Manager) TCIAudioDo(fn func(TCIAudioController) error) error {
return m.exec(func(b Backend) error {
tc, ok := b.(TCIAudioController)
if !ok {
return fmt.Errorf("active CAT backend is not a TCI radio")
}
return fn(tc)
})
}
+189
View File
@@ -0,0 +1,189 @@
package cat
// Playing a recorded message to the radio over TCI — the voice keyer's path.
//
// The same exchange the tone probe established, with a WAV in place of the
// sine: the radio asks for a frame, we answer with the next slice of the
// message, and it sets the pace. What is added here is the conversion, because
// a recording is whatever the microphone gave it — 16-bit, often mono, often
// not 48 kHz — and the radio wants interleaved float32 at the stream's rate.
//
// The message is converted ONCE, up front, rather than per frame. A voice
// message is a few hundred kilobytes; resampling it inside the callback would
// put arithmetic on the path that has 21 ms to answer, and a late frame is a
// gap in what goes out.
import (
"encoding/binary"
"fmt"
"math"
"time"
)
// tciTXFirstAskTimeout is how long to wait for the radio to ask for the first
// frame before giving up.
//
// It answers within a frame or two when it is going to answer at all, so this
// is generous. When it stays quiet the cause is always the same — the transmit
// audio source is the microphone rather than TCI — and a fifth of a second of
// carrier is a cheap way to find that out.
const tciTXFirstAskTimeout = 200 * time.Millisecond
// PlayTXAudio sends one message and returns when it has all been handed over,
// or when stop is closed.
//
// The PTT is NOT touched here. The voice keyer keys before calling and unkeys
// after, exactly as it does with a sound card, so the transmission is bracketed
// by the same code whichever way the audio travels.
func (t *TCI) PlayTXAudio(pcm []byte, rate, ch, bits int, stop <-chan struct{}) error {
t.mu.Lock()
connected := t.conn != nil
t.mu.Unlock()
if !connected {
return fmt.Errorf("not connected to the radio")
}
t.audio.mu.Lock()
outRate := t.audio.rate
t.audio.mu.Unlock()
if outRate <= 0 {
outRate = 48000
}
mono := decodeToMono(pcm, ch, bits)
if len(mono) == 0 {
return fmt.Errorf("the message is empty")
}
if rate > 0 && rate != outRate {
mono = resampleLinear(mono, rate, outRate)
}
// Served from here on. The callback does nothing but copy and interleave,
// which is what keeps it inside the frame interval.
pos := 0
done := make(chan struct{})
var closed bool
t.setTXFeed(func(samples int) []byte {
if samples <= 0 {
samples = 2048
}
pairs := samples / 2
if pos >= len(mono) {
if !closed {
closed = true
close(done)
}
return nil
}
payload := make([]byte, samples*4)
le := binary.LittleEndian
for i := 0; i < pairs; i++ {
var v float32
if pos < len(mono) {
v = mono[pos]
pos++
}
bits := math.Float32bits(v)
le.PutUint32(payload[(i*2)*4:], bits) // left
le.PutUint32(payload[(i*2+1)*4:], bits) // right
}
return payload
})
defer t.setTXFeed(nil)
// Nothing asked for in a fifth of a second means nothing is listening.
// Reported plainly: the message would otherwise go out as silence, and a
// voice keyer that transmits silence is worse than one that refuses.
deadline := time.Now().Add(tciTXFirstAskTimeout)
for time.Now().Before(deadline) {
t.audio.mu.Lock()
asked := t.audio.txSent > 0
t.audio.mu.Unlock()
if asked {
break
}
select {
case <-stop:
return nil
case <-time.After(10 * time.Millisecond):
}
}
t.audio.mu.Lock()
asked := t.audio.txSent
t.audio.mu.Unlock()
if asked == 0 {
return fmt.Errorf("the radio did not ask for any audio — set its transmit audio source to TCI instead of the microphone")
}
// The radio drains the message at real time, so this waits for the feed to
// run out. The cap is the message's own length with a second to spare: a
// radio that stops asking mid-message must not hold the transmitter up.
limit := time.Duration(float64(len(mono))/float64(outRate)*float64(time.Second)) + time.Second
select {
case <-done:
case <-stop:
case <-time.After(limit):
debugLog.Printf("TCI: the radio stopped asking for audio before the message ended")
}
return nil
}
// decodeToMono turns interleaved PCM into one channel of -1…1 floats.
func decodeToMono(pcm []byte, ch, bits int) []float32 {
if ch <= 0 {
ch = 1
}
switch bits {
case 16:
frame := ch * 2
out := make([]float32, 0, len(pcm)/frame+1)
for i := 0; i+frame <= len(pcm); i += frame {
var sum float32
for c := 0; c < ch; c++ {
v := int16(uint16(pcm[i+c*2]) | uint16(pcm[i+c*2+1])<<8)
sum += float32(v) / 32768
}
out = append(out, sum/float32(ch))
}
return out
case 8:
// Unsigned, centred on 128 — the one format where silence is not zero.
out := make([]float32, 0, len(pcm)/ch+1)
for i := 0; i+ch <= len(pcm); i += ch {
var sum float32
for c := 0; c < ch; c++ {
sum += (float32(pcm[i+c]) - 128) / 128
}
out = append(out, sum/float32(ch))
}
return out
}
return nil
}
// resampleLinear moves samples from one rate to another.
//
// Linear interpolation, which is crude and entirely adequate here: a voice
// recording at 16 kHz going to 48 kHz is being INTERPOLATED, and interpolation
// invents no frequencies to alias. Going the other way would want a filter
// first, but a message recorded above the radio's stream rate is not a case
// that arises — the recorder works at 16 kHz and radios stream at 48.
func resampleLinear(in []float32, from, to int) []float32 {
if from <= 0 || to <= 0 || from == to || len(in) == 0 {
return in
}
ratio := float64(from) / float64(to)
n := int(float64(len(in)) / ratio)
out := make([]float32, n)
for i := 0; i < n; i++ {
src := float64(i) * ratio
j := int(src)
frac := float32(src - float64(j))
if j+1 < len(in) {
out[i] = in[j]*(1-frac) + in[j+1]*frac
} else {
out[i] = in[len(in)-1]
}
}
return out
}
+117
View File
@@ -0,0 +1,117 @@
package cat
// Sending audio TO the radio over TCI.
//
// Three transmissions on a real SunSDR settled how this works, and none of it
// was guessable from the documentation:
//
// 1. The radio asks for audio only when the transmission is the CLIENT'S. With
// the operator keying the microphone it sent 282 receive frames and nothing
// else, over six seconds.
// 2. It asks only when its TRANSMIT AUDIO SOURCE is TCI rather than the
// microphone. This first read as "digital modes only" — SSB produced
// nothing four times over, DIGU answered at once — but the mode was a
// coincidence: ExpertSDR3 keeps that source setting per mode, and it was on
// the microphone in SSB. Which is why nothing is refused on the strength of
// the mode: the radio is asked, and it answers by asking or by staying
// quiet.
// 3. The chrono is a REQUEST, not a clock to follow. It carries no payload —
// the message itself is the ask — and it names the size it wants in the
// header's length field: 2048 samples, two channels interleaved, arriving
// 47 times a second. Which is 1024 sample-pairs at 48 kHz, exactly real
// time, measured rather than assumed.
//
// So audio is sent in ANSWER to chrono, never on a timer of our own. A timer
// was the first attempt and the radio ignored every frame of it: 234 sent, none
// used. Answering the request is what makes the difference, and it also means
// the radio sets the pace — no drift, no buffer to tune.
//
// All of it was established with a tone probe — key the radio, push a sine,
// watch — which is gone now that it has served its purpose: it answered the
// three questions above, confirmed 80 W out on a real SunSDR, and had no
// business in front of an operator once the voice keyer worked. What is left is
// the exchange it discovered, with tci_tx_play.go supplying the message.
import (
"encoding/binary"
"fmt"
"time"
"github.com/gorilla/websocket"
)
// sendBinaryFrame writes one TCI binary frame: the 16-word header the radio's
// own frames carry, then the payload.
func (t *TCI) sendBinaryFrame(stype, rx, rate, length int, payload []byte) error {
t.mu.Lock()
c := t.conn
t.mu.Unlock()
if c == nil {
return fmt.Errorf("tci: not connected")
}
buf := make([]byte, tciHeaderBytes+len(payload))
le := binary.LittleEndian
le.PutUint32(buf[0:], uint32(rx))
le.PutUint32(buf[4:], uint32(rate))
// format=3, codec=0: mirrored from what this radio SENDS. The field is
// documented as an enumeration whose numbering did not survive contact with
// the firmware — the receive stream answers 3 for four-byte floats — so the
// only defensible choice is to speak back exactly what was spoken to us.
le.PutUint32(buf[8:], 3)
le.PutUint32(buf[12:], 0)
le.PutUint32(buf[16:], 0) // crc — the radio sends 0 and does not check ours
le.PutUint32(buf[20:], uint32(length))
le.PutUint32(buf[24:], uint32(stype))
copy(buf[tciHeaderBytes:], payload)
t.wmu.Lock()
defer t.wmu.Unlock()
_ = c.SetWriteDeadline(time.Now().Add(3 * time.Second))
return c.WriteMessage(websocket.BinaryMessage, buf)
}
// serveChrono answers one request for transmit audio.
//
// Called from the reader goroutine, so it does the least it can: take the
// frame from whatever is feeding, and write it. A feed that has run out returns
// nil and the request is counted rather than answered with silence — silence
// would be indistinguishable from a working stream on a meter.
func (t *TCI) serveChrono(rate, samples int) {
t.audio.mu.Lock()
feed := t.audio.txFeed
t.audio.mu.Unlock()
if feed == nil {
return
}
if samples <= 0 {
samples = 2048
}
payload := feed(samples)
if payload == nil {
// Answered with silence rather than left unanswered. TCI 2.0 §3.4: "the
// client may not send a response or may send a signal with zero counts,
// which corresponds to no signal - this option is preferable."
payload = make([]byte, samples*4)
t.audio.mu.Lock()
t.audio.txShort++
t.audio.mu.Unlock()
}
if rate <= 0 {
rate = 48000
}
if err := t.sendBinaryFrame(tciStreamTXAudio, 0, rate, samples, payload); err != nil {
debugLog.Printf("TCI: could not send transmit audio: %v", err)
return
}
t.audio.mu.Lock()
t.audio.txSent++
t.audio.mu.Unlock()
}
// setTXFeed installs (or clears) the source of transmit audio.
func (t *TCI) setTXFeed(fn func(samples int) []byte) {
t.audio.mu.Lock()
t.audio.txFeed = fn
t.audio.txSent, t.audio.txShort = 0, 0
t.audio.mu.Unlock()
}
BIN
View File
Binary file not shown.
+1 -1
View File
@@ -21,7 +21,7 @@ import (
const (
// appVersion is stamped on every heartbeat (and could feed the About box).
appVersion = "0.26.14"
appVersion = "0.26.16"
// posthogHost is the PostHog ingestion endpoint. EU cloud by default; change
// to https://us.i.posthog.com for a US project.