hamradiotools

Nexus 1.9.2

Released 29 August 2026. The current release is 1.24.0. Download it · Every version

Added

  • Amplifier controls in every cockpit.

    With an amplifier configured — SPE Expert or Elecraft KPA500/KPA1500 — each operating screen's header carries a compact strip: Standby/Operate, band down and up, and power out, so you can put the amp in line or move it a band without leaving the screen you are working on. It appears only when an amplifier is set up, and nothing at all is added for the stations that have none. Operate reads from the amplifier itself rather than from your click, so the button always shows where the amplifier actually is, even if you press the front-panel key instead. Both controls are refused while you are transmitting: changing band on a keyed amplifier can damage it, and dropping to standby mid-over does not stop anything — the exciter keeps keying and the drive passes straight through. Nexus will never switch your amplifier off; that command does not exist in the code. Asked for by KD9TAW. ⚠️ The Elecraft side has never been run on hardware — it is written from Elecraft's own programmer's references and is waiting on someone with a KPA to try it. The SPE side has been tested against a 1.5K-FA.

  • Nexus can read your amplifier.

    Put an SPE Expert (1.3K-FA, 1.5K-FA, 2K-FA) or an Elecraft KPA500/KPA1500 on its own serial port, set it under Settings ▸ Radio ▸ Amplifier, and place the Amplifier pane in Connect: power out, SWR at the antenna and before the tuner, supply volts and current, PA temperature, and the amplifier's own alarms and warnings. It only ever asks — the one thing Nexus can send this amplifier is the six-byte status request, so it cannot switch it off, put it in standby, change band or start a tune. Readings clear the moment a poll goes unanswered, because a stale wattage beside a dead link is worse than no wattage; an alarm code this build has never seen still shows as a fault rather than going quiet; and the port is checked against everything else on the station, since a serial port opens only once and an amplifier typed onto the CAT port takes the radio down. Verified against a real 1.5K-FA. The Elecraft side is written from Elecraft's published references and has not been run on hardware.

  • A zero-beat light in the CW cockpit.

    The scope has always drawn a marker at your CW pitch; now the app measures the tone actually coming in and tells you where it sits against it. A light comes on when you are on pitch, and beside it a needle and a signed offset in Hz say which way and how far off you are — being 80 Hz out no longer looks the same as being 400 Hz out. How close counts as on pitch follows your rig's CW filter (25 Hz behind the usual 500 Hz one, tighter behind a narrow filter). When several signals are in the passband it follows the one nearest your marker rather than the loudest, so it does not jump to a strong station 300 Hz away while you are closing in. On a dead band it goes quiet and stays quiet: a reading only appears once a tone has held the same frequency long enough to be a signal you are tuning rather than the loudest thing the receiver happened to hear, so the needle does not twitch at band noise between overs. It waits about a fifth of a second before showing a new signal, which is the price of that stillness, and it keeps tracking live once it has one — the wait is paid when a station starts sending, never between their letters. It is a display only; it never touches your dial. Asked for by KD9TAW.

  • Tune can key at its own power.

    Set a tune power under Settings ▸ Digital ▸ Transmit & Sequencing and a tune-up keys at that level instead of whatever you are running. It can only ever turn the rig down — it keys at whichever is lower, your setting or your current power — so it can never lift you past a per-mode limit. Left empty it does nothing at all, which is the default. Asked for by an operator running an MFJ-1786 loop, where a hundred watts into an untuned loop is not what you want while the tuner hunts.

  • Split contacts log both frequencies.

    A contact worked split now records the frequency you transmitted on and the frequency you received on, as separate ADIF fields, so a logger you export to sees the same pair you actually used. Both values were already on screen and were being thrown away at the log boundary. Not split? Nothing changes and no second frequency is written — an invented one would be a false claim of split that every logger downstream believes. Satellite passes are deliberately unchanged for now; doing them properly means moving the band field too, which reaches every award and needed-list in Nexus. Asked for by swann.

  • PSK has a log strip.

    The PSK cockpit now carries the same log strip as CW and Phone, PSK31 and QPSK31 are in the mode list, and the manual log form stops pre-filling FT8 for a PSK contact. It logs the sub-mode you actually worked, so a QPSK31 QSO records QPSK31. Reported by KR4FQG.

  • A callsign card in the FT8 cockpit.

    Clicking a call in the roster now shows what you already know about that station — previous contacts with band, mode and date, what is unconfirmed, the new-one flags, distance and bearing, and your private note — the same card CW and Phone have had. It follows the station you are working as well as one you click, so it is populated through a normal run rather than only when you go looking. Reported by KR4FQG.

  • RTTY starts receiving when you open it.

    Every other decode mode auto-arms; RTTY was the one that waited for you to press Arm RX, which is a likely reason behind "RTTY is not decoding". Same Settings opt-out as PSK, APRS and SSTV.

  • Tell Nexus what your serial cable actually does.

    Two new declarations under Settings ▸ Radio ▸ Rig & CAT ▸ Advanced — serial handshake, and the keying line's state at startup. Both default to Auto, which is exactly today's behaviour, so nothing about your station changes by upgrading. They exist for one fault: a rig that keys the transmitter the moment Nexus opens. If that is not happening to you, leave them alone.

  • The app sizes itself to a high-resolution screen on Linux.

    On a 4K panel, or a small sharp laptop screen, Nexus used to open at 1:1 with the screen's pixels and read half the size the same app does on Windows — Windows and macOS tell an app how dense the screen is, and Linux does not. Nexus now asks the display for its physical size on first launch and lifts its own zoom ceiling to match, so the interface can grow to the screen instead of being pinned at 100%. It only ever makes things bigger, only on a screen that genuinely needs it, and only once: set the zoom yourself under Settings ▸ Appearance and your choice stands for good. On an ordinary monitor, and on Windows and macOS, nothing changes at all. Reported by an operator running 1920×1080 on a twelve-inch panel.

Fixed

  • Soundcard CW keying now uses a data mode, so the audio actually reaches the transmitter.

    With the CW keyer set to Soundcard, Nexus put the radio into plain USB or LSB. On the common Icom and default Yaesu wiring, plain SSB takes its transmit audio from the MIC socket, so the rig keyed and radiated nothing — the red light came on and no signal went out. It now commands the DATA submode (DATA-U/DATA-L), which is what FT8, PSK31, RTTY and SSTV have always done; soundcard CW was the one path that did not. ⚠️ If your interface feeds the rig's MIC socket and soundcard CW was working for you, tick "plain SSB for data modes" on that radio in Settings and it will work again — that is the same switch the other four modes already use. Reported by an FTX-1 operator.

  • The CW cockpit and Settings now warn when CAT keying is unproven on your radio.

    On the Yaesu FTX-1, the keying command Hamlib sends differs from the one it uses on other radios, and it reports success either way — so if nothing goes out, Nexus has no way to know and the keyer error line stays dark. Rather than let you lose an evening to it, both the Settings picker and the cockpit's live keyer switch now say it is unproven on that radio and point you at the serial keyline, WinKeyer or the soundcard keyer. It is a notice, not a block: nothing stops you using CAT keying, and nothing about what Nexus sends has changed.

  • Switching to the Soundcard keyer says that it takes the radio out of CW.

    It always did — a keyed audio tone needs the rig in an SSB or data mode — but nothing on screen said so, so the mode change looked like a bug. The Settings hint and the cockpit's keyer switch now both say it, and say that CW mode comes back when you pick another keyer.

  • The Work button left the markers behind.

    Clicking Work on a station card in the Classic layout's Stations pane started the QSO but did not move your RX and TX markers onto the station, so you answered someone while still listening and transmitting on a different offset. Band Activity's double-click and the Roster table both moved them; the card did not, which made one gesture behave three ways in the same cockpit. Both marks now follow the station, RX always and TX unless you have Hold Tx Freq set, the same as WSJT-X. Reported by bitslave.

  • eQSL sync blamed your password when the password was fine.

    If the download came back as anything other than the log file, Nexus told you to check your username and password — at a point where the login had already succeeded. Uploads kept working, because they never go through that check, so the app was simultaneously proving your credentials were right and telling you they were wrong. It now says what actually happened and states plainly that it is not a login problem. The LoTW sync carried the identical fault and has been corrected the same way, before anyone hit it — that is the one that matters for award credit.

  • The SWR and ALC meters had no warning band.

    The bar that should turn amber as SWR climbs was painting the same red as the "too hot" band, so there was nothing between "fine" and "trouble" — you saw a problem arriving rather than coming. Same on the ALC and the S-meter. The colour was written correctly and never took effect. Green, amber, red now, as intended.

  • A paused waterfall looked like an error.

    Same cause as the meters: the pause indicator was reaching for an amber that never applied, and painted the error red instead.

  • Small print was rendering at full size in twenty places.

    Text written to be small — panel tags, beacon and band rows, the CW decoder's status and age lines, the waterfall pop-out controls, APRS beacon titles — was silently falling back to body size, so it sat at 14px beside neighbours at 12px and 11px. Now the size it was always meant to be.

  • Colours that could not follow the theme now do.

    Around eighty surfaces were painting hard-coded colours rather than the app's own palette: reds that did not match each other (three different ones), greens likewise, and ambers that stayed identical whether you were in light or dark. Worst of it was on the light theme, where several recessed areas — the connections log, the waterfall pop-out, the SSTV picture area, the CW decode strip — rendered as grey slabs because their colour was a fixed black wash chosen against a dark panel. The settings search results list had no background at all and drew straight over whatever was behind it.

  • eQSL sync refused a perfectly good download.

    eQSL stopped starting its InBox export with the words Nexus was looking for, so the sync rejected the file and told you your credentials were wrong — at a point where the login had already succeeded. Nexus now checks the markers that identify an eQSL export rather than a sentence eQSL can reword whenever it likes. An HTML error page is still refused, which is what that check was for in the first place. Found, diagnosed and fixed by KR8MER.

  • Dropdowns were white on white on Linux.

    Every dropdown in the app drew as a white box with the app's own pale text on it, which on a dark theme meant you could not read what was selected — including the radio and sound-card pickers in step 2 of first-time setup, so a new operator could not see what they were choosing while setting the app up. Linux draws form controls with its own widget theme unless an app takes them over, and Nexus never did; Windows and macOS were never affected, which is why the same build looked fine there. Every dropdown is now drawn by Nexus itself, in your theme, with its own arrow. Reported by an operator on Ubuntu 24.04 with an FT-991A, and by M0LHJ on Fedora. The open list is a separate problem and is not fixed here: Linux draws that as a system menu that an application cannot style at all, so the list you see after clicking may still use your desktop's colours.

  • Grey slabs in the light theme.

    On the light theme, the CW decode strip, the copilot strip, and the SSTV image area, drop zone and progress bar all drew as flat mid-grey blocks — the SSTV picture area worst of all, since it is the largest. Each was asking the stylesheet for a colour that had never been defined, and falling back to a fixed dark wash that happens to look right on the dark theme and muddy on the light one. They now use the same recessed colour the rest of the app uses, which is defined for both themes. The dark theme is unchanged.

  • Four Settings dropdowns were the wrong size.

    The rotator model, pounce threshold, WAV recording and propagation engine pickers rendered noticeably shorter than every other control on the same page. They now match.

  • A guessed radio model no longer looks confirmed.

    When auto-detect proves the port but has to guess the model — an FT-991A answers a probe meant for an FTDX10 — the wizard marks the model box for you to confirm. That marking had never actually appeared on screen, on any platform, so the one moment you could catch a wrong radio passed by silently. The box is now outlined.

  • APRS reported packets that were never there.

    With the squelch open, the APRS panel counted plain noise as packets failing their checksum — about one every four seconds — so within moments of listening you were told the channel was full of traffic Nexus could not read. Nothing was wrong with the decoder; the counter simply accepted any burst of noise that happened to look frame-shaped. It now checks that what it found could actually be an AX.25 frame before counting it. A real packet that genuinely fails its checksum is still counted, which is what that reading is for. APRS also writes to the diagnostic log now — it was the only mode that said nothing at all, so an APRS problem arrived with a log that talked exclusively about FT8. Reported by swinn.

  • "Check your network" was the answer to problems that had nothing to do with your network.

    Every connector — QRZ, ClubLog, LoTW, eQSL, HRDLog and the rest — reported a rejected secure connection as a network failure. So if antivirus or a company proxy inspects your HTTPS traffic, Nexus would fail to upload while your browser worked perfectly and every check you could run said your connection was fine. It now tells the two apart and says when interception is the likely cause. A genuinely unreachable network still says so. Raised by lz2aov.

  • Test CAT told you to close other software when the real problem was permissions.

    On Linux, a serial port your user is not permitted to open reports "permission denied" — and Nexus answered with the advice for a busy port, telling you to close WSJT-X and flrig. No amount of closing programs grants a group membership, so you would close everything, test again, fail again, and reasonably conclude Nexus does not do CAT on Linux. It now recognises a permission refusal and gives the actual cure: add yourself to the dialout group, then log out and back in — the second half matters, because a group does not apply to a session already running. A port another program really is holding still gets the advice it always did. Reported from Ubuntu 24.04 with an FT-991A that worked on Windows on the same machine.

  • Audio broke up badly on Windows, and worse on battery.

    The sound-card callback and the code feeding it were fighting over the same lock, and the feeding side held it while copying a whole transmission — so when Windows slowed the cores down on battery, the callback missed its deadline and put silence on the air. That is why raising Nexus's priority helped, and why WSJT-X on the same laptop was clean. The two no longer share a lock. There are underrun counters behind it now, so the next report comes with a number instead of an impression. Reported by KR8MER.

  • No decodes, and a log that grew to 294 MB.

    Nexus opened the playback side of your sound card at startup whether or not you ever transmitted. On a codec that cannot record and play at the same time, that open wedges, and the error retried forever — millions of identical lines — while capture kept working, so the waterfall looked healthy and nothing ever decoded. Nexus now opens the playback side when you first transmit, so a station that is only listening opens it at all. Repeated errors are also capped: 22 lines with a count on each, instead of filling your disk. Reported by M0LHJ, who did the diagnostic work that found it.

  • Your radio's audio card could be missing from the transmit list on macOS.

    A card that only plays and does not record was dropped from the list entirely, so the rig keyed with no audio and nothing on screen explained why. Two operators hit it on different interfaces. Reported by pvanderp and crabtreejw; the cause was found by M7HNF-Ian, whose diagnosis this fix follows.

  • Some sound cards were refused outright, with no audio at all.

    Nexus understood four of the ten audio formats a card can report. A card whose natural format was any of the other six got an "unsupported format" message and silence — not a setting you could change, and nothing you did wrong. All formats are carried now. If you have an interface that never worked with Nexus and you could never find out why, it is worth another try.

  • Nexus no longer adopts a rigctld that is driving a different radio.

    When a rigctld was already listening on a radio's CAT port, Nexus connected through it — which is the right thing, and is how it shares a rig with WSJT-X — but it never checked WHICH radio that daemon was attached to. On a two-radio station a stray daemon left over from the other rig would be adopted, and from then on every frequency read, every band change and every keying command went to the wrong radio, with the app showing the other rig's dial as though it were yours. Nexus now asks first: it reads the daemon's own arguments (which carry the model AND the serial device) and falls back to asking the daemon over the protocol when there is no local process to read. On a mismatch it refuses, says which radio that daemon is actually driving, and tells you how to fix it — stop the daemon, or give this radio its own rigctld port. Sharing a daemon that IS this radio's is unchanged, and so is coexisting with an external NET-rigctl station.

  • Tune could drop out repeatedly, defeating an automatic antenna tuner.

    On a station with an auto-tuning antenna — a magnetic loop especially — the tune carrier was interrupted over and over. A tuner reads those gaps as a match, beeps, and stops before it has found one. Reported by an operator running an MFJ-1786 with an IC-7300. Two causes, both ours. Nexus queued only about 40 ms of tune audio ahead of the sound card while asking the radio for meter readings on the same thread — and a radio that is slow to answer stalls that thread for longer than the audio it has queued, so the carrier stops while the rig stays keyed. Separately, if you had a per-mode power limit set, Nexus re-sent that limit to the radio every twenty milliseconds for the whole tune. The meters now stand down during a tune-up, the power limit is not re-sent under a live carrier, and the carrier runs a quarter of a second ahead instead of forty milliseconds.

  • A rig could key the transmitter the moment Nexus started.

    With PTT set to serial RTS and no separate PTT port, Nexus told the rig-control daemon nothing about the RTS line at all, so its idle state fell to whatever the driver happened to do — on some cables, keyed. Two earlier attempts guessed at this from other settings. Rather than guess a third time, the handshake and the keying line's startup state are now things you can state outright (see Added). Reported by VK6MO.

  • Sharing your radio could report success for a mode it never set.

    If another program — VarAC, FreeDV, WSJT-X, a logger — asked Nexus for a data mode while Nexus was in a voice section, Nexus answered "done" and left the rig in plain SSB. Your data signal went out on a voice emission with nothing on screen to say so. It now answers honestly, and the connection log names the mode that was asked for and the mode your radio is actually in. Reading the mode back is honest too, on an FM calling channel, an APRS park or during an SSTV picture. Reported by rogerloxton. This is a behaviour change you may notice — see Changed.

  • A refused PTT from another program said nothing at all.

    Nexus starts with transmit disarmed, so the first time an external program tried to key, it was refused silently — which is why pressing PTT in Phone appeared to "wake it up". The refusal now says why, in the connection log.

  • Sharing could fill your connection log at one line a second, forever.

    If something else already owned the sharing port — your own rigctld, a second copy of Nexus — Nexus retried the bind every second and logged every failure, without ever telling you plainly. It now says so once, names the port and the likely cause, and backs off.

  • Callsigns could be given the wrong country and a nonsense state.

    WL7E showed as Alaska with a state of "CA". The country comes from the callsign prefix and the state came from the FCC index, which is a mailing address — and nothing compared the two. An Alaskan or Hawaiian call now gets AK or HI, and a station outside the US gets no state at all. This also fed the Worked All States "new one" cue, so it was quietly wrong there too. Reported by KR4FQG.

  • Settings you changed while editing another radio were thrown away.

    If you clicked Edit on a radio you were not currently using and then changed anything station-wide — QRZ auto-upload, for instance — the panel said it saved and nothing was written. This seam has now cost a keying port, three Flex fields, an OmniRig slot and an Icom submode; the save itself is fixed rather than another field name being added to it. Reported by barnburner6503.

  • Your own transmissions vanished from the Rx Frequency pane.

    Stopping transmit, changing band or switching radio wiped the record of overs you had already sent — including when Nexus did it for you, with nobody touching anything. Reported by Luk73.

  • "Calling you" alerts only fired after the contact had ended.

    An alert for the station you were working was suppressed — and the station answering your CQ becomes that station in the same instant the decode arrives, so the alert you actually wanted was the one that got suppressed. Your own callsign now always raises the alert, as it does in WSJT-X. Reported by KR4FQG.

  • LoTW confirmations were recorded as paper QSL cards.

    A LoTW report carries an ordinary confirmation field, and Nexus read it as a card — so the log claimed a card you never received, and an ADIF export carried that claim on to any other logger. A hand-imported third-party ADIF still counts as a card, correctly. Award credit is unaffected either way. Reported by rgoiko.

  • A QSL-sent mark could not be undone.

    Mis-click the QSL menu and it was permanent. There is a clear entry now, and the clear sticks even if you later re-import a log that still says the card went out. Reported by rgoiko.

  • WSPR could beacon where nobody was listening.

    The transmit marker could be dragged anywhere from 200 Hz to 4 kHz, while every WSPR decoder searches 1400-1600 Hz. You could beacon all night and be heard by nobody. It is held inside the sub-band now, matching WSJT-X. FT8, FT4 and FST4W are unchanged. Reported by akhepcat.

  • A repeated report went unanswered, and a station returning late got silence.

    When a contact closed, Nexus dropped the station immediately — so a partner who repeated their report got nothing back, and a station the run had given up on that came back with RR73 never got a closing 73. WSJT-X keeps the station and answers both. Nexus now does too, bounded to a single over within three minutes, without re-entering the contact. Reported by KR4FQG and bitslave.

  • The update notice did nothing on Linux, and silenced itself.

    Clicking Download dismissed the pop-up and opened nothing — and it recorded the version as dismissed anyway, so the notice never came back. Anyone who clicked it once stopped being told about updates entirely. It now opens the page or tells you it could not and gives you the link to copy.

  • A radio's audio could be refused by the monitor while working everywhere else.

    On Linux the headphone monitor picked the first audio format the card advertised rather than the one it could actually use, and reported "unsupported format" for a card the rest of Nexus was using happily. Reported by MW0CQU.

  • Satellite elements stopped updating.

    The upstream catalogue renamed the status that marks a satellite as being in orbit, and the mirror correctly refused to publish rather than ship an empty list — which meant satellite data quietly went stale instead. Fixed, and the check now refuses an unknown status by name rather than only noticing when everything moves at once.

  • Icom DATA submode was quietly reset whenever you edited a radio you were not using.

    If you run more than one radio and set an Icom to DATA2 or DATA3, opening that radio's entry and saving it put it back to DATA1 without saying so. The panel reported success, the setting was gone, and the next time you keyed that rig it was in the wrong submode. The per-radio Edit form saves through a patch that the app and the radio settings each describe separately, and the submode was missing from one of the two descriptions. Anything missing there is filled in with a default rather than your value. It travels with the patch now, and there is a check that fails the build if the two descriptions ever disagree again — the same drift has cost a keying port, a Flex address and an OmniRig slot before this.

Changed

  • A station that is only receiving no longer opens your sound card's playback side at all.

    You will see this if you look — the playback stream simply is not there until the first time you transmit. That is the fix for the wedged-open problem above, and it also means one fewer thing holding your card if you share it with another program.

  • Your sound card may open at a different rate or format after this update.

    The audio library Nexus uses now picks the card's default differently — 48 kHz where it might previously have chosen 44.1 kHz, and a different sample format on some cards. Nexus adapts either way and you should not hear a difference, but if you keep notes on your station's audio settings, this is the release where a number may move on its own.

  • Backup and Restore moved to their own Config tab.

    They existed, but sat under Radio ▸ Transmit limits & sharing, which is why operators did not find them: backing up a whole station has nothing to do with transmit limits. Same controls, same behaviour — a findable home, and search keywords wide enough to survive a panic ("backup", "restore", "factory", "defaults", "start over").

  • Sharing your radio now refuses a mode it cannot set, where it used to accept it silently.

    This is the honest half of the fix above, and it is the one you might feel: a program configured for a data mode while Nexus sits in Phone now gets a real error instead of a quiet success. Put Nexus in the section that matches what the other program is doing — Digital for FT8 and VarAC-style data, Phone for voice, CW for CW — and it goes through exactly as before. If you do see a refusal, the connection log names both modes so you can see which end to change.

  • The spots filter says what it does.

    "Heard near me" is now "Heard on my continent", which is what the filter has always actually done. Asked about by barnburner6503, who went looking for a Europe-only filter that was already there and switched on.

  • Grids by band lists the VUCC bands by default.

    Grids count toward an award on 50 MHz and up and nowhere else, so on an HF station the list was mostly grids that count toward nothing, with the few bands that do count buried in it. It now starts at 6 m and up, with an All bands button if you want the full count back. The VUCC box above it has always been VHF-only and is unchanged. Suggested by NT9E.

  • macOS: the Settings pickers can now tell two identical radios apart.

    On a station with two rigs that use the same bridge and codec chips, every serial port carried the same product label ("CP2105 Dual USB to UART Bridge Controller", eight times) and both sound cards enumerated as "USB Audio Device" — with only a positional " #2" between them, assigned by enumeration order. Moving one rig to a different USB socket therefore swapped which radio each saved name referred to, silently, with nothing to warn you. Nexus now reads USB topology on macOS and uses it three ways. Rig auto-detect and the CAT port probe no longer see the same physical port twice: a Silicon Labs bridge is offered once by Apple's driver (cu.usbserial-…) and again by the vendor's (cu.SLAB_USBtoUART…), two names no rule could pair, so Detect listed each rig twice and the baud sweep spent a full ladder probing a port it had already tried. Saving now warns if you picked the half of a dual bridge that carries no CAT — the most convincing way to make a working radio look dead, because the port opens and the writes succeed and nothing ever answers. And saving warns if the sound card you chose is inside the other radio. Both warnings are advisory and never block a save: an unusual-but-correct station is still yours to configure. The port list itself is unchanged — every port you could pick before, you can still pick.

Every version